From: Chet Ramey Date: Tue, 6 Mar 2012 02:16:53 +0000 (-0500) Subject: commit bash-20120216 snapshot X-Git-Tag: bash-4.3-alpha~81 X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=1f6ec1a86f688bb09a79232b6fb4a30a19ae71b2;p=thirdparty%2Fbash.git commit bash-20120216 snapshot --- diff --git a/CWRU/CWRU.chlog b/CWRU/CWRU.chlog index 07c37a439..8ae4f8e9c 100644 --- a/CWRU/CWRU.chlog +++ b/CWRU/CWRU.chlog @@ -13333,3 +13333,65 @@ builtins/declare.def assignment like declare a[b]=c. Fixes seg fault resulting from a being an already-declared local associative array variable in a function. Ubuntu bash bug 928900. + + 2/14 + ---- + +execute_cmd.c + - execute_command_internal: if redirections into or out of a loop fail, + don't try to free ofifo_list unless saved_fifo is non-zero. It's + only valid if saved_fifo is set + + 2/15 + ---- +{arrayfunc,braces,variables}.c + - last_command_exit_value: make sure it's set before any calls to + report_error, since -e will cause that to exit the shell + +builtins/common.c + - get_job_by_name: call internal_error instead of report_error so this + doesn't exit the shell + + 2/18 + ---- +builtins/evalstring.c + - parse_and_execute: make sure the file descriptor to be redirected to + is 1 before calling cat_file. One fix for bug reported by Dan Douglas + + +parse.y + - read_token_word: don't return NUMBER if a string of all digits + resolves to a number that overflows the bounds of an intmax_t. + Other fix for bug reported by Dan Douglas + + 2/19 + ---- +lib/sh/strtrans.c + - ansicstr: use 0x7f as the boundary for characters that translate + directly from ASCII to unicode (\u and \U escapes) instead of + UCHAR_MAX, since everything >= 0x80 requires more than one byte. + Bug and fix from John Kearney + +builtins/printf.def + - tescape: ditto for printf \u and \U escape sequences + + 2/20 + ---- +lib/sh/unicode.c + - u32toutf8: fix to handle encodings up to six bytes long correctly + (though technically UTF-8 only has characters up to 4 bytes long). + Report and fix from John Kearney + - u32toutf8: first argument is now an unsigned 32-bit quantity, + changed callers (u32cconv) to pass c instead of wc + - u32reset: new function, resets local static state to uninitialized + (locale information, currently) + +lib/sh/strtrans.c + - ansicstr: allocate a new string 6 times longer than the original + if the \U escape sequence appears, to accommodate UTF-8 chars + of maximum theoretical length + +locale.c + - call u32reset whenever LC_CTYPE/LC_ALL/LANG is changed to reset the + cached locale information used by u32cconv. From a report from + John Kearney diff --git a/CWRU/CWRU.chlog~ b/CWRU/CWRU.chlog~ index 9558f698a..76b126833 100644 --- a/CWRU/CWRU.chlog~ +++ b/CWRU/CWRU.chlog~ @@ -13277,15 +13277,29 @@ lib/readline/isearch.c builtins/gen-helpfiles.c - new file: reads struct builtin and writes the long docs to files in the `helpdirs' subdirectory. The filename is given in the - previously-unused `handle' member of the struct builtin + previously-unused `handle' member of the struct builtin. Links + with `tmpbuiltins.o', which is created by Makefile to have the + right long documentation. When not cross-compiling, gets the + right #defines based on configuration options from config.h instead + of trying to parse conditional parts of def files. Fixes + shortcoming pointed out by Andreas Schwab builtins/Makefile.in - tmpbuiltins.c: new generated file, created to enable creation of separate helpfiles based on correct #defines instead of trying to parse conditional parts of def files - - gen-helpfiles: new program to generate helpfiles + - gen-helpfiles: new program to generate helpfiles, links with + tmpbuiltins.o - HELPFILES_TARGET: new target, substituted by configure to `helpdoc' if separate helpfiles requested + - targets: new target, libbuiltins.a and $(HELPFILES_TARGET) + - CREATED_OBJECTS: new variable, holds created object files for + make clean; changed make clean to remove created objects + - helpdoc: changed to call gen-helpfiles instead of mkbuiltins + +Makefile.in + - when building libbuiltins.a, recursively call make with `targets' + argument to make sure separate helpfiles get built configure.in - substitute `helpdoc' as value of HELPFILES_TARGET if @@ -13296,4 +13310,87 @@ builtins/mkbuiltins.c for function implementing a particular builtin to struct builtin and to write document file name to `handle' member of struct builtin - no longer writes separate helpfiles; that is left to gen-helpfiles - + + 2/8 + --- +subst.c + - make sure last_command_exit_value is set to a non-zero value before + any calls to report_error, since `-e' set will short-circuit + report_error. Fixes bug reported by Ewan Mellor + + +variables.c + - make_local_array_variable: added second argument; if non-zero, + function will return an existing local associative array variable + instead of insisting on an indexed array + +variable.h,subst.c + - make_local_array_variable: changed prototype and caller + +builtins/declare.def + - declare_internal: add second arg to call to make_local_array_variable; + making_array_special, which indicates we're processing an + assignment like declare a[b]=c. Fixes seg fault resulting from + a being an already-declared local associative array variable in a + function. Ubuntu bash bug 928900. + + 2/14 + ---- + +execute_cmd.c + - execute_command_internal: if redirections into or out of a loop fail, + don't try to free ofifo_list unless saved_fifo is non-zero. It's + only valid if saved_fifo is set + + 2/15 + ---- +{arrayfunc,braces,variables}.c + - last_command_exit_value: make sure it's set before any calls to + report_error, since -e will cause that to exit the shell + +builtins/common.c + - get_job_by_name: call internal_error instead of report_error so this + doesn't exit the shell + + 2/18 + ---- +builtins/evalstring.c + - parse_and_execute: make sure the file descriptor to be redirected to + is 1 before calling cat_file. One fix for bug reported by Dan Douglas + + +parse.y + - read_token_word: don't return NUMBER if a string of all digits + resolves to a number that overflows the bounds of an intmax_t. + Other fix for bug reported by Dan Douglas + + 2/19 + ---- +lib/sh/strtrans.c + - ansicstr: use 0x7f as the boundary for characters that translate + directly from ASCII to unicode (\u and \U escapes) instead of + UCHAR_MAX, since everything >= 0x80 requires more than one byte. + Bug and fix from John Kearney + +builtins/printf.def + - tescape: ditto for printf \u and \U escape sequences + + 2/20 + ---- +lib/sh/unicode.c + - u32toutf8: fix to handle encodings up to six bytes long correctly + (though technically UTF-8 only has characters up to 4 bytes long). + Report and fix from John Kearney + - u32toutf8: first argument is now an unsigned 32-bit quantity, + changed callers (u32cconv) to pass c instead of wc + - u32reset: new function, resets local static state to uninitialized + (locale information, currently) + +lib/sh/strtrans.c + - ansicstr: allocate a new string 6 times longer than the original + if the \U escape sequence appears, to accommodate UTF-8 chars + of maximum theoretical length + +locale.c + - call u32reset whenever LC_CTYPE/LC_ALL/LANG is changed to reset the + cached locale information used by u32cconv diff --git a/MANIFEST b/MANIFEST index 3bc17a869..376966d62 100644 --- a/MANIFEST +++ b/MANIFEST @@ -792,6 +792,7 @@ tests/assoc3.sub f tests/assoc4.sub f tests/assoc5.sub f tests/assoc6.sub f +tests/assoc7.sub f tests/braces.tests f tests/braces.right f tests/builtins.tests f @@ -857,6 +858,7 @@ tests/dstack2.right f tests/errors.tests f tests/errors.right f tests/errors1.sub f +tests/errors2.sub f tests/execscript f tests/exec.right f tests/exec1.sub f 755 diff --git a/arrayfunc.c b/arrayfunc.c index d46e3b021..0ae89dacf 100644 --- a/arrayfunc.c +++ b/arrayfunc.c @@ -326,6 +326,7 @@ find_or_make_array_variable (name, flags) } else if ((flags & 2) && array_p (var)) { + last_command_exit_value = 1; report_error (_("%s: cannot convert indexed to associative array"), name); return ((SHELL_VAR *)NULL); } @@ -507,6 +508,7 @@ assign_compound_array_list (var, nlist, flags) if (ALL_ELEMENT_SUB (w[1]) && len == 2) { + last_command_exit_value = 1; if (assoc_p (var)) report_error (_("%s: invalid associative array key"), w); else @@ -551,6 +553,7 @@ assign_compound_array_list (var, nlist, flags) } else if (assoc_p (var)) { + last_command_exit_value = 1; report_error (_("%s: %s: must use subscript when assigning associative array"), var->name, w); continue; } diff --git a/braces.c b/braces.c index 1735ae5a4..91dde81f1 100644 --- a/braces.c +++ b/braces.c @@ -47,6 +47,8 @@ extern int asprintf __P((char **, const char *, ...)) __attribute__((__format__ (printf, 2, 3))); +extern int last_command_exit_value; + /* Basic idea: Segregate the text into 3 sections: preamble (stuff before an open brace), @@ -172,6 +174,7 @@ brace_expand (text) if (text[j] == brace_arg_separator) { /* { */ strvec_dispose (result); + last_command_exit_value = 1; report_error ("no closing `%c' in %s", '}', text); throw_to_top_level (); } diff --git a/builtins/common.c b/builtins/common.c index d6c4cdd13..fb2b38cdd 100644 --- a/builtins/common.c +++ b/builtins/common.c @@ -628,7 +628,7 @@ get_job_by_name (name, flags) if (this_shell_builtin) builtin_error (_("%s: ambiguous job spec"), name); else - report_error (_("%s: ambiguous job spec"), name); + internal_error (_("%s: ambiguous job spec"), name); return (DUP_JOB); } else diff --git a/builtins/evalstring.c b/builtins/evalstring.c index 333a56ee0..fe812dee0 100644 --- a/builtins/evalstring.c +++ b/builtins/evalstring.c @@ -1,6 +1,6 @@ /* evalstring.c - evaluate a string as one or more shell commands. */ -/* Copyright (C) 1996-2010 Free Software Foundation, Inc. +/* Copyright (C) 1996-2012 Free Software Foundation, Inc. This file is part of GNU Bash, the Bourne Again SHell. @@ -308,7 +308,8 @@ parse_and_execute (string, from_file, flags) command->value.Simple->words == 0 && command->value.Simple->redirects && command->value.Simple->redirects->next == 0 && - command->value.Simple->redirects->instruction == r_input_direction) + command->value.Simple->redirects->instruction == r_input_direction && + command->value.Simple->redirects->redirector.dest == 1) { int r; r = cat_file (command->value.Simple->redirects); @@ -317,7 +318,6 @@ parse_and_execute (string, from_file, flags) else last_result = execute_command_internal (command, 0, NO_PIPE, NO_PIPE, bitmap); - dispose_command (command); dispose_fd_bitmap (bitmap); discard_unwind_frame ("pe_dispose"); diff --git a/builtins/evalstring.c~ b/builtins/evalstring.c~ new file mode 100644 index 000000000..86e9aaa25 --- /dev/null +++ b/builtins/evalstring.c~ @@ -0,0 +1,508 @@ +/* evalstring.c - evaluate a string as one or more shell commands. */ + +/* Copyright (C) 1996-2010 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include + +#if defined (HAVE_UNISTD_H) +# ifdef _MINIX +# include +# endif +# include +#endif + +#include +#include + +#include + +#include "filecntl.h" +#include "../bashansi.h" + +#include "../shell.h" +#include "../jobs.h" +#include "../builtins.h" +#include "../flags.h" +#include "../input.h" +#include "../execute_cmd.h" +#include "../redir.h" +#include "../trap.h" +#include "../bashintl.h" + +#include + +#if defined (HISTORY) +# include "../bashhist.h" +#endif + +#include "common.h" +#include "builtext.h" + +#if !defined (errno) +extern int errno; +#endif + +#define IS_BUILTIN(s) (builtin_address_internal(s, 0) != (struct builtin *)NULL) + +extern int indirection_level, subshell_environment; +extern int line_number; +extern int current_token, shell_eof_token; +extern int last_command_exit_value; +extern int running_trap; +extern int loop_level; +extern int executing_list; +extern int comsub_ignore_return; +extern int posixly_correct; +extern sh_builtin_func_t *this_shell_builtin; + +int parse_and_execute_level = 0; + +static int cat_file __P((REDIRECT *)); + +#define PE_TAG "parse_and_execute top" +#define PS_TAG "parse_string top" + +#if defined (HISTORY) +static void +set_history_remembering () +{ + remember_on_history = enable_history_list; +} +#endif + +/* How to force parse_and_execute () to clean up after itself. */ +void +parse_and_execute_cleanup () +{ + if (running_trap) + { + run_trap_cleanup (running_trap - 1); + unfreeze_jobs_list (); + } + + if (have_unwind_protects ()) + run_unwind_frame (PE_TAG); + else + parse_and_execute_level = 0; /* XXX */ +} + +static void +parse_prologue (string, flags, tag) + char *string; + int flags; + char *tag; +{ + char *orig_string; + int x; + + orig_string = string; + /* Unwind protect this invocation of parse_and_execute (). */ + begin_unwind_frame (tag); + unwind_protect_int (parse_and_execute_level); + unwind_protect_jmp_buf (top_level); + unwind_protect_int (indirection_level); + unwind_protect_int (line_number); + unwind_protect_int (loop_level); + unwind_protect_int (executing_list); + unwind_protect_int (comsub_ignore_return); + if (flags & (SEVAL_NONINT|SEVAL_INTERACT)) + unwind_protect_int (interactive); + +#if defined (HISTORY) + if (parse_and_execute_level == 0) + add_unwind_protect (set_history_remembering, (char *)NULL); + else + unwind_protect_int (remember_on_history); /* can be used in scripts */ +# if defined (BANG_HISTORY) + if (interactive_shell) + unwind_protect_int (history_expansion_inhibited); +# endif /* BANG_HISTORY */ +#endif /* HISTORY */ + + if (interactive_shell) + { + x = get_current_prompt_level (); + add_unwind_protect (set_current_prompt_level, x); + } + + add_unwind_protect (pop_stream, (char *)NULL); + if (orig_string && ((flags & SEVAL_NOFREE) == 0)) + add_unwind_protect (xfree, orig_string); + end_unwind_frame (); + + if (flags & (SEVAL_NONINT|SEVAL_INTERACT)) + interactive = (flags & SEVAL_NONINT) ? 0 : 1; + +#if defined (HISTORY) + if (flags & SEVAL_NOHIST) + bash_history_disable (); +#endif /* HISTORY */ +} + +/* Parse and execute the commands in STRING. Returns whatever + execute_command () returns. This frees STRING. FLAGS is a + flags word; look in common.h for the possible values. Actions + are: + (flags & SEVAL_NONINT) -> interactive = 0; + (flags & SEVAL_INTERACT) -> interactive = 1; + (flags & SEVAL_NOHIST) -> call bash_history_disable () + (flags & SEVAL_NOFREE) -> don't free STRING when finished + (flags & SEVAL_RESETLINE) -> reset line_number to 1 +*/ + +int +parse_and_execute (string, from_file, flags) + char *string; + const char *from_file; + int flags; +{ + int code, lreset; + volatile int should_jump_to_top_level, last_result; + COMMAND *volatile command; + + parse_prologue (string, flags, PE_TAG); + + parse_and_execute_level++; + + lreset = flags & SEVAL_RESETLINE; + + /* Reset the line number if the caller wants us to. If we don't reset the + line number, we have to subtract one, because we will add one just + before executing the next command (resetting the line number sets it to + 0; the first line number is 1). */ + push_stream (lreset); + if (lreset == 0) + line_number--; + + indirection_level++; + + code = should_jump_to_top_level = 0; + last_result = EXECUTION_SUCCESS; + + with_input_from_string (string, from_file); + while (*(bash_input.location.string)) + { + command = (COMMAND *)NULL; + + if (interrupt_state) + { + last_result = EXECUTION_FAILURE; + break; + } + + /* Provide a location for functions which `longjmp (top_level)' to + jump to. This prevents errors in substitution from restarting + the reader loop directly, for example. */ + code = setjmp (top_level); + + if (code) + { + should_jump_to_top_level = 0; + switch (code) + { + case FORCE_EOF: + case ERREXIT: + case EXITPROG: + if (command) + run_unwind_frame ("pe_dispose"); + /* Remember to call longjmp (top_level) after the old + value for it is restored. */ + should_jump_to_top_level = 1; + goto out; + + case DISCARD: + if (command) + run_unwind_frame ("pe_dispose"); + last_result = last_command_exit_value = EXECUTION_FAILURE; /* XXX */ + if (subshell_environment) + { + should_jump_to_top_level = 1; + goto out; + } + else + { +#if 0 + dispose_command (command); /* pe_dispose does this */ +#endif + continue; + } + + default: + command_error ("parse_and_execute", CMDERR_BADJUMP, code, 0); + break; + } + } + + if (parse_command () == 0) + { + if ((flags & SEVAL_PARSEONLY) || (interactive_shell == 0 && read_but_dont_execute)) + { + last_result = EXECUTION_SUCCESS; + dispose_command (global_command); + global_command = (COMMAND *)NULL; + } + else if (command = global_command) + { + struct fd_bitmap *bitmap; + + bitmap = new_fd_bitmap (FD_BITMAP_SIZE); + begin_unwind_frame ("pe_dispose"); + add_unwind_protect (dispose_fd_bitmap, bitmap); + add_unwind_protect (dispose_command, command); /* XXX */ + + global_command = (COMMAND *)NULL; + + if ((subshell_environment & SUBSHELL_COMSUB) && comsub_ignore_return) + command->flags |= CMD_IGNORE_RETURN; + +#if defined (ONESHOT) + /* + * IF + * we were invoked as `bash -c' (startup_state == 2) AND + * parse_and_execute has not been called recursively AND + * we're not running a trap AND + * we have parsed the full command (string == '\0') AND + * we're not going to run the exit trap AND + * we have a simple command without redirections AND + * the command is not being timed AND + * the command's return status is not being inverted + * THEN + * tell the execution code that we don't need to fork + */ + if (startup_state == 2 && parse_and_execute_level == 1 && + running_trap == 0 && + *bash_input.location.string == '\0' && + command->type == cm_simple && + signal_is_trapped (EXIT_TRAP) == 0 && + command->redirects == 0 && command->value.Simple->redirects == 0 && + ((command->flags & CMD_TIME_PIPELINE) == 0) && + ((command->flags & CMD_INVERT_RETURN) == 0)) + { + command->flags |= CMD_NO_FORK; + command->value.Simple->flags |= CMD_NO_FORK; + } +#endif /* ONESHOT */ + + /* See if this is a candidate for $( type == cm_simple && !command->redirects && + (command->flags & CMD_TIME_PIPELINE) == 0 && + command->value.Simple->words == 0 && + command->value.Simple->redirects && + command->value.Simple->redirects->next == 0 && + command->value.Simple->redirects->instruction == r_input_direction && + command->value.Simple->redirects->redirector.dest == 1) + { + int r; + r = cat_file (command->value.Simple->redirects); + last_result = (r < 0) ? EXECUTION_FAILURE : EXECUTION_SUCCESS; + } + else + last_result = execute_command_internal + (command, 0, NO_PIPE, NO_PIPE, bitmap); + dispose_command (command); + dispose_fd_bitmap (bitmap); + discard_unwind_frame ("pe_dispose"); + } + } + else + { + last_result = EXECUTION_FAILURE; + + if (interactive_shell == 0 && this_shell_builtin && + (this_shell_builtin == source_builtin || this_shell_builtin == eval_builtin) && + last_command_exit_value == EX_BADSYNTAX && posixly_correct) + { + should_jump_to_top_level = 1; + code = ERREXIT; + last_command_exit_value = EX_BADUSAGE; + } + + /* Since we are shell compatible, syntax errors in a script + abort the execution of the script. Right? */ + break; + } + } + + out: + + run_unwind_frame (PE_TAG); + + if (interrupt_state && parse_and_execute_level == 0) + { + /* An interrupt during non-interactive execution in an + interactive shell (e.g. via $PROMPT_COMMAND) should + not cause the shell to exit. */ + interactive = interactive_shell; + throw_to_top_level (); + } + + if (should_jump_to_top_level) + jump_to_top_level (code); + + return (last_result); +} + +/* Parse a command contained in STRING according to FLAGS and return the + number of characters consumed from the string. If non-NULL, set *ENDP + to the position in the string where the parse ended. Used to validate + command substitutions during parsing to obey Posix rules about finding + the end of the command and balancing parens. */ +int +parse_string (string, from_file, flags, endp) + char *string; + const char *from_file; + int flags; + char **endp; +{ + int code, nc; + volatile int should_jump_to_top_level; + COMMAND *volatile command, *oglobal; + char *ostring; + + parse_prologue (string, flags, PS_TAG); + + /* Reset the line number if the caller wants us to. If we don't reset the + line number, we have to subtract one, because we will add one just + before executing the next command (resetting the line number sets it to + 0; the first line number is 1). */ + push_stream (0); + + code = should_jump_to_top_level = 0; + oglobal = global_command; + ostring = string; + + with_input_from_string (string, from_file); + while (*(bash_input.location.string)) + { + command = (COMMAND *)NULL; + +#if 0 + if (interrupt_state) + break; +#endif + + /* Provide a location for functions which `longjmp (top_level)' to + jump to. */ + code = setjmp (top_level); + + if (code) + { +#if defined (DEBUG) +itrace("parse_string: longjmp executed: code = %d", code); +#endif + should_jump_to_top_level = 0; + switch (code) + { + case FORCE_EOF: + case ERREXIT: + case EXITPROG: + case DISCARD: /* XXX */ + if (command) + dispose_command (command); + /* Remember to call longjmp (top_level) after the old + value for it is restored. */ + should_jump_to_top_level = 1; + goto out; + + default: + command_error ("parse_string", CMDERR_BADJUMP, code, 0); + break; + } + } + + if (parse_command () == 0) + { + dispose_command (global_command); + global_command = (COMMAND *)NULL; + } + else + { + if ((flags & SEVAL_NOLONGJMP) == 0) + { + should_jump_to_top_level = 1; + code = DISCARD; + } + else + reset_parser (); /* XXX - sets token_to_read */ + break; + } + + if (current_token == yacc_EOF || current_token == shell_eof_token) + break; + } + + out: + + global_command = oglobal; + nc = bash_input.location.string - ostring; + if (endp) + *endp = bash_input.location.string; + + run_unwind_frame (PS_TAG); + + if (should_jump_to_top_level) + jump_to_top_level (code); + + return (nc); +} + +/* Handle a $( < file ) command substitution. This expands the filename, + returning errors as appropriate, then just cats the file to the standard + output. */ +static int +cat_file (r) + REDIRECT *r; +{ + char *fn; + int fd, rval; + + if (r->instruction != r_input_direction) + return -1; + + /* Get the filename. */ + if (posixly_correct && !interactive_shell) + disallow_filename_globbing++; + fn = redirection_expand (r->redirectee.filename); + if (posixly_correct && !interactive_shell) + disallow_filename_globbing--; + + if (fn == 0) + { + redirection_error (r, AMBIGUOUS_REDIRECT); + return -1; + } + + fd = open(fn, O_RDONLY); + if (fd < 0) + { + file_error (fn); + free (fn); + return -1; + } + + rval = zcatfd (fd, 1, fn); + + free (fn); + close (fd); + + return (rval); +} diff --git a/builtins/printf.def b/builtins/printf.def index a5162aa8a..27cb235a8 100644 --- a/builtins/printf.def +++ b/builtins/printf.def @@ -870,7 +870,7 @@ tescape (estart, cp, lenp, sawc) *cp = '\\'; return 0; } - if (uvalue <= UCHAR_MAX) + if (uvalue <= 0x7f) /* <= 0x7f translates directly */ *cp = uvalue; else { diff --git a/builtins/printf.def~ b/builtins/printf.def~ new file mode 100644 index 000000000..a5162aa8a --- /dev/null +++ b/builtins/printf.def~ @@ -0,0 +1,1252 @@ +This file is printf.def, from which is created printf.c. +It implements the builtin "printf" in Bash. + +Copyright (C) 1997-2010 Free Software Foundation, Inc. + +This file is part of GNU Bash, the Bourne Again SHell. + +Bash is free software: you can redistribute it and/or modify +it under the terms of the GNU General Public License as published by +the Free Software Foundation, either version 3 of the License, or +(at your option) any later version. + +Bash is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with Bash. If not, see . + +$PRODUCES printf.c + +$BUILTIN printf +$FUNCTION printf_builtin +$SHORT_DOC printf [-v var] format [arguments] +Formats and prints ARGUMENTS under control of the FORMAT. + +Options: + -v var assign the output to shell variable VAR rather than + display it on the standard output + +FORMAT is a character string which contains three types of objects: plain +characters, which are simply copied to standard output; character escape +sequences, which are converted and copied to the standard output; and +format specifications, each of which causes printing of the next successive +argument. + +In addition to the standard format specifications described in printf(1), +printf interprets: + + %b expand backslash escape sequences in the corresponding argument + %q quote the argument in a way that can be reused as shell input + %(fmt)T output the date-time string resulting from using FMT as a format + string for strftime(3) + +Exit Status: +Returns success unless an invalid option is given or a write or assignment +error occurs. +$END + +#include + +#include "../bashtypes.h" + +#include +#if defined (HAVE_LIMITS_H) +# include +#else + /* Assume 32-bit ints. */ +# define INT_MAX 2147483647 +# define INT_MIN (-2147483647-1) +#endif + +#if defined (PREFER_STDARG) +# include +#else +# include +#endif + +#include +#include + +#ifdef HAVE_INTTYPES_H +# include +#endif + +#include "posixtime.h" +#include "../bashansi.h" +#include "../bashintl.h" + +#define NEED_STRFTIME_DECL + +#include "../shell.h" +#include "shmbutil.h" +#include "stdc.h" +#include "bashgetopt.h" +#include "common.h" + +#if defined (PRI_MACROS_BROKEN) +# undef PRIdMAX +#endif + +#if !defined (PRIdMAX) +# if HAVE_LONG_LONG +# define PRIdMAX "lld" +# else +# define PRIdMAX "ld" +# endif +#endif + +#if !defined (errno) +extern int errno; +#endif + +#define PC(c) \ + do { \ + char b[2]; \ + tw++; \ + b[0] = c; b[1] = '\0'; \ + if (vflag) \ + vbadd (b, 1); \ + else \ + putchar (c); \ + } while (0) + +#define PF(f, func) \ + do { \ + int nw; \ + clearerr (stdout); \ + if (have_fieldwidth && have_precision) \ + nw = vflag ? vbprintf (f, fieldwidth, precision, func) : printf (f, fieldwidth, precision, func); \ + else if (have_fieldwidth) \ + nw = vflag ? vbprintf (f, fieldwidth, func) : printf (f, fieldwidth, func); \ + else if (have_precision) \ + nw = vflag ? vbprintf (f, precision, func) : printf (f, precision, func); \ + else \ + nw = vflag ? vbprintf (f, func) : printf (f, func); \ + tw += nw; \ + if (ferror (stdout)) \ + { \ + sh_wrerror (); \ + clearerr (stdout); \ + return (EXECUTION_FAILURE); \ + } \ + } while (0) + +/* We free the buffer used by mklong() if it's `too big'. */ +#define PRETURN(value) \ + do \ + { \ + if (vflag) \ + { \ + bind_printf_variable (vname, vbuf, 0); \ + stupidly_hack_special_variables (vname); \ + } \ + if (conv_bufsize > 4096 ) \ + { \ + free (conv_buf); \ + conv_bufsize = 0; \ + conv_buf = 0; \ + } \ + if (vbsize > 4096) \ + { \ + free (vbuf); \ + vbsize = 0; \ + vbuf = 0; \ + } \ + else if (vbuf) \ + vbuf[0] = 0; \ + terminate_immediately--; \ + if (ferror (stdout) == 0) \ + fflush (stdout); \ + if (ferror (stdout)) \ + { \ + sh_wrerror (); \ + clearerr (stdout); \ + return (EXECUTION_FAILURE); \ + } \ + return (value); \ + } \ + while (0) + +#define SKIP1 "#'-+ 0" +#define LENMODS "hjlLtz" + +extern time_t shell_start_time; + +#if !HAVE_ASPRINTF +extern int asprintf __P((char **, const char *, ...)) __attribute__((__format__ (printf, 2, 3))); +#endif + +#if !HAVE_VSNPRINTF +extern int vsnprintf __P((char *, size_t, const char *, va_list)) __attribute__((__format__ (printf, 3, 0))); +#endif + +static void printf_erange __P((char *)); +static int printstr __P((char *, char *, int, int, int)); +static int tescape __P((char *, char *, int *, int *)); +static char *bexpand __P((char *, int, int *, int *)); +static char *vbadd __P((char *, int)); +static int vbprintf __P((const char *, ...)) __attribute__((__format__ (printf, 1, 2))); +static char *mklong __P((char *, char *, size_t)); +static int getchr __P((void)); +static char *getstr __P((void)); +static int getint __P((void)); +static intmax_t getintmax __P((void)); +static uintmax_t getuintmax __P((void)); +static SHELL_VAR *bind_printf_variable __P((char *, char *, int)); + +#if defined (HAVE_LONG_DOUBLE) && HAVE_DECL_STRTOLD && !defined(STRTOLD_BROKEN) +typedef long double floatmax_t; +# define FLOATMAX_CONV "L" +# define strtofltmax strtold +#else +typedef double floatmax_t; +# define FLOATMAX_CONV "" +# define strtofltmax strtod +#endif +static floatmax_t getfloatmax __P((void)); + +static intmax_t asciicode __P((void)); + +static WORD_LIST *garglist; +static int retval; +static int conversion_error; + +/* printf -v var support */ +static int vflag = 0; +static char *vbuf, *vname; +static size_t vbsize; +static int vblen; + +static intmax_t tw; + +static char *conv_buf; +static size_t conv_bufsize; + +int +printf_builtin (list) + WORD_LIST *list; +{ + int ch, fieldwidth, precision; + int have_fieldwidth, have_precision; + char convch, thisch, nextch, *format, *modstart, *fmt, *start; +#if defined (HANDLE_MULTIBYTE) + char mbch[25]; /* 25 > MB_LEN_MAX, plus can handle 4-byte UTF-8 and large Unicode characters*/ + int mbind, mblen; +#endif + + conversion_error = 0; + retval = EXECUTION_SUCCESS; + + vflag = 0; + + reset_internal_getopt (); + while ((ch = internal_getopt (list, "v:")) != -1) + { + switch (ch) + { + case 'v': + vname = list_optarg; +#if defined (ARRAY_VARS) + if (legal_identifier (vname) || valid_array_reference (vname)) +#else + if (legal_identifier (vname)) +#endif + { + vflag = 1; + if (vbsize == 0) + vbuf = xmalloc (vbsize = 16); + vblen = 0; + if (vbuf) + vbuf[0] = 0; + } + else + { + sh_invalidid (vname); + return (EX_USAGE); + } + break; + default: + builtin_usage (); + return (EX_USAGE); + } + } + list = loptend; /* skip over possible `--' */ + + if (list == 0) + { + builtin_usage (); + return (EX_USAGE); + } + + if (list->word->word == 0 || list->word->word[0] == '\0') + return (EXECUTION_SUCCESS); + + format = list->word->word; + tw = 0; + + garglist = list->next; + + /* If the format string is empty after preprocessing, return immediately. */ + if (format == 0 || *format == 0) + return (EXECUTION_SUCCESS); + + terminate_immediately++; + + /* Basic algorithm is to scan the format string for conversion + specifications -- once one is found, find out if the field + width or precision is a '*'; if it is, gather up value. Note, + format strings are reused as necessary to use up the provided + arguments, arguments of zero/null string are provided to use + up the format string. */ + do + { + tw = 0; + /* find next format specification */ + for (fmt = format; *fmt; fmt++) + { + precision = fieldwidth = 0; + have_fieldwidth = have_precision = 0; + + if (*fmt == '\\') + { + fmt++; + /* A NULL third argument to tescape means to bypass the + special processing for arguments to %b. */ +#if defined (HANDLE_MULTIBYTE) + /* Accommodate possible use of \u or \U, which can result in + multibyte characters */ + memset (mbch, '\0', sizeof (mbch)); + fmt += tescape (fmt, mbch, &mblen, (int *)NULL); + for (mbind = 0; mbind < mblen; mbind++) + PC (mbch[mbind]); +#else + fmt += tescape (fmt, &nextch, (int *)NULL, (int *)NULL); + PC (nextch); +#endif + fmt--; /* for loop will increment it for us again */ + continue; + } + + if (*fmt != '%') + { + PC (*fmt); + continue; + } + + /* ASSERT(*fmt == '%') */ + start = fmt++; + + if (*fmt == '%') /* %% prints a % */ + { + PC ('%'); + continue; + } + + /* found format specification, skip to field width */ + for (; *fmt && strchr(SKIP1, *fmt); ++fmt) + ; + + /* Skip optional field width. */ + if (*fmt == '*') + { + fmt++; + have_fieldwidth = 1; + fieldwidth = getint (); + } + else + while (DIGIT (*fmt)) + fmt++; + + /* Skip optional '.' and precision */ + if (*fmt == '.') + { + ++fmt; + if (*fmt == '*') + { + fmt++; + have_precision = 1; + precision = getint (); + } + else + { + /* Negative precisions are allowed but treated as if the + precision were missing; I would like to allow a leading + `+' in the precision number as an extension, but lots + of asprintf/fprintf implementations get this wrong. */ +#if 0 + if (*fmt == '-' || *fmt == '+') +#else + if (*fmt == '-') +#endif + fmt++; + while (DIGIT (*fmt)) + fmt++; + } + } + + /* skip possible format modifiers */ + modstart = fmt; + while (*fmt && strchr (LENMODS, *fmt)) + fmt++; + + if (*fmt == 0) + { + builtin_error (_("`%s': missing format character"), start); + PRETURN (EXECUTION_FAILURE); + } + + convch = *fmt; + thisch = modstart[0]; + nextch = modstart[1]; + modstart[0] = convch; + modstart[1] = '\0'; + + switch(convch) + { + case 'c': + { + char p; + + p = getchr (); + PF(start, p); + break; + } + + case 's': + { + char *p; + + p = getstr (); + PF(start, p); + break; + } + + case '(': + { + char *timefmt, timebuf[128], *t; + int n; + intmax_t arg; + time_t secs; + struct tm *tm; + + modstart[1] = nextch; /* restore char after left paren */ + timefmt = xmalloc (strlen (fmt) + 3); + fmt++; /* skip over left paren */ + for (t = timefmt, n = 1; *fmt; ) + { + if (*fmt == '(') + n++; + else if (*fmt == ')') + n--; + if (n == 0) + break; + *t++ = *fmt++; + } + *t = '\0'; + if (*++fmt != 'T') + { + builtin_warning (_("`%c': invalid time format specification"), *fmt); + fmt = start; + free (timefmt); + PC (*fmt); + continue; + } + if (timefmt[0] == '\0') + { + timefmt[0] = '%'; + timefmt[1] = 'X'; /* locale-specific current time - should we use `+'? */ + timefmt[2] = '\0'; + } + /* argument is seconds since the epoch with special -1 and -2 */ + arg = getintmax (); + if (arg == -1) + secs = NOW; /* roughly date +%s */ + else if (arg == -2) + secs = shell_start_time; /* roughly $SECONDS */ + else + secs = arg; +#if defined (HAVE_TZSET) + sv_tz ("TZ"); /* XXX -- just make sure */ +#endif + tm = localtime (&secs); + n = strftime (timebuf, sizeof (timebuf), timefmt, tm); + free (timefmt); + if (n == 0) + timebuf[0] = '\0'; + else + timebuf[sizeof(timebuf) - 1] = '\0'; + /* convert to %s format that preserves fieldwidth and precision */ + modstart[0] = 's'; + modstart[1] = '\0'; + n = printstr (start, timebuf, strlen (timebuf), fieldwidth, precision); /* XXX - %s for now */ + if (n < 0) + { + if (ferror (stdout) == 0) + { + sh_wrerror (); + clearerr (stdout); + } + PRETURN (EXECUTION_FAILURE); + } + break; + } + + case 'n': + { + char *var; + + var = getstr (); + if (var && *var) + { + if (legal_identifier (var)) + bind_var_to_int (var, tw); + else + { + sh_invalidid (var); + PRETURN (EXECUTION_FAILURE); + } + } + break; + } + + case 'b': /* expand escapes in argument */ + { + char *p, *xp; + int rlen, r; + + p = getstr (); + ch = rlen = r = 0; + xp = bexpand (p, strlen (p), &ch, &rlen); + + if (xp) + { + /* Have to use printstr because of possible NUL bytes + in XP -- printf does not handle that well. */ + r = printstr (start, xp, rlen, fieldwidth, precision); + if (r < 0) + { + if (ferror (stdout) == 0) + { + sh_wrerror (); + clearerr (stdout); + } + retval = EXECUTION_FAILURE; + } + free (xp); + } + + if (ch || r < 0) + PRETURN (retval); + break; + } + + case 'q': /* print with shell quoting */ + { + char *p, *xp; + int r; + + r = 0; + p = getstr (); + if (p && *p == 0) /* XXX - getstr never returns null */ + xp = savestring ("''"); + else if (ansic_shouldquote (p)) + xp = ansic_quote (p, 0, (int *)0); + else + xp = sh_backslash_quote (p, 0); + if (xp) + { + /* Use printstr to get fieldwidth and precision right. */ + r = printstr (start, xp, strlen (xp), fieldwidth, precision); + if (r < 0) + { + sh_wrerror (); + clearerr (stdout); + } + free (xp); + } + + if (r < 0) + PRETURN (EXECUTION_FAILURE); + break; + } + + case 'd': + case 'i': + { + char *f; + long p; + intmax_t pp; + + p = pp = getintmax (); + if (p != pp) + { + f = mklong (start, PRIdMAX, sizeof (PRIdMAX) - 2); + PF (f, pp); + } + else + { + /* Optimize the common case where the integer fits + in "long". This also works around some long + long and/or intmax_t library bugs in the common + case, e.g. glibc 2.2 x86. */ + f = mklong (start, "l", 1); + PF (f, p); + } + break; + } + + case 'o': + case 'u': + case 'x': + case 'X': + { + char *f; + unsigned long p; + uintmax_t pp; + + p = pp = getuintmax (); + if (p != pp) + { + f = mklong (start, PRIdMAX, sizeof (PRIdMAX) - 2); + PF (f, pp); + } + else + { + f = mklong (start, "l", 1); + PF (f, p); + } + break; + } + + case 'e': + case 'E': + case 'f': + case 'F': + case 'g': + case 'G': +#if defined (HAVE_PRINTF_A_FORMAT) + case 'a': + case 'A': +#endif + { + char *f; + floatmax_t p; + + p = getfloatmax (); + f = mklong (start, FLOATMAX_CONV, sizeof(FLOATMAX_CONV) - 1); + PF (f, p); + break; + } + + /* We don't output unrecognized format characters; we print an + error message and return a failure exit status. */ + default: + builtin_error (_("`%c': invalid format character"), convch); + PRETURN (EXECUTION_FAILURE); + } + + modstart[0] = thisch; + modstart[1] = nextch; + } + + if (ferror (stdout)) + { + /* PRETURN will print error message. */ + PRETURN (EXECUTION_FAILURE); + } + } + while (garglist && garglist != list->next); + + if (conversion_error) + retval = EXECUTION_FAILURE; + + PRETURN (retval); +} + +static void +printf_erange (s) + char *s; +{ + builtin_error (_("warning: %s: %s"), s, strerror(ERANGE)); +} + +/* We duplicate a lot of what printf(3) does here. */ +static int +printstr (fmt, string, len, fieldwidth, precision) + char *fmt; /* format */ + char *string; /* expanded string argument */ + int len; /* length of expanded string */ + int fieldwidth; /* argument for width of `*' */ + int precision; /* argument for precision of `*' */ +{ +#if 0 + char *s; +#endif + int padlen, nc, ljust, i; + int fw, pr; /* fieldwidth and precision */ + intmax_t mfw, mpr; + + if (string == 0 || len == 0) + return 0; + +#if 0 + s = fmt; +#endif + if (*fmt == '%') + fmt++; + + ljust = fw = 0; + pr = -1; + mfw = 0; + mpr = -1; + + /* skip flags */ + while (strchr (SKIP1, *fmt)) + { + if (*fmt == '-') + ljust = 1; + fmt++; + } + + /* get fieldwidth, if present. rely on caller to clamp fieldwidth at INT_MAX */ + if (*fmt == '*') + { + fmt++; + fw = fieldwidth; + if (fw < 0) + { + fw = -fw; + ljust = 1; + } + } + else if (DIGIT (*fmt)) + { + mfw = *fmt++ - '0'; + while (DIGIT (*fmt)) + mfw = (mfw * 10) + (*fmt++ - '0'); + /* Error if fieldwidth > INT_MAX here? */ + fw = (mfw < 0 || mfw > INT_MAX) ? INT_MAX : mfw; + } + + /* get precision, if present */ + if (*fmt == '.') + { + fmt++; + if (*fmt == '*') + { + fmt++; + pr = precision; + } + else if (DIGIT (*fmt)) + { + mpr = *fmt++ - '0'; + while (DIGIT (*fmt)) + mpr = (mpr * 10) + (*fmt++ - '0'); + /* Error if precision > INT_MAX here? */ + pr = (mpr < 0 || mpr > INT_MAX) ? INT_MAX : mpr; + } + } + +#if 0 + /* If we remove this, get rid of `s'. */ + if (*fmt != 'b' && *fmt != 'q') + { + internal_error ("format parsing problem: %s", s); + fw = pr = 0; + } +#endif + + /* chars from string to print */ + nc = (pr >= 0 && pr <= len) ? pr : len; + + padlen = fw - nc; + if (padlen < 0) + padlen = 0; + if (ljust) + padlen = -padlen; + + /* leading pad characters */ + for (; padlen > 0; padlen--) + PC (' '); + + /* output NC characters from STRING */ + for (i = 0; i < nc; i++) + PC (string[i]); + + /* output any necessary trailing padding */ + for (; padlen < 0; padlen++) + PC (' '); + + return (ferror (stdout) ? -1 : 0); +} + +/* Convert STRING by expanding the escape sequences specified by the + POSIX standard for printf's `%b' format string. If SAWC is non-null, + perform the processing appropriate for %b arguments. In particular, + recognize `\c' and use that as a string terminator. If we see \c, set + *SAWC to 1 before returning. LEN is the length of STRING. */ + +/* Translate a single backslash-escape sequence starting at ESTART (the + character after the backslash) and return the number of characters + consumed by the sequence. CP is the place to return the translated + value. *SAWC is set to 1 if the escape sequence was \c, since that means + to short-circuit the rest of the processing. If SAWC is null, we don't + do the \c short-circuiting, and \c is treated as an unrecognized escape + sequence; we also bypass the other processing specific to %b arguments. */ +static int +tescape (estart, cp, lenp, sawc) + char *estart; + char *cp; + int *lenp, *sawc; +{ + register char *p; + int temp, c, evalue; + unsigned long uvalue; + + p = estart; + if (lenp) + *lenp = 1; + + switch (c = *p++) + { +#if defined (__STDC__) + case 'a': *cp = '\a'; break; +#else + case 'a': *cp = '\007'; break; +#endif + + case 'b': *cp = '\b'; break; + + case 'e': + case 'E': *cp = '\033'; break; /* ESC -- non-ANSI */ + + case 'f': *cp = '\f'; break; + + case 'n': *cp = '\n'; break; + + case 'r': *cp = '\r'; break; + + case 't': *cp = '\t'; break; + + case 'v': *cp = '\v'; break; + + /* The octal escape sequences are `\0' followed by up to three octal + digits (if SAWC), or `\' followed by up to three octal digits (if + !SAWC). As an extension, we allow the latter form even if SAWC. */ + case '0': case '1': case '2': case '3': + case '4': case '5': case '6': case '7': + evalue = OCTVALUE (c); + for (temp = 2 + (!evalue && !!sawc); ISOCTAL (*p) && temp--; p++) + evalue = (evalue * 8) + OCTVALUE (*p); + *cp = evalue & 0xFF; + break; + + /* And, as another extension, we allow \xNN, where each N is a + hex digit. */ + case 'x': + for (temp = 2, evalue = 0; ISXDIGIT ((unsigned char)*p) && temp--; p++) + evalue = (evalue * 16) + HEXVALUE (*p); + if (p == estart + 1) + { + builtin_error (_("missing hex digit for \\x")); + *cp = '\\'; + return 0; + } + *cp = evalue & 0xFF; + break; + +#if defined (HANDLE_MULTIBYTE) + case 'u': + case 'U': + temp = (c == 'u') ? 4 : 8; /* \uNNNN \UNNNNNNNN */ + for (uvalue = 0; ISXDIGIT ((unsigned char)*p) && temp--; p++) + uvalue = (uvalue * 16) + HEXVALUE (*p); + if (p == estart + 1) + { + builtin_error (_("missing unicode digit for \\%c"), c); + *cp = '\\'; + return 0; + } + if (uvalue <= UCHAR_MAX) + *cp = uvalue; + else + { + temp = u32cconv (uvalue, cp); + cp[temp] = '\0'; + if (lenp) + *lenp = temp; + } + break; +#endif + + case '\\': /* \\ -> \ */ + *cp = c; + break; + + /* SAWC == 0 means that \', \", and \? are recognized as escape + sequences, though the only processing performed is backslash + removal. */ + case '\'': case '"': case '?': + if (!sawc) + *cp = c; + else + { + *cp = '\\'; + return 0; + } + break; + + case 'c': + if (sawc) + { + *sawc = 1; + break; + } + /* other backslash escapes are passed through unaltered */ + default: + *cp = '\\'; + return 0; + } + return (p - estart); +} + +static char * +bexpand (string, len, sawc, lenp) + char *string; + int len, *sawc, *lenp; +{ + int temp; + char *ret, *r, *s, c; +#if defined (HANDLE_MULTIBYTE) + char mbch[25]; + int mbind, mblen; +#endif + + if (string == 0 || len == 0) + { + if (sawc) + *sawc = 0; + if (lenp) + *lenp = 0; + return ((char *)NULL); + } + + ret = (char *)xmalloc (len + 1); + for (r = ret, s = string; s && *s; ) + { + c = *s++; + if (c != '\\' || *s == '\0') + { + *r++ = c; + continue; + } + temp = 0; +#if defined (HANDLE_MULTIBYTE) + memset (mbch, '\0', sizeof (mbch)); + s += tescape (s, mbch, &mblen, &temp); +#else + s += tescape (s, &c, (int *)NULL, &temp); +#endif + if (temp) + { + if (sawc) + *sawc = 1; + break; + } + +#if defined (HANDLE_MULTIBYTE) + for (mbind = 0; mbind < mblen; mbind++) + *r++ = mbch[mbind]; +#else + *r++ = c; +#endif + } + + *r = '\0'; + if (lenp) + *lenp = r - ret; + return ret; +} + +static char * +vbadd (buf, blen) + char *buf; + int blen; +{ + size_t nlen; + + nlen = vblen + blen + 1; + if (nlen >= vbsize) + { + vbsize = ((nlen + 63) >> 6) << 6; + vbuf = (char *)xrealloc (vbuf, vbsize); + } + + if (blen == 1) + vbuf[vblen++] = buf[0]; + else if (blen > 1) + { + FASTCOPY (buf, vbuf + vblen, blen); + vblen += blen; + } + vbuf[vblen] = '\0'; + +#ifdef DEBUG + if (strlen (vbuf) != vblen) + internal_error ("printf:vbadd: vblen (%d) != strlen (vbuf) (%d)", vblen, (int)strlen (vbuf)); +#endif + + return vbuf; +} + +static int +#if defined (PREFER_STDARG) +vbprintf (const char *format, ...) +#else +vbprintf (format, va_alist) + const char *format; + va_dcl +#endif +{ + va_list args; + size_t nlen; + int blen; + + SH_VA_START (args, format); + blen = vsnprintf (vbuf + vblen, vbsize - vblen, format, args); + va_end (args); + + nlen = vblen + blen + 1; + if (nlen >= vbsize) + { + vbsize = ((nlen + 63) >> 6) << 6; + vbuf = (char *)xrealloc (vbuf, vbsize); + SH_VA_START (args, format); + blen = vsnprintf (vbuf + vblen, vbsize - vblen, format, args); + va_end (args); + } + + vblen += blen; + vbuf[vblen] = '\0'; + +#ifdef DEBUG + if (strlen (vbuf) != vblen) + internal_error ("printf:vbadd: vblen (%d) != strlen (vbuf) (%d)", vblen, (int)strlen (vbuf)); +#endif + + return (blen); +} + +static char * +mklong (str, modifiers, mlen) + char *str; + char *modifiers; + size_t mlen; +{ + size_t len, slen; + + slen = strlen (str); + len = slen + mlen + 1; + + if (len > conv_bufsize) + { + conv_bufsize = (((len + 1023) >> 10) << 10); + conv_buf = (char *)xrealloc (conv_buf, conv_bufsize); + } + + FASTCOPY (str, conv_buf, slen - 1); + FASTCOPY (modifiers, conv_buf + slen - 1, mlen); + + conv_buf[len - 2] = str[slen - 1]; + conv_buf[len - 1] = '\0'; + return (conv_buf); +} + +static int +getchr () +{ + int ret; + + if (garglist == 0) + return ('\0'); + + ret = (int)garglist->word->word[0]; + garglist = garglist->next; + return ret; +} + +static char * +getstr () +{ + char *ret; + + if (garglist == 0) + return (""); + + ret = garglist->word->word; + garglist = garglist->next; + return ret; +} + +static int +getint () +{ + intmax_t ret; + + ret = getintmax (); + + if (garglist == 0) + return ret; + + if (ret > INT_MAX) + { + printf_erange (garglist->word->word); + ret = INT_MAX; + } + else if (ret < INT_MIN) + { + printf_erange (garglist->word->word); + ret = INT_MIN; + } + + return ((int)ret); +} + +static intmax_t +getintmax () +{ + intmax_t ret; + char *ep; + + if (garglist == 0) + return (0); + + if (garglist->word->word[0] == '\'' || garglist->word->word[0] == '"') + return asciicode (); + + errno = 0; + ret = strtoimax (garglist->word->word, &ep, 0); + + if (*ep) + { + sh_invalidnum (garglist->word->word); + /* POSIX.2 says ``...a diagnostic message shall be written to standard + error, and the utility shall not exit with a zero exit status, but + shall continue processing any remaining operands and shall write the + value accumulated at the time the error was detected to standard + output.'' Yecch. */ +#if 0 + ret = 0; /* return partially-converted value from strtoimax */ +#endif + conversion_error = 1; + } + else if (errno == ERANGE) + printf_erange (garglist->word->word); + + garglist = garglist->next; + return (ret); +} + +static uintmax_t +getuintmax () +{ + uintmax_t ret; + char *ep; + + if (garglist == 0) + return (0); + + if (garglist->word->word[0] == '\'' || garglist->word->word[0] == '"') + return asciicode (); + + errno = 0; + ret = strtoumax (garglist->word->word, &ep, 0); + + if (*ep) + { + sh_invalidnum (garglist->word->word); + /* Same POSIX.2 conversion error requirements as getintmax(). */ + ret = 0; + conversion_error = 1; + } + else if (errno == ERANGE) + printf_erange (garglist->word->word); + + garglist = garglist->next; + return (ret); +} + +static floatmax_t +getfloatmax () +{ + floatmax_t ret; + char *ep; + + if (garglist == 0) + return (0); + + if (garglist->word->word[0] == '\'' || garglist->word->word[0] == '"') + return asciicode (); + + errno = 0; + ret = strtofltmax (garglist->word->word, &ep); + + if (*ep) + { + sh_invalidnum (garglist->word->word); + /* Same thing about POSIX.2 conversion error requirements. */ + ret = 0; + conversion_error = 1; + } + else if (errno == ERANGE) + printf_erange (garglist->word->word); + + garglist = garglist->next; + return (ret); +} + +/* NO check is needed for garglist here. */ +static intmax_t +asciicode () +{ + register intmax_t ch; +#if defined (HANDLE_MULTIBYTE) + wchar_t wc; + size_t mblength, slen; +#endif + DECLARE_MBSTATE; + +#if defined (HANDLE_MULTIBYTE) + slen = strlen (garglist->word->word+1); + mblength = MBLEN (garglist->word->word+1, slen); + if (mblength > 1) + { + mblength = mbtowc (&wc, garglist->word->word+1, slen); + ch = wc; /* XXX */ + } + else +#endif + ch = (unsigned char)garglist->word->word[1]; + + garglist = garglist->next; + return (ch); +} + +static SHELL_VAR * +bind_printf_variable (name, value, flags) + char *name; + char *value; + int flags; +{ +#if defined (ARRAY_VARS) + if (valid_array_reference (name) == 0) + return (bind_variable (name, value, flags)); + else + return (assign_array_element (name, value, flags)); +#else /* !ARRAY_VARS */ + return bind_variable (name, value, flags); +#endif /* !ARRAY_VARS */ +} diff --git a/execute_cmd.c b/execute_cmd.c index c6f98906f..19a685da0 100644 --- a/execute_cmd.c +++ b/execute_cmd.c @@ -693,7 +693,8 @@ execute_command_internal (command, asynchronous, pipe_in, pipe_out, redirection_undo_list = (REDIRECT *)NULL; dispose_exec_redirects (); #if defined (PROCESS_SUBSTITUTION) - free (ofifo_list); + if (saved_fifo) + free (ofifo_list); #endif return (last_command_exit_value = EXECUTION_FAILURE); } diff --git a/execute_cmd.c~ b/execute_cmd.c~ new file mode 100644 index 000000000..05896100c --- /dev/null +++ b/execute_cmd.c~ @@ -0,0 +1,5336 @@ +/* execute_cmd.c -- Execute a COMMAND structure. */ + +/* Copyright (C) 1987-2012 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include "config.h" + +#if !defined (__GNUC__) && !defined (HAVE_ALLOCA_H) && defined (_AIX) + #pragma alloca +#endif /* _AIX && RISC6000 && !__GNUC__ */ + +#include +#include "chartypes.h" +#include "bashtypes.h" +#if !defined (_MINIX) && defined (HAVE_SYS_FILE_H) +# include +#endif +#include "filecntl.h" +#include "posixstat.h" +#include +#ifndef _MINIX +# include +#endif + +#if defined (HAVE_UNISTD_H) +# include +#endif + +#include "posixtime.h" + +#if defined (HAVE_SYS_RESOURCE_H) && !defined (RLIMTYPE) +# include +#endif + +#if defined (HAVE_SYS_TIMES_H) && defined (HAVE_TIMES) +# include +#endif + +#include + +#if !defined (errno) +extern int errno; +#endif + +#define NEED_FPURGE_DECL + +#include "bashansi.h" +#include "bashintl.h" + +#include "memalloc.h" +#include "shell.h" +#include /* use <...> so we pick it up from the build directory */ +#include "flags.h" +#include "builtins.h" +#include "hashlib.h" +#include "jobs.h" +#include "execute_cmd.h" +#include "findcmd.h" +#include "redir.h" +#include "trap.h" +#include "pathexp.h" +#include "hashcmd.h" + +#if defined (COND_COMMAND) +# include "test.h" +#endif + +#include "builtins/common.h" +#include "builtins/builtext.h" /* list of builtins */ + +#include +#include + +#if defined (BUFFERED_INPUT) +# include "input.h" +#endif + +#if defined (ALIAS) +# include "alias.h" +#endif + +#if defined (HISTORY) +# include "bashhist.h" +#endif + +extern int dollar_dollar_pid; +extern int posixly_correct; +extern int expand_aliases; +extern int autocd; +extern int breaking, continuing, loop_level; +extern int parse_and_execute_level, running_trap, sourcelevel; +extern int command_string_index, line_number; +extern int dot_found_in_search; +extern int already_making_children; +extern int tempenv_assign_error; +extern char *the_printed_command, *shell_name; +extern pid_t last_command_subst_pid; +extern sh_builtin_func_t *last_shell_builtin, *this_shell_builtin; +extern char **subshell_argv, **subshell_envp; +extern int subshell_argc; +extern time_t shell_start_time; +#if 0 +extern char *glob_argv_flags; +#endif + +extern int close __P((int)); + +/* Static functions defined and used in this file. */ +static void close_pipes __P((int, int)); +static void do_piping __P((int, int)); +static void bind_lastarg __P((char *)); +static int shell_control_structure __P((enum command_type)); +static void cleanup_redirects __P((REDIRECT *)); + +#if defined (JOB_CONTROL) +static int restore_signal_mask __P((sigset_t *)); +#endif + +static void async_redirect_stdin __P((void)); + +static int builtin_status __P((int)); + +static int execute_for_command __P((FOR_COM *)); +#if defined (SELECT_COMMAND) +static int displen __P((const char *)); +static int print_index_and_element __P((int, int, WORD_LIST *)); +static void indent __P((int, int)); +static void print_select_list __P((WORD_LIST *, int, int, int)); +static char *select_query __P((WORD_LIST *, int, char *, int)); +static int execute_select_command __P((SELECT_COM *)); +#endif +#if defined (DPAREN_ARITHMETIC) +static int execute_arith_command __P((ARITH_COM *)); +#endif +#if defined (COND_COMMAND) +static int execute_cond_node __P((COND_COM *)); +static int execute_cond_command __P((COND_COM *)); +#endif +#if defined (COMMAND_TIMING) +static int mkfmt __P((char *, int, int, time_t, int)); +static void print_formatted_time __P((FILE *, char *, + time_t, int, time_t, int, + time_t, int, int)); +static int time_command __P((COMMAND *, int, int, int, struct fd_bitmap *)); +#endif +#if defined (ARITH_FOR_COMMAND) +static intmax_t eval_arith_for_expr __P((WORD_LIST *, int *)); +static int execute_arith_for_command __P((ARITH_FOR_COM *)); +#endif +static int execute_case_command __P((CASE_COM *)); +static int execute_while_command __P((WHILE_COM *)); +static int execute_until_command __P((WHILE_COM *)); +static int execute_while_or_until __P((WHILE_COM *, int)); +static int execute_if_command __P((IF_COM *)); +static int execute_null_command __P((REDIRECT *, int, int, int)); +static void fix_assignment_words __P((WORD_LIST *)); +static int execute_simple_command __P((SIMPLE_COM *, int, int, int, struct fd_bitmap *)); +static int execute_builtin __P((sh_builtin_func_t *, WORD_LIST *, int, int)); +static int execute_function __P((SHELL_VAR *, WORD_LIST *, int, struct fd_bitmap *, int, int)); +static int execute_builtin_or_function __P((WORD_LIST *, sh_builtin_func_t *, + SHELL_VAR *, + REDIRECT *, struct fd_bitmap *, int)); +static void execute_subshell_builtin_or_function __P((WORD_LIST *, REDIRECT *, + sh_builtin_func_t *, + SHELL_VAR *, + int, int, int, + struct fd_bitmap *, + int)); +static int execute_disk_command __P((WORD_LIST *, REDIRECT *, char *, + int, int, int, struct fd_bitmap *, int)); + +static char *getinterp __P((char *, int, int *)); +static void initialize_subshell __P((void)); +static int execute_in_subshell __P((COMMAND *, int, int, int, struct fd_bitmap *)); +#if defined (COPROCESS_SUPPORT) +static int execute_coproc __P((COMMAND *, int, int, struct fd_bitmap *)); +#endif + +static int execute_pipeline __P((COMMAND *, int, int, int, struct fd_bitmap *)); + +static int execute_connection __P((COMMAND *, int, int, int, struct fd_bitmap *)); + +static int execute_intern_function __P((WORD_DESC *, FUNCTION_DEF *)); + +/* Set to 1 if fd 0 was the subject of redirection to a subshell. Global + so that reader_loop can set it to zero before executing a command. */ +int stdin_redir; + +/* The name of the command that is currently being executed. + `test' needs this, for example. */ +char *this_command_name; + +/* The printed representation of the currently-executing command (same as + the_printed_command), except when a trap is being executed. Useful for + a debugger to know where exactly the program is currently executing. */ +char *the_printed_command_except_trap; + +/* For catching RETURN in a function. */ +int return_catch_flag; +int return_catch_value; +procenv_t return_catch; + +/* The value returned by the last synchronous command. */ +int last_command_exit_value; + +/* Whether or not the last command (corresponding to last_command_exit_value) + was terminated by a signal, and, if so, which one. */ +int last_command_exit_signal; + +/* The list of redirections to perform which will undo the redirections + that I made in the shell. */ +REDIRECT *redirection_undo_list = (REDIRECT *)NULL; + +/* The list of redirections to perform which will undo the internal + redirections performed by the `exec' builtin. These are redirections + that must be undone even when exec discards redirection_undo_list. */ +REDIRECT *exec_redirection_undo_list = (REDIRECT *)NULL; + +/* When greater than zero, value is the `level' of builtins we are + currently executing (e.g. `eval echo a' would have it set to 2). */ +int executing_builtin = 0; + +/* Non-zero if we are executing a command list (a;b;c, etc.) */ +int executing_list = 0; + +/* Non-zero if failing commands in a command substitution should not exit the + shell even if -e is set. Used to pass the CMD_IGNORE_RETURN flag down to + commands run in command substitutions by parse_and_execute. */ +int comsub_ignore_return = 0; + +/* Non-zero if we have just forked and are currently running in a subshell + environment. */ +int subshell_environment; + +/* Count of nested subshells, like SHLVL. Available via $BASH_SUBSHELL */ +int subshell_level = 0; + +/* Currently-executing shell function. */ +SHELL_VAR *this_shell_function; + +/* If non-zero, matches in case and [[ ... ]] are case-insensitive */ +int match_ignore_case = 0; + +int executing_command_builtin = 0; + +struct stat SB; /* used for debugging */ + +static int special_builtin_failed; + +static COMMAND *currently_executing_command; + +/* The line number that the currently executing function starts on. */ +static int function_line_number; + +/* XXX - set to 1 if we're running the DEBUG trap and we want to show the line + number containing the function name. Used by executing_line_number to + report the correct line number. Kind of a hack. */ +static int showing_function_line; + +static int line_number_for_err_trap; + +/* A sort of function nesting level counter */ +int funcnest = 0; +int funcnest_max = 0; /* XXX - bash-4.2 */ + +int lastpipe_opt = 0; + +struct fd_bitmap *current_fds_to_close = (struct fd_bitmap *)NULL; + +#define FD_BITMAP_DEFAULT_SIZE 32 + +/* Functions to allocate and deallocate the structures used to pass + information from the shell to its children about file descriptors + to close. */ +struct fd_bitmap * +new_fd_bitmap (size) + int size; +{ + struct fd_bitmap *ret; + + ret = (struct fd_bitmap *)xmalloc (sizeof (struct fd_bitmap)); + + ret->size = size; + + if (size) + { + ret->bitmap = (char *)xmalloc (size); + memset (ret->bitmap, '\0', size); + } + else + ret->bitmap = (char *)NULL; + return (ret); +} + +void +dispose_fd_bitmap (fdbp) + struct fd_bitmap *fdbp; +{ + FREE (fdbp->bitmap); + free (fdbp); +} + +void +close_fd_bitmap (fdbp) + struct fd_bitmap *fdbp; +{ + register int i; + + if (fdbp) + { + for (i = 0; i < fdbp->size; i++) + if (fdbp->bitmap[i]) + { + close (i); + fdbp->bitmap[i] = 0; + } + } +} + +/* Return the line number of the currently executing command. */ +int +executing_line_number () +{ + if (executing && showing_function_line == 0 && + (variable_context == 0 || interactive_shell == 0) && + currently_executing_command) + { +#if defined (COND_COMMAND) + if (currently_executing_command->type == cm_cond) + return currently_executing_command->value.Cond->line; +#endif +#if defined (DPAREN_ARITHMETIC) + else if (currently_executing_command->type == cm_arith) + return currently_executing_command->value.Arith->line; +#endif +#if defined (ARITH_FOR_COMMAND) + else if (currently_executing_command->type == cm_arith_for) + return currently_executing_command->value.ArithFor->line; +#endif + + return line_number; + } + else + return line_number; +} + +/* Execute the command passed in COMMAND. COMMAND is exactly what + read_command () places into GLOBAL_COMMAND. See "command.h" for the + details of the command structure. + + EXECUTION_SUCCESS or EXECUTION_FAILURE are the only possible + return values. Executing a command with nothing in it returns + EXECUTION_SUCCESS. */ +int +execute_command (command) + COMMAND *command; +{ + struct fd_bitmap *bitmap; + int result; + + current_fds_to_close = (struct fd_bitmap *)NULL; + bitmap = new_fd_bitmap (FD_BITMAP_DEFAULT_SIZE); + begin_unwind_frame ("execute-command"); + add_unwind_protect (dispose_fd_bitmap, (char *)bitmap); + + /* Just do the command, but not asynchronously. */ + result = execute_command_internal (command, 0, NO_PIPE, NO_PIPE, bitmap); + + dispose_fd_bitmap (bitmap); + discard_unwind_frame ("execute-command"); + +#if defined (PROCESS_SUBSTITUTION) + /* don't unlink fifos if we're in a shell function; wait until the function + returns. */ + if (variable_context == 0) + unlink_fifo_list (); +#endif /* PROCESS_SUBSTITUTION */ + + QUIT; + return (result); +} + +/* Return 1 if TYPE is a shell control structure type. */ +static int +shell_control_structure (type) + enum command_type type; +{ + switch (type) + { +#if defined (ARITH_FOR_COMMAND) + case cm_arith_for: +#endif +#if defined (SELECT_COMMAND) + case cm_select: +#endif +#if defined (DPAREN_ARITHMETIC) + case cm_arith: +#endif +#if defined (COND_COMMAND) + case cm_cond: +#endif + case cm_case: + case cm_while: + case cm_until: + case cm_if: + case cm_for: + case cm_group: + case cm_function_def: + return (1); + + default: + return (0); + } +} + +/* A function to use to unwind_protect the redirection undo list + for loops. */ +static void +cleanup_redirects (list) + REDIRECT *list; +{ + do_redirections (list, RX_ACTIVE); + dispose_redirects (list); +} + +#if 0 +/* Function to unwind_protect the redirections for functions and builtins. */ +static void +cleanup_func_redirects (list) + REDIRECT *list; +{ + do_redirections (list, RX_ACTIVE); +} +#endif + +void +dispose_exec_redirects () +{ + if (exec_redirection_undo_list) + { + dispose_redirects (exec_redirection_undo_list); + exec_redirection_undo_list = (REDIRECT *)NULL; + } +} + +#if defined (JOB_CONTROL) +/* A function to restore the signal mask to its proper value when the shell + is interrupted or errors occur while creating a pipeline. */ +static int +restore_signal_mask (set) + sigset_t *set; +{ + return (sigprocmask (SIG_SETMASK, set, (sigset_t *)NULL)); +} +#endif /* JOB_CONTROL */ + +#ifdef DEBUG +/* A debugging function that can be called from gdb, for instance. */ +void +open_files () +{ + register int i; + int f, fd_table_size; + + fd_table_size = getdtablesize (); + + fprintf (stderr, "pid %ld open files:", (long)getpid ()); + for (i = 3; i < fd_table_size; i++) + { + if ((f = fcntl (i, F_GETFD, 0)) != -1) + fprintf (stderr, " %d (%s)", i, f ? "close" : "open"); + } + fprintf (stderr, "\n"); +} +#endif + +static void +async_redirect_stdin () +{ + int fd; + + fd = open ("/dev/null", O_RDONLY); + if (fd > 0) + { + dup2 (fd, 0); + close (fd); + } + else if (fd < 0) + internal_error (_("cannot redirect standard input from /dev/null: %s"), strerror (errno)); +} + +#define DESCRIBE_PID(pid) do { if (interactive) describe_pid (pid); } while (0) + +/* Execute the command passed in COMMAND, perhaps doing it asynchrounously. + COMMAND is exactly what read_command () places into GLOBAL_COMMAND. + ASYNCHROUNOUS, if non-zero, says to do this command in the background. + PIPE_IN and PIPE_OUT are file descriptors saying where input comes + from and where it goes. They can have the value of NO_PIPE, which means + I/O is stdin/stdout. + FDS_TO_CLOSE is a list of file descriptors to close once the child has + been forked. This list often contains the unusable sides of pipes, etc. + + EXECUTION_SUCCESS or EXECUTION_FAILURE are the only possible + return values. Executing a command with nothing in it returns + EXECUTION_SUCCESS. */ +int +execute_command_internal (command, asynchronous, pipe_in, pipe_out, + fds_to_close) + COMMAND *command; + int asynchronous; + int pipe_in, pipe_out; + struct fd_bitmap *fds_to_close; +{ + int exec_result, user_subshell, invert, ignore_return, was_error_trap; + REDIRECT *my_undo_list, *exec_undo_list; + volatile int last_pid; + volatile int save_line_number; +#if defined (PROCESS_SUBSTITUTION) + volatile int ofifo, nfifo, osize, saved_fifo; + volatile char *ofifo_list; +#endif + +#if 0 + if (command == 0 || breaking || continuing || read_but_dont_execute) + return (EXECUTION_SUCCESS); +#else + if (breaking || continuing) + return (last_command_exit_value); + if (command == 0 || read_but_dont_execute) + return (EXECUTION_SUCCESS); +#endif + + QUIT; + run_pending_traps (); + +#if 0 + if (running_trap == 0) +#endif + currently_executing_command = command; + + invert = (command->flags & CMD_INVERT_RETURN) != 0; + + /* If we're inverting the return value and `set -e' has been executed, + we don't want a failing command to inadvertently cause the shell + to exit. */ + if (exit_immediately_on_error && invert) /* XXX */ + command->flags |= CMD_IGNORE_RETURN; /* XXX */ + + exec_result = EXECUTION_SUCCESS; + + /* If a command was being explicitly run in a subshell, or if it is + a shell control-structure, and it has a pipe, then we do the command + in a subshell. */ + if (command->type == cm_subshell && (command->flags & CMD_NO_FORK)) + return (execute_in_subshell (command, asynchronous, pipe_in, pipe_out, fds_to_close)); + +#if defined (COPROCESS_SUPPORT) + if (command->type == cm_coproc) + return (execute_coproc (command, pipe_in, pipe_out, fds_to_close)); +#endif + + user_subshell = command->type == cm_subshell || ((command->flags & CMD_WANT_SUBSHELL) != 0); + + if (command->type == cm_subshell || + (command->flags & (CMD_WANT_SUBSHELL|CMD_FORCE_SUBSHELL)) || + (shell_control_structure (command->type) && + (pipe_out != NO_PIPE || pipe_in != NO_PIPE || asynchronous))) + { + pid_t paren_pid; + + /* Fork a subshell, turn off the subshell bit, turn off job + control and call execute_command () on the command again. */ + line_number_for_err_trap = line_number; + paren_pid = make_child (savestring (make_command_string (command)), + asynchronous); + if (paren_pid == 0) + exit (execute_in_subshell (command, asynchronous, pipe_in, pipe_out, fds_to_close)); + /* NOTREACHED */ + else + { + close_pipes (pipe_in, pipe_out); + +#if defined (PROCESS_SUBSTITUTION) && defined (HAVE_DEV_FD) + if (variable_context == 0) /* wait until shell function completes */ + unlink_fifo_list (); +#endif + /* If we are part of a pipeline, and not the end of the pipeline, + then we should simply return and let the last command in the + pipe be waited for. If we are not in a pipeline, or are the + last command in the pipeline, then we wait for the subshell + and return its exit status as usual. */ + if (pipe_out != NO_PIPE) + return (EXECUTION_SUCCESS); + + stop_pipeline (asynchronous, (COMMAND *)NULL); + + if (asynchronous == 0) + { + was_error_trap = signal_is_trapped (ERROR_TRAP) && signal_is_ignored (ERROR_TRAP) == 0; + invert = (command->flags & CMD_INVERT_RETURN) != 0; + ignore_return = (command->flags & CMD_IGNORE_RETURN) != 0; + + exec_result = wait_for (paren_pid); + + /* If we have to, invert the return value. */ + if (invert) + exec_result = ((exec_result == EXECUTION_SUCCESS) + ? EXECUTION_FAILURE + : EXECUTION_SUCCESS); + + last_command_exit_value = exec_result; + if (user_subshell && was_error_trap && ignore_return == 0 && invert == 0 && exec_result != EXECUTION_SUCCESS) + { + save_line_number = line_number; + line_number = line_number_for_err_trap; + run_error_trap (); + line_number = save_line_number; + } + + if (user_subshell && ignore_return == 0 && invert == 0 && exit_immediately_on_error && exec_result != EXECUTION_SUCCESS) + { + run_pending_traps (); + jump_to_top_level (ERREXIT); + } + + return (last_command_exit_value); + } + else + { + DESCRIBE_PID (paren_pid); + + run_pending_traps (); + + return (EXECUTION_SUCCESS); + } + } + } + +#if defined (COMMAND_TIMING) + if (command->flags & CMD_TIME_PIPELINE) + { + if (asynchronous) + { + command->flags |= CMD_FORCE_SUBSHELL; + exec_result = execute_command_internal (command, 1, pipe_in, pipe_out, fds_to_close); + } + else + { + exec_result = time_command (command, asynchronous, pipe_in, pipe_out, fds_to_close); +#if 0 + if (running_trap == 0) +#endif + currently_executing_command = (COMMAND *)NULL; + } + return (exec_result); + } +#endif /* COMMAND_TIMING */ + + if (shell_control_structure (command->type) && command->redirects) + stdin_redir = stdin_redirects (command->redirects); + +#if defined (PROCESS_SUBSTITUTION) + if (variable_context != 0) + { + ofifo = num_fifos (); + ofifo_list = copy_fifo_list (&osize); + saved_fifo = 1; + } + else + saved_fifo = 0; +#endif + + /* Handle WHILE FOR CASE etc. with redirections. (Also '&' input + redirection.) */ + if (do_redirections (command->redirects, RX_ACTIVE|RX_UNDOABLE) != 0) + { + cleanup_redirects (redirection_undo_list); + redirection_undo_list = (REDIRECT *)NULL; + dispose_exec_redirects (); +#if defined (PROCESS_SUBSTITUTION) + if (saved_fifo) + free (ofifo_list); +#endif + return (last_command_exit_value = EXECUTION_FAILURE); + } + + if (redirection_undo_list) + { + /* XXX - why copy here? */ + my_undo_list = (REDIRECT *)copy_redirects (redirection_undo_list); + dispose_redirects (redirection_undo_list); + redirection_undo_list = (REDIRECT *)NULL; + } + else + my_undo_list = (REDIRECT *)NULL; + + if (exec_redirection_undo_list) + { + /* XXX - why copy here? */ + exec_undo_list = (REDIRECT *)copy_redirects (exec_redirection_undo_list); + dispose_redirects (exec_redirection_undo_list); + exec_redirection_undo_list = (REDIRECT *)NULL; + } + else + exec_undo_list = (REDIRECT *)NULL; + + if (my_undo_list || exec_undo_list) + begin_unwind_frame ("loop_redirections"); + + if (my_undo_list) + add_unwind_protect ((Function *)cleanup_redirects, my_undo_list); + + if (exec_undo_list) + add_unwind_protect ((Function *)dispose_redirects, exec_undo_list); + + ignore_return = (command->flags & CMD_IGNORE_RETURN) != 0; + + QUIT; + + switch (command->type) + { + case cm_simple: + { + save_line_number = line_number; + /* We can't rely on variables retaining their values across a + call to execute_simple_command if a longjmp occurs as the + result of a `return' builtin. This is true for sure with gcc. */ +#if defined (RECYCLES_PIDS) + last_made_pid = NO_PID; +#endif + last_pid = last_made_pid; + was_error_trap = signal_is_trapped (ERROR_TRAP) && signal_is_ignored (ERROR_TRAP) == 0; + + if (ignore_return && command->value.Simple) + command->value.Simple->flags |= CMD_IGNORE_RETURN; + if (command->flags & CMD_STDIN_REDIR) + command->value.Simple->flags |= CMD_STDIN_REDIR; + + line_number_for_err_trap = line_number = command->value.Simple->line; + exec_result = + execute_simple_command (command->value.Simple, pipe_in, pipe_out, + asynchronous, fds_to_close); + line_number = save_line_number; + + /* The temporary environment should be used for only the simple + command immediately following its definition. */ + dispose_used_env_vars (); + +#if (defined (ultrix) && defined (mips)) || defined (C_ALLOCA) + /* Reclaim memory allocated with alloca () on machines which + may be using the alloca emulation code. */ + (void) alloca (0); +#endif /* (ultrix && mips) || C_ALLOCA */ + + /* If we forked to do the command, then we must wait_for () + the child. */ + + /* XXX - this is something to watch out for if there are problems + when the shell is compiled without job control. */ + if (already_making_children && pipe_out == NO_PIPE && + last_made_pid != last_pid) + { + stop_pipeline (asynchronous, (COMMAND *)NULL); + + if (asynchronous) + { + DESCRIBE_PID (last_made_pid); + } + else +#if !defined (JOB_CONTROL) + /* Do not wait for asynchronous processes started from + startup files. */ + if (last_made_pid != last_asynchronous_pid) +#endif + /* When executing a shell function that executes other + commands, this causes the last simple command in + the function to be waited for twice. This also causes + subshells forked to execute builtin commands (e.g., in + pipelines) to be waited for twice. */ + exec_result = wait_for (last_made_pid); + } + } + + /* 2009/02/13 -- pipeline failure is processed elsewhere. This handles + only the failure of a simple command. */ + if (was_error_trap && ignore_return == 0 && invert == 0 && pipe_in == NO_PIPE && pipe_out == NO_PIPE && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + line_number = line_number_for_err_trap; + run_error_trap (); + line_number = save_line_number; + } + + if (ignore_return == 0 && invert == 0 && + ((posixly_correct && interactive == 0 && special_builtin_failed) || + (exit_immediately_on_error && pipe_in == NO_PIPE && pipe_out == NO_PIPE && exec_result != EXECUTION_SUCCESS))) + { + last_command_exit_value = exec_result; + run_pending_traps (); + jump_to_top_level (ERREXIT); + } + + break; + + case cm_for: + if (ignore_return) + command->value.For->flags |= CMD_IGNORE_RETURN; + exec_result = execute_for_command (command->value.For); + break; + +#if defined (ARITH_FOR_COMMAND) + case cm_arith_for: + if (ignore_return) + command->value.ArithFor->flags |= CMD_IGNORE_RETURN; + exec_result = execute_arith_for_command (command->value.ArithFor); + break; +#endif + +#if defined (SELECT_COMMAND) + case cm_select: + if (ignore_return) + command->value.Select->flags |= CMD_IGNORE_RETURN; + exec_result = execute_select_command (command->value.Select); + break; +#endif + + case cm_case: + if (ignore_return) + command->value.Case->flags |= CMD_IGNORE_RETURN; + exec_result = execute_case_command (command->value.Case); + break; + + case cm_while: + if (ignore_return) + command->value.While->flags |= CMD_IGNORE_RETURN; + exec_result = execute_while_command (command->value.While); + break; + + case cm_until: + if (ignore_return) + command->value.While->flags |= CMD_IGNORE_RETURN; + exec_result = execute_until_command (command->value.While); + break; + + case cm_if: + if (ignore_return) + command->value.If->flags |= CMD_IGNORE_RETURN; + exec_result = execute_if_command (command->value.If); + break; + + case cm_group: + + /* This code can be executed from either of two paths: an explicit + '{}' command, or via a function call. If we are executed via a + function call, we have already taken care of the function being + executed in the background (down there in execute_simple_command ()), + and this command should *not* be marked as asynchronous. If we + are executing a regular '{}' group command, and asynchronous == 1, + we must want to execute the whole command in the background, so we + need a subshell, and we want the stuff executed in that subshell + (this group command) to be executed in the foreground of that + subshell (i.e. there will not be *another* subshell forked). + + What we do is to force a subshell if asynchronous, and then call + execute_command_internal again with asynchronous still set to 1, + but with the original group command, so the printed command will + look right. + + The code above that handles forking off subshells will note that + both subshell and async are on, and turn off async in the child + after forking the subshell (but leave async set in the parent, so + the normal call to describe_pid is made). This turning off + async is *crucial*; if it is not done, this will fall into an + infinite loop of executions through this spot in subshell after + subshell until the process limit is exhausted. */ + + if (asynchronous) + { + command->flags |= CMD_FORCE_SUBSHELL; + exec_result = + execute_command_internal (command, 1, pipe_in, pipe_out, + fds_to_close); + } + else + { + if (ignore_return && command->value.Group->command) + command->value.Group->command->flags |= CMD_IGNORE_RETURN; + exec_result = + execute_command_internal (command->value.Group->command, + asynchronous, pipe_in, pipe_out, + fds_to_close); + } + break; + + case cm_connection: + exec_result = execute_connection (command, asynchronous, + pipe_in, pipe_out, fds_to_close); + break; + +#if defined (DPAREN_ARITHMETIC) + case cm_arith: + was_error_trap = signal_is_trapped (ERROR_TRAP) && signal_is_ignored (ERROR_TRAP) == 0; + if (ignore_return) + command->value.Arith->flags |= CMD_IGNORE_RETURN; + line_number_for_err_trap = save_line_number = line_number; + exec_result = execute_arith_command (command->value.Arith); + line_number = save_line_number; + + if (was_error_trap && ignore_return == 0 && invert == 0 && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + save_line_number = line_number; + line_number = line_number_for_err_trap; + run_error_trap (); + line_number = save_line_number; + } + + if (ignore_return == 0 && invert == 0 && exit_immediately_on_error && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + run_pending_traps (); + jump_to_top_level (ERREXIT); + } + + break; +#endif + +#if defined (COND_COMMAND) + case cm_cond: + was_error_trap = signal_is_trapped (ERROR_TRAP) && signal_is_ignored (ERROR_TRAP) == 0; + if (ignore_return) + command->value.Cond->flags |= CMD_IGNORE_RETURN; + + line_number_for_err_trap = save_line_number = line_number; + exec_result = execute_cond_command (command->value.Cond); + line_number = save_line_number; + + if (was_error_trap && ignore_return == 0 && invert == 0 && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + save_line_number = line_number; + line_number = line_number_for_err_trap; + run_error_trap (); + line_number = save_line_number; + } + + if (ignore_return == 0 && invert == 0 && exit_immediately_on_error && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + run_pending_traps (); + jump_to_top_level (ERREXIT); + } + + break; +#endif + + case cm_function_def: + exec_result = execute_intern_function (command->value.Function_def->name, + command->value.Function_def); + break; + + default: + command_error ("execute_command", CMDERR_BADTYPE, command->type, 0); + } + + if (my_undo_list) + { + do_redirections (my_undo_list, RX_ACTIVE); + dispose_redirects (my_undo_list); + } + + if (exec_undo_list) + dispose_redirects (exec_undo_list); + + if (my_undo_list || exec_undo_list) + discard_unwind_frame ("loop_redirections"); + +#if defined (PROCESS_SUBSTITUTION) + if (saved_fifo) + { + nfifo = num_fifos (); + if (nfifo > ofifo) + close_new_fifos (ofifo_list, osize); + free (ofifo_list); + } +#endif + + /* Invert the return value if we have to */ + if (invert) + exec_result = (exec_result == EXECUTION_SUCCESS) + ? EXECUTION_FAILURE + : EXECUTION_SUCCESS; + +#if defined (DPAREN_ARITHMETIC) || defined (COND_COMMAND) + /* This is where we set PIPESTATUS from the exit status of the appropriate + compound commands (the ones that look enough like simple commands to + cause confusion). We might be able to optimize by not doing this if + subshell_environment != 0. */ + switch (command->type) + { +# if defined (DPAREN_ARITHMETIC) + case cm_arith: +# endif +# if defined (COND_COMMAND) + case cm_cond: +# endif + set_pipestatus_from_exit (exec_result); + break; + } +#endif + + last_command_exit_value = exec_result; + run_pending_traps (); +#if 0 + if (running_trap == 0) +#endif + currently_executing_command = (COMMAND *)NULL; + + return (last_command_exit_value); +} + +#if defined (COMMAND_TIMING) + +#if defined (HAVE_GETRUSAGE) && defined (HAVE_GETTIMEOFDAY) +extern struct timeval *difftimeval __P((struct timeval *, struct timeval *, struct timeval *)); +extern struct timeval *addtimeval __P((struct timeval *, struct timeval *, struct timeval *)); +extern int timeval_to_cpu __P((struct timeval *, struct timeval *, struct timeval *)); +#endif + +#define POSIX_TIMEFORMAT "real %2R\nuser %2U\nsys %2S" +#define BASH_TIMEFORMAT "\nreal\t%3lR\nuser\t%3lU\nsys\t%3lS" + +static const int precs[] = { 0, 100, 10, 1 }; + +/* Expand one `%'-prefixed escape sequence from a time format string. */ +static int +mkfmt (buf, prec, lng, sec, sec_fraction) + char *buf; + int prec, lng; + time_t sec; + int sec_fraction; +{ + time_t min; + char abuf[INT_STRLEN_BOUND(time_t) + 1]; + int ind, aind; + + ind = 0; + abuf[sizeof(abuf) - 1] = '\0'; + + /* If LNG is non-zero, we want to decompose SEC into minutes and seconds. */ + if (lng) + { + min = sec / 60; + sec %= 60; + aind = sizeof(abuf) - 2; + do + abuf[aind--] = (min % 10) + '0'; + while (min /= 10); + aind++; + while (abuf[aind]) + buf[ind++] = abuf[aind++]; + buf[ind++] = 'm'; + } + + /* Now add the seconds. */ + aind = sizeof (abuf) - 2; + do + abuf[aind--] = (sec % 10) + '0'; + while (sec /= 10); + aind++; + while (abuf[aind]) + buf[ind++] = abuf[aind++]; + + /* We want to add a decimal point and PREC places after it if PREC is + nonzero. PREC is not greater than 3. SEC_FRACTION is between 0 + and 999. */ + if (prec != 0) + { + buf[ind++] = '.'; + for (aind = 1; aind <= prec; aind++) + { + buf[ind++] = (sec_fraction / precs[aind]) + '0'; + sec_fraction %= precs[aind]; + } + } + + if (lng) + buf[ind++] = 's'; + buf[ind] = '\0'; + + return (ind); +} + +/* Interpret the format string FORMAT, interpolating the following escape + sequences: + %[prec][l][RUS] + + where the optional `prec' is a precision, meaning the number of + characters after the decimal point, the optional `l' means to format + using minutes and seconds (MMmNN[.FF]s), like the `times' builtin', + and the last character is one of + + R number of seconds of `real' time + U number of seconds of `user' time + S number of seconds of `system' time + + An occurrence of `%%' in the format string is translated to a `%'. The + result is printed to FP, a pointer to a FILE. The other variables are + the seconds and thousandths of a second of real, user, and system time, + resectively. */ +static void +print_formatted_time (fp, format, rs, rsf, us, usf, ss, ssf, cpu) + FILE *fp; + char *format; + time_t rs; + int rsf; + time_t us; + int usf; + time_t ss; + int ssf, cpu; +{ + int prec, lng, len; + char *str, *s, ts[INT_STRLEN_BOUND (time_t) + sizeof ("mSS.FFFF")]; + time_t sum; + int sum_frac; + int sindex, ssize; + + len = strlen (format); + ssize = (len + 64) - (len % 64); + str = (char *)xmalloc (ssize); + sindex = 0; + + for (s = format; *s; s++) + { + if (*s != '%' || s[1] == '\0') + { + RESIZE_MALLOCED_BUFFER (str, sindex, 1, ssize, 64); + str[sindex++] = *s; + } + else if (s[1] == '%') + { + s++; + RESIZE_MALLOCED_BUFFER (str, sindex, 1, ssize, 64); + str[sindex++] = *s; + } + else if (s[1] == 'P') + { + s++; +#if 0 + /* clamp CPU usage at 100% */ + if (cpu > 10000) + cpu = 10000; +#endif + sum = cpu / 100; + sum_frac = (cpu % 100) * 10; + len = mkfmt (ts, 2, 0, sum, sum_frac); + RESIZE_MALLOCED_BUFFER (str, sindex, len, ssize, 64); + strcpy (str + sindex, ts); + sindex += len; + } + else + { + prec = 3; /* default is three places past the decimal point. */ + lng = 0; /* default is to not use minutes or append `s' */ + s++; + if (DIGIT (*s)) /* `precision' */ + { + prec = *s++ - '0'; + if (prec > 3) prec = 3; + } + if (*s == 'l') /* `length extender' */ + { + lng = 1; + s++; + } + if (*s == 'R' || *s == 'E') + len = mkfmt (ts, prec, lng, rs, rsf); + else if (*s == 'U') + len = mkfmt (ts, prec, lng, us, usf); + else if (*s == 'S') + len = mkfmt (ts, prec, lng, ss, ssf); + else + { + internal_error (_("TIMEFORMAT: `%c': invalid format character"), *s); + free (str); + return; + } + RESIZE_MALLOCED_BUFFER (str, sindex, len, ssize, 64); + strcpy (str + sindex, ts); + sindex += len; + } + } + + str[sindex] = '\0'; + fprintf (fp, "%s\n", str); + fflush (fp); + + free (str); +} + +static int +time_command (command, asynchronous, pipe_in, pipe_out, fds_to_close) + COMMAND *command; + int asynchronous, pipe_in, pipe_out; + struct fd_bitmap *fds_to_close; +{ + int rv, posix_time, old_flags, nullcmd; + time_t rs, us, ss; + int rsf, usf, ssf; + int cpu; + char *time_format; + +#if defined (HAVE_GETRUSAGE) && defined (HAVE_GETTIMEOFDAY) + struct timeval real, user, sys; + struct timeval before, after; +# if defined (HAVE_STRUCT_TIMEZONE) + struct timezone dtz; /* posix doesn't define this */ +# endif + struct rusage selfb, selfa, kidsb, kidsa; /* a = after, b = before */ +#else +# if defined (HAVE_TIMES) + clock_t tbefore, tafter, real, user, sys; + struct tms before, after; +# endif +#endif + +#if defined (HAVE_GETRUSAGE) && defined (HAVE_GETTIMEOFDAY) +# if defined (HAVE_STRUCT_TIMEZONE) + gettimeofday (&before, &dtz); +# else + gettimeofday (&before, (void *)NULL); +# endif /* !HAVE_STRUCT_TIMEZONE */ + getrusage (RUSAGE_SELF, &selfb); + getrusage (RUSAGE_CHILDREN, &kidsb); +#else +# if defined (HAVE_TIMES) + tbefore = times (&before); +# endif +#endif + + posix_time = command && (command->flags & CMD_TIME_POSIX); + + nullcmd = (command == 0) || (command->type == cm_simple && command->value.Simple->words == 0 && command->value.Simple->redirects == 0); + if (posixly_correct && nullcmd) + { +#if defined (HAVE_GETRUSAGE) + selfb.ru_utime.tv_sec = kidsb.ru_utime.tv_sec = selfb.ru_stime.tv_sec = kidsb.ru_stime.tv_sec = 0; + selfb.ru_utime.tv_usec = kidsb.ru_utime.tv_usec = selfb.ru_stime.tv_usec = kidsb.ru_stime.tv_usec = 0; + before.tv_sec = shell_start_time; + before.tv_usec = 0; +#else + before.tms_utime = before.tms_stime = before.tms_cutime = before.tms_cstime = 0; + tbefore = shell_start_time; +#endif + } + + old_flags = command->flags; + command->flags &= ~(CMD_TIME_PIPELINE|CMD_TIME_POSIX); + rv = execute_command_internal (command, asynchronous, pipe_in, pipe_out, fds_to_close); + command->flags = old_flags; + + rs = us = ss = 0; + rsf = usf = ssf = cpu = 0; + +#if defined (HAVE_GETRUSAGE) && defined (HAVE_GETTIMEOFDAY) +# if defined (HAVE_STRUCT_TIMEZONE) + gettimeofday (&after, &dtz); +# else + gettimeofday (&after, (void *)NULL); +# endif /* !HAVE_STRUCT_TIMEZONE */ + getrusage (RUSAGE_SELF, &selfa); + getrusage (RUSAGE_CHILDREN, &kidsa); + + difftimeval (&real, &before, &after); + timeval_to_secs (&real, &rs, &rsf); + + addtimeval (&user, difftimeval(&after, &selfb.ru_utime, &selfa.ru_utime), + difftimeval(&before, &kidsb.ru_utime, &kidsa.ru_utime)); + timeval_to_secs (&user, &us, &usf); + + addtimeval (&sys, difftimeval(&after, &selfb.ru_stime, &selfa.ru_stime), + difftimeval(&before, &kidsb.ru_stime, &kidsa.ru_stime)); + timeval_to_secs (&sys, &ss, &ssf); + + cpu = timeval_to_cpu (&real, &user, &sys); +#else +# if defined (HAVE_TIMES) + tafter = times (&after); + + real = tafter - tbefore; + clock_t_to_secs (real, &rs, &rsf); + + user = (after.tms_utime - before.tms_utime) + (after.tms_cutime - before.tms_cutime); + clock_t_to_secs (user, &us, &usf); + + sys = (after.tms_stime - before.tms_stime) + (after.tms_cstime - before.tms_cstime); + clock_t_to_secs (sys, &ss, &ssf); + + cpu = (real == 0) ? 0 : ((user + sys) * 10000) / real; + +# else + rs = us = ss = 0; + rsf = usf = ssf = cpu = 0; +# endif +#endif + + if (posix_time) + time_format = POSIX_TIMEFORMAT; + else if ((time_format = get_string_value ("TIMEFORMAT")) == 0) + { + if (posixly_correct && nullcmd) + time_format = "user\t%2lU\nsys\t%2lS"; + else + time_format = BASH_TIMEFORMAT; + } + if (time_format && *time_format) + print_formatted_time (stderr, time_format, rs, rsf, us, usf, ss, ssf, cpu); + + return rv; +} +#endif /* COMMAND_TIMING */ + +/* Execute a command that's supposed to be in a subshell. This must be + called after make_child and we must be running in the child process. + The caller will return or exit() immediately with the value this returns. */ +static int +execute_in_subshell (command, asynchronous, pipe_in, pipe_out, fds_to_close) + COMMAND *command; + int asynchronous; + int pipe_in, pipe_out; + struct fd_bitmap *fds_to_close; +{ + int user_subshell, return_code, function_value, should_redir_stdin, invert; + int ois, user_coproc; + int result; + volatile COMMAND *tcom; + + USE_VAR(user_subshell); + USE_VAR(user_coproc); + USE_VAR(invert); + USE_VAR(tcom); + USE_VAR(asynchronous); + + subshell_level++; + should_redir_stdin = (asynchronous && (command->flags & CMD_STDIN_REDIR) && + pipe_in == NO_PIPE && + stdin_redirects (command->redirects) == 0); + + invert = (command->flags & CMD_INVERT_RETURN) != 0; + user_subshell = command->type == cm_subshell || ((command->flags & CMD_WANT_SUBSHELL) != 0); + user_coproc = command->type == cm_coproc; + + command->flags &= ~(CMD_FORCE_SUBSHELL | CMD_WANT_SUBSHELL | CMD_INVERT_RETURN); + + /* If a command is asynchronous in a subshell (like ( foo ) & or + the special case of an asynchronous GROUP command where the + the subshell bit is turned on down in case cm_group: below), + turn off `asynchronous', so that two subshells aren't spawned. + XXX - asynchronous used to be set to 0 in this block, but that + means that setup_async_signals was never run. Now it's set to + 0 after subshell_environment is set appropriately and setup_async_signals + is run. + + This seems semantically correct to me. For example, + ( foo ) & seems to say ``do the command `foo' in a subshell + environment, but don't wait for that subshell to finish'', + and "{ foo ; bar ; } &" seems to me to be like functions or + builtins in the background, which executed in a subshell + environment. I just don't see the need to fork two subshells. */ + + /* Don't fork again, we are already in a subshell. A `doubly + async' shell is not interactive, however. */ + if (asynchronous) + { +#if defined (JOB_CONTROL) + /* If a construct like ( exec xxx yyy ) & is given while job + control is active, we want to prevent exec from putting the + subshell back into the original process group, carefully + undoing all the work we just did in make_child. */ + original_pgrp = -1; +#endif /* JOB_CONTROL */ + ois = interactive_shell; + interactive_shell = 0; + /* This test is to prevent alias expansion by interactive shells that + run `(command) &' but to allow scripts that have enabled alias + expansion with `shopt -s expand_alias' to continue to expand + aliases. */ + if (ois != interactive_shell) + expand_aliases = 0; + } + + /* Subshells are neither login nor interactive. */ + login_shell = interactive = 0; + + if (user_subshell) + subshell_environment = SUBSHELL_PAREN; + else + { + subshell_environment = 0; /* XXX */ + if (asynchronous) + subshell_environment |= SUBSHELL_ASYNC; + if (pipe_in != NO_PIPE || pipe_out != NO_PIPE) + subshell_environment |= SUBSHELL_PIPE; + if (user_coproc) + subshell_environment |= SUBSHELL_COPROC; + } + + reset_terminating_signals (); /* in sig.c */ + /* Cancel traps, in trap.c. */ + /* Reset the signal handlers in the child, but don't free the + trap strings. Set a flag noting that we have to free the + trap strings if we run trap to change a signal disposition. */ + reset_signal_handlers (); + subshell_environment |= SUBSHELL_RESETTRAP; + + /* Make sure restore_original_signals doesn't undo the work done by + make_child to ensure that asynchronous children are immune to SIGINT + and SIGQUIT. Turn off asynchronous to make sure more subshells are + not spawned. */ + if (asynchronous) + { + setup_async_signals (); + asynchronous = 0; + } + +#if defined (JOB_CONTROL) + set_sigchld_handler (); +#endif /* JOB_CONTROL */ + + set_sigint_handler (); + +#if defined (JOB_CONTROL) + /* Delete all traces that there were any jobs running. This is + only for subshells. */ + without_job_control (); +#endif /* JOB_CONTROL */ + + if (fds_to_close) + close_fd_bitmap (fds_to_close); + + do_piping (pipe_in, pipe_out); + +#if defined (COPROCESS_SUPPORT) + coproc_closeall (); +#endif + + /* If this is a user subshell, set a flag if stdin was redirected. + This is used later to decide whether to redirect fd 0 to + /dev/null for async commands in the subshell. This adds more + sh compatibility, but I'm not sure it's the right thing to do. */ + if (user_subshell) + { + stdin_redir = stdin_redirects (command->redirects); + restore_default_signal (0); + } + + /* If this is an asynchronous command (command &), we want to + redirect the standard input from /dev/null in the absence of + any specific redirection involving stdin. */ + if (should_redir_stdin && stdin_redir == 0) + async_redirect_stdin (); + + /* Do redirections, then dispose of them before recursive call. */ + if (command->redirects) + { + if (do_redirections (command->redirects, RX_ACTIVE) != 0) + exit (invert ? EXECUTION_SUCCESS : EXECUTION_FAILURE); + + dispose_redirects (command->redirects); + command->redirects = (REDIRECT *)NULL; + } + + if (command->type == cm_subshell) + tcom = command->value.Subshell->command; + else if (user_coproc) + tcom = command->value.Coproc->command; + else + tcom = command; + + if (command->flags & CMD_TIME_PIPELINE) + tcom->flags |= CMD_TIME_PIPELINE; + if (command->flags & CMD_TIME_POSIX) + tcom->flags |= CMD_TIME_POSIX; + + /* Make sure the subshell inherits any CMD_IGNORE_RETURN flag. */ + if ((command->flags & CMD_IGNORE_RETURN) && tcom != command) + tcom->flags |= CMD_IGNORE_RETURN; + + /* If this is a simple command, tell execute_disk_command that it + might be able to get away without forking and simply exec. + This means things like ( sleep 10 ) will only cause one fork. + If we're timing the command or inverting its return value, however, + we cannot do this optimization. */ + if ((user_subshell || user_coproc) && (tcom->type == cm_simple || tcom->type == cm_subshell) && + ((tcom->flags & CMD_TIME_PIPELINE) == 0) && + ((tcom->flags & CMD_INVERT_RETURN) == 0)) + { + tcom->flags |= CMD_NO_FORK; + if (tcom->type == cm_simple) + tcom->value.Simple->flags |= CMD_NO_FORK; + } + + invert = (tcom->flags & CMD_INVERT_RETURN) != 0; + tcom->flags &= ~CMD_INVERT_RETURN; + + result = setjmp (top_level); + + /* If we're inside a function while executing this subshell, we + need to handle a possible `return'. */ + function_value = 0; + if (return_catch_flag) + function_value = setjmp (return_catch); + + /* If we're going to exit the shell, we don't want to invert the return + status. */ + if (result == EXITPROG) + invert = 0, return_code = last_command_exit_value; + else if (result) + return_code = EXECUTION_FAILURE; + else if (function_value) + return_code = return_catch_value; + else + return_code = execute_command_internal ((COMMAND *)tcom, asynchronous, NO_PIPE, NO_PIPE, fds_to_close); + + /* If we are asked to, invert the return value. */ + if (invert) + return_code = (return_code == EXECUTION_SUCCESS) ? EXECUTION_FAILURE + : EXECUTION_SUCCESS; + + /* If we were explicitly placed in a subshell with (), we need + to do the `shell cleanup' things, such as running traps[0]. */ + if (user_subshell && signal_is_trapped (0)) + { + last_command_exit_value = return_code; + return_code = run_exit_trap (); + } + + subshell_level--; + return (return_code); + /* NOTREACHED */ +} + +#if defined (COPROCESS_SUPPORT) +#define COPROC_MAX 16 + +typedef struct cpelement + { + struct cpelement *next; + struct coproc *coproc; + } +cpelement_t; + +typedef struct cplist + { + struct cpelement *head; + struct cpelement *tail; + int ncoproc; + } +cplist_t; + +static struct cpelement *cpe_alloc __P((struct coproc *)); +static void cpe_dispose __P((struct cpelement *)); +static struct cpelement *cpl_add __P((struct coproc *)); +static struct cpelement *cpl_delete __P((pid_t)); +static void cpl_reap __P((void)); +static void cpl_flush __P((void)); +static void cpl_closeall __P((void)); +static struct cpelement *cpl_search __P((pid_t)); +static struct cpelement *cpl_searchbyname __P((const char *)); +static void cpl_prune __P((void)); + +static void coproc_free __P((struct coproc *)); + +/* Will go away when there is fully-implemented support for multiple coprocs. */ +Coproc sh_coproc = { 0, NO_PID, -1, -1, 0, 0, 0, 0 }; + +cplist_t coproc_list = {0, 0, 0}; + +/* Functions to manage the list of coprocs */ + +static struct cpelement * +cpe_alloc (cp) + Coproc *cp; +{ + struct cpelement *cpe; + + cpe = (struct cpelement *)xmalloc (sizeof (struct cpelement)); + cpe->coproc = cp; + cpe->next = (struct cpelement *)0; + return cpe; +} + +static void +cpe_dispose (cpe) + struct cpelement *cpe; +{ + free (cpe); +} + +static struct cpelement * +cpl_add (cp) + Coproc *cp; +{ + struct cpelement *cpe; + + cpe = cpe_alloc (cp); + + if (coproc_list.head == 0) + { + coproc_list.head = coproc_list.tail = cpe; + coproc_list.ncoproc = 0; /* just to make sure */ + } + else + { + coproc_list.tail->next = cpe; + coproc_list.tail = cpe; + } + coproc_list.ncoproc++; + + return cpe; +} + +static struct cpelement * +cpl_delete (pid) + pid_t pid; +{ + struct cpelement *prev, *p; + + for (prev = p = coproc_list.head; p; prev = p, p = p->next) + if (p->coproc->c_pid == pid) + { + prev->next = p->next; /* remove from list */ + break; + } + + if (p == 0) + return 0; /* not found */ + +#if defined (DEBUG) + itrace("cpl_delete: deleting %d", pid); +#endif + + /* Housekeeping in the border cases. */ + if (p == coproc_list.head) + coproc_list.head = coproc_list.head->next; + else if (p == coproc_list.tail) + coproc_list.tail = prev; + + coproc_list.ncoproc--; + if (coproc_list.ncoproc == 0) + coproc_list.head = coproc_list.tail = 0; + else if (coproc_list.ncoproc == 1) + coproc_list.tail = coproc_list.head; /* just to make sure */ + + return (p); +} + +static void +cpl_reap () +{ + struct cpelement *p, *next, *nh, *nt; + + /* Build a new list by removing dead coprocs and fix up the coproc_list + pointers when done. */ + nh = nt = next = (struct cpelement *)0; + for (p = coproc_list.head; p; p = next) + { + next = p->next; + if (p->coproc->c_flags & COPROC_DEAD) + { + coproc_list.ncoproc--; /* keep running count, fix up pointers later */ + +#if defined (DEBUG) + itrace("cpl_reap: deleting %d", p->coproc->c_pid); +#endif + + coproc_dispose (p->coproc); + cpe_dispose (p); + } + else if (nh == 0) + nh = nt = p; + else + { + nt->next = p; + nt = nt->next; + } + } + + if (coproc_list.ncoproc == 0) + coproc_list.head = coproc_list.tail = 0; + else + { + if (nt) + nt->next = 0; + coproc_list.head = nh; + coproc_list.tail = nt; + if (coproc_list.ncoproc == 1) + coproc_list.tail = coproc_list.head; /* just to make sure */ + } +} + +/* Clear out the list of saved statuses */ +static void +cpl_flush () +{ + struct cpelement *cpe, *p; + + for (cpe = coproc_list.head; cpe; ) + { + p = cpe; + cpe = cpe->next; + + coproc_dispose (p->coproc); + cpe_dispose (p); + } + + coproc_list.head = coproc_list.tail = 0; + coproc_list.ncoproc = 0; +} + +static void +cpl_closeall () +{ + struct cpelement *cpe; + + for (cpe = coproc_list.head; cpe; cpe = cpe->next) + coproc_close (cpe->coproc); +} + +static void +cpl_fdchk (fd) + int fd; +{ + struct cpelement *cpe; + + for (cpe = coproc_list.head; cpe; cpe = cpe->next) + coproc_checkfd (cpe->coproc, fd); +} + +/* Search for PID in the list of coprocs; return the cpelement struct if + found. If not found, return NULL. */ +static struct cpelement * +cpl_search (pid) + pid_t pid; +{ + struct cpelement *cpe; + + for (cpe = coproc_list.head ; cpe; cpe = cpe->next) + if (cpe->coproc->c_pid == pid) + return cpe; + return (struct cpelement *)NULL; +} + +/* Search for the coproc named NAME in the list of coprocs; return the + cpelement struct if found. If not found, return NULL. */ +static struct cpelement * +cpl_searchbyname (name) + const char *name; +{ + struct cpelement *cp; + + for (cp = coproc_list.head ; cp; cp = cp->next) + if (STREQ (cp->coproc->c_name, name)) + return cp; + return (struct cpelement *)NULL; +} + +#if 0 +static void +cpl_prune () +{ + struct cpelement *cp; + + while (coproc_list.head && coproc_list.ncoproc > COPROC_MAX) + { + cp = coproc_list.head; + coproc_list.head = coproc_list.head->next; + coproc_dispose (cp->coproc); + cpe_dispose (cp); + coproc_list.ncoproc--; + } +} +#endif + +/* These currently use a single global "shell coproc" but are written in a + way to not preclude additional coprocs later (using the list management + package above). */ + +struct coproc * +getcoprocbypid (pid) + pid_t pid; +{ +#if MULTIPLE_COPROCS + struct cpelement *p; + + p = cpl_search (pid); + return (p ? p->coproc : 0); +#else + return (pid == sh_coproc.c_pid ? &sh_coproc : 0); +#endif +} + +struct coproc * +getcoprocbyname (name) + const char *name; +{ +#if MULTIPLE_COPROCS + struct cpelement *p; + + p = cpl_searchbyname (name); + return (p ? p->coproc : 0); +#else + return ((sh_coproc.c_name && STREQ (sh_coproc.c_name, name)) ? &sh_coproc : 0); +#endif +} + +void +coproc_init (cp) + struct coproc *cp; +{ + cp->c_name = 0; + cp->c_pid = NO_PID; + cp->c_rfd = cp->c_wfd = -1; + cp->c_rsave = cp->c_wsave = -1; + cp->c_flags = cp->c_status = 0; +} + +struct coproc * +coproc_alloc (name, pid) + char *name; + pid_t pid; +{ + struct coproc *cp; + +#if MULTIPLE_COPROCS + cp = (struct coproc *)xmalloc (sizeof (struct coproc)); +#else + cp = &sh_coproc; +#endif + coproc_init (cp); + + cp->c_name = savestring (name); + cp->c_pid = pid; + +#if MULTIPLE_COPROCS + cpl_add (cp); +#endif + + return (cp); +} + +static void +coproc_free (cp) + struct coproc *cp; +{ + free (cp); +} + +void +coproc_dispose (cp) + struct coproc *cp; +{ + if (cp == 0) + return; + + coproc_unsetvars (cp); + FREE (cp->c_name); + coproc_close (cp); +#if MULTIPLE_COPROCS + coproc_free (cp); +#else + coproc_init (cp); +#endif +} + +/* Placeholder for now. Will require changes for multiple coprocs */ +void +coproc_flush () +{ +#if MULTIPLE_COPROCS + cpl_flush (); +#else + coproc_dispose (&sh_coproc); +#endif +} + +void +coproc_close (cp) + struct coproc *cp; +{ + if (cp->c_rfd >= 0) + { + close (cp->c_rfd); + cp->c_rfd = -1; + } + if (cp->c_wfd >= 0) + { + close (cp->c_wfd); + cp->c_wfd = -1; + } + cp->c_rsave = cp->c_wsave = -1; +} + +void +coproc_closeall () +{ +#if MULTIPLE_COPROCS + cpl_closeall (); +#else + coproc_close (&sh_coproc); /* XXX - will require changes for multiple coprocs */ +#endif +} + +void +coproc_reap () +{ +#if MULTIPLE_COPROCS + cpl_reap (); +#else + struct coproc *cp; + + cp = &sh_coproc; /* XXX - will require changes for multiple coprocs */ + if (cp && (cp->c_flags & COPROC_DEAD)) + coproc_dispose (cp); +#endif +} + +void +coproc_rclose (cp, fd) + struct coproc *cp; + int fd; +{ + if (cp->c_rfd >= 0 && cp->c_rfd == fd) + { + close (cp->c_rfd); + cp->c_rfd = -1; + } +} + +void +coproc_wclose (cp, fd) + struct coproc *cp; + int fd; +{ + if (cp->c_wfd >= 0 && cp->c_wfd == fd) + { + close (cp->c_wfd); + cp->c_wfd = -1; + } +} + +void +coproc_checkfd (cp, fd) + struct coproc *cp; + int fd; +{ + int update; + + update = 0; + if (cp->c_rfd >= 0 && cp->c_rfd == fd) + update = cp->c_rfd = -1; + if (cp->c_wfd >= 0 && cp->c_wfd == fd) + update = cp->c_wfd = -1; + if (update) + coproc_setvars (cp); +} + +void +coproc_fdchk (fd) + int fd; +{ +#if MULTIPLE_COPROCS + cpl_fdchk (fd); +#else + coproc_checkfd (&sh_coproc, fd); +#endif +} + +void +coproc_fdclose (cp, fd) + struct coproc *cp; + int fd; +{ + coproc_rclose (cp, fd); + coproc_wclose (cp, fd); + coproc_setvars (cp); +} + +void +coproc_fdsave (cp) + struct coproc *cp; +{ + cp->c_rsave = cp->c_rfd; + cp->c_wsave = cp->c_wfd; +} + +void +coproc_fdrestore (cp) + struct coproc *cp; +{ + cp->c_rfd = cp->c_rsave; + cp->c_wfd = cp->c_wsave; +} + +void +coproc_pidchk (pid, status) + pid_t pid; +{ + struct coproc *cp; + +#if MULTIPLE_COPROCS + struct cpelement *cpe; + + cpe = cpl_delete (pid); + cp = cpe ? cpe->coproc : 0; +#else + cp = getcoprocbypid (pid); +#endif + if (cp) + { +#if 0 + itrace("coproc_pidchk: pid %d has died", pid); +#endif + cp->c_status = status; + cp->c_flags |= COPROC_DEAD; + cp->c_flags &= ~COPROC_RUNNING; +#if MULTIPLE_COPROCS + coproc_dispose (cp); +#else + coproc_unsetvars (cp); +#endif + } +} + +void +coproc_setvars (cp) + struct coproc *cp; +{ + SHELL_VAR *v; + char *namevar, *t; + int l; +#if defined (ARRAY_VARS) + arrayind_t ind; +#endif + + if (cp->c_name == 0) + return; + + l = strlen (cp->c_name); + namevar = xmalloc (l + 16); + +#if defined (ARRAY_VARS) + v = find_variable (cp->c_name); + if (v == 0) + v = make_new_array_variable (cp->c_name); + if (array_p (v) == 0) + v = convert_var_to_array (v); + + t = itos (cp->c_rfd); + ind = 0; + v = bind_array_variable (cp->c_name, ind, t, 0); + free (t); + + t = itos (cp->c_wfd); + ind = 1; + bind_array_variable (cp->c_name, ind, t, 0); + free (t); +#else + sprintf (namevar, "%s_READ", cp->c_name); + t = itos (cp->c_rfd); + bind_variable (namevar, t, 0); + free (t); + sprintf (namevar, "%s_WRITE", cp->c_name); + t = itos (cp->c_wfd); + bind_variable (namevar, t, 0); + free (t); +#endif + + sprintf (namevar, "%s_PID", cp->c_name); + t = itos (cp->c_pid); + bind_variable (namevar, t, 0); + free (t); + + free (namevar); +} + +void +coproc_unsetvars (cp) + struct coproc *cp; +{ + int l; + char *namevar; + + if (cp->c_name == 0) + return; + + l = strlen (cp->c_name); + namevar = xmalloc (l + 16); + + sprintf (namevar, "%s_PID", cp->c_name); + unbind_variable (namevar); + +#if defined (ARRAY_VARS) + unbind_variable (cp->c_name); +#else + sprintf (namevar, "%s_READ", cp->c_name); + unbind_variable (namevar); + sprintf (namevar, "%s_WRITE", cp->c_name); + unbind_variable (namevar); +#endif + + free (namevar); +} + +static int +execute_coproc (command, pipe_in, pipe_out, fds_to_close) + COMMAND *command; + int pipe_in, pipe_out; + struct fd_bitmap *fds_to_close; +{ + int rpipe[2], wpipe[2], estat; + pid_t coproc_pid; + Coproc *cp; + char *tcmd; + + /* XXX -- can be removed after changes to handle multiple coprocs */ +#if !MULTIPLE_COPROCS + if (sh_coproc.c_pid != NO_PID) + internal_warning ("execute_coproc: coproc [%d:%s] still exists", sh_coproc.c_pid, sh_coproc.c_name); + coproc_init (&sh_coproc); +#endif + + command_string_index = 0; + tcmd = make_command_string (command); + + sh_openpipe ((int *)&rpipe); /* 0 = parent read, 1 = child write */ + sh_openpipe ((int *)&wpipe); /* 0 = child read, 1 = parent write */ + + coproc_pid = make_child (savestring (tcmd), 1); + if (coproc_pid == 0) + { + close (rpipe[0]); + close (wpipe[1]); + + estat = execute_in_subshell (command, 1, wpipe[0], rpipe[1], fds_to_close); + + fflush (stdout); + fflush (stderr); + + exit (estat); + } + + close (rpipe[1]); + close (wpipe[0]); + + cp = coproc_alloc (command->value.Coproc->name, coproc_pid); + cp->c_rfd = rpipe[0]; + cp->c_wfd = wpipe[1]; + + SET_CLOSE_ON_EXEC (cp->c_rfd); + SET_CLOSE_ON_EXEC (cp->c_wfd); + + coproc_setvars (cp); + +#if 0 + itrace ("execute_coproc: [%d] %s", coproc_pid, the_printed_command); +#endif + + close_pipes (pipe_in, pipe_out); +#if defined (PROCESS_SUBSTITUTION) && defined (HAVE_DEV_FD) + unlink_fifo_list (); +#endif + stop_pipeline (1, (COMMAND *)NULL); + DESCRIBE_PID (coproc_pid); + run_pending_traps (); + + return (EXECUTION_SUCCESS); +} +#endif + +static void +restore_stdin (s) + int s; +{ + dup2 (s, 0); + close (s); +} + +/* Catch-all cleanup function for lastpipe code for unwind-protects */ +static void +lastpipe_cleanup (s) + int s; +{ + unfreeze_jobs_list (); +} + +static int +execute_pipeline (command, asynchronous, pipe_in, pipe_out, fds_to_close) + COMMAND *command; + int asynchronous, pipe_in, pipe_out; + struct fd_bitmap *fds_to_close; +{ + int prev, fildes[2], new_bitmap_size, dummyfd, ignore_return, exec_result; + int lstdin, lastpipe_flag, lastpipe_jid; + COMMAND *cmd; + struct fd_bitmap *fd_bitmap; + pid_t lastpid; + +#if defined (JOB_CONTROL) + sigset_t set, oset; + BLOCK_CHILD (set, oset); +#endif /* JOB_CONTROL */ + + ignore_return = (command->flags & CMD_IGNORE_RETURN) != 0; + + prev = pipe_in; + cmd = command; + + while (cmd && cmd->type == cm_connection && + cmd->value.Connection && cmd->value.Connection->connector == '|') + { + /* Make a pipeline between the two commands. */ + if (pipe (fildes) < 0) + { + sys_error (_("pipe error")); +#if defined (JOB_CONTROL) + terminate_current_pipeline (); + kill_current_pipeline (); + UNBLOCK_CHILD (oset); +#endif /* JOB_CONTROL */ + last_command_exit_value = EXECUTION_FAILURE; + /* The unwind-protects installed below will take care + of closing all of the open file descriptors. */ + throw_to_top_level (); + return (EXECUTION_FAILURE); /* XXX */ + } + + /* Here is a problem: with the new file close-on-exec + code, the read end of the pipe (fildes[0]) stays open + in the first process, so that process will never get a + SIGPIPE. There is no way to signal the first process + that it should close fildes[0] after forking, so it + remains open. No SIGPIPE is ever sent because there + is still a file descriptor open for reading connected + to the pipe. We take care of that here. This passes + around a bitmap of file descriptors that must be + closed after making a child process in execute_simple_command. */ + + /* We need fd_bitmap to be at least as big as fildes[0]. + If fildes[0] is less than fds_to_close->size, then + use fds_to_close->size. */ + new_bitmap_size = (fildes[0] < fds_to_close->size) + ? fds_to_close->size + : fildes[0] + 8; + + fd_bitmap = new_fd_bitmap (new_bitmap_size); + + /* Now copy the old information into the new bitmap. */ + xbcopy ((char *)fds_to_close->bitmap, (char *)fd_bitmap->bitmap, fds_to_close->size); + + /* And mark the pipe file descriptors to be closed. */ + fd_bitmap->bitmap[fildes[0]] = 1; + + /* In case there are pipe or out-of-processes errors, we + want all these file descriptors to be closed when + unwind-protects are run, and the storage used for the + bitmaps freed up. */ + begin_unwind_frame ("pipe-file-descriptors"); + add_unwind_protect (dispose_fd_bitmap, fd_bitmap); + add_unwind_protect (close_fd_bitmap, fd_bitmap); + if (prev >= 0) + add_unwind_protect (close, prev); + dummyfd = fildes[1]; + add_unwind_protect (close, dummyfd); + +#if defined (JOB_CONTROL) + add_unwind_protect (restore_signal_mask, &oset); +#endif /* JOB_CONTROL */ + + if (ignore_return && cmd->value.Connection->first) + cmd->value.Connection->first->flags |= CMD_IGNORE_RETURN; + execute_command_internal (cmd->value.Connection->first, asynchronous, + prev, fildes[1], fd_bitmap); + + if (prev >= 0) + close (prev); + + prev = fildes[0]; + close (fildes[1]); + + dispose_fd_bitmap (fd_bitmap); + discard_unwind_frame ("pipe-file-descriptors"); + + cmd = cmd->value.Connection->second; + } + + lastpid = last_made_pid; + + /* Now execute the rightmost command in the pipeline. */ + if (ignore_return && cmd) + cmd->flags |= CMD_IGNORE_RETURN; + +#if defined (JOB_CONTROL) + lastpipe_flag = 0; + begin_unwind_frame ("lastpipe-exec"); + lstdin = -1; + /* If the `lastpipe' option is set with shopt, and job control is not + enabled, execute the last element of non-async pipelines in the + current shell environment. */ + if (lastpipe_opt && job_control == 0 && asynchronous == 0 && pipe_out == NO_PIPE && prev > 0) + { + lstdin = move_to_high_fd (0, 0, 255); + if (lstdin > 0) + { + do_piping (prev, pipe_out); + prev = NO_PIPE; + add_unwind_protect (restore_stdin, lstdin); + lastpipe_flag = 1; + freeze_jobs_list (); + lastpipe_jid = stop_pipeline (0, (COMMAND *)NULL); /* XXX */ + add_unwind_protect (lastpipe_cleanup, lastpipe_jid); + } + if (cmd) + cmd->flags |= CMD_LASTPIPE; + } + if (prev >= 0) + add_unwind_protect (close, prev); +#endif + + exec_result = execute_command_internal (cmd, asynchronous, prev, pipe_out, fds_to_close); + +#if defined (JOB_CONTROL) + if (lstdin > 0) + restore_stdin (lstdin); +#endif + + if (prev >= 0) + close (prev); + +#if defined (JOB_CONTROL) + UNBLOCK_CHILD (oset); +#endif + + QUIT; + + if (lastpipe_flag) + { +#if defined (JOB_CONTROL) + append_process (savestring (the_printed_command), dollar_dollar_pid, exec_result, lastpipe_jid); +#endif + lstdin = wait_for (lastpid); +#if defined (JOB_CONTROL) + exec_result = job_exit_status (lastpipe_jid); +#endif + unfreeze_jobs_list (); + } + + discard_unwind_frame ("lastpipe-exec"); + + return (exec_result); +} + +static int +execute_connection (command, asynchronous, pipe_in, pipe_out, fds_to_close) + COMMAND *command; + int asynchronous, pipe_in, pipe_out; + struct fd_bitmap *fds_to_close; +{ + COMMAND *tc, *second; + int ignore_return, exec_result, was_error_trap, invert; + volatile int save_line_number; + + ignore_return = (command->flags & CMD_IGNORE_RETURN) != 0; + + switch (command->value.Connection->connector) + { + /* Do the first command asynchronously. */ + case '&': + tc = command->value.Connection->first; + if (tc == 0) + return (EXECUTION_SUCCESS); + + if (ignore_return) + tc->flags |= CMD_IGNORE_RETURN; + tc->flags |= CMD_AMPERSAND; + + /* If this shell was compiled without job control support, + if we are currently in a subshell via `( xxx )', or if job + control is not active then the standard input for an + asynchronous command is forced to /dev/null. */ +#if defined (JOB_CONTROL) + if ((subshell_environment || !job_control) && !stdin_redir) +#else + if (!stdin_redir) +#endif /* JOB_CONTROL */ + tc->flags |= CMD_STDIN_REDIR; + + exec_result = execute_command_internal (tc, 1, pipe_in, pipe_out, fds_to_close); + QUIT; + + if (tc->flags & CMD_STDIN_REDIR) + tc->flags &= ~CMD_STDIN_REDIR; + + second = command->value.Connection->second; + if (second) + { + if (ignore_return) + second->flags |= CMD_IGNORE_RETURN; + + exec_result = execute_command_internal (second, asynchronous, pipe_in, pipe_out, fds_to_close); + } + + break; + + /* Just call execute command on both sides. */ + case ';': + if (ignore_return) + { + if (command->value.Connection->first) + command->value.Connection->first->flags |= CMD_IGNORE_RETURN; + if (command->value.Connection->second) + command->value.Connection->second->flags |= CMD_IGNORE_RETURN; + } + executing_list++; + QUIT; + execute_command (command->value.Connection->first); + QUIT; + exec_result = execute_command_internal (command->value.Connection->second, + asynchronous, pipe_in, pipe_out, + fds_to_close); + executing_list--; + break; + + case '|': + was_error_trap = signal_is_trapped (ERROR_TRAP) && signal_is_ignored (ERROR_TRAP) == 0; + invert = (command->flags & CMD_INVERT_RETURN) != 0; + ignore_return = (command->flags & CMD_IGNORE_RETURN) != 0; + + line_number_for_err_trap = line_number; + exec_result = execute_pipeline (command, asynchronous, pipe_in, pipe_out, fds_to_close); + + if (was_error_trap && ignore_return == 0 && invert == 0 && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + save_line_number = line_number; + line_number = line_number_for_err_trap; + run_error_trap (); + line_number = save_line_number; + } + + if (ignore_return == 0 && invert == 0 && exit_immediately_on_error && exec_result != EXECUTION_SUCCESS) + { + last_command_exit_value = exec_result; + run_pending_traps (); + jump_to_top_level (ERREXIT); + } + + break; + + case AND_AND: + case OR_OR: + if (asynchronous) + { + /* If we have something like `a && b &' or `a || b &', run the + && or || stuff in a subshell. Force a subshell and just call + execute_command_internal again. Leave asynchronous on + so that we get a report from the parent shell about the + background job. */ + command->flags |= CMD_FORCE_SUBSHELL; + exec_result = execute_command_internal (command, 1, pipe_in, pipe_out, fds_to_close); + break; + } + + /* Execute the first command. If the result of that is successful + and the connector is AND_AND, or the result is not successful + and the connector is OR_OR, then execute the second command, + otherwise return. */ + + executing_list++; + if (command->value.Connection->first) + command->value.Connection->first->flags |= CMD_IGNORE_RETURN; + + exec_result = execute_command (command->value.Connection->first); + QUIT; + if (((command->value.Connection->connector == AND_AND) && + (exec_result == EXECUTION_SUCCESS)) || + ((command->value.Connection->connector == OR_OR) && + (exec_result != EXECUTION_SUCCESS))) + { + if (ignore_return && command->value.Connection->second) + command->value.Connection->second->flags |= CMD_IGNORE_RETURN; + + exec_result = execute_command (command->value.Connection->second); + } + executing_list--; + break; + + default: + command_error ("execute_connection", CMDERR_BADCONN, command->value.Connection->connector, 0); + jump_to_top_level (DISCARD); + exec_result = EXECUTION_FAILURE; + } + + return exec_result; +} + +#define REAP() \ + do \ + { \ + if (!interactive_shell) \ + reap_dead_jobs (); \ + } \ + while (0) + +/* Execute a FOR command. The syntax is: FOR word_desc IN word_list; + DO command; DONE */ +static int +execute_for_command (for_command) + FOR_COM *for_command; +{ + register WORD_LIST *releaser, *list; + SHELL_VAR *v; + char *identifier; + int retval, save_line_number; +#if 0 + SHELL_VAR *old_value = (SHELL_VAR *)NULL; /* Remember the old value of x. */ +#endif + + save_line_number = line_number; + if (check_identifier (for_command->name, 1) == 0) + { + if (posixly_correct && interactive_shell == 0) + { + last_command_exit_value = EX_BADUSAGE; + jump_to_top_level (ERREXIT); + } + return (EXECUTION_FAILURE); + } + + loop_level++; + identifier = for_command->name->word; + + list = releaser = expand_words_no_vars (for_command->map_list); + + begin_unwind_frame ("for"); + add_unwind_protect (dispose_words, releaser); + +#if 0 + if (lexical_scoping) + { + old_value = copy_variable (find_variable (identifier)); + if (old_value) + add_unwind_protect (dispose_variable, old_value); + } +#endif + + if (for_command->flags & CMD_IGNORE_RETURN) + for_command->action->flags |= CMD_IGNORE_RETURN; + + for (retval = EXECUTION_SUCCESS; list; list = list->next) + { + QUIT; + + line_number = for_command->line; + + /* Remember what this command looks like, for debugger. */ + command_string_index = 0; + print_for_command_head (for_command); + + if (echo_command_at_execute) + xtrace_print_for_command_head (for_command); + + /* Save this command unless it's a trap command and we're not running + a debug trap. */ +#if 0 + if (signal_in_progress (DEBUG_TRAP) == 0 && (this_command_name == 0 || (STREQ (this_command_name, "trap") == 0))) +#else + if (signal_in_progress (DEBUG_TRAP) == 0 && running_trap == 0) +#endif + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = savestring (the_printed_command); + } + + retval = run_debug_trap (); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode && retval != EXECUTION_SUCCESS) + continue; +#endif + + this_command_name = (char *)NULL; + v = bind_variable (identifier, list->word->word, 0); + if (readonly_p (v) || noassign_p (v)) + { + line_number = save_line_number; + if (readonly_p (v) && interactive_shell == 0 && posixly_correct) + { + last_command_exit_value = EXECUTION_FAILURE; + jump_to_top_level (FORCE_EOF); + } + else + { + dispose_words (releaser); + discard_unwind_frame ("for"); + loop_level--; + return (EXECUTION_FAILURE); + } + } + retval = execute_command (for_command->action); + REAP (); + QUIT; + + if (breaking) + { + breaking--; + break; + } + + if (continuing) + { + continuing--; + if (continuing) + break; + } + } + + loop_level--; + line_number = save_line_number; + +#if 0 + if (lexical_scoping) + { + if (!old_value) + unbind_variable (identifier); + else + { + SHELL_VAR *new_value; + + new_value = bind_variable (identifier, value_cell(old_value), 0); + new_value->attributes = old_value->attributes; + dispose_variable (old_value); + } + } +#endif + + dispose_words (releaser); + discard_unwind_frame ("for"); + return (retval); +} + +#if defined (ARITH_FOR_COMMAND) +/* Execute an arithmetic for command. The syntax is + + for (( init ; step ; test )) + do + body + done + + The execution should be exactly equivalent to + + eval \(\( init \)\) + while eval \(\( test \)\) ; do + body; + eval \(\( step \)\) + done +*/ +static intmax_t +eval_arith_for_expr (l, okp) + WORD_LIST *l; + int *okp; +{ + WORD_LIST *new; + intmax_t expresult; + int r; + + new = expand_words_no_vars (l); + if (new) + { + if (echo_command_at_execute) + xtrace_print_arith_cmd (new); + this_command_name = "(("; /* )) for expression error messages */ + + command_string_index = 0; + print_arith_command (new); + if (signal_in_progress (DEBUG_TRAP) == 0) + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = savestring (the_printed_command); + } + + r = run_debug_trap (); + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ +#if defined (DEBUGGER) + if (debugging_mode == 0 || r == EXECUTION_SUCCESS) + expresult = evalexp (new->word->word, okp); + else + { + expresult = 0; + if (okp) + *okp = 1; + } +#else + expresult = evalexp (new->word->word, okp); +#endif + dispose_words (new); + } + else + { + expresult = 0; + if (okp) + *okp = 1; + } + return (expresult); +} + +static int +execute_arith_for_command (arith_for_command) + ARITH_FOR_COM *arith_for_command; +{ + intmax_t expresult; + int expok, body_status, arith_lineno, save_lineno; + + body_status = EXECUTION_SUCCESS; + loop_level++; + save_lineno = line_number; + + if (arith_for_command->flags & CMD_IGNORE_RETURN) + arith_for_command->action->flags |= CMD_IGNORE_RETURN; + + this_command_name = "(("; /* )) for expression error messages */ + + /* save the starting line number of the command so we can reset + line_number before executing each expression -- for $LINENO + and the DEBUG trap. */ + line_number = arith_lineno = arith_for_command->line; + if (variable_context && interactive_shell) + line_number -= function_line_number; + + /* Evaluate the initialization expression. */ + expresult = eval_arith_for_expr (arith_for_command->init, &expok); + if (expok == 0) + { + line_number = save_lineno; + return (EXECUTION_FAILURE); + } + + while (1) + { + /* Evaluate the test expression. */ + line_number = arith_lineno; + expresult = eval_arith_for_expr (arith_for_command->test, &expok); + line_number = save_lineno; + + if (expok == 0) + { + body_status = EXECUTION_FAILURE; + break; + } + REAP (); + if (expresult == 0) + break; + + /* Execute the body of the arithmetic for command. */ + QUIT; + body_status = execute_command (arith_for_command->action); + QUIT; + + /* Handle any `break' or `continue' commands executed by the body. */ + if (breaking) + { + breaking--; + break; + } + + if (continuing) + { + continuing--; + if (continuing) + break; + } + + /* Evaluate the step expression. */ + line_number = arith_lineno; + expresult = eval_arith_for_expr (arith_for_command->step, &expok); + line_number = save_lineno; + + if (expok == 0) + { + body_status = EXECUTION_FAILURE; + break; + } + } + + loop_level--; + line_number = save_lineno; + + return (body_status); +} +#endif + +#if defined (SELECT_COMMAND) +static int LINES, COLS, tabsize; + +#define RP_SPACE ") " +#define RP_SPACE_LEN 2 + +/* XXX - does not handle numbers > 1000000 at all. */ +#define NUMBER_LEN(s) \ +((s < 10) ? 1 \ + : ((s < 100) ? 2 \ + : ((s < 1000) ? 3 \ + : ((s < 10000) ? 4 \ + : ((s < 100000) ? 5 \ + : 6))))) + +static int +displen (s) + const char *s; +{ +#if defined (HANDLE_MULTIBYTE) + wchar_t *wcstr; + size_t wclen, slen; + + wcstr = 0; + slen = mbstowcs (wcstr, s, 0); + if (slen == -1) + slen = 0; + wcstr = (wchar_t *)xmalloc (sizeof (wchar_t) * (slen + 1)); + mbstowcs (wcstr, s, slen + 1); + wclen = wcswidth (wcstr, slen); + free (wcstr); + return ((int)wclen); +#else + return (STRLEN (s)); +#endif +} + +static int +print_index_and_element (len, ind, list) + int len, ind; + WORD_LIST *list; +{ + register WORD_LIST *l; + register int i; + + if (list == 0) + return (0); + for (i = ind, l = list; l && --i; l = l->next) + ; + if (l == 0) /* don't think this can happen */ + return (0); + fprintf (stderr, "%*d%s%s", len, ind, RP_SPACE, l->word->word); + return (displen (l->word->word)); +} + +static void +indent (from, to) + int from, to; +{ + while (from < to) + { + if ((to / tabsize) > (from / tabsize)) + { + putc ('\t', stderr); + from += tabsize - from % tabsize; + } + else + { + putc (' ', stderr); + from++; + } + } +} + +static void +print_select_list (list, list_len, max_elem_len, indices_len) + WORD_LIST *list; + int list_len, max_elem_len, indices_len; +{ + int ind, row, elem_len, pos, cols, rows; + int first_column_indices_len, other_indices_len; + + if (list == 0) + { + putc ('\n', stderr); + return; + } + + cols = max_elem_len ? COLS / max_elem_len : 1; + if (cols == 0) + cols = 1; + rows = list_len ? list_len / cols + (list_len % cols != 0) : 1; + cols = list_len ? list_len / rows + (list_len % rows != 0) : 1; + + if (rows == 1) + { + rows = cols; + cols = 1; + } + + first_column_indices_len = NUMBER_LEN (rows); + other_indices_len = indices_len; + + for (row = 0; row < rows; row++) + { + ind = row; + pos = 0; + while (1) + { + indices_len = (pos == 0) ? first_column_indices_len : other_indices_len; + elem_len = print_index_and_element (indices_len, ind + 1, list); + elem_len += indices_len + RP_SPACE_LEN; + ind += rows; + if (ind >= list_len) + break; + indent (pos + elem_len, pos + max_elem_len); + pos += max_elem_len; + } + putc ('\n', stderr); + } +} + +/* Print the elements of LIST, one per line, preceded by an index from 1 to + LIST_LEN. Then display PROMPT and wait for the user to enter a number. + If the number is between 1 and LIST_LEN, return that selection. If EOF + is read, return a null string. If a blank line is entered, or an invalid + number is entered, the loop is executed again. */ +static char * +select_query (list, list_len, prompt, print_menu) + WORD_LIST *list; + int list_len; + char *prompt; + int print_menu; +{ + int max_elem_len, indices_len, len; + intmax_t reply; + WORD_LIST *l; + char *repl_string, *t; + +#if 0 + t = get_string_value ("LINES"); + LINES = (t && *t) ? atoi (t) : 24; +#endif + t = get_string_value ("COLUMNS"); + COLS = (t && *t) ? atoi (t) : 80; + +#if 0 + t = get_string_value ("TABSIZE"); + tabsize = (t && *t) ? atoi (t) : 8; + if (tabsize <= 0) + tabsize = 8; +#else + tabsize = 8; +#endif + + max_elem_len = 0; + for (l = list; l; l = l->next) + { + len = displen (l->word->word); + if (len > max_elem_len) + max_elem_len = len; + } + indices_len = NUMBER_LEN (list_len); + max_elem_len += indices_len + RP_SPACE_LEN + 2; + + while (1) + { + if (print_menu) + print_select_list (list, list_len, max_elem_len, indices_len); + fprintf (stderr, "%s", prompt); + fflush (stderr); + QUIT; + + if (read_builtin ((WORD_LIST *)NULL) != EXECUTION_SUCCESS) + { + putchar ('\n'); + return ((char *)NULL); + } + repl_string = get_string_value ("REPLY"); + if (*repl_string == 0) + { + print_menu = 1; + continue; + } + if (legal_number (repl_string, &reply) == 0) + return ""; + if (reply < 1 || reply > list_len) + return ""; + + for (l = list; l && --reply; l = l->next) + ; + return (l->word->word); /* XXX - can't be null? */ + } +} + +/* Execute a SELECT command. The syntax is: + SELECT word IN list DO command_list DONE + Only `break' or `return' in command_list will terminate + the command. */ +static int +execute_select_command (select_command) + SELECT_COM *select_command; +{ + WORD_LIST *releaser, *list; + SHELL_VAR *v; + char *identifier, *ps3_prompt, *selection; + int retval, list_len, show_menu, save_line_number; + + if (check_identifier (select_command->name, 1) == 0) + return (EXECUTION_FAILURE); + + save_line_number = line_number; + line_number = select_command->line; + + command_string_index = 0; + print_select_command_head (select_command); + + if (echo_command_at_execute) + xtrace_print_select_command_head (select_command); + +#if 0 + if (signal_in_progress (DEBUG_TRAP) == 0 && (this_command_name == 0 || (STREQ (this_command_name, "trap") == 0))) +#else + if (signal_in_progress (DEBUG_TRAP) == 0 && running_trap == 0) +#endif + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = savestring (the_printed_command); + } + + retval = run_debug_trap (); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode && retval != EXECUTION_SUCCESS) + return (EXECUTION_SUCCESS); +#endif + + loop_level++; + identifier = select_command->name->word; + + /* command and arithmetic substitution, parameter and variable expansion, + word splitting, pathname expansion, and quote removal. */ + list = releaser = expand_words_no_vars (select_command->map_list); + list_len = list_length (list); + if (list == 0 || list_len == 0) + { + if (list) + dispose_words (list); + line_number = save_line_number; + return (EXECUTION_SUCCESS); + } + + begin_unwind_frame ("select"); + add_unwind_protect (dispose_words, releaser); + + if (select_command->flags & CMD_IGNORE_RETURN) + select_command->action->flags |= CMD_IGNORE_RETURN; + + retval = EXECUTION_SUCCESS; + show_menu = 1; + + while (1) + { + line_number = select_command->line; + ps3_prompt = get_string_value ("PS3"); + if (ps3_prompt == 0) + ps3_prompt = "#? "; + + QUIT; + selection = select_query (list, list_len, ps3_prompt, show_menu); + QUIT; + if (selection == 0) + { + /* select_query returns EXECUTION_FAILURE if the read builtin + fails, so we want to return failure in this case. */ + retval = EXECUTION_FAILURE; + break; + } + + v = bind_variable (identifier, selection, 0); + if (readonly_p (v) || noassign_p (v)) + { + if (readonly_p (v) && interactive_shell == 0 && posixly_correct) + { + last_command_exit_value = EXECUTION_FAILURE; + jump_to_top_level (FORCE_EOF); + } + else + { + dispose_words (releaser); + discard_unwind_frame ("select"); + loop_level--; + line_number = save_line_number; + return (EXECUTION_FAILURE); + } + } + + retval = execute_command (select_command->action); + + REAP (); + QUIT; + + if (breaking) + { + breaking--; + break; + } + + if (continuing) + { + continuing--; + if (continuing) + break; + } + +#if defined (KSH_COMPATIBLE_SELECT) + show_menu = 0; + selection = get_string_value ("REPLY"); + if (selection && *selection == '\0') + show_menu = 1; +#endif + } + + loop_level--; + line_number = save_line_number; + + dispose_words (releaser); + discard_unwind_frame ("select"); + return (retval); +} +#endif /* SELECT_COMMAND */ + +/* Execute a CASE command. The syntax is: CASE word_desc IN pattern_list ESAC. + The pattern_list is a linked list of pattern clauses; each clause contains + some patterns to compare word_desc against, and an associated command to + execute. */ +static int +execute_case_command (case_command) + CASE_COM *case_command; +{ + register WORD_LIST *list; + WORD_LIST *wlist, *es; + PATTERN_LIST *clauses; + char *word, *pattern; + int retval, match, ignore_return, save_line_number; + + save_line_number = line_number; + line_number = case_command->line; + + command_string_index = 0; + print_case_command_head (case_command); + + if (echo_command_at_execute) + xtrace_print_case_command_head (case_command); + +#if 0 + if (signal_in_progress (DEBUG_TRAP) == 0 && (this_command_name == 0 || (STREQ (this_command_name, "trap") == 0))) +#else + if (signal_in_progress (DEBUG_TRAP) == 0 && running_trap == 0) +#endif + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = savestring (the_printed_command); + } + + retval = run_debug_trap(); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode && retval != EXECUTION_SUCCESS) + { + line_number = save_line_number; + return (EXECUTION_SUCCESS); + } +#endif + + wlist = expand_word_unsplit (case_command->word, 0); + word = wlist ? string_list (wlist) : savestring (""); + dispose_words (wlist); + + retval = EXECUTION_SUCCESS; + ignore_return = case_command->flags & CMD_IGNORE_RETURN; + + begin_unwind_frame ("case"); + add_unwind_protect ((Function *)xfree, word); + +#define EXIT_CASE() goto exit_case_command + + for (clauses = case_command->clauses; clauses; clauses = clauses->next) + { + QUIT; + for (list = clauses->patterns; list; list = list->next) + { + es = expand_word_leave_quoted (list->word, 0); + + if (es && es->word && es->word->word && *(es->word->word)) + pattern = quote_string_for_globbing (es->word->word, QGLOB_CVTNULL); + else + { + pattern = (char *)xmalloc (1); + pattern[0] = '\0'; + } + + /* Since the pattern does not undergo quote removal (as per + Posix.2, section 3.9.4.3), the strmatch () call must be able + to recognize backslashes as escape characters. */ + match = strmatch (pattern, word, FNMATCH_EXTFLAG|FNMATCH_IGNCASE) != FNM_NOMATCH; + free (pattern); + + dispose_words (es); + + if (match) + { + do + { + if (clauses->action && ignore_return) + clauses->action->flags |= CMD_IGNORE_RETURN; + retval = execute_command (clauses->action); + } + while ((clauses->flags & CASEPAT_FALLTHROUGH) && (clauses = clauses->next)); + if (clauses == 0 || (clauses->flags & CASEPAT_TESTNEXT) == 0) + EXIT_CASE (); + else + break; + } + + QUIT; + } + } + +exit_case_command: + free (word); + discard_unwind_frame ("case"); + line_number = save_line_number; + return (retval); +} + +#define CMD_WHILE 0 +#define CMD_UNTIL 1 + +/* The WHILE command. Syntax: WHILE test DO action; DONE. + Repeatedly execute action while executing test produces + EXECUTION_SUCCESS. */ +static int +execute_while_command (while_command) + WHILE_COM *while_command; +{ + return (execute_while_or_until (while_command, CMD_WHILE)); +} + +/* UNTIL is just like WHILE except that the test result is negated. */ +static int +execute_until_command (while_command) + WHILE_COM *while_command; +{ + return (execute_while_or_until (while_command, CMD_UNTIL)); +} + +/* The body for both while and until. The only difference between the + two is that the test value is treated differently. TYPE is + CMD_WHILE or CMD_UNTIL. The return value for both commands should + be EXECUTION_SUCCESS if no commands in the body are executed, and + the status of the last command executed in the body otherwise. */ +static int +execute_while_or_until (while_command, type) + WHILE_COM *while_command; + int type; +{ + int return_value, body_status; + + body_status = EXECUTION_SUCCESS; + loop_level++; + + while_command->test->flags |= CMD_IGNORE_RETURN; + if (while_command->flags & CMD_IGNORE_RETURN) + while_command->action->flags |= CMD_IGNORE_RETURN; + + while (1) + { + return_value = execute_command (while_command->test); + REAP (); + + /* Need to handle `break' in the test when we would break out of the + loop. The job control code will set `breaking' to loop_level + when a job in a loop is stopped with SIGTSTP. If the stopped job + is in the loop test, `breaking' will not be reset unless we do + this, and the shell will cease to execute commands. */ + if (type == CMD_WHILE && return_value != EXECUTION_SUCCESS) + { + if (breaking) + breaking--; + break; + } + if (type == CMD_UNTIL && return_value == EXECUTION_SUCCESS) + { + if (breaking) + breaking--; + break; + } + + QUIT; + body_status = execute_command (while_command->action); + QUIT; + + if (breaking) + { + breaking--; + break; + } + + if (continuing) + { + continuing--; + if (continuing) + break; + } + } + loop_level--; + + return (body_status); +} + +/* IF test THEN command [ELSE command]. + IF also allows ELIF in the place of ELSE IF, but + the parser makes *that* stupidity transparent. */ +static int +execute_if_command (if_command) + IF_COM *if_command; +{ + int return_value, save_line_number; + + save_line_number = line_number; + if_command->test->flags |= CMD_IGNORE_RETURN; + return_value = execute_command (if_command->test); + line_number = save_line_number; + + if (return_value == EXECUTION_SUCCESS) + { + QUIT; + + if (if_command->true_case && (if_command->flags & CMD_IGNORE_RETURN)) + if_command->true_case->flags |= CMD_IGNORE_RETURN; + + return (execute_command (if_command->true_case)); + } + else + { + QUIT; + + if (if_command->false_case && (if_command->flags & CMD_IGNORE_RETURN)) + if_command->false_case->flags |= CMD_IGNORE_RETURN; + + return (execute_command (if_command->false_case)); + } +} + +#if defined (DPAREN_ARITHMETIC) +static int +execute_arith_command (arith_command) + ARITH_COM *arith_command; +{ + int expok, save_line_number, retval; + intmax_t expresult; + WORD_LIST *new; + char *exp; + + expresult = 0; + + save_line_number = line_number; + this_command_name = "(("; /* )) */ + line_number = arith_command->line; + /* If we're in a function, update the line number information. */ + if (variable_context && interactive_shell) + line_number -= function_line_number; + + command_string_index = 0; + print_arith_command (arith_command->exp); + + if (signal_in_progress (DEBUG_TRAP) == 0) + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = savestring (the_printed_command); + } + + /* Run the debug trap before each arithmetic command, but do it after we + update the line number information and before we expand the various + words in the expression. */ + retval = run_debug_trap (); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode && retval != EXECUTION_SUCCESS) + { + line_number = save_line_number; + return (EXECUTION_SUCCESS); + } +#endif + + new = expand_words_no_vars (arith_command->exp); + + /* If we're tracing, make a new word list with `((' at the front and `))' + at the back and print it. */ + if (echo_command_at_execute) + xtrace_print_arith_cmd (new); + + if (new) + { + exp = new->next ? string_list (new) : new->word->word; + expresult = evalexp (exp, &expok); + line_number = save_line_number; + if (exp != new->word->word) + free (exp); + dispose_words (new); + } + else + { + expresult = 0; + expok = 1; + } + + if (expok == 0) + return (EXECUTION_FAILURE); + + return (expresult == 0 ? EXECUTION_FAILURE : EXECUTION_SUCCESS); +} +#endif /* DPAREN_ARITHMETIC */ + +#if defined (COND_COMMAND) + +static char * const nullstr = ""; + +/* XXX - can COND ever be NULL when this is called? */ +static int +execute_cond_node (cond) + COND_COM *cond; +{ + int result, invert, patmatch, rmatch, mflags, ignore; + char *arg1, *arg2; + + invert = (cond->flags & CMD_INVERT_RETURN); + ignore = (cond->flags & CMD_IGNORE_RETURN); + if (ignore) + { + if (cond->left) + cond->left->flags |= CMD_IGNORE_RETURN; + if (cond->right) + cond->right->flags |= CMD_IGNORE_RETURN; + } + + if (cond->type == COND_EXPR) + result = execute_cond_node (cond->left); + else if (cond->type == COND_OR) + { + result = execute_cond_node (cond->left); + if (result != EXECUTION_SUCCESS) + result = execute_cond_node (cond->right); + } + else if (cond->type == COND_AND) + { + result = execute_cond_node (cond->left); + if (result == EXECUTION_SUCCESS) + result = execute_cond_node (cond->right); + } + else if (cond->type == COND_UNARY) + { + if (ignore) + comsub_ignore_return++; + arg1 = cond_expand_word (cond->left->op, 0); + if (ignore) + comsub_ignore_return--; + if (arg1 == 0) + arg1 = nullstr; + if (echo_command_at_execute) + xtrace_print_cond_term (cond->type, invert, cond->op, arg1, (char *)NULL); + result = unary_test (cond->op->word, arg1) ? EXECUTION_SUCCESS : EXECUTION_FAILURE; + if (arg1 != nullstr) + free (arg1); + } + else if (cond->type == COND_BINARY) + { + rmatch = 0; + patmatch = (((cond->op->word[1] == '=') && (cond->op->word[2] == '\0') && + (cond->op->word[0] == '!' || cond->op->word[0] == '=')) || + (cond->op->word[0] == '=' && cond->op->word[1] == '\0')); +#if defined (COND_REGEXP) + rmatch = (cond->op->word[0] == '=' && cond->op->word[1] == '~' && + cond->op->word[2] == '\0'); +#endif + + if (ignore) + comsub_ignore_return++; + arg1 = cond_expand_word (cond->left->op, 0); + if (ignore) + comsub_ignore_return--; + if (arg1 == 0) + arg1 = nullstr; + if (ignore) + comsub_ignore_return++; + arg2 = cond_expand_word (cond->right->op, + (rmatch && shell_compatibility_level > 31) ? 2 : (patmatch ? 1 : 0)); + if (ignore) + comsub_ignore_return--; + if (arg2 == 0) + arg2 = nullstr; + + if (echo_command_at_execute) + xtrace_print_cond_term (cond->type, invert, cond->op, arg1, arg2); + +#if defined (COND_REGEXP) + if (rmatch) + { + mflags = SHMAT_PWARN; +#if defined (ARRAY_VARS) + mflags |= SHMAT_SUBEXP; +#endif + + result = sh_regmatch (arg1, arg2, mflags); + } + else +#endif /* COND_REGEXP */ + { + int oe; + oe = extended_glob; + extended_glob = 1; + result = binary_test (cond->op->word, arg1, arg2, TEST_PATMATCH|TEST_ARITHEXP|TEST_LOCALE) + ? EXECUTION_SUCCESS + : EXECUTION_FAILURE; + extended_glob = oe; + } + if (arg1 != nullstr) + free (arg1); + if (arg2 != nullstr) + free (arg2); + } + else + { + command_error ("execute_cond_node", CMDERR_BADTYPE, cond->type, 0); + jump_to_top_level (DISCARD); + result = EXECUTION_FAILURE; + } + + if (invert) + result = (result == EXECUTION_SUCCESS) ? EXECUTION_FAILURE : EXECUTION_SUCCESS; + + return result; +} + +static int +execute_cond_command (cond_command) + COND_COM *cond_command; +{ + int retval, save_line_number; + + retval = EXECUTION_SUCCESS; + save_line_number = line_number; + + this_command_name = "[["; + line_number = cond_command->line; + /* If we're in a function, update the line number information. */ + if (variable_context && interactive_shell) + line_number -= function_line_number; + command_string_index = 0; + print_cond_command (cond_command); + + if (signal_in_progress (DEBUG_TRAP) == 0) + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = savestring (the_printed_command); + } + + /* Run the debug trap before each conditional command, but do it after we + update the line number information. */ + retval = run_debug_trap (); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode && retval != EXECUTION_SUCCESS) + { + line_number = save_line_number; + return (EXECUTION_SUCCESS); + } +#endif + +#if 0 + debug_print_cond_command (cond_command); +#endif + + last_command_exit_value = retval = execute_cond_node (cond_command); + line_number = save_line_number; + return (retval); +} +#endif /* COND_COMMAND */ + +static void +bind_lastarg (arg) + char *arg; +{ + SHELL_VAR *var; + + if (arg == 0) + arg = ""; + var = bind_variable ("_", arg, 0); + VUNSETATTR (var, att_exported); +} + +/* Execute a null command. Fork a subshell if the command uses pipes or is + to be run asynchronously. This handles all the side effects that are + supposed to take place. */ +static int +execute_null_command (redirects, pipe_in, pipe_out, async) + REDIRECT *redirects; + int pipe_in, pipe_out, async; +{ + int r; + int forcefork; + REDIRECT *rd; + + for (forcefork = 0, rd = redirects; rd; rd = rd->next) + forcefork += rd->rflags & REDIR_VARASSIGN; + + if (forcefork || pipe_in != NO_PIPE || pipe_out != NO_PIPE || async) + { + /* We have a null command, but we really want a subshell to take + care of it. Just fork, do piping and redirections, and exit. */ + if (make_child ((char *)NULL, async) == 0) + { + /* Cancel traps, in trap.c. */ + restore_original_signals (); /* XXX */ + + do_piping (pipe_in, pipe_out); + +#if defined (COPROCESS_SUPPORT) + coproc_closeall (); +#endif + + subshell_environment = 0; + if (async) + subshell_environment |= SUBSHELL_ASYNC; + if (pipe_in != NO_PIPE || pipe_out != NO_PIPE) + subshell_environment |= SUBSHELL_PIPE; + + if (do_redirections (redirects, RX_ACTIVE) == 0) + exit (EXECUTION_SUCCESS); + else + exit (EXECUTION_FAILURE); + } + else + { + close_pipes (pipe_in, pipe_out); +#if defined (PROCESS_SUBSTITUTION) && defined (HAVE_DEV_FD) + unlink_fifo_list (); +#endif + return (EXECUTION_SUCCESS); + } + } + else + { + /* Even if there aren't any command names, pretend to do the + redirections that are specified. The user expects the side + effects to take place. If the redirections fail, then return + failure. Otherwise, if a command substitution took place while + expanding the command or a redirection, return the value of that + substitution. Otherwise, return EXECUTION_SUCCESS. */ + + r = do_redirections (redirects, RX_ACTIVE|RX_UNDOABLE); + cleanup_redirects (redirection_undo_list); + redirection_undo_list = (REDIRECT *)NULL; + + if (r != 0) + return (EXECUTION_FAILURE); + else if (last_command_subst_pid != NO_PID) + return (last_command_exit_value); + else + return (EXECUTION_SUCCESS); + } +} + +/* This is a hack to suppress word splitting for assignment statements + given as arguments to builtins with the ASSIGNMENT_BUILTIN flag set. */ +static void +fix_assignment_words (words) + WORD_LIST *words; +{ + WORD_LIST *w, *wcmd; + struct builtin *b; + int assoc; + + if (words == 0) + return; + + b = 0; + assoc = 0; + + wcmd = words; + for (w = words; w; w = w->next) + if (w->word->flags & W_ASSIGNMENT) + { + if (b == 0) + { + /* Posix (post-2008) says that `command' doesn't change whether + or not the builtin it shadows is a `declaration command', even + though it removes other special builtin properties. In Posix + mode, we skip over one or more instances of `command' and + deal with the next word as the assignment builtin. */ + while (posixly_correct && wcmd && wcmd->word && wcmd->word->word && STREQ (wcmd->word->word, "command")) + wcmd = wcmd->next; + b = builtin_address_internal (wcmd->word->word, 0); + if (b == 0 || (b->flags & ASSIGNMENT_BUILTIN) == 0) + return; + else if (b && (b->flags & ASSIGNMENT_BUILTIN)) + wcmd->word->flags |= W_ASSNBLTIN; + } + w->word->flags |= (W_NOSPLIT|W_NOGLOB|W_TILDEEXP|W_ASSIGNARG); +#if defined (ARRAY_VARS) + if (assoc) + w->word->flags |= W_ASSIGNASSOC; +#endif + } +#if defined (ARRAY_VARS) + /* Note that we saw an associative array option to a builtin that takes + assignment statements. This is a bit of a kludge. */ + else if (w->word->word[0] == '-' && strchr (w->word->word, 'A')) + { + if (b == 0) + { + while (posixly_correct && wcmd && wcmd->word && wcmd->word->word && STREQ (wcmd->word->word, "command")) + wcmd = wcmd->next; + b = builtin_address_internal (wcmd->word->word, 0); + if (b == 0 || (b->flags & ASSIGNMENT_BUILTIN) == 0) + return; + else if (b && (b->flags & ASSIGNMENT_BUILTIN)) + wcmd->word->flags |= W_ASSNBLTIN; + } + if (wcmd->word->flags & W_ASSNBLTIN) + assoc = 1; + } +#endif +} + +/* Return 1 if the file found by searching $PATH for PATHNAME, defaulting + to PATHNAME, is a directory. Used by the autocd code below. */ +static int +is_dirname (pathname) + char *pathname; +{ + char *temp; + int ret; + + temp = search_for_command (pathname, 0); + ret = (temp ? file_isdir (temp) : file_isdir (pathname)); + free (temp); + return ret; +} + +/* The meaty part of all the executions. We have to start hacking the + real execution of commands here. Fork a process, set things up, + execute the command. */ +static int +execute_simple_command (simple_command, pipe_in, pipe_out, async, fds_to_close) + SIMPLE_COM *simple_command; + int pipe_in, pipe_out, async; + struct fd_bitmap *fds_to_close; +{ + WORD_LIST *words, *lastword; + char *command_line, *lastarg, *temp; + int first_word_quoted, result, builtin_is_special, already_forked, dofork; + pid_t old_last_async_pid; + sh_builtin_func_t *builtin; + SHELL_VAR *func; + volatile int old_builtin, old_command_builtin; + + result = EXECUTION_SUCCESS; + special_builtin_failed = builtin_is_special = 0; + command_line = (char *)0; + + QUIT; + + /* If we're in a function, update the line number information. */ + if (variable_context && interactive_shell && sourcelevel == 0) + line_number -= function_line_number; + + /* Remember what this command line looks like at invocation. */ + command_string_index = 0; + print_simple_command (simple_command); + +#if 0 + if (signal_in_progress (DEBUG_TRAP) == 0 && (this_command_name == 0 || (STREQ (this_command_name, "trap") == 0))) +#else + if (signal_in_progress (DEBUG_TRAP) == 0 && running_trap == 0) +#endif + { + FREE (the_printed_command_except_trap); + the_printed_command_except_trap = the_printed_command ? savestring (the_printed_command) : (char *)0; + } + + /* Run the debug trap before each simple command, but do it after we + update the line number information. */ + result = run_debug_trap (); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode && result != EXECUTION_SUCCESS) + return (EXECUTION_SUCCESS); +#endif + + first_word_quoted = + simple_command->words ? (simple_command->words->word->flags & W_QUOTED) : 0; + + last_command_subst_pid = NO_PID; + old_last_async_pid = last_asynchronous_pid; + + already_forked = dofork = 0; + + /* If we're in a pipeline or run in the background, set DOFORK so we + make the child early, before word expansion. This keeps assignment + statements from affecting the parent shell's environment when they + should not. */ + dofork = pipe_in != NO_PIPE || pipe_out != NO_PIPE || async; + + /* Something like `%2 &' should restart job 2 in the background, not cause + the shell to fork here. */ + if (dofork && pipe_in == NO_PIPE && pipe_out == NO_PIPE && + simple_command->words && simple_command->words->word && + simple_command->words->word->word && + (simple_command->words->word->word[0] == '%')) + dofork = 0; + + if (dofork) + { + /* Do this now, because execute_disk_command will do it anyway in the + vast majority of cases. */ + maybe_make_export_env (); + + /* Don't let a DEBUG trap overwrite the command string to be saved with + the process/job associated with this child. */ + if (make_child (savestring (the_printed_command_except_trap), async) == 0) + { + already_forked = 1; + simple_command->flags |= CMD_NO_FORK; + + subshell_environment = SUBSHELL_FORK; + if (pipe_in != NO_PIPE || pipe_out != NO_PIPE) + subshell_environment |= SUBSHELL_PIPE; + if (async) + subshell_environment |= SUBSHELL_ASYNC; + + /* We need to do this before piping to handle some really + pathological cases where one of the pipe file descriptors + is < 2. */ + if (fds_to_close) + close_fd_bitmap (fds_to_close); + + do_piping (pipe_in, pipe_out); + pipe_in = pipe_out = NO_PIPE; +#if defined (COPROCESS_SUPPORT) + coproc_closeall (); +#endif + + last_asynchronous_pid = old_last_async_pid; + } + else + { + /* Don't let simple commands that aren't the last command in a + pipeline change $? for the rest of the pipeline (or at all). */ + if (pipe_out != NO_PIPE) + result = last_command_exit_value; + close_pipes (pipe_in, pipe_out); +#if defined (PROCESS_SUBSTITUTION) && defined (HAVE_DEV_FD) + unlink_fifo_list (); +#endif + command_line = (char *)NULL; /* don't free this. */ + bind_lastarg ((char *)NULL); + return (result); + } + } + + /* If we are re-running this as the result of executing the `command' + builtin, do not expand the command words a second time. */ + if ((simple_command->flags & CMD_INHIBIT_EXPANSION) == 0) + { + current_fds_to_close = fds_to_close; + fix_assignment_words (simple_command->words); + /* Pass the ignore return flag down to command substitutions */ + if (simple_command->flags & CMD_IGNORE_RETURN) /* XXX */ + comsub_ignore_return++; + words = expand_words (simple_command->words); + if (simple_command->flags & CMD_IGNORE_RETURN) + comsub_ignore_return--; + current_fds_to_close = (struct fd_bitmap *)NULL; + } + else + words = copy_word_list (simple_command->words); + + /* It is possible for WORDS not to have anything left in it. + Perhaps all the words consisted of `$foo', and there was + no variable `$foo'. */ + if (words == 0) + { + this_command_name = 0; + result = execute_null_command (simple_command->redirects, + pipe_in, pipe_out, + already_forked ? 0 : async); + if (already_forked) + exit (result); + else + { + bind_lastarg ((char *)NULL); + set_pipestatus_from_exit (result); + return (result); + } + } + + lastarg = (char *)NULL; + + begin_unwind_frame ("simple-command"); + + if (echo_command_at_execute) + xtrace_print_word_list (words, 1); + + builtin = (sh_builtin_func_t *)NULL; + func = (SHELL_VAR *)NULL; + if ((simple_command->flags & CMD_NO_FUNCTIONS) == 0) + { + /* Posix.2 says special builtins are found before functions. We + don't set builtin_is_special anywhere other than here, because + this path is followed only when the `command' builtin is *not* + being used, and we don't want to exit the shell if a special + builtin executed with `command builtin' fails. `command' is not + a special builtin. */ + if (posixly_correct) + { + builtin = find_special_builtin (words->word->word); + if (builtin) + builtin_is_special = 1; + } + if (builtin == 0) + func = find_function (words->word->word); + } + + /* In POSIX mode, assignment errors in the temporary environment cause a + non-interactive shell to exit. */ + if (builtin_is_special && interactive_shell == 0 && tempenv_assign_error) + { + last_command_exit_value = EXECUTION_FAILURE; + jump_to_top_level (ERREXIT); + } + + add_unwind_protect (dispose_words, words); + QUIT; + + /* Bind the last word in this command to "$_" after execution. */ + for (lastword = words; lastword->next; lastword = lastword->next) + ; + lastarg = lastword->word->word; + +#if defined (JOB_CONTROL) + /* Is this command a job control related thing? */ + if (words->word->word[0] == '%' && already_forked == 0) + { + this_command_name = async ? "bg" : "fg"; + last_shell_builtin = this_shell_builtin; + this_shell_builtin = builtin_address (this_command_name); + result = (*this_shell_builtin) (words); + goto return_result; + } + + /* One other possiblilty. The user may want to resume an existing job. + If they do, find out whether this word is a candidate for a running + job. */ + if (job_control && already_forked == 0 && async == 0 && + !first_word_quoted && + !words->next && + words->word->word[0] && + !simple_command->redirects && + pipe_in == NO_PIPE && + pipe_out == NO_PIPE && + (temp = get_string_value ("auto_resume"))) + { + int job, jflags, started_status; + + jflags = JM_STOPPED|JM_FIRSTMATCH; + if (STREQ (temp, "exact")) + jflags |= JM_EXACT; + else if (STREQ (temp, "substring")) + jflags |= JM_SUBSTRING; + else + jflags |= JM_PREFIX; + job = get_job_by_name (words->word->word, jflags); + if (job != NO_JOB) + { + run_unwind_frame ("simple-command"); + this_command_name = "fg"; + last_shell_builtin = this_shell_builtin; + this_shell_builtin = builtin_address ("fg"); + + started_status = start_job (job, 1); + return ((started_status < 0) ? EXECUTION_FAILURE : started_status); + } + } +#endif /* JOB_CONTROL */ + +run_builtin: + /* Remember the name of this command globally. */ + this_command_name = words->word->word; + + QUIT; + + /* This command could be a shell builtin or a user-defined function. + We have already found special builtins by this time, so we do not + set builtin_is_special. If this is a function or builtin, and we + have pipes, then fork a subshell in here. Otherwise, just execute + the command directly. */ + if (func == 0 && builtin == 0) + builtin = find_shell_builtin (this_command_name); + + last_shell_builtin = this_shell_builtin; + this_shell_builtin = builtin; + + if (builtin || func) + { + if (builtin) + { + old_builtin = executing_builtin; + old_command_builtin = executing_command_builtin; + unwind_protect_int (executing_builtin); /* modified in execute_builtin */ + unwind_protect_int (executing_command_builtin); /* ditto */ + } + if (already_forked) + { + /* reset_terminating_signals (); */ /* XXX */ + /* Reset the signal handlers in the child, but don't free the + trap strings. Set a flag noting that we have to free the + trap strings if we run trap to change a signal disposition. */ + reset_signal_handlers (); + subshell_environment |= SUBSHELL_RESETTRAP; + + if (async) + { + if ((simple_command->flags & CMD_STDIN_REDIR) && + pipe_in == NO_PIPE && + (stdin_redirects (simple_command->redirects) == 0)) + async_redirect_stdin (); + setup_async_signals (); + } + + subshell_level++; + execute_subshell_builtin_or_function + (words, simple_command->redirects, builtin, func, + pipe_in, pipe_out, async, fds_to_close, + simple_command->flags); + subshell_level--; + } + else + { + result = execute_builtin_or_function + (words, builtin, func, simple_command->redirects, fds_to_close, + simple_command->flags); + if (builtin) + { + if (result > EX_SHERRBASE) + { + result = builtin_status (result); + if (builtin_is_special) + special_builtin_failed = 1; + } + /* In POSIX mode, if there are assignment statements preceding + a special builtin, they persist after the builtin + completes. */ + if (posixly_correct && builtin_is_special && temporary_env) + merge_temporary_env (); + } + else /* function */ + { + if (result == EX_USAGE) + result = EX_BADUSAGE; + else if (result > EX_SHERRBASE) + result = EXECUTION_FAILURE; + } + + set_pipestatus_from_exit (result); + + goto return_result; + } + } + + if (autocd && interactive && words->word && is_dirname (words->word->word)) + { + words = make_word_list (make_word ("cd"), words); + xtrace_print_word_list (words, 0); + goto run_builtin; + } + + if (command_line == 0) + command_line = savestring (the_printed_command_except_trap); + +#if defined (PROCESS_SUBSTITUTION) + if ((subshell_environment & SUBSHELL_COMSUB) && (simple_command->flags & CMD_NO_FORK) && fifos_pending() > 0) + simple_command->flags &= ~CMD_NO_FORK; +#endif + + result = execute_disk_command (words, simple_command->redirects, command_line, + pipe_in, pipe_out, async, fds_to_close, + simple_command->flags); + + return_result: + bind_lastarg (lastarg); + FREE (command_line); + dispose_words (words); + if (builtin) + { + executing_builtin = old_builtin; + executing_command_builtin = old_command_builtin; + } + discard_unwind_frame ("simple-command"); + this_command_name = (char *)NULL; /* points to freed memory now */ + return (result); +} + +/* Translate the special builtin exit statuses. We don't really need a + function for this; it's a placeholder for future work. */ +static int +builtin_status (result) + int result; +{ + int r; + + switch (result) + { + case EX_USAGE: + r = EX_BADUSAGE; + break; + case EX_REDIRFAIL: + case EX_BADSYNTAX: + case EX_BADASSIGN: + case EX_EXPFAIL: + r = EXECUTION_FAILURE; + break; + default: + r = EXECUTION_SUCCESS; + break; + } + return (r); +} + +static int +execute_builtin (builtin, words, flags, subshell) + sh_builtin_func_t *builtin; + WORD_LIST *words; + int flags, subshell; +{ + int old_e_flag, result, eval_unwind; + int isbltinenv; + char *error_trap; + + error_trap = 0; + old_e_flag = exit_immediately_on_error; + /* The eval builtin calls parse_and_execute, which does not know about + the setting of flags, and always calls the execution functions with + flags that will exit the shell on an error if -e is set. If the + eval builtin is being called, and we're supposed to ignore the exit + value of the command, we turn the -e flag off ourselves and disable + the ERR trap, then restore them when the command completes. This is + also a problem (as below) for the command and source/. builtins. */ + if (subshell == 0 && (flags & CMD_IGNORE_RETURN) && + (builtin == eval_builtin || builtin == command_builtin || builtin == source_builtin)) + { + begin_unwind_frame ("eval_builtin"); + unwind_protect_int (exit_immediately_on_error); + error_trap = TRAP_STRING (ERROR_TRAP); + if (error_trap) + { + error_trap = savestring (error_trap); + add_unwind_protect (xfree, error_trap); + add_unwind_protect (set_error_trap, error_trap); + restore_default_signal (ERROR_TRAP); + } + exit_immediately_on_error = 0; + eval_unwind = 1; + } + else + eval_unwind = 0; + + /* The temporary environment for a builtin is supposed to apply to + all commands executed by that builtin. Currently, this is a + problem only with the `unset', `source' and `eval' builtins. */ + + isbltinenv = (builtin == source_builtin || builtin == eval_builtin || builtin == unset_builtin); + + if (isbltinenv) + { + if (subshell == 0) + begin_unwind_frame ("builtin_env"); + + if (temporary_env) + { + push_scope (VC_BLTNENV, temporary_env); + if (subshell == 0) + add_unwind_protect (pop_scope, (flags & CMD_COMMAND_BUILTIN) ? 0 : "1"); + temporary_env = (HASH_TABLE *)NULL; + } + } + + /* `return' does a longjmp() back to a saved environment in execute_function. + If a variable assignment list preceded the command, and the shell is + running in POSIX mode, we need to merge that into the shell_variables + table, since `return' is a POSIX special builtin. */ + if (posixly_correct && subshell == 0 && builtin == return_builtin && temporary_env) + { + begin_unwind_frame ("return_temp_env"); + add_unwind_protect (merge_temporary_env, (char *)NULL); + } + + executing_builtin++; + executing_command_builtin |= builtin == command_builtin; + result = ((*builtin) (words->next)); + + /* This shouldn't happen, but in case `return' comes back instead of + longjmp'ing, we need to unwind. */ + if (posixly_correct && subshell == 0 && builtin == return_builtin && temporary_env) + discard_unwind_frame ("return_temp_env"); + + if (subshell == 0 && isbltinenv) + run_unwind_frame ("builtin_env"); + + if (eval_unwind) + { + exit_immediately_on_error += old_e_flag; + if (error_trap) + { + set_error_trap (error_trap); + xfree (error_trap); + } + discard_unwind_frame ("eval_builtin"); + } + + return (result); +} + +static int +execute_function (var, words, flags, fds_to_close, async, subshell) + SHELL_VAR *var; + WORD_LIST *words; + int flags; + struct fd_bitmap *fds_to_close; + int async, subshell; +{ + int return_val, result; + COMMAND *tc, *fc, *save_current; + char *debug_trap, *error_trap, *return_trap; +#if defined (ARRAY_VARS) + SHELL_VAR *funcname_v, *nfv, *bash_source_v, *bash_lineno_v; + ARRAY *funcname_a; + volatile ARRAY *bash_source_a; + volatile ARRAY *bash_lineno_a; +#endif + FUNCTION_DEF *shell_fn; + char *sfile, *t; + + USE_VAR(fc); + + if (funcnest_max > 0 && funcnest >= funcnest_max) + { + internal_error (_("%s: maximum function nesting level exceeded (%d)"), var->name, funcnest); + funcnest = 0; /* XXX - should we reset it somewhere else? */ + jump_to_top_level (DISCARD); + } + +#if defined (ARRAY_VARS) + GET_ARRAY_FROM_VAR ("FUNCNAME", funcname_v, funcname_a); + GET_ARRAY_FROM_VAR ("BASH_SOURCE", bash_source_v, bash_source_a); + GET_ARRAY_FROM_VAR ("BASH_LINENO", bash_lineno_v, bash_lineno_a); +#endif + + tc = (COMMAND *)copy_command (function_cell (var)); + if (tc && (flags & CMD_IGNORE_RETURN)) + tc->flags |= CMD_IGNORE_RETURN; + + if (subshell == 0) + { + begin_unwind_frame ("function_calling"); + push_context (var->name, subshell, temporary_env); + add_unwind_protect (pop_context, (char *)NULL); + unwind_protect_int (line_number); + unwind_protect_int (return_catch_flag); + unwind_protect_jmp_buf (return_catch); + add_unwind_protect (dispose_command, (char *)tc); + unwind_protect_pointer (this_shell_function); + unwind_protect_int (loop_level); + unwind_protect_int (funcnest); + } + else + push_context (var->name, subshell, temporary_env); /* don't unwind-protect for subshells */ + + temporary_env = (HASH_TABLE *)NULL; + + this_shell_function = var; + make_funcname_visible (1); + + debug_trap = TRAP_STRING(DEBUG_TRAP); + error_trap = TRAP_STRING(ERROR_TRAP); + return_trap = TRAP_STRING(RETURN_TRAP); + + /* The order of the unwind protects for debug_trap, error_trap and + return_trap is important here! unwind-protect commands are run + in reverse order of registration. If this causes problems, take + out the xfree unwind-protect calls and live with the small memory leak. */ + + /* function_trace_mode != 0 means that all functions inherit the DEBUG trap. + if the function has the trace attribute set, it inherits the DEBUG trap */ + if (debug_trap && ((trace_p (var) == 0) && function_trace_mode == 0)) + { + if (subshell == 0) + { + debug_trap = savestring (debug_trap); + add_unwind_protect (xfree, debug_trap); + add_unwind_protect (set_debug_trap, debug_trap); + } + restore_default_signal (DEBUG_TRAP); + } + + /* error_trace_mode != 0 means that functions inherit the ERR trap. */ + if (error_trap && error_trace_mode == 0) + { + if (subshell == 0) + { + error_trap = savestring (error_trap); + add_unwind_protect (xfree, error_trap); + add_unwind_protect (set_error_trap, error_trap); + } + restore_default_signal (ERROR_TRAP); + } + + /* Shell functions inherit the RETURN trap if function tracing is on + globally or on individually for this function. */ +#if 0 + if (return_trap && ((trace_p (var) == 0) && function_trace_mode == 0)) +#else + if (return_trap && (signal_in_progress (DEBUG_TRAP) || ((trace_p (var) == 0) && function_trace_mode == 0))) +#endif + { + if (subshell == 0) + { + return_trap = savestring (return_trap); + add_unwind_protect (xfree, return_trap); + add_unwind_protect (set_return_trap, return_trap); + } + restore_default_signal (RETURN_TRAP); + } + + funcnest++; +#if defined (ARRAY_VARS) + /* This is quite similar to the code in shell.c and elsewhere. */ + shell_fn = find_function_def (this_shell_function->name); + sfile = shell_fn ? shell_fn->source_file : ""; + array_push ((ARRAY *)funcname_a, this_shell_function->name); + + array_push ((ARRAY *)bash_source_a, sfile); + t = itos (executing_line_number ()); + array_push ((ARRAY *)bash_lineno_a, t); + free (t); +#endif + + /* The temporary environment for a function is supposed to apply to + all commands executed within the function body. */ + + remember_args (words->next, 1); + + /* Update BASH_ARGV and BASH_ARGC */ + if (debugging_mode) + push_args (words->next); + + /* Number of the line on which the function body starts. */ + line_number = function_line_number = tc->line; + +#if defined (JOB_CONTROL) + if (subshell) + stop_pipeline (async, (COMMAND *)NULL); +#endif + + fc = tc; + + return_catch_flag++; + return_val = setjmp (return_catch); + + if (return_val) + { + result = return_catch_value; + /* Run the RETURN trap in the function's context. */ + save_current = currently_executing_command; + run_return_trap (); + currently_executing_command = save_current; + } + else + { + /* Run the debug trap here so we can trap at the start of a function's + execution rather than the execution of the body's first command. */ + showing_function_line = 1; + save_current = currently_executing_command; + result = run_debug_trap (); +#if defined (DEBUGGER) + /* In debugging mode, if the DEBUG trap returns a non-zero status, we + skip the command. */ + if (debugging_mode == 0 || result == EXECUTION_SUCCESS) + { + showing_function_line = 0; + currently_executing_command = save_current; + result = execute_command_internal (fc, 0, NO_PIPE, NO_PIPE, fds_to_close); + + /* Run the RETURN trap in the function's context */ + save_current = currently_executing_command; + run_return_trap (); + currently_executing_command = save_current; + } +#else + result = execute_command_internal (fc, 0, NO_PIPE, NO_PIPE, fds_to_close); + + save_current = currently_executing_command; + run_return_trap (); + currently_executing_command = save_current; +#endif + showing_function_line = 0; + } + + /* Restore BASH_ARGC and BASH_ARGV */ + if (debugging_mode) + pop_args (); + + if (subshell == 0) + run_unwind_frame ("function_calling"); + +#if defined (ARRAY_VARS) + /* These two variables cannot be unset, and cannot be affected by the + function. */ + array_pop ((ARRAY *)bash_source_a); + array_pop ((ARRAY *)bash_lineno_a); + + /* FUNCNAME can be unset, and so can potentially be changed by the + function. */ + GET_ARRAY_FROM_VAR ("FUNCNAME", nfv, funcname_a); + if (nfv == funcname_v) + array_pop (funcname_a); +#endif + + if (variable_context == 0 || this_shell_function == 0) + { + make_funcname_visible (0); +#if defined (PROCESS_SUBSTITUTION) + unlink_fifo_list (); +#endif + } + + return (result); +} + +/* A convenience routine for use by other parts of the shell to execute + a particular shell function. */ +int +execute_shell_function (var, words) + SHELL_VAR *var; + WORD_LIST *words; +{ + int ret; + struct fd_bitmap *bitmap; + + bitmap = new_fd_bitmap (FD_BITMAP_DEFAULT_SIZE); + begin_unwind_frame ("execute-shell-function"); + add_unwind_protect (dispose_fd_bitmap, (char *)bitmap); + + ret = execute_function (var, words, 0, bitmap, 0, 0); + + dispose_fd_bitmap (bitmap); + discard_unwind_frame ("execute-shell-function"); + + return ret; +} + +/* Execute a shell builtin or function in a subshell environment. This + routine does not return; it only calls exit(). If BUILTIN is non-null, + it points to a function to call to execute a shell builtin; otherwise + VAR points at the body of a function to execute. WORDS is the arguments + to the command, REDIRECTS specifies redirections to perform before the + command is executed. */ +static void +execute_subshell_builtin_or_function (words, redirects, builtin, var, + pipe_in, pipe_out, async, fds_to_close, + flags) + WORD_LIST *words; + REDIRECT *redirects; + sh_builtin_func_t *builtin; + SHELL_VAR *var; + int pipe_in, pipe_out, async; + struct fd_bitmap *fds_to_close; + int flags; +{ + int result, r, funcvalue; +#if defined (JOB_CONTROL) + int jobs_hack; + + jobs_hack = (builtin == jobs_builtin) && + ((subshell_environment & SUBSHELL_ASYNC) == 0 || pipe_out != NO_PIPE); +#endif + + /* A subshell is neither a login shell nor interactive. */ + login_shell = interactive = 0; + + if (async) + subshell_environment |= SUBSHELL_ASYNC; + if (pipe_in != NO_PIPE || pipe_out != NO_PIPE) + subshell_environment |= SUBSHELL_PIPE; + + maybe_make_export_env (); /* XXX - is this needed? */ + +#if defined (JOB_CONTROL) + /* Eradicate all traces of job control after we fork the subshell, so + all jobs begun by this subshell are in the same process group as + the shell itself. */ + + /* Allow the output of `jobs' to be piped. */ + if (jobs_hack) + kill_current_pipeline (); + else + without_job_control (); + + set_sigchld_handler (); +#endif /* JOB_CONTROL */ + + set_sigint_handler (); + + if (fds_to_close) + close_fd_bitmap (fds_to_close); + + do_piping (pipe_in, pipe_out); + + if (do_redirections (redirects, RX_ACTIVE) != 0) + exit (EXECUTION_FAILURE); + + if (builtin) + { + /* Give builtins a place to jump back to on failure, + so we don't go back up to main(). */ + result = setjmp (top_level); + + /* Give the return builtin a place to jump to when executed in a subshell + or pipeline */ + funcvalue = 0; + if (return_catch_flag && builtin == return_builtin) + funcvalue = setjmp (return_catch); + + if (result == EXITPROG) + exit (last_command_exit_value); + else if (result) + exit (EXECUTION_FAILURE); + else if (funcvalue) + exit (return_catch_value); + else + { + r = execute_builtin (builtin, words, flags, 1); + fflush (stdout); + if (r == EX_USAGE) + r = EX_BADUSAGE; + exit (r); + } + } + else + { + r = execute_function (var, words, flags, fds_to_close, async, 1); + fflush (stdout); + exit (r); + } +} + +/* Execute a builtin or function in the current shell context. If BUILTIN + is non-null, it is the builtin command to execute, otherwise VAR points + to the body of a function. WORDS are the command's arguments, REDIRECTS + are the redirections to perform. FDS_TO_CLOSE is the usual bitmap of + file descriptors to close. + + If BUILTIN is exec_builtin, the redirections specified in REDIRECTS are + not undone before this function returns. */ +static int +execute_builtin_or_function (words, builtin, var, redirects, + fds_to_close, flags) + WORD_LIST *words; + sh_builtin_func_t *builtin; + SHELL_VAR *var; + REDIRECT *redirects; + struct fd_bitmap *fds_to_close; + int flags; +{ + int result; + REDIRECT *saved_undo_list; +#if defined (PROCESS_SUBSTITUTION) + int ofifo, nfifo, osize; + char *ofifo_list; +#endif + + +#if defined (PROCESS_SUBSTITUTION) + ofifo = num_fifos (); + ofifo_list = copy_fifo_list (&osize); +#endif + + if (do_redirections (redirects, RX_ACTIVE|RX_UNDOABLE) != 0) + { + cleanup_redirects (redirection_undo_list); + redirection_undo_list = (REDIRECT *)NULL; + dispose_exec_redirects (); +#if defined (PROCESS_SUBSTITUTION) + free (ofifo_list); +#endif + return (EX_REDIRFAIL); /* was EXECUTION_FAILURE */ + } + + saved_undo_list = redirection_undo_list; + + /* Calling the "exec" builtin changes redirections forever. */ + if (builtin == exec_builtin) + { + dispose_redirects (saved_undo_list); + saved_undo_list = exec_redirection_undo_list; + exec_redirection_undo_list = (REDIRECT *)NULL; + } + else + dispose_exec_redirects (); + + if (saved_undo_list) + { + begin_unwind_frame ("saved redirects"); + add_unwind_protect (cleanup_redirects, (char *)saved_undo_list); + } + + redirection_undo_list = (REDIRECT *)NULL; + + if (builtin) + result = execute_builtin (builtin, words, flags, 0); + else + result = execute_function (var, words, flags, fds_to_close, 0, 0); + + /* We do this before undoing the effects of any redirections. */ + fflush (stdout); + fpurge (stdout); + if (ferror (stdout)) + clearerr (stdout); + + /* If we are executing the `command' builtin, but this_shell_builtin is + set to `exec_builtin', we know that we have something like + `command exec [redirection]', since otherwise `exec' would have + overwritten the shell and we wouldn't get here. In this case, we + want to behave as if the `command' builtin had not been specified + and preserve the redirections. */ + if (builtin == command_builtin && this_shell_builtin == exec_builtin) + { + int discard; + + discard = 0; + if (saved_undo_list) + { + dispose_redirects (saved_undo_list); + discard = 1; + } + redirection_undo_list = exec_redirection_undo_list; + saved_undo_list = exec_redirection_undo_list = (REDIRECT *)NULL; + if (discard) + discard_unwind_frame ("saved redirects"); + } + + if (saved_undo_list) + { + redirection_undo_list = saved_undo_list; + discard_unwind_frame ("saved redirects"); + } + + if (redirection_undo_list) + { + cleanup_redirects (redirection_undo_list); + redirection_undo_list = (REDIRECT *)NULL; + } + +#if defined (PROCESS_SUBSTITUTION) + /* Close any FIFOs created by this builtin or function. */ + nfifo = num_fifos (); + if (nfifo > ofifo) + close_new_fifos (ofifo_list, osize); + free (ofifo_list); +#endif + + return (result); +} + +void +setup_async_signals () +{ +#if defined (__BEOS__) + set_signal_handler (SIGHUP, SIG_IGN); /* they want csh-like behavior */ +#endif + +#if defined (JOB_CONTROL) + if (job_control == 0) +#endif + { + set_signal_handler (SIGINT, SIG_IGN); + set_signal_ignored (SIGINT); + set_signal_handler (SIGQUIT, SIG_IGN); + set_signal_ignored (SIGQUIT); + } +} + +/* Execute a simple command that is hopefully defined in a disk file + somewhere. + + 1) fork () + 2) connect pipes + 3) look up the command + 4) do redirections + 5) execve () + 6) If the execve failed, see if the file has executable mode set. + If so, and it isn't a directory, then execute its contents as + a shell script. + + Note that the filename hashing stuff has to take place up here, + in the parent. This is probably why the Bourne style shells + don't handle it, since that would require them to go through + this gnarly hair, for no good reason. + + NOTE: callers expect this to fork or exit(). */ + +/* Name of a shell function to call when a command name is not found. */ +#ifndef NOTFOUND_HOOK +# define NOTFOUND_HOOK "command_not_found_handle" +#endif + +static int +execute_disk_command (words, redirects, command_line, pipe_in, pipe_out, + async, fds_to_close, cmdflags) + WORD_LIST *words; + REDIRECT *redirects; + char *command_line; + int pipe_in, pipe_out, async; + struct fd_bitmap *fds_to_close; + int cmdflags; +{ + char *pathname, *command, **args; + int nofork, result; + pid_t pid; + SHELL_VAR *hookf; + WORD_LIST *wl; + + nofork = (cmdflags & CMD_NO_FORK); /* Don't fork, just exec, if no pipes */ + pathname = words->word->word; + + result = EXECUTION_SUCCESS; +#if defined (RESTRICTED_SHELL) + command = (char *)NULL; + if (restricted && mbschr (pathname, '/')) + { + internal_error (_("%s: restricted: cannot specify `/' in command names"), + pathname); + result = last_command_exit_value = EXECUTION_FAILURE; + + /* If we're not going to fork below, we must already be in a child + process or a context in which it's safe to call exit(2). */ + if (nofork && pipe_in == NO_PIPE && pipe_out == NO_PIPE) + exit (last_command_exit_value); + else + goto parent_return; + } +#endif /* RESTRICTED_SHELL */ + + command = search_for_command (pathname, 1); + + if (command) + { + maybe_make_export_env (); + put_command_name_into_env (command); + } + + /* We have to make the child before we check for the non-existence + of COMMAND, since we want the error messages to be redirected. */ + /* If we can get away without forking and there are no pipes to deal with, + don't bother to fork, just directly exec the command. */ + if (nofork && pipe_in == NO_PIPE && pipe_out == NO_PIPE) + pid = 0; + else + pid = make_child (savestring (command_line), async); + + if (pid == 0) + { + int old_interactive; + +#if 0 + /* This has been disabled for the time being. */ +#if !defined (ARG_MAX) || ARG_MAX >= 10240 + if (posixly_correct == 0) + put_gnu_argv_flags_into_env ((long)getpid (), glob_argv_flags); +#endif +#endif + + reset_terminating_signals (); /* XXX */ + /* Cancel traps, in trap.c. */ + restore_original_signals (); + + /* restore_original_signals may have undone the work done + by make_child to ensure that SIGINT and SIGQUIT are ignored + in asynchronous children. */ + if (async) + { + if ((cmdflags & CMD_STDIN_REDIR) && + pipe_in == NO_PIPE && + (stdin_redirects (redirects) == 0)) + async_redirect_stdin (); + setup_async_signals (); + } + + /* This functionality is now provided by close-on-exec of the + file descriptors manipulated by redirection and piping. + Some file descriptors still need to be closed in all children + because of the way bash does pipes; fds_to_close is a + bitmap of all such file descriptors. */ + if (fds_to_close) + close_fd_bitmap (fds_to_close); + + do_piping (pipe_in, pipe_out); + + old_interactive = interactive; + if (async) + interactive = 0; + + subshell_environment = SUBSHELL_FORK; + + if (redirects && (do_redirections (redirects, RX_ACTIVE) != 0)) + { +#if defined (PROCESS_SUBSTITUTION) + /* Try to remove named pipes that may have been created as the + result of redirections. */ + unlink_fifo_list (); +#endif /* PROCESS_SUBSTITUTION */ + exit (EXECUTION_FAILURE); + } + + if (async) + interactive = old_interactive; + + if (command == 0) + { +itrace("not found"); + hookf = find_function (NOTFOUND_HOOK); + if (hookf == 0) + { + /* Make sure filenames are displayed using printable characters */ + if (ansic_shouldquote (pathname)) + pathname = ansic_quote (pathname, 0, NULL); + internal_error (_("%s: command not found"), pathname); + exit (EX_NOTFOUND); /* Posix.2 says the exit status is 127 */ + } + + wl = make_word_list (make_word (NOTFOUND_HOOK), words); + exit (execute_shell_function (hookf, wl)); + } + + /* Execve expects the command name to be in args[0]. So we + leave it there, in the same format that the user used to + type it in. */ + args = strvec_from_word_list (words, 0, 0, (int *)NULL); + exit (shell_execve (command, args, export_env)); + } + else + { +parent_return: + QUIT; + + /* Make sure that the pipes are closed in the parent. */ + close_pipes (pipe_in, pipe_out); +#if defined (PROCESS_SUBSTITUTION) && defined (HAVE_DEV_FD) + if (variable_context == 0) + unlink_fifo_list (); +#endif + FREE (command); + return (result); + } +} + +/* CPP defines to decide whether a particular index into the #! line + corresponds to a valid interpreter name or argument character, or + whitespace. The MSDOS define is to allow \r to be treated the same + as \n. */ + +#if !defined (MSDOS) +# define STRINGCHAR(ind) \ + (ind < sample_len && !whitespace (sample[ind]) && sample[ind] != '\n') +# define WHITECHAR(ind) \ + (ind < sample_len && whitespace (sample[ind])) +#else /* MSDOS */ +# define STRINGCHAR(ind) \ + (ind < sample_len && !whitespace (sample[ind]) && sample[ind] != '\n' && sample[ind] != '\r') +# define WHITECHAR(ind) \ + (ind < sample_len && whitespace (sample[ind])) +#endif /* MSDOS */ + +static char * +getinterp (sample, sample_len, endp) + char *sample; + int sample_len, *endp; +{ + register int i; + char *execname; + int start; + + /* Find the name of the interpreter to exec. */ + for (i = 2; i < sample_len && whitespace (sample[i]); i++) + ; + + for (start = i; STRINGCHAR(i); i++) + ; + + execname = substring (sample, start, i); + + if (endp) + *endp = i; + return execname; +} + +#if !defined (HAVE_HASH_BANG_EXEC) +/* If the operating system on which we're running does not handle + the #! executable format, then help out. SAMPLE is the text read + from the file, SAMPLE_LEN characters. COMMAND is the name of + the script; it and ARGS, the arguments given by the user, will + become arguments to the specified interpreter. ENV is the environment + to pass to the interpreter. + + The word immediately following the #! is the interpreter to execute. + A single argument to the interpreter is allowed. */ + +static int +execute_shell_script (sample, sample_len, command, args, env) + char *sample; + int sample_len; + char *command; + char **args, **env; +{ + char *execname, *firstarg; + int i, start, size_increment, larry; + + /* Find the name of the interpreter to exec. */ + execname = getinterp (sample, sample_len, &i); + size_increment = 1; + + /* Now the argument, if any. */ + for (firstarg = (char *)NULL, start = i; WHITECHAR(i); i++) + ; + + /* If there is more text on the line, then it is an argument for the + interpreter. */ + + if (STRINGCHAR(i)) + { + for (start = i; STRINGCHAR(i); i++) + ; + firstarg = substring ((char *)sample, start, i); + size_increment = 2; + } + + larry = strvec_len (args) + size_increment; + args = strvec_resize (args, larry + 1); + + for (i = larry - 1; i; i--) + args[i] = args[i - size_increment]; + + args[0] = execname; + if (firstarg) + { + args[1] = firstarg; + args[2] = command; + } + else + args[1] = command; + + args[larry] = (char *)NULL; + + return (shell_execve (execname, args, env)); +} +#undef STRINGCHAR +#undef WHITECHAR + +#endif /* !HAVE_HASH_BANG_EXEC */ + +static void +initialize_subshell () +{ +#if defined (ALIAS) + /* Forget about any aliases that we knew of. We are in a subshell. */ + delete_all_aliases (); +#endif /* ALIAS */ + +#if defined (HISTORY) + /* Forget about the history lines we have read. This is a non-interactive + subshell. */ + history_lines_this_session = 0; +#endif + +#if defined (JOB_CONTROL) + /* Forget about the way job control was working. We are in a subshell. */ + without_job_control (); + set_sigchld_handler (); + init_job_stats (); +#endif /* JOB_CONTROL */ + + /* Reset the values of the shell flags and options. */ + reset_shell_flags (); + reset_shell_options (); + reset_shopt_options (); + + /* Zero out builtin_env, since this could be a shell script run from a + sourced file with a temporary environment supplied to the `source/.' + builtin. Such variables are not supposed to be exported (empirical + testing with sh and ksh). Just throw it away; don't worry about a + memory leak. */ + if (vc_isbltnenv (shell_variables)) + shell_variables = shell_variables->down; + + clear_unwind_protect_list (0); + /* XXX -- are there other things we should be resetting here? */ + parse_and_execute_level = 0; /* nothing left to restore it */ + + /* We're no longer inside a shell function. */ + variable_context = return_catch_flag = funcnest = 0; + + executing_list = 0; /* XXX */ + + /* If we're not interactive, close the file descriptor from which we're + reading the current shell script. */ + if (interactive_shell == 0) + unset_bash_input (0); +} + +#if defined (HAVE_SETOSTYPE) && defined (_POSIX_SOURCE) +# define SETOSTYPE(x) __setostype(x) +#else +# define SETOSTYPE(x) +#endif + +#define READ_SAMPLE_BUF(file, buf, len) \ + do \ + { \ + fd = open(file, O_RDONLY); \ + if (fd >= 0) \ + { \ + len = read (fd, buf, 80); \ + close (fd); \ + } \ + else \ + len = -1; \ + } \ + while (0) + +/* Call execve (), handling interpreting shell scripts, and handling + exec failures. */ +int +shell_execve (command, args, env) + char *command; + char **args, **env; +{ + int larray, i, fd; + char sample[80]; + int sample_len; + + SETOSTYPE (0); /* Some systems use for USG/POSIX semantics */ + execve (command, args, env); + i = errno; /* error from execve() */ + CHECK_TERMSIG; + SETOSTYPE (1); + + /* If we get to this point, then start checking out the file. + Maybe it is something we can hack ourselves. */ + if (i != ENOEXEC) + { + if (file_isdir (command)) +#if defined (EISDIR) + internal_error (_("%s: %s"), command, strerror (EISDIR)); +#else + internal_error (_("%s: is a directory"), command); +#endif + else if (executable_file (command) == 0) + { + errno = i; + file_error (command); + } + /* errors not involving the path argument to execve. */ + else if (i == E2BIG || i == ENOMEM) + { + errno = i; + file_error (command); + } + else + { + /* The file has the execute bits set, but the kernel refuses to + run it for some reason. See why. */ +#if defined (HAVE_HASH_BANG_EXEC) + READ_SAMPLE_BUF (command, sample, sample_len); + sample[sample_len - 1] = '\0'; + if (sample_len > 2 && sample[0] == '#' && sample[1] == '!') + { + char *interp; + int ilen; + + interp = getinterp (sample, sample_len, (int *)NULL); + ilen = strlen (interp); + errno = i; + if (interp[ilen - 1] == '\r') + { + interp = xrealloc (interp, ilen + 2); + interp[ilen - 1] = '^'; + interp[ilen] = 'M'; + interp[ilen + 1] = '\0'; + } + sys_error (_("%s: %s: bad interpreter"), command, interp ? interp : ""); + FREE (interp); + return (EX_NOEXEC); + } +#endif + errno = i; + file_error (command); + } + return ((i == ENOENT) ? EX_NOTFOUND : EX_NOEXEC); /* XXX Posix.2 says that exit status is 126 */ + } + + /* This file is executable. + If it begins with #!, then help out people with losing operating + systems. Otherwise, check to see if it is a binary file by seeing + if the contents of the first line (or up to 80 characters) are in the + ASCII set. If it's a text file, execute the contents as shell commands, + otherwise return 126 (EX_BINARY_FILE). */ + READ_SAMPLE_BUF (command, sample, sample_len); + + if (sample_len == 0) + return (EXECUTION_SUCCESS); + + /* Is this supposed to be an executable script? + If so, the format of the line is "#! interpreter [argument]". + A single argument is allowed. The BSD kernel restricts + the length of the entire line to 32 characters (32 bytes + being the size of the BSD exec header), but we allow 80 + characters. */ + if (sample_len > 0) + { +#if !defined (HAVE_HASH_BANG_EXEC) + if (sample_len > 2 && sample[0] == '#' && sample[1] == '!') + return (execute_shell_script (sample, sample_len, command, args, env)); + else +#endif + if (check_binary_file (sample, sample_len)) + { + internal_error (_("%s: cannot execute binary file: %s"), command, strerror (i)); + return (EX_BINARY_FILE); + } + } + + /* We have committed to attempting to execute the contents of this file + as shell commands. */ + + initialize_subshell (); + + set_sigint_handler (); + + /* Insert the name of this shell into the argument list. */ + larray = strvec_len (args) + 1; + args = strvec_resize (args, larray + 1); + + for (i = larray - 1; i; i--) + args[i] = args[i - 1]; + + args[0] = shell_name; + args[1] = command; + args[larray] = (char *)NULL; + + if (args[0][0] == '-') + args[0]++; + +#if defined (RESTRICTED_SHELL) + if (restricted) + change_flag ('r', FLAG_OFF); +#endif + + if (subshell_argv) + { + /* Can't free subshell_argv[0]; that is shell_name. */ + for (i = 1; i < subshell_argc; i++) + free (subshell_argv[i]); + free (subshell_argv); + } + + dispose_command (currently_executing_command); /* XXX */ + currently_executing_command = (COMMAND *)NULL; + + subshell_argc = larray; + subshell_argv = args; + subshell_envp = env; + + unbind_args (); /* remove the positional parameters */ + + longjmp (subshell_top_level, 1); + /*NOTREACHED*/ +} + +static int +execute_intern_function (name, funcdef) + WORD_DESC *name; + FUNCTION_DEF *funcdef; +{ + SHELL_VAR *var; + + if (check_identifier (name, posixly_correct) == 0) + { + if (posixly_correct && interactive_shell == 0) + { + last_command_exit_value = EX_BADUSAGE; + jump_to_top_level (ERREXIT); + } + return (EXECUTION_FAILURE); + } + + /* Posix interpretation 383 */ + if (posixly_correct && find_special_builtin (name->word)) + { + internal_error (_("`%s': is a special builtin"), name->word); + last_command_exit_value = EX_BADUSAGE; + jump_to_top_level (ERREXIT); + } + + var = find_function (name->word); + if (var && (readonly_p (var) || noassign_p (var))) + { + if (readonly_p (var)) + internal_error (_("%s: readonly function"), var->name); + return (EXECUTION_FAILURE); + } + +#if defined (DEBUGGER) + bind_function_def (name->word, funcdef); +#endif + + bind_function (name->word, funcdef->command); + return (EXECUTION_SUCCESS); +} + +#if defined (INCLUDE_UNUSED) +#if defined (PROCESS_SUBSTITUTION) +void +close_all_files () +{ + register int i, fd_table_size; + + fd_table_size = getdtablesize (); + if (fd_table_size > 256) /* clamp to a reasonable value */ + fd_table_size = 256; + + for (i = 3; i < fd_table_size; i++) + close (i); +} +#endif /* PROCESS_SUBSTITUTION */ +#endif + +static void +close_pipes (in, out) + int in, out; +{ + if (in >= 0) + close (in); + if (out >= 0) + close (out); +} + +static void +dup_error (oldd, newd) + int oldd, newd; +{ + sys_error (_("cannot duplicate fd %d to fd %d"), oldd, newd); +} + +/* Redirect input and output to be from and to the specified pipes. + NO_PIPE and REDIRECT_BOTH are handled correctly. */ +static void +do_piping (pipe_in, pipe_out) + int pipe_in, pipe_out; +{ + if (pipe_in != NO_PIPE) + { + if (dup2 (pipe_in, 0) < 0) + dup_error (pipe_in, 0); + if (pipe_in > 0) + close (pipe_in); +#ifdef __CYGWIN__ + /* Let stdio know the fd may have changed from text to binary mode. */ + freopen (NULL, "r", stdin); +#endif /* __CYGWIN__ */ + } + if (pipe_out != NO_PIPE) + { + if (pipe_out != REDIRECT_BOTH) + { + if (dup2 (pipe_out, 1) < 0) + dup_error (pipe_out, 1); + if (pipe_out == 0 || pipe_out > 1) + close (pipe_out); + } + else + { + if (dup2 (1, 2) < 0) + dup_error (1, 2); + } +#ifdef __CYGWIN__ + /* Let stdio know the fd may have changed from text to binary mode, and + make sure to preserve stdout line buffering. */ + freopen (NULL, "w", stdout); + sh_setlinebuf (stdout); +#endif /* __CYGWIN__ */ + } +} diff --git a/externs.h b/externs.h index 57f7117f5..834a4ecde 100644 --- a/externs.h +++ b/externs.h @@ -463,6 +463,7 @@ extern unsigned int fsleep __P((unsigned int, unsigned int)); /* declarations for functions defined in lib/sh/unicode.c */ extern int u32cconv __P((unsigned long, char *)); +extern void u32reset __P((void)); /* declarations for functions defined in lib/sh/winsize.c */ extern void get_new_window_size __P((int, int *, int *)); diff --git a/externs.h~ b/externs.h~ new file mode 100644 index 000000000..57f7117f5 --- /dev/null +++ b/externs.h~ @@ -0,0 +1,500 @@ +/* externs.h -- extern function declarations which do not appear in their + own header file. */ + +/* Copyright (C) 1993-2010 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +/* Make sure that this is included *after* config.h! */ + +#if !defined (_EXTERNS_H_) +# define _EXTERNS_H_ + +#include "stdc.h" + +/* Functions from expr.c. */ +extern intmax_t evalexp __P((char *, int *)); + +/* Functions from print_cmd.c. */ +#define FUNC_MULTILINE 0x01 +#define FUNC_EXTERNAL 0x02 + +extern char *make_command_string __P((COMMAND *)); +extern char *named_function_string __P((char *, COMMAND *, int)); + +extern void print_command __P((COMMAND *)); +extern void print_simple_command __P((SIMPLE_COM *)); +extern void print_word_list __P((WORD_LIST *, char *)); + +/* debugger support */ +extern void print_for_command_head __P((FOR_COM *)); +#if defined (SELECT_COMMAND) +extern void print_select_command_head __P((SELECT_COM *)); +#endif +extern void print_case_command_head __P((CASE_COM *)); +#if defined (DPAREN_ARITHMETIC) +extern void print_arith_command __P((WORD_LIST *)); +#endif +#if defined (COND_COMMAND) +extern void print_cond_command __P((COND_COM *)); +#endif + +/* set -x support */ +extern void xtrace_init __P((void)); +#ifdef NEED_XTRACE_SET_DECL +extern void xtrace_set __P((int, FILE *)); +#endif +extern void xtrace_fdchk __P((int)); +extern void xtrace_reset __P((void)); +extern char *indirection_level_string __P((void)); +extern void xtrace_print_assignment __P((char *, char *, int, int)); +extern void xtrace_print_word_list __P((WORD_LIST *, int)); +extern void xtrace_print_for_command_head __P((FOR_COM *)); +#if defined (SELECT_COMMAND) +extern void xtrace_print_select_command_head __P((SELECT_COM *)); +#endif +extern void xtrace_print_case_command_head __P((CASE_COM *)); +#if defined (DPAREN_ARITHMETIC) +extern void xtrace_print_arith_cmd __P((WORD_LIST *)); +#endif +#if defined (COND_COMMAND) +extern void xtrace_print_cond_term __P((int, int, WORD_DESC *, char *, char *)); +#endif + +/* Functions from shell.c. */ +extern void exit_shell __P((int)) __attribute__((__noreturn__)); +extern void sh_exit __P((int)) __attribute__((__noreturn__)); +extern void disable_priv_mode __P((void)); +extern void unbind_args __P((void)); + +#if defined (RESTRICTED_SHELL) +extern int shell_is_restricted __P((char *)); +extern int maybe_make_restricted __P((char *)); +#endif + +extern void unset_bash_input __P((int)); +extern void get_current_user_info __P((void)); + +/* Functions from eval.c. */ +extern int reader_loop __P((void)); +extern int parse_command __P((void)); +extern int read_command __P((void)); + +/* Functions from braces.c. */ +#if defined (BRACE_EXPANSION) +extern char **brace_expand __P((char *)); +#endif + +/* Miscellaneous functions from parse.y */ +extern int yyparse __P((void)); +extern int return_EOF __P((void)); +extern char *xparse_dolparen __P((char *, char *, int *, int)); +extern void reset_parser __P((void)); +extern WORD_LIST *parse_string_to_word_list __P((char *, int, const char *)); + +extern int parser_in_command_position __P((void)); + +extern void free_pushed_string_input __P((void)); + +extern char *decode_prompt_string __P((char *)); + +extern int get_current_prompt_level __P((void)); +extern void set_current_prompt_level __P((int)); + +#if defined (HISTORY) +extern char *history_delimiting_chars __P((const char *)); +#endif + +/* Declarations for functions defined in locale.c */ +extern void set_default_locale __P((void)); +extern void set_default_locale_vars __P((void)); +extern int set_locale_var __P((char *, char *)); +extern int set_lang __P((char *, char *)); +extern void set_default_lang __P((void)); +extern char *get_locale_var __P((char *)); +extern char *localetrans __P((char *, int, int *)); +extern char *mk_msgstr __P((char *, int *)); +extern char *localeexpand __P((char *, int, int, int, int *)); + +/* Declarations for functions defined in list.c. */ +extern void list_walk __P((GENERIC_LIST *, sh_glist_func_t *)); +extern void wlist_walk __P((WORD_LIST *, sh_icpfunc_t *)); +extern GENERIC_LIST *list_reverse (); +extern int list_length (); +extern GENERIC_LIST *list_append (); +extern GENERIC_LIST *list_remove (); + +/* Declarations for functions defined in stringlib.c */ +extern int find_string_in_alist __P((char *, STRING_INT_ALIST *, int)); +extern char *find_token_in_alist __P((int, STRING_INT_ALIST *, int)); +extern int find_index_in_alist __P((char *, STRING_INT_ALIST *, int)); + +extern char *substring __P((const char *, int, int)); +extern char *strsub __P((char *, char *, char *, int)); +extern char *strcreplace __P((char *, int, char *, int)); +extern void strip_leading __P((char *)); +extern void strip_trailing __P((char *, int, int)); +extern void xbcopy __P((char *, char *, int)); + +/* Functions from version.c. */ +extern char *shell_version_string __P((void)); +extern void show_shell_version __P((int)); + +/* Functions from the bash library, lib/sh/libsh.a. These should really + go into a separate include file. */ + +/* declarations for functions defined in lib/sh/casemod.c */ +extern char *sh_modcase __P((const char *, char *, int)); + +/* Defines for flags argument to sh_modcase. These need to agree with what's + in lib/sh/casemode.c */ +#define CASE_LOWER 0x0001 +#define CASE_UPPER 0x0002 +#define CASE_CAPITALIZE 0x0004 +#define CASE_UNCAP 0x0008 +#define CASE_TOGGLE 0x0010 +#define CASE_TOGGLEALL 0x0020 +#define CASE_UPFIRST 0x0040 +#define CASE_LOWFIRST 0x0080 + +#define CASE_USEWORDS 0x1000 + +/* declarations for functions defined in lib/sh/clktck.c */ +extern long get_clk_tck __P((void)); + +/* declarations for functions defined in lib/sh/clock.c */ +extern void clock_t_to_secs (); +extern void print_clock_t (); + +/* Declarations for functions defined in lib/sh/dprintf.c */ +#if !defined (HAVE_DPRINTF) +extern void dprintf __P((int, const char *, ...)) __attribute__((__format__ (printf, 2, 3))); +#endif + +/* Declarations for functions defined in lib/sh/fmtulong.c */ +#define FL_PREFIX 0x01 /* add 0x, 0X, or 0 prefix as appropriate */ +#define FL_ADDBASE 0x02 /* add base# prefix to converted value */ +#define FL_HEXUPPER 0x04 /* use uppercase when converting to hex */ +#define FL_UNSIGNED 0x08 /* don't add any sign */ + +extern char *fmtulong __P((unsigned long int, int, char *, size_t, int)); + +/* Declarations for functions defined in lib/sh/fmtulong.c */ +#if defined (HAVE_LONG_LONG) +extern char *fmtullong __P((unsigned long long int, int, char *, size_t, int)); +#endif + +/* Declarations for functions defined in lib/sh/fmtumax.c */ +extern char *fmtumax __P((uintmax_t, int, char *, size_t, int)); + +/* Declarations for functions defined in lib/sh/fnxform.c */ +extern char *fnx_fromfs __P((char *, size_t)); +extern char *fnx_tofs __P((char *, size_t)); + +/* Declarations for functions defined in lib/sh/fpurge.c */ + +#if defined NEED_FPURGE_DECL +#if !HAVE_DECL_FPURGE + +#if HAVE_FPURGE +# define fpurge _bash_fpurge +#endif +extern int fpurge __P((FILE *stream)); + +#endif /* HAVE_DECL_FPURGE */ +#endif /* NEED_FPURGE_DECL */ + +/* Declarations for functions defined in lib/sh/getcwd.c */ +#if !defined (HAVE_GETCWD) +extern char *getcwd __P((char *, size_t)); +#endif + +/* Declarations for functions defined in lib/sh/input_avail.c */ +extern int input_avail __P((int)); + +/* Declarations for functions defined in lib/sh/itos.c */ +extern char *inttostr __P((intmax_t, char *, size_t)); +extern char *itos __P((intmax_t)); +extern char *uinttostr __P((uintmax_t, char *, size_t)); +extern char *uitos __P((uintmax_t)); + +/* declarations for functions defined in lib/sh/makepath.c */ +#define MP_DOTILDE 0x01 +#define MP_DOCWD 0x02 +#define MP_RMDOT 0x04 +#define MP_IGNDOT 0x08 + +extern char *sh_makepath __P((const char *, const char *, int)); + +/* declarations for functions defined in lib/sh/mbscasecmp.c */ +#if !defined (HAVE_MBSCASECMP) +extern char *mbscasecmp __P((const char *, const char *)); +#endif + +/* declarations for functions defined in lib/sh/mbschr.c */ +#if !defined (HAVE_MBSCHR) +extern char *mbschr __P((const char *, int)); +#endif + +/* declarations for functions defined in lib/sh/mbscmp.c */ +#if !defined (HAVE_MBSCMP) +extern char *mbscmp __P((const char *, const char *)); +#endif + +/* declarations for functions defined in lib/sh/netconn.c */ +extern int isnetconn __P((int)); + +/* declarations for functions defined in lib/sh/netopen.c */ +extern int netopen __P((char *)); + +/* Declarations for functions defined in lib/sh/oslib.c */ + +#if !defined (HAVE_DUP2) || defined (DUP2_BROKEN) +extern int dup2 __P((int, int)); +#endif + +#if !defined (HAVE_GETDTABLESIZE) +extern int getdtablesize __P((void)); +#endif /* !HAVE_GETDTABLESIZE */ + +#if !defined (HAVE_GETHOSTNAME) +extern int gethostname __P((char *, int)); +#endif /* !HAVE_GETHOSTNAME */ + +extern int getmaxgroups __P((void)); +extern long getmaxchild __P((void)); + +/* declarations for functions defined in lib/sh/pathcanon.c */ +#define PATH_CHECKDOTDOT 0x0001 +#define PATH_CHECKEXISTS 0x0002 +#define PATH_HARDPATH 0x0004 +#define PATH_NOALLOC 0x0008 + +extern char *sh_canonpath __P((char *, int)); + +/* declarations for functions defined in lib/sh/pathphys.c */ +extern char *sh_physpath __P((char *, int)); +extern char *sh_realpath __P((const char *, char *)); + +/* declarations for functions defined in lib/sh/setlinebuf.c */ +#ifdef NEED_SH_SETLINEBUF_DECL +extern int sh_setlinebuf __P((FILE *)); +#endif + +/* declarations for functions defined in lib/sh/shaccess.c */ +extern int sh_eaccess __P((char *, int)); + +/* declarations for functions defined in lib/sh/shmatch.c */ +extern int sh_regmatch __P((const char *, const char *, int)); + +/* defines for flags argument to sh_regmatch. */ +#define SHMAT_SUBEXP 0x001 /* save subexpressions in SH_REMATCH */ +#define SHMAT_PWARN 0x002 /* print a warning message on invalid regexp */ + +/* declarations for functions defined in lib/sh/shmbchar.c */ +extern size_t mbstrlen __P((const char *)); +extern char *mbsmbchar __P((const char *)); + +/* declarations for functions defined in lib/sh/shquote.c */ +extern char *sh_single_quote __P((const char *)); +extern char *sh_double_quote __P((const char *)); +extern char *sh_mkdoublequoted __P((const char *, int, int)); +extern char *sh_un_double_quote __P((char *)); +extern char *sh_backslash_quote __P((char *, const char *)); +extern char *sh_backslash_quote_for_double_quotes __P((char *)); +extern int sh_contains_shell_metas __P((char *)); + +/* declarations for functions defined in lib/sh/spell.c */ +extern int spname __P((char *, char *)); +extern char *dirspell __P((char *)); + +/* declarations for functions defined in lib/sh/strcasecmp.c */ +#if !defined (HAVE_STRCASECMP) +extern int strncasecmp __P((const char *, const char *, int)); +extern int strcasecmp __P((const char *, const char *)); +#endif /* HAVE_STRCASECMP */ + +/* declarations for functions defined in lib/sh/strcasestr.c */ +#if ! HAVE_STRCASESTR +extern char *strcasestr __P((const char *, const char *)); +#endif + +/* declarations for functions defined in lib/sh/strchrnul.c */ +#if ! HAVE_STRCHRNUL +extern char *strchrnul __P((const char *, int)); +#endif + +/* declarations for functions defined in lib/sh/strerror.c */ +#if !defined (HAVE_STRERROR) && !defined (strerror) +extern char *strerror __P((int)); +#endif + +/* declarations for functions defined in lib/sh/strftime.c */ +#if !defined (HAVE_STRFTIME) && defined (NEED_STRFTIME_DECL) +extern size_t strftime __P((char *, size_t, const char *, const struct tm *)); +#endif + +/* declarations for functions and structures defined in lib/sh/stringlist.c */ + +/* This is a general-purpose argv-style array struct. */ +typedef struct _list_of_strings { + char **list; + int list_size; + int list_len; +} STRINGLIST; + +typedef int sh_strlist_map_func_t __P((char *)); + +extern STRINGLIST *strlist_create __P((int)); +extern STRINGLIST *strlist_resize __P((STRINGLIST *, int)); +extern void strlist_flush __P((STRINGLIST *)); +extern void strlist_dispose __P((STRINGLIST *)); +extern int strlist_remove __P((STRINGLIST *, char *)); +extern STRINGLIST *strlist_copy __P((STRINGLIST *)); +extern STRINGLIST *strlist_merge __P((STRINGLIST *, STRINGLIST *)); +extern STRINGLIST *strlist_append __P((STRINGLIST *, STRINGLIST *)); +extern STRINGLIST *strlist_prefix_suffix __P((STRINGLIST *, char *, char *)); +extern void strlist_print __P((STRINGLIST *, char *)); +extern void strlist_walk __P((STRINGLIST *, sh_strlist_map_func_t *)); +extern void strlist_sort __P((STRINGLIST *)); + +/* declarations for functions defined in lib/sh/stringvec.c */ + +extern char **strvec_create __P((int)); +extern char **strvec_resize __P((char **, int)); +extern void strvec_flush __P((char **)); +extern void strvec_dispose __P((char **)); +extern int strvec_remove __P((char **, char *)); +extern int strvec_len __P((char **)); +extern int strvec_search __P((char **, char *)); +extern char **strvec_copy __P((char **)); +extern int strvec_strcmp __P((char **, char **)); +extern void strvec_sort __P((char **)); + +extern char **strvec_from_word_list __P((WORD_LIST *, int, int, int *)); +extern WORD_LIST *strvec_to_word_list __P((char **, int, int)); + +/* declarations for functions defined in lib/sh/strnlen.c */ +#if !defined (HAVE_STRNLEN) +extern size_t strnlen __P((const char *, size_t)); +#endif + +/* declarations for functions defined in lib/sh/strpbrk.c */ +#if !defined (HAVE_STRPBRK) +extern char *strpbrk __P((const char *, const char *)); +#endif + +/* declarations for functions defined in lib/sh/strtod.c */ +#if !defined (HAVE_STRTOD) +extern double strtod __P((const char *, char **)); +#endif + +/* declarations for functions defined in lib/sh/strtol.c */ +#if !HAVE_DECL_STRTOL +extern long strtol __P((const char *, char **, int)); +#endif + +/* declarations for functions defined in lib/sh/strtoll.c */ +#if defined (HAVE_LONG_LONG) && !HAVE_DECL_STRTOLL +extern long long strtoll __P((const char *, char **, int)); +#endif + +/* declarations for functions defined in lib/sh/strtoul.c */ +#if !HAVE_DECL_STRTOUL +extern unsigned long strtoul __P((const char *, char **, int)); +#endif + +/* declarations for functions defined in lib/sh/strtoull.c */ +#if defined (HAVE_LONG_LONG) && !HAVE_DECL_STRTOULL +extern unsigned long long strtoull __P((const char *, char **, int)); +#endif + +/* declarations for functions defined in lib/sh/strimax.c */ +#if !HAVE_DECL_STRTOIMAX +extern intmax_t strtoimax __P((const char *, char **, int)); +#endif + +/* declarations for functions defined in lib/sh/strumax.c */ +#if !HAVE_DECL_STRTOUMAX +extern uintmax_t strtoumax __P((const char *, char **, int)); +#endif + +/* declarations for functions defined in lib/sh/strtrans.c */ +extern char *ansicstr __P((char *, int, int, int *, int *)); +extern char *ansic_quote __P((char *, int, int *)); +extern int ansic_shouldquote __P((const char *)); +extern char *ansiexpand __P((char *, int, int, int *)); + +/* declarations for functions defined in lib/sh/timeval.c. No prototypes + so we don't have to count on having a definition of struct timeval in + scope when this file is included. */ +extern void timeval_to_secs (); +extern void print_timeval (); + +/* declarations for functions defined in lib/sh/tmpfile.c */ +#define MT_USETMPDIR 0x0001 +#define MT_READWRITE 0x0002 +#define MT_USERANDOM 0x0004 + +extern char *sh_mktmpname __P((char *, int)); +extern int sh_mktmpfd __P((char *, int, char **)); +/* extern FILE *sh_mktmpfp __P((char *, int, char **)); */ + +/* declarations for functions defined in lib/sh/uconvert.c */ +extern int uconvert __P((char *, long *, long *)); + +/* declarations for functions defined in lib/sh/ufuncs.c */ +extern unsigned int falarm __P((unsigned int, unsigned int)); +extern unsigned int fsleep __P((unsigned int, unsigned int)); + +/* declarations for functions defined in lib/sh/unicode.c */ +extern int u32cconv __P((unsigned long, char *)); + +/* declarations for functions defined in lib/sh/winsize.c */ +extern void get_new_window_size __P((int, int *, int *)); + +/* declarations for functions defined in lib/sh/zcatfd.c */ +extern int zcatfd __P((int, int, char *)); + +/* declarations for functions defined in lib/sh/zgetline.c */ +extern ssize_t zgetline __P((int, char **, size_t *, int)); + +/* declarations for functions defined in lib/sh/zmapfd.c */ +extern int zmapfd __P((int, char **, char *)); + +/* declarations for functions defined in lib/sh/zread.c */ +extern ssize_t zread __P((int, char *, size_t)); +extern ssize_t zreadretry __P((int, char *, size_t)); +extern ssize_t zreadintr __P((int, char *, size_t)); +extern ssize_t zreadc __P((int, char *)); +extern ssize_t zreadcintr __P((int, char *)); +extern void zreset __P((void)); +extern void zsyncfd __P((int)); + +/* declarations for functions defined in lib/sh/zwrite.c */ +extern int zwrite __P((int, char *, size_t)); + +/* declarations for functions defined in lib/glob/gmisc.c */ +extern int match_pattern_char __P((char *, char *)); +extern int umatchlen __P((char *, size_t)); + +#if defined (HANDLE_MULTIBYTE) +extern int match_pattern_wchar __P((wchar_t *, wchar_t *)); +extern int wmatchlen __P((wchar_t *, size_t)); +#endif + +#endif /* _EXTERNS_H_ */ diff --git a/lib/readline/util.c b/lib/readline/util.c index 1271e26fd..5882371a7 100644 --- a/lib/readline/util.c +++ b/lib/readline/util.c @@ -533,6 +533,9 @@ _rl_settracefp (fp) #if HAVE_DECL_AUDIT_USER_TTY && defined (ENABLE_TTY_AUDIT_SUPPORT) +#include +#include +#include /* Report STRING to the audit system. */ void diff --git a/lib/sh/strtrans.c b/lib/sh/strtrans.c index 676548c92..8691c054f 100644 --- a/lib/sh/strtrans.c +++ b/lib/sh/strtrans.c @@ -60,7 +60,10 @@ ansicstr (string, len, flags, sawc, rlen) return ((char *)NULL); #if defined (HANDLE_MULTIBYTE) - ret = (char *)xmalloc (4*len + 1); + if (strstr (string, "\\U") != 0) + ret = (char *)xmalloc (6*len + 1); + else + ret = (char *)xmalloc (4*len + 1); #else ret = (char *)xmalloc (2*len + 1); /* 2*len for possible CTLESC */ #endif @@ -147,7 +150,7 @@ ansicstr (string, len, flags, sawc, rlen) *r++ = '\\'; /* c remains unchanged */ break; } - else if (v <= UCHAR_MAX) + else if (v <= 0x7f) /* <= 0x7f translates directly */ { c = v; break; diff --git a/lib/sh/strtrans.c~ b/lib/sh/strtrans.c~ new file mode 100644 index 000000000..79878c65c --- /dev/null +++ b/lib/sh/strtrans.c~ @@ -0,0 +1,374 @@ +/* strtrans.c - Translate and untranslate strings with ANSI-C escape sequences. */ + +/* Copyright (C) 2000-2011 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include + +#if defined (HAVE_UNISTD_H) +# include +#endif + +#include +#include +#include + +#include "shell.h" + +#include "shmbchar.h" +#include "shmbutil.h" + +#ifdef ESC +#undef ESC +#endif +#define ESC '\033' /* ASCII */ + +/* Convert STRING by expanding the escape sequences specified by the + ANSI C standard. If SAWC is non-null, recognize `\c' and use that + as a string terminator. If we see \c, set *SAWC to 1 before + returning. LEN is the length of STRING. If (FLAGS&1) is non-zero, + that we're translating a string for `echo -e', and therefore should not + treat a single quote as a character that may be escaped with a backslash. + If (FLAGS&2) is non-zero, we're expanding for the parser and want to + quote CTLESC and CTLNUL with CTLESC. If (flags&4) is non-zero, we want + to remove the backslash before any unrecognized escape sequence. */ +char * +ansicstr (string, len, flags, sawc, rlen) + char *string; + int len, flags, *sawc, *rlen; +{ + int c, temp; + char *ret, *r, *s; + unsigned long v; + + if (string == 0 || *string == '\0') + return ((char *)NULL); + +#if defined (HANDLE_MULTIBYTE) + ret = (char *)xmalloc (6*len + 1); +#else + ret = (char *)xmalloc (2*len + 1); /* 2*len for possible CTLESC */ +#endif + for (r = ret, s = string; s && *s; ) + { + c = *s++; + if (c != '\\' || *s == '\0') + *r++ = c; + else + { + switch (c = *s++) + { +#if defined (__STDC__) + case 'a': c = '\a'; break; + case 'v': c = '\v'; break; +#else + case 'a': c = (int) 0x07; break; + case 'v': c = (int) 0x0B; break; +#endif + case 'b': c = '\b'; break; + case 'e': case 'E': /* ESC -- non-ANSI */ + c = ESC; break; + case 'f': c = '\f'; break; + case 'n': c = '\n'; break; + case 'r': c = '\r'; break; + case 't': c = '\t'; break; + case '1': case '2': case '3': + case '4': case '5': case '6': + case '7': +#if 1 + if (flags & 1) + { + *r++ = '\\'; + break; + } + /*FALLTHROUGH*/ +#endif + case '0': + /* If (FLAGS & 1), we're translating a string for echo -e (or + the equivalent xpg_echo option), so we obey the SUSv3/ + POSIX-2001 requirement and accept 0-3 octal digits after + a leading `0'. */ + temp = 2 + ((flags & 1) && (c == '0')); + for (c -= '0'; ISOCTAL (*s) && temp--; s++) + c = (c * 8) + OCTVALUE (*s); + c &= 0xFF; + break; + case 'x': /* Hex digit -- non-ANSI */ + if ((flags & 2) && *s == '{') + { + flags |= 16; /* internal flag value */ + s++; + } + /* Consume at least two hex characters */ + for (temp = 2, c = 0; ISXDIGIT ((unsigned char)*s) && temp--; s++) + c = (c * 16) + HEXVALUE (*s); + /* DGK says that after a `\x{' ksh93 consumes ISXDIGIT chars + until a non-xdigit or `}', so potentially more than two + chars are consumed. */ + if (flags & 16) + { + for ( ; ISXDIGIT ((unsigned char)*s); s++) + c = (c * 16) + HEXVALUE (*s); + flags &= ~16; + if (*s == '}') + s++; + } + /* \x followed by non-hex digits is passed through unchanged */ + else if (temp == 2) + { + *r++ = '\\'; + c = 'x'; + } + c &= 0xFF; + break; +#if defined (HANDLE_MULTIBYTE) + case 'u': + case 'U': + temp = (c == 'u') ? 4 : 8; /* \uNNNN \UNNNNNNNN */ + for (v = 0; ISXDIGIT ((unsigned char)*s) && temp--; s++) + v = (v * 16) + HEXVALUE (*s); + if (temp == ((c == 'u') ? 4 : 8)) + { + *r++ = '\\'; /* c remains unchanged */ + break; + } + else if (v <= 0x7f) /* <= 0x7f translates directly */ + { + c = v; + break; + } + else + { + temp = u32cconv (v, r); + r += temp; + continue; + } +#endif + case '\\': + break; + case '\'': case '"': case '?': + if (flags & 1) + *r++ = '\\'; + break; + case 'c': + if (sawc) + { + *sawc = 1; + *r = '\0'; + if (rlen) + *rlen = r - ret; + return ret; + } + else if ((flags & 1) == 0 && *s == 0) + ; /* pass \c through */ + else if ((flags & 1) == 0 && (c = *s)) + { + s++; + if ((flags & 2) && c == '\\' && c == *s) + s++; /* Posix requires $'\c\\' do backslash escaping */ + c = TOCTRL(c); + break; + } + /*FALLTHROUGH*/ + default: + if ((flags & 4) == 0) + *r++ = '\\'; + break; + } + if ((flags & 2) && (c == CTLESC || c == CTLNUL)) + *r++ = CTLESC; + *r++ = c; + } + } + *r = '\0'; + if (rlen) + *rlen = r - ret; + return ret; +} + +/* Take a string STR, possibly containing non-printing characters, and turn it + into a $'...' ANSI-C style quoted string. Returns a new string. */ +char * +ansic_quote (str, flags, rlen) + char *str; + int flags, *rlen; +{ + char *r, *ret, *s; + int l, rsize, sindex; + unsigned char c; + size_t slen, clen; +#if defined (HANDLE_MULTIBYTE) + int b; + wchar_t wc; +#endif + + if (str == 0 || *str == 0) + return ((char *)0); + + l = strlen (str); + rsize = 4 * l + 4; + r = ret = (char *)xmalloc (rsize); + + *r++ = '$'; + *r++ = '\''; + + s = str; + slen = strlen (str); + + for (s = str; c = *s; s++) + { + l = 1; /* 1 == add backslash; 0 == no backslash */ + clen = 1; + + switch (c) + { + case ESC: c = 'E'; break; +#ifdef __STDC__ + case '\a': c = 'a'; break; + case '\v': c = 'v'; break; +#else + case 0x07: c = 'a'; break; + case 0x0b: c = 'v'; break; +#endif + + case '\b': c = 'b'; break; + case '\f': c = 'f'; break; + case '\n': c = 'n'; break; + case '\r': c = 'r'; break; + case '\t': c = 't'; break; + case '\\': + case '\'': + break; + default: +#if defined (HANDLE_MULTIBYTE) + b = is_basic (c); + if ((b == 0 && ((clen = mbrtowc (&wc, s, MB_CUR_MAX, 0)) < 0 || iswprint (wc) == 0)) || + (b == 1 && ISPRINT (c) == 0)) +#else + if (ISPRINT (c) == 0) +#endif + { + *r++ = '\\'; + *r++ = TOCHAR ((c >> 6) & 07); + *r++ = TOCHAR ((c >> 3) & 07); + *r++ = TOCHAR (c & 07); + continue; + } + l = 0; + break; + } + if (l) + *r++ = '\\'; + + if (clen == 1) + *r++ = c; + else + for (b = 0; b < (int)clen; c = b ? *++s : c) + *r++ = c; + } + + *r++ = '\''; + *r = '\0'; + if (rlen) + *rlen = r - ret; + return ret; +} + +#if defined (HANDLE_MULTIBYTE) +int +ansic_wshouldquote (string) + const char *string; +{ + const wchar_t *wcs; + wchar_t wcc; + + wchar_t *wcstr = NULL; + size_t wclen, slen; + + + slen = mbstowcs (wcstr, string, 0); + + if (slen == -1) + slen = 0; + wcstr = (wchar_t *)xmalloc (sizeof (wchar_t) * (slen + 1)); + mbstowcs (wcstr, string, slen + 1); + + for (wcs = wcstr; wcc = *wcs; wcs++) + if (iswprint(wcc) == 0) + { + free (wcstr); + return 1; + } + + free (wcstr); + return 0; +} +#endif + +/* return 1 if we need to quote with $'...' because of non-printing chars. */ +int +ansic_shouldquote (string) + const char *string; +{ + const char *s; + unsigned char c; + + if (string == 0) + return 0; + + for (s = string; c = *s; s++) + { +#if defined (HANDLE_MULTIBYTE) + if (is_basic (c) == 0) + return (ansic_wshouldquote (s)); +#endif + if (ISPRINT (c) == 0) + return 1; + } + + return 0; +} + +/* $'...' ANSI-C expand the portion of STRING between START and END and + return the result. The result cannot be longer than the input string. */ +char * +ansiexpand (string, start, end, lenp) + char *string; + int start, end, *lenp; +{ + char *temp, *t; + int len, tlen; + + temp = (char *)xmalloc (end - start + 1); + for (tlen = 0, len = start; len < end; ) + temp[tlen++] = string[len++]; + temp[tlen] = '\0'; + + if (*temp) + { + t = ansicstr (temp, tlen, 2, (int *)NULL, lenp); + free (temp); + return (t); + } + else + { + if (lenp) + *lenp = 0; + return (temp); + } +} diff --git a/lib/sh/unicode.c b/lib/sh/unicode.c index d34fa08a4..5bb4702c9 100644 --- a/lib/sh/unicode.c +++ b/lib/sh/unicode.c @@ -84,6 +84,13 @@ stub_charset () } #endif +void +u32reset () +{ + u32init = 0; + utf8locale = 0; +} + /* u32toascii ? */ int u32tochar (wc, s) @@ -114,28 +121,61 @@ u32tochar (wc, s) return l; } +/* Convert unsigned 32-bit int to utf-8 character string */ int u32toutf8 (wc, s) - wchar_t wc; + u_bits32_t wc; char *s; { int l; - l = (wc < 0x0080) ? 1 : ((wc < 0x0800) ? 2 : 3); - if (wc < 0x0080) - s[0] = (unsigned char)wc; + { + s[0] = (char)wc; + l = 1; + } else if (wc < 0x0800) { s[0] = (wc >> 6) | 0xc0; s[1] = (wc & 0x3f) | 0x80; + l = 2; } - else + else if (wc < 0x10000) { s[0] = (wc >> 12) | 0xe0; s[1] = ((wc >> 6) & 0x3f) | 0x80; s[2] = (wc & 0x3f) | 0x80; + l = 3; } + else if (wc < 0x200000) + { + s[0] = (wc >> 18) | 0xf0; + s[1] = ((wc >> 12) & 0x3f) | 0x80; + s[2] = ((wc >> 6) & 0x3f) | 0x80; + s[3] = (wc & 0x3f) | 0x80; + l = 4; + } + /* Strictly speaking, UTF-8 doesn't have characters longer than 4 bytes */ + else if (wc < 0x04000000) + { + s[0] = (wc >> 24) | 0xf8; + s[1] = ((wc >> 18) & 0x3f) | 0x80; + s[2] = ((wc >> 12) & 0x3f) | 0x80; + s[3] = ((wc >> 6) & 0x3f) | 0x80; + s[4] = (wc & 0x3f) | 0x80; + l = 5; + } + else + { + s[0] = (wc >> 30) | 0xf8; + s[1] = ((wc >> 24) & 0x3f) | 0x80; + s[2] = ((wc >> 18) & 0x3f) | 0x80; + s[3] = ((wc >> 12) & 0x3f) | 0x80; + s[4] = ((wc >> 6) & 0x3f) | 0x80; + s[5] = (wc & 0x3f) | 0x80; + l = 6; + } + s[l] = '\0'; return l; } @@ -171,7 +211,7 @@ u32cconv (c, s) codeset = nl_langinfo (CODESET); if (STREQ (codeset, "UTF-8")) { - n = u32toutf8 (wc, s); + n = u32toutf8 (c, s); return n; } #endif @@ -198,7 +238,7 @@ u32cconv (c, s) if (utf8locale) { - n = u32toutf8 (wc, s); + n = u32toutf8 (c, s); return n; } @@ -208,7 +248,7 @@ u32cconv (c, s) return n; } - n = u32toutf8 (wc, s); + n = u32toutf8 (c, s); optr = obuf; obytesleft = sizeof (obuf); diff --git a/locale.c b/locale.c index 105555914..a452ac527 100644 --- a/locale.c +++ b/locale.c @@ -103,6 +103,7 @@ set_default_locale_vars () setlocale (LC_CTYPE, lc_all); locale_setblanks (); locale_mb_cur_max = MB_CUR_MAX; + u32reset (); } # endif @@ -206,6 +207,7 @@ set_locale_var (var, value) } locale_setblanks (); locale_mb_cur_max = MB_CUR_MAX; + u32reset (); return r; #else return (1); @@ -221,6 +223,7 @@ set_locale_var (var, value) x = setlocale (LC_CTYPE, get_locale_var ("LC_CTYPE")); locale_setblanks (); locale_mb_cur_max = MB_CUR_MAX; + u32reset (); } # endif } @@ -353,6 +356,7 @@ reset_locale_vars () locale_setblanks (); locale_mb_cur_max = MB_CUR_MAX; + u32reset (); #endif return 1; diff --git a/locale.c~ b/locale.c~ new file mode 100644 index 000000000..105555914 --- /dev/null +++ b/locale.c~ @@ -0,0 +1,563 @@ +/* locale.c - Miscellaneous internationalization functions. */ + +/* Copyright (C) 1996-2009 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include "config.h" + +#include "bashtypes.h" + +#if defined (HAVE_UNISTD_H) +# include +#endif + +#if HAVE_LANGINFO_CODESET +# include +#endif + +#include "bashintl.h" +#include "bashansi.h" +#include +#include "chartypes.h" +#include + +#include "shell.h" +#include "input.h" /* For bash_input */ + +#ifndef errno +extern int errno; +#endif + +int locale_utf8locale; /* unused for now */ +int locale_mb_cur_max; /* value of MB_CUR_MAX for current locale (LC_CTYPE) */ + +extern int dump_translatable_strings, dump_po_strings; + +/* The current locale when the program begins */ +static char *default_locale; + +/* The current domain for textdomain(3). */ +static char *default_domain; +static char *default_dir; + +/* tracks the value of LC_ALL; used to override values for other locale + categories */ +static char *lc_all; + +/* tracks the value of LC_ALL; used to provide defaults for locale + categories */ +static char *lang; + +/* Called to reset all of the locale variables to their appropriate values + if (and only if) LC_ALL has not been assigned a value. */ +static int reset_locale_vars __P((void)); + +static void locale_setblanks __P((void)); +static int locale_isutf8 __P((char *)); + +/* Set the value of default_locale and make the current locale the + system default locale. This should be called very early in main(). */ +void +set_default_locale () +{ +#if defined (HAVE_SETLOCALE) + default_locale = setlocale (LC_ALL, ""); + if (default_locale) + default_locale = savestring (default_locale); +#endif /* HAVE_SETLOCALE */ + bindtextdomain (PACKAGE, LOCALEDIR); + textdomain (PACKAGE); + + locale_mb_cur_max = MB_CUR_MAX; +} + +/* Set default values for LC_CTYPE, LC_COLLATE, LC_MESSAGES, LC_NUMERIC and + LC_TIME if they are not specified in the environment, but LC_ALL is. This + should be called from main() after parsing the environment. */ +void +set_default_locale_vars () +{ + char *val; + +#if defined (HAVE_SETLOCALE) + +# if defined (LC_CTYPE) + val = get_string_value ("LC_CTYPE"); + if (val == 0 && lc_all && *lc_all) + { + setlocale (LC_CTYPE, lc_all); + locale_setblanks (); + locale_mb_cur_max = MB_CUR_MAX; + } +# endif + +# if defined (LC_COLLATE) + val = get_string_value ("LC_COLLATE"); + if (val == 0 && lc_all && *lc_all) + setlocale (LC_COLLATE, lc_all); +# endif /* LC_COLLATE */ + +# if defined (LC_MESSAGES) + val = get_string_value ("LC_MESSAGES"); + if (val == 0 && lc_all && *lc_all) + setlocale (LC_MESSAGES, lc_all); +# endif /* LC_MESSAGES */ + +# if defined (LC_NUMERIC) + val = get_string_value ("LC_NUMERIC"); + if (val == 0 && lc_all && *lc_all) + setlocale (LC_NUMERIC, lc_all); +# endif /* LC_NUMERIC */ + +# if defined (LC_TIME) + val = get_string_value ("LC_TIME"); + if (val == 0 && lc_all && *lc_all) + setlocale (LC_TIME, lc_all); +# endif /* LC_TIME */ + +#endif /* HAVE_SETLOCALE */ + + val = get_string_value ("TEXTDOMAIN"); + if (val && *val) + { + FREE (default_domain); + default_domain = savestring (val); +#if 0 + /* Don't want to override the shell's textdomain as the default */ + textdomain (default_domain); +#endif + } + + val = get_string_value ("TEXTDOMAINDIR"); + if (val && *val) + { + FREE (default_dir); + default_dir = savestring (val); + if (default_domain && *default_domain) + bindtextdomain (default_domain, default_dir); + } +} + +/* Set one of the locale categories (specified by VAR) to VALUE. Returns 1 + if successful, 0 otherwise. */ +int +set_locale_var (var, value) + char *var, *value; +{ + int r; + char *x; + + x = ""; + errno = 0; + if (var[0] == 'T' && var[10] == 0) /* TEXTDOMAIN */ + { + FREE (default_domain); + default_domain = value ? savestring (value) : (char *)NULL; +#if 0 + /* Don't want to override the shell's textdomain as the default */ + textdomain (default_domain); +#endif + return (1); + } + else if (var[0] == 'T') /* TEXTDOMAINDIR */ + { + FREE (default_dir); + default_dir = value ? savestring (value) : (char *)NULL; + if (default_domain && *default_domain) + bindtextdomain (default_domain, default_dir); + return (1); + } + + /* var[0] == 'L' && var[1] == 'C' && var[2] == '_' */ + + else if (var[3] == 'A') /* LC_ALL */ + { + FREE (lc_all); + if (value) + lc_all = savestring (value); + else + { + lc_all = (char *)xmalloc (1); + lc_all[0] = '\0'; + } +#if defined (HAVE_SETLOCALE) + r = *lc_all ? ((x = setlocale (LC_ALL, lc_all)) != 0) : reset_locale_vars (); + if (x == 0) + { + if (errno == 0) + internal_warning(_("setlocale: LC_ALL: cannot change locale (%s)"), lc_all); + else + internal_warning(_("setlocale: LC_ALL: cannot change locale (%s): %s"), lc_all, strerror (errno)); + } + locale_setblanks (); + locale_mb_cur_max = MB_CUR_MAX; + return r; +#else + return (1); +#endif + } + +#if defined (HAVE_SETLOCALE) + else if (var[3] == 'C' && var[4] == 'T') /* LC_CTYPE */ + { +# if defined (LC_CTYPE) + if (lc_all == 0 || *lc_all == '\0') + { + x = setlocale (LC_CTYPE, get_locale_var ("LC_CTYPE")); + locale_setblanks (); + locale_mb_cur_max = MB_CUR_MAX; + } +# endif + } + else if (var[3] == 'C' && var[4] == 'O') /* LC_COLLATE */ + { +# if defined (LC_COLLATE) + if (lc_all == 0 || *lc_all == '\0') + x = setlocale (LC_COLLATE, get_locale_var ("LC_COLLATE")); +# endif /* LC_COLLATE */ + } + else if (var[3] == 'M' && var[4] == 'E') /* LC_MESSAGES */ + { +# if defined (LC_MESSAGES) + if (lc_all == 0 || *lc_all == '\0') + x = setlocale (LC_MESSAGES, get_locale_var ("LC_MESSAGES")); +# endif /* LC_MESSAGES */ + } + else if (var[3] == 'N' && var[4] == 'U') /* LC_NUMERIC */ + { +# if defined (LC_NUMERIC) + if (lc_all == 0 || *lc_all == '\0') + x = setlocale (LC_NUMERIC, get_locale_var ("LC_NUMERIC")); +# endif /* LC_NUMERIC */ + } + else if (var[3] == 'T' && var[4] == 'I') /* LC_TIME */ + { +# if defined (LC_TIME) + if (lc_all == 0 || *lc_all == '\0') + x = setlocale (LC_TIME, get_locale_var ("LC_TIME")); +# endif /* LC_TIME */ + } +#endif /* HAVE_SETLOCALE */ + + if (x == 0) + { + if (errno == 0) + internal_warning(_("setlocale: %s: cannot change locale (%s)"), var, get_locale_var (var)); + else + internal_warning(_("setlocale: %s: cannot change locale (%s): %s"), var, get_locale_var (var), strerror (errno)); + } + + return (x != 0); +} + +/* Called when LANG is assigned a value. Tracks value in `lang'. Calls + reset_locale_vars() to reset any default values if LC_ALL is unset or + null. */ +int +set_lang (var, value) + char *var, *value; +{ + FREE (lang); + if (value) + lang = savestring (value); + else + { + lang = (char *)xmalloc (1); + lang[0] = '\0'; + } + + return ((lc_all == 0 || *lc_all == 0) ? reset_locale_vars () : 0); +} + +/* Set default values for LANG and LC_ALL. Default values for all other + locale-related variables depend on these. */ +void +set_default_lang () +{ + char *v; + + v = get_string_value ("LC_ALL"); + set_locale_var ("LC_ALL", v); + + v = get_string_value ("LANG"); + set_lang ("LANG", v); +} + +/* Get the value of one of the locale variables (LC_MESSAGES, LC_CTYPE). + The precedence is as POSIX.2 specifies: LC_ALL has precedence over + the specific locale variables, and LANG, if set, is used as the default. */ +char * +get_locale_var (var) + char *var; +{ + char *locale; + + locale = lc_all; + + if (locale == 0 || *locale == 0) + locale = get_string_value (var); /* XXX - mem leak? */ + if (locale == 0 || *locale == 0) + locale = lang; + if (locale == 0 || *locale == 0) +#if 0 + locale = default_locale; /* system-dependent; not really portable. should it be "C"? */ +#else + locale = ""; +#endif + return (locale); +} + +/* Called to reset all of the locale variables to their appropriate values + if (and only if) LC_ALL has not been assigned a value. DO NOT CALL THIS + IF LC_ALL HAS BEEN ASSIGNED A VALUE. */ +static int +reset_locale_vars () +{ + char *t; +#if defined (HAVE_SETLOCALE) + if (lang == 0 || *lang == '\0') + maybe_make_export_env (); /* trust that this will change environment for setlocale */ + if (setlocale (LC_ALL, lang ? lang : "") == 0) + return 0; + +# if defined (LC_CTYPE) + t = setlocale (LC_CTYPE, get_locale_var ("LC_CTYPE")); +# endif +# if defined (LC_COLLATE) + t = setlocale (LC_COLLATE, get_locale_var ("LC_COLLATE")); +# endif +# if defined (LC_MESSAGES) + t = setlocale (LC_MESSAGES, get_locale_var ("LC_MESSAGES")); +# endif +# if defined (LC_NUMERIC) + t = setlocale (LC_NUMERIC, get_locale_var ("LC_NUMERIC")); +# endif +# if defined (LC_TIME) + t = setlocale (LC_TIME, get_locale_var ("LC_TIME")); +# endif + + locale_setblanks (); + locale_mb_cur_max = MB_CUR_MAX; + +#endif + return 1; +} + +/* Translate the contents of STRING, a $"..." quoted string, according + to the current locale. In the `C' or `POSIX' locale, or if gettext() + is not available, the passed string is returned unchanged. The + length of the translated string is returned in LENP, if non-null. */ +char * +localetrans (string, len, lenp) + char *string; + int len, *lenp; +{ + char *locale, *t; + char *translated; + int tlen; + + /* Don't try to translate null strings. */ + if (string == 0 || *string == 0) + { + if (lenp) + *lenp = 0; + return ((char *)NULL); + } + + locale = get_locale_var ("LC_MESSAGES"); + + /* If we don't have setlocale() or the current locale is `C' or `POSIX', + just return the string. If we don't have gettext(), there's no use + doing anything else. */ + if (locale == 0 || locale[0] == '\0' || + (locale[0] == 'C' && locale[1] == '\0') || STREQ (locale, "POSIX")) + { + t = (char *)xmalloc (len + 1); + strcpy (t, string); + if (lenp) + *lenp = len; + return (t); + } + + /* Now try to translate it. */ + if (default_domain && *default_domain) + translated = dgettext (default_domain, string); + else + translated = string; + + if (translated == string) /* gettext returns its argument if untranslatable */ + { + t = (char *)xmalloc (len + 1); + strcpy (t, string); + if (lenp) + *lenp = len; + } + else + { + tlen = strlen (translated); + t = (char *)xmalloc (tlen + 1); + strcpy (t, translated); + if (lenp) + *lenp = tlen; + } + return (t); +} + +/* Change a bash string into a string suitable for inclusion in a `po' file. + This backslash-escapes `"' and `\' and changes newlines into \\\n"\n". */ +char * +mk_msgstr (string, foundnlp) + char *string; + int *foundnlp; +{ + register int c, len; + char *result, *r, *s; + + for (len = 0, s = string; s && *s; s++) + { + len++; + if (*s == '"' || *s == '\\') + len++; + else if (*s == '\n') + len += 5; + } + + r = result = (char *)xmalloc (len + 3); + *r++ = '"'; + + for (s = string; s && (c = *s); s++) + { + if (c == '\n') /* -> \n"" */ + { + *r++ = '\\'; + *r++ = 'n'; + *r++ = '"'; + *r++ = '\n'; + *r++ = '"'; + if (foundnlp) + *foundnlp = 1; + continue; + } + if (c == '"' || c == '\\') + *r++ = '\\'; + *r++ = c; + } + + *r++ = '"'; + *r++ = '\0'; + + return result; +} + +/* $"..." -- Translate the portion of STRING between START and END + according to current locale using gettext (if available) and return + the result. The caller will take care of leaving the quotes intact. + The string will be left without the leading `$' by the caller. + If translation is performed, the translated string will be double-quoted + by the caller. The length of the translated string is returned in LENP, + if non-null. */ +char * +localeexpand (string, start, end, lineno, lenp) + char *string; + int start, end, lineno, *lenp; +{ + int len, tlen, foundnl; + char *temp, *t, *t2; + + temp = (char *)xmalloc (end - start + 1); + for (tlen = 0, len = start; len < end; ) + temp[tlen++] = string[len++]; + temp[tlen] = '\0'; + + /* If we're just dumping translatable strings, don't do anything with the + string itself, but if we're dumping in `po' file format, convert it into + a form more palatable to gettext(3) and friends by quoting `"' and `\' + with backslashes and converting into `\n""'. If we find a + newline in TEMP, we first output a `msgid ""' line and then the + translated string; otherwise we output the `msgid' and translated + string all on one line. */ + if (dump_translatable_strings) + { + if (dump_po_strings) + { + foundnl = 0; + t = mk_msgstr (temp, &foundnl); + t2 = foundnl ? "\"\"\n" : ""; + + printf ("#: %s:%d\nmsgid %s%s\nmsgstr \"\"\n", + yy_input_name (), lineno, t2, t); + free (t); + } + else + printf ("\"%s\"\n", temp); + + if (lenp) + *lenp = tlen; + return (temp); + } + else if (*temp) + { + t = localetrans (temp, tlen, &len); + free (temp); + if (lenp) + *lenp = len; + return (t); + } + else + { + if (lenp) + *lenp = 0; + return (temp); + } +} + +/* Set every character in the character class to be a shell break + character for the lexical analyzer when the locale changes. */ +static void +locale_setblanks () +{ + int x; + + for (x = 0; x < sh_syntabsiz; x++) + { + if (isblank (x)) + sh_syntaxtab[x] |= CSHBRK|CBLANK; + else if (member (x, shell_break_chars)) + { + sh_syntaxtab[x] |= CSHBRK; + sh_syntaxtab[x] &= ~CBLANK; + } + else + sh_syntaxtab[x] &= ~(CSHBRK|CBLANK); + } +} + +static int +locale_isutf8 (lspec) + char *lspec; +{ + char *cp; + +#if HAVE_LANGINFO_CODESET + cp = nl_langinfo (CODESET); + return (STREQ (cp, "UTF-8") || STREQ (cp, "utf8")); +#else + /* Take a shot */ + return (strstr (lspec, "UTF-8") || strstr (lspec, "utf8")); +#endif +} diff --git a/parse.y b/parse.y index 40f3ee068..45f687f93 100644 --- a/parse.y +++ b/parse.y @@ -4670,10 +4670,10 @@ got_token: last_read_token == GREATER_AND)) { if (legal_number (token, &lvalue) && (int)lvalue == lvalue) - yylval.number = lvalue; - else - yylval.number = -1; - return (NUMBER); + { + yylval.number = lvalue; + return (NUMBER); + } } /* Check for special case tokens. */ diff --git a/parse.y~ b/parse.y~ new file mode 100644 index 000000000..d57f78309 --- /dev/null +++ b/parse.y~ @@ -0,0 +1,6086 @@ +/* parse.y - Yacc grammar for bash. */ + +/* Copyright (C) 1989-2010 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +%{ +#include "config.h" + +#include "bashtypes.h" +#include "bashansi.h" + +#include "filecntl.h" + +#if defined (HAVE_UNISTD_H) +# include +#endif + +#if defined (HAVE_LOCALE_H) +# include +#endif + +#include +#include "chartypes.h" +#include + +#include "memalloc.h" + +#include "bashintl.h" + +#define NEED_STRFTIME_DECL /* used in externs.h */ + +#include "shell.h" +#include "trap.h" +#include "flags.h" +#include "parser.h" +#include "mailcheck.h" +#include "test.h" +#include "builtins.h" +#include "builtins/common.h" +#include "builtins/builtext.h" + +#include "shmbutil.h" + +#if defined (READLINE) +# include "bashline.h" +# include +#endif /* READLINE */ + +#if defined (HISTORY) +# include "bashhist.h" +# include +#endif /* HISTORY */ + +#if defined (JOB_CONTROL) +# include "jobs.h" +#endif /* JOB_CONTROL */ + +#if defined (ALIAS) +# include "alias.h" +#else +typedef void *alias_t; +#endif /* ALIAS */ + +#if defined (PROMPT_STRING_DECODE) +# ifndef _MINIX +# include +# endif +# include +# if defined (TM_IN_SYS_TIME) +# include +# include +# endif /* TM_IN_SYS_TIME */ +# include "maxpath.h" +#endif /* PROMPT_STRING_DECODE */ + +#define RE_READ_TOKEN -99 +#define NO_EXPANSION -100 + +#ifdef DEBUG +# define YYDEBUG 1 +#else +# define YYDEBUG 0 +#endif + +#if defined (HANDLE_MULTIBYTE) +# define last_shell_getc_is_singlebyte \ + ((shell_input_line_index > 1) \ + ? shell_input_line_property[shell_input_line_index - 1] \ + : 1) +# define MBTEST(x) ((x) && last_shell_getc_is_singlebyte) +#else +# define last_shell_getc_is_singlebyte 1 +# define MBTEST(x) ((x)) +#endif + +#if defined (EXTENDED_GLOB) +extern int extended_glob; +#endif + +extern int eof_encountered; +extern int no_line_editing, running_under_emacs; +extern int current_command_number; +extern int sourcelevel, parse_and_execute_level; +extern int posixly_correct; +extern int last_command_exit_value; +extern pid_t last_command_subst_pid; +extern char *shell_name, *current_host_name; +extern char *dist_version; +extern int patch_level; +extern int dump_translatable_strings, dump_po_strings; +extern sh_builtin_func_t *last_shell_builtin, *this_shell_builtin; +#if defined (BUFFERED_INPUT) +extern int bash_input_fd_changed; +#endif + +extern int errno; +/* **************************************************************** */ +/* */ +/* "Forward" declarations */ +/* */ +/* **************************************************************** */ + +#ifdef DEBUG +static void debug_parser __P((int)); +#endif + +static int yy_getc __P((void)); +static int yy_ungetc __P((int)); + +#if defined (READLINE) +static int yy_readline_get __P((void)); +static int yy_readline_unget __P((int)); +#endif + +static int yy_string_get __P((void)); +static int yy_string_unget __P((int)); +static void rewind_input_string __P((void)); +static int yy_stream_get __P((void)); +static int yy_stream_unget __P((int)); + +static int shell_getc __P((int)); +static void shell_ungetc __P((int)); +static void discard_until __P((int)); + +#if defined (ALIAS) || defined (DPAREN_ARITHMETIC) +static void push_string __P((char *, int, alias_t *)); +static void pop_string __P((void)); +static void free_string_list __P((void)); +#endif + +static char *read_a_line __P((int)); + +static int reserved_word_acceptable __P((int)); +static int yylex __P((void)); +static int alias_expand_token __P((char *)); +static int time_command_acceptable __P((void)); +static int special_case_tokens __P((char *)); +static int read_token __P((int)); +static char *parse_matched_pair __P((int, int, int, int *, int)); +static char *parse_comsub __P((int, int, int, int *, int)); +#if defined (ARRAY_VARS) +static char *parse_compound_assignment __P((int *)); +#endif +#if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND) +static int parse_dparen __P((int)); +static int parse_arith_cmd __P((char **, int)); +#endif +#if defined (COND_COMMAND) +static void cond_error __P((void)); +static COND_COM *cond_expr __P((void)); +static COND_COM *cond_or __P((void)); +static COND_COM *cond_and __P((void)); +static COND_COM *cond_term __P((void)); +static int cond_skip_newlines __P((void)); +static COMMAND *parse_cond_command __P((void)); +#endif +#if defined (ARRAY_VARS) +static int token_is_assignment __P((char *, int)); +static int token_is_ident __P((char *, int)); +#endif +static int read_token_word __P((int)); +static void discard_parser_constructs __P((int)); + +static char *error_token_from_token __P((int)); +static char *error_token_from_text __P((void)); +static void print_offending_line __P((void)); +static void report_syntax_error __P((char *)); + +static void handle_eof_input_unit __P((void)); +static void prompt_again __P((void)); +#if 0 +static void reset_readline_prompt __P((void)); +#endif +static void print_prompt __P((void)); + +#if defined (HANDLE_MULTIBYTE) +static void set_line_mbstate __P((void)); +static char *shell_input_line_property = NULL; +#else +# define set_line_mbstate() +#endif + +extern int yyerror __P((const char *)); + +#ifdef DEBUG +extern int yydebug; +#endif + +/* Default prompt strings */ +char *primary_prompt = PPROMPT; +char *secondary_prompt = SPROMPT; + +/* PROMPT_STRING_POINTER points to one of these, never to an actual string. */ +char *ps1_prompt, *ps2_prompt; + +/* Handle on the current prompt string. Indirectly points through + ps1_ or ps2_prompt. */ +char **prompt_string_pointer = (char **)NULL; +char *current_prompt_string; + +/* Non-zero means we expand aliases in commands. */ +int expand_aliases = 0; + +/* If non-zero, the decoded prompt string undergoes parameter and + variable substitution, command substitution, arithmetic substitution, + string expansion, process substitution, and quote removal in + decode_prompt_string. */ +int promptvars = 1; + +/* If non-zero, $'...' and $"..." are expanded when they appear within + a ${...} expansion, even when the expansion appears within double + quotes. */ +int extended_quote = 1; + +/* The number of lines read from input while creating the current command. */ +int current_command_line_count; + +/* The number of lines in a command saved while we run parse_and_execute */ +int saved_command_line_count; + +/* The token that currently denotes the end of parse. */ +int shell_eof_token; + +/* The token currently being read. */ +int current_token; + +/* The current parser state. */ +int parser_state; + +/* Variables to manage the task of reading here documents, because we need to + defer the reading until after a complete command has been collected. */ +static REDIRECT *redir_stack[10]; +int need_here_doc; + +/* Where shell input comes from. History expansion is performed on each + line when the shell is interactive. */ +static char *shell_input_line = (char *)NULL; +static int shell_input_line_index; +static int shell_input_line_size; /* Amount allocated for shell_input_line. */ +static int shell_input_line_len; /* strlen (shell_input_line) */ + +/* Either zero or EOF. */ +static int shell_input_line_terminator; + +/* The line number in a script on which a function definition starts. */ +static int function_dstart; + +/* The line number in a script on which a function body starts. */ +static int function_bstart; + +/* The line number in a script at which an arithmetic for command starts. */ +static int arith_for_lineno; + +/* The decoded prompt string. Used if READLINE is not defined or if + editing is turned off. Analogous to current_readline_prompt. */ +static char *current_decoded_prompt; + +/* The last read token, or NULL. read_token () uses this for context + checking. */ +static int last_read_token; + +/* The token read prior to last_read_token. */ +static int token_before_that; + +/* The token read prior to token_before_that. */ +static int two_tokens_ago; + +static int global_extglob; + +/* The line number in a script where the word in a `case WORD', `select WORD' + or `for WORD' begins. This is a nested command maximum, since the array + index is decremented after a case, select, or for command is parsed. */ +#define MAX_CASE_NEST 128 +static int word_lineno[MAX_CASE_NEST]; +static int word_top = -1; + +/* If non-zero, it is the token that we want read_token to return + regardless of what text is (or isn't) present to be read. This + is reset by read_token. If token_to_read == WORD or + ASSIGNMENT_WORD, yylval.word should be set to word_desc_to_read. */ +static int token_to_read; +static WORD_DESC *word_desc_to_read; + +static REDIRECTEE source; +static REDIRECTEE redir; +%} + +%union { + WORD_DESC *word; /* the word that we read. */ + int number; /* the number that we read. */ + WORD_LIST *word_list; + COMMAND *command; + REDIRECT *redirect; + ELEMENT element; + PATTERN_LIST *pattern; +} + +/* Reserved words. Members of the first group are only recognized + in the case that they are preceded by a list_terminator. Members + of the second group are for [[...]] commands. Members of the + third group are recognized only under special circumstances. */ +%token IF THEN ELSE ELIF FI CASE ESAC FOR SELECT WHILE UNTIL DO DONE FUNCTION COPROC +%token COND_START COND_END COND_ERROR +%token IN BANG TIME TIMEOPT TIMEIGN + +/* More general tokens. yylex () knows how to make these. */ +%token WORD ASSIGNMENT_WORD REDIR_WORD +%token NUMBER +%token ARITH_CMD ARITH_FOR_EXPRS +%token COND_CMD +%token AND_AND OR_OR GREATER_GREATER LESS_LESS LESS_AND LESS_LESS_LESS +%token GREATER_AND SEMI_SEMI SEMI_AND SEMI_SEMI_AND +%token LESS_LESS_MINUS AND_GREATER AND_GREATER_GREATER LESS_GREATER +%token GREATER_BAR BAR_AND + +/* The types that the various syntactical units return. */ + +%type inputunit command pipeline pipeline_command +%type list list0 list1 compound_list simple_list simple_list1 +%type simple_command shell_command +%type for_command select_command case_command group_command +%type arith_command +%type cond_command +%type arith_for_command +%type coproc +%type function_def function_body if_command elif_clause subshell +%type redirection redirection_list +%type simple_command_element +%type word_list pattern +%type pattern_list case_clause_sequence case_clause +%type timespec +%type list_terminator + +%start inputunit + +%left '&' ';' '\n' yacc_EOF +%left AND_AND OR_OR +%right '|' BAR_AND +%% + +inputunit: simple_list simple_list_terminator + { + /* Case of regular command. Discard the error + safety net,and return the command just parsed. */ + global_command = $1; + eof_encountered = 0; + /* discard_parser_constructs (0); */ + if (parser_state & PST_CMDSUBST) + parser_state |= PST_EOFTOKEN; + YYACCEPT; + } + | '\n' + { + /* Case of regular command, but not a very + interesting one. Return a NULL command. */ + global_command = (COMMAND *)NULL; + if (parser_state & PST_CMDSUBST) + parser_state |= PST_EOFTOKEN; + YYACCEPT; + } + | error '\n' + { + /* Error during parsing. Return NULL command. */ + global_command = (COMMAND *)NULL; + eof_encountered = 0; + /* discard_parser_constructs (1); */ + if (interactive && parse_and_execute_level == 0) + { + YYACCEPT; + } + else + { + YYABORT; + } + } + | yacc_EOF + { + /* Case of EOF seen by itself. Do ignoreeof or + not. */ + global_command = (COMMAND *)NULL; + handle_eof_input_unit (); + YYACCEPT; + } + ; + +word_list: WORD + { $$ = make_word_list ($1, (WORD_LIST *)NULL); } + | word_list WORD + { $$ = make_word_list ($2, $1); } + ; + +redirection: '>' WORD + { + source.dest = 1; + redir.filename = $2; + $$ = make_redirection (source, r_output_direction, redir, 0); + } + | '<' WORD + { + source.dest = 0; + redir.filename = $2; + $$ = make_redirection (source, r_input_direction, redir, 0); + } + | NUMBER '>' WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_output_direction, redir, 0); + } + | NUMBER '<' WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_input_direction, redir, 0); + } + | REDIR_WORD '>' WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_output_direction, redir, REDIR_VARASSIGN); + } + | REDIR_WORD '<' WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_input_direction, redir, REDIR_VARASSIGN); + } + | GREATER_GREATER WORD + { + source.dest = 1; + redir.filename = $2; + $$ = make_redirection (source, r_appending_to, redir, 0); + } + | NUMBER GREATER_GREATER WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_appending_to, redir, 0); + } + | REDIR_WORD GREATER_GREATER WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_appending_to, redir, REDIR_VARASSIGN); + } + | GREATER_BAR WORD + { + source.dest = 1; + redir.filename = $2; + $$ = make_redirection (source, r_output_force, redir, 0); + } + | NUMBER GREATER_BAR WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_output_force, redir, 0); + } + | REDIR_WORD GREATER_BAR WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_output_force, redir, REDIR_VARASSIGN); + } + | LESS_GREATER WORD + { + source.dest = 0; + redir.filename = $2; + $$ = make_redirection (source, r_input_output, redir, 0); + } + | NUMBER LESS_GREATER WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_input_output, redir, 0); + } + | REDIR_WORD LESS_GREATER WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_input_output, redir, REDIR_VARASSIGN); + } + | LESS_LESS WORD + { + source.dest = 0; + redir.filename = $2; + $$ = make_redirection (source, r_reading_until, redir, 0); + redir_stack[need_here_doc++] = $$; + } + | NUMBER LESS_LESS WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_reading_until, redir, 0); + redir_stack[need_here_doc++] = $$; + } + | REDIR_WORD LESS_LESS WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_reading_until, redir, REDIR_VARASSIGN); + redir_stack[need_here_doc++] = $$; + } + | LESS_LESS_MINUS WORD + { + source.dest = 0; + redir.filename = $2; + $$ = make_redirection (source, r_deblank_reading_until, redir, 0); + redir_stack[need_here_doc++] = $$; + } + | NUMBER LESS_LESS_MINUS WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_deblank_reading_until, redir, 0); + redir_stack[need_here_doc++] = $$; + } + | REDIR_WORD LESS_LESS_MINUS WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_deblank_reading_until, redir, REDIR_VARASSIGN); + redir_stack[need_here_doc++] = $$; + } + | LESS_LESS_LESS WORD + { + source.dest = 0; + redir.filename = $2; + $$ = make_redirection (source, r_reading_string, redir, 0); + } + | NUMBER LESS_LESS_LESS WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_reading_string, redir, 0); + } + | REDIR_WORD LESS_LESS_LESS WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_reading_string, redir, REDIR_VARASSIGN); + } + | LESS_AND NUMBER + { + source.dest = 0; + redir.dest = $2; + $$ = make_redirection (source, r_duplicating_input, redir, 0); + } + | NUMBER LESS_AND NUMBER + { + source.dest = $1; + redir.dest = $3; + $$ = make_redirection (source, r_duplicating_input, redir, 0); + } + | REDIR_WORD LESS_AND NUMBER + { + source.filename = $1; + redir.dest = $3; + $$ = make_redirection (source, r_duplicating_input, redir, REDIR_VARASSIGN); + } + | GREATER_AND NUMBER + { + source.dest = 1; + redir.dest = $2; + $$ = make_redirection (source, r_duplicating_output, redir, 0); + } + | NUMBER GREATER_AND NUMBER + { + source.dest = $1; + redir.dest = $3; + $$ = make_redirection (source, r_duplicating_output, redir, 0); + } + | REDIR_WORD GREATER_AND NUMBER + { + source.filename = $1; + redir.dest = $3; + $$ = make_redirection (source, r_duplicating_output, redir, REDIR_VARASSIGN); + } + | LESS_AND WORD + { + source.dest = 0; + redir.filename = $2; + $$ = make_redirection (source, r_duplicating_input_word, redir, 0); + } + | NUMBER LESS_AND WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_duplicating_input_word, redir, 0); + } + | REDIR_WORD LESS_AND WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_duplicating_input_word, redir, REDIR_VARASSIGN); + } + | GREATER_AND WORD + { + source.dest = 1; + redir.filename = $2; + $$ = make_redirection (source, r_duplicating_output_word, redir, 0); + } + | NUMBER GREATER_AND WORD + { + source.dest = $1; + redir.filename = $3; + $$ = make_redirection (source, r_duplicating_output_word, redir, 0); + } + | REDIR_WORD GREATER_AND WORD + { + source.filename = $1; + redir.filename = $3; + $$ = make_redirection (source, r_duplicating_output_word, redir, REDIR_VARASSIGN); + } + | GREATER_AND '-' + { + source.dest = 1; + redir.dest = 0; + $$ = make_redirection (source, r_close_this, redir, 0); + } + | NUMBER GREATER_AND '-' + { + source.dest = $1; + redir.dest = 0; + $$ = make_redirection (source, r_close_this, redir, 0); + } + | REDIR_WORD GREATER_AND '-' + { + source.filename = $1; + redir.dest = 0; + $$ = make_redirection (source, r_close_this, redir, REDIR_VARASSIGN); + } + | LESS_AND '-' + { + source.dest = 0; + redir.dest = 0; + $$ = make_redirection (source, r_close_this, redir, 0); + } + | NUMBER LESS_AND '-' + { + source.dest = $1; + redir.dest = 0; + $$ = make_redirection (source, r_close_this, redir, 0); + } + | REDIR_WORD LESS_AND '-' + { + source.filename = $1; + redir.dest = 0; + $$ = make_redirection (source, r_close_this, redir, REDIR_VARASSIGN); + } + | AND_GREATER WORD + { + source.dest = 1; + redir.filename = $2; + $$ = make_redirection (source, r_err_and_out, redir, 0); + } + | AND_GREATER_GREATER WORD + { + source.dest = 1; + redir.filename = $2; + $$ = make_redirection (source, r_append_err_and_out, redir, 0); + } + ; + +simple_command_element: WORD + { $$.word = $1; $$.redirect = 0; } + | ASSIGNMENT_WORD + { $$.word = $1; $$.redirect = 0; } + | redirection + { $$.redirect = $1; $$.word = 0; } + ; + +redirection_list: redirection + { + $$ = $1; + } + | redirection_list redirection + { + register REDIRECT *t; + + for (t = $1; t->next; t = t->next) + ; + t->next = $2; + $$ = $1; + } + ; + +simple_command: simple_command_element + { $$ = make_simple_command ($1, (COMMAND *)NULL); } + | simple_command simple_command_element + { $$ = make_simple_command ($2, $1); } + ; + +command: simple_command + { $$ = clean_simple_command ($1); } + | shell_command + { $$ = $1; } + | shell_command redirection_list + { + COMMAND *tc; + + tc = $1; + if (tc->redirects) + { + register REDIRECT *t; + for (t = tc->redirects; t->next; t = t->next) + ; + t->next = $2; + } + else + tc->redirects = $2; + $$ = $1; + } + | function_def + { $$ = $1; } + | coproc + { $$ = $1; } + ; + +shell_command: for_command + { $$ = $1; } + | case_command + { $$ = $1; } + | WHILE compound_list DO compound_list DONE + { $$ = make_while_command ($2, $4); } + | UNTIL compound_list DO compound_list DONE + { $$ = make_until_command ($2, $4); } + | select_command + { $$ = $1; } + | if_command + { $$ = $1; } + | subshell + { $$ = $1; } + | group_command + { $$ = $1; } + | arith_command + { $$ = $1; } + | cond_command + { $$ = $1; } + | arith_for_command + { $$ = $1; } + ; + +for_command: FOR WORD newline_list DO compound_list DONE + { + $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD newline_list '{' compound_list '}' + { + $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD ';' newline_list DO compound_list DONE + { + $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD ';' newline_list '{' compound_list '}' + { + $$ = make_for_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD newline_list IN word_list list_terminator newline_list DO compound_list DONE + { + $$ = make_for_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD newline_list IN word_list list_terminator newline_list '{' compound_list '}' + { + $$ = make_for_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD newline_list IN list_terminator newline_list DO compound_list DONE + { + $$ = make_for_command ($2, (WORD_LIST *)NULL, $8, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | FOR WORD newline_list IN list_terminator newline_list '{' compound_list '}' + { + $$ = make_for_command ($2, (WORD_LIST *)NULL, $8, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + ; + +arith_for_command: FOR ARITH_FOR_EXPRS list_terminator newline_list DO compound_list DONE + { + $$ = make_arith_for_command ($2, $6, arith_for_lineno); + if (word_top > 0) word_top--; + } + | FOR ARITH_FOR_EXPRS list_terminator newline_list '{' compound_list '}' + { + $$ = make_arith_for_command ($2, $6, arith_for_lineno); + if (word_top > 0) word_top--; + } + | FOR ARITH_FOR_EXPRS DO compound_list DONE + { + $$ = make_arith_for_command ($2, $4, arith_for_lineno); + if (word_top > 0) word_top--; + } + | FOR ARITH_FOR_EXPRS '{' compound_list '}' + { + $$ = make_arith_for_command ($2, $4, arith_for_lineno); + if (word_top > 0) word_top--; + } + ; + +select_command: SELECT WORD newline_list DO list DONE + { + $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | SELECT WORD newline_list '{' list '}' + { + $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $5, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | SELECT WORD ';' newline_list DO list DONE + { + $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | SELECT WORD ';' newline_list '{' list '}' + { + $$ = make_select_command ($2, add_string_to_list ("\"$@\"", (WORD_LIST *)NULL), $6, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | SELECT WORD newline_list IN word_list list_terminator newline_list DO list DONE + { + $$ = make_select_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | SELECT WORD newline_list IN word_list list_terminator newline_list '{' list '}' + { + $$ = make_select_command ($2, REVERSE_LIST ($5, WORD_LIST *), $9, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + ; + +case_command: CASE WORD newline_list IN newline_list ESAC + { + $$ = make_case_command ($2, (PATTERN_LIST *)NULL, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | CASE WORD newline_list IN case_clause_sequence newline_list ESAC + { + $$ = make_case_command ($2, $5, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + | CASE WORD newline_list IN case_clause ESAC + { + $$ = make_case_command ($2, $5, word_lineno[word_top]); + if (word_top > 0) word_top--; + } + ; + +function_def: WORD '(' ')' newline_list function_body + { $$ = make_function_def ($1, $5, function_dstart, function_bstart); } + + | FUNCTION WORD '(' ')' newline_list function_body + { $$ = make_function_def ($2, $6, function_dstart, function_bstart); } + + | FUNCTION WORD newline_list function_body + { $$ = make_function_def ($2, $4, function_dstart, function_bstart); } + ; + +function_body: shell_command + { $$ = $1; } + | shell_command redirection_list + { + COMMAND *tc; + + tc = $1; + /* According to Posix.2 3.9.5, redirections + specified after the body of a function should + be attached to the function and performed when + the function is executed, not as part of the + function definition command. */ + /* XXX - I don't think it matters, but we might + want to change this in the future to avoid + problems differentiating between a function + definition with a redirection and a function + definition containing a single command with a + redirection. The two are semantically equivalent, + though -- the only difference is in how the + command printing code displays the redirections. */ + if (tc->redirects) + { + register REDIRECT *t; + for (t = tc->redirects; t->next; t = t->next) + ; + t->next = $2; + } + else + tc->redirects = $2; + $$ = $1; + } + ; + +subshell: '(' compound_list ')' + { + $$ = make_subshell_command ($2); + $$->flags |= CMD_WANT_SUBSHELL; + } + ; + +coproc: COPROC shell_command + { + $$ = make_coproc_command ("COPROC", $2); + $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL; + } + | COPROC shell_command redirection_list + { + COMMAND *tc; + + tc = $2; + if (tc->redirects) + { + register REDIRECT *t; + for (t = tc->redirects; t->next; t = t->next) + ; + t->next = $3; + } + else + tc->redirects = $3; + $$ = make_coproc_command ("COPROC", $2); + $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL; + } + | COPROC WORD shell_command + { + $$ = make_coproc_command ($2->word, $3); + $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL; + } + | COPROC WORD shell_command redirection_list + { + COMMAND *tc; + + tc = $3; + if (tc->redirects) + { + register REDIRECT *t; + for (t = tc->redirects; t->next; t = t->next) + ; + t->next = $4; + } + else + tc->redirects = $4; + $$ = make_coproc_command ($2->word, $3); + $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL; + } + | COPROC simple_command + { + $$ = make_coproc_command ("COPROC", clean_simple_command ($2)); + $$->flags |= CMD_WANT_SUBSHELL|CMD_COPROC_SUBSHELL; + } + ; + +if_command: IF compound_list THEN compound_list FI + { $$ = make_if_command ($2, $4, (COMMAND *)NULL); } + | IF compound_list THEN compound_list ELSE compound_list FI + { $$ = make_if_command ($2, $4, $6); } + | IF compound_list THEN compound_list elif_clause FI + { $$ = make_if_command ($2, $4, $5); } + ; + + +group_command: '{' compound_list '}' + { $$ = make_group_command ($2); } + ; + +arith_command: ARITH_CMD + { $$ = make_arith_command ($1); } + ; + +cond_command: COND_START COND_CMD COND_END + { $$ = $2; } + ; + +elif_clause: ELIF compound_list THEN compound_list + { $$ = make_if_command ($2, $4, (COMMAND *)NULL); } + | ELIF compound_list THEN compound_list ELSE compound_list + { $$ = make_if_command ($2, $4, $6); } + | ELIF compound_list THEN compound_list elif_clause + { $$ = make_if_command ($2, $4, $5); } + ; + +case_clause: pattern_list + | case_clause_sequence pattern_list + { $2->next = $1; $$ = $2; } + ; + +pattern_list: newline_list pattern ')' compound_list + { $$ = make_pattern_list ($2, $4); } + | newline_list pattern ')' newline_list + { $$ = make_pattern_list ($2, (COMMAND *)NULL); } + | newline_list '(' pattern ')' compound_list + { $$ = make_pattern_list ($3, $5); } + | newline_list '(' pattern ')' newline_list + { $$ = make_pattern_list ($3, (COMMAND *)NULL); } + ; + +case_clause_sequence: pattern_list SEMI_SEMI + { $$ = $1; } + | case_clause_sequence pattern_list SEMI_SEMI + { $2->next = $1; $$ = $2; } + | pattern_list SEMI_AND + { $1->flags |= CASEPAT_FALLTHROUGH; $$ = $1; } + | case_clause_sequence pattern_list SEMI_AND + { $2->flags |= CASEPAT_FALLTHROUGH; $2->next = $1; $$ = $2; } + | pattern_list SEMI_SEMI_AND + { $1->flags |= CASEPAT_TESTNEXT; $$ = $1; } + | case_clause_sequence pattern_list SEMI_SEMI_AND + { $2->flags |= CASEPAT_TESTNEXT; $2->next = $1; $$ = $2; } + ; + +pattern: WORD + { $$ = make_word_list ($1, (WORD_LIST *)NULL); } + | pattern '|' WORD + { $$ = make_word_list ($3, $1); } + ; + +/* A list allows leading or trailing newlines and + newlines as operators (equivalent to semicolons). + It must end with a newline or semicolon. + Lists are used within commands such as if, for, while. */ + +list: newline_list list0 + { + $$ = $2; + if (need_here_doc) + gather_here_documents (); + } + ; + +compound_list: list + | newline_list list1 + { + $$ = $2; + } + ; + +list0: list1 '\n' newline_list + | list1 '&' newline_list + { + if ($1->type == cm_connection) + $$ = connect_async_list ($1, (COMMAND *)NULL, '&'); + else + $$ = command_connect ($1, (COMMAND *)NULL, '&'); + } + | list1 ';' newline_list + + ; + +list1: list1 AND_AND newline_list list1 + { $$ = command_connect ($1, $4, AND_AND); } + | list1 OR_OR newline_list list1 + { $$ = command_connect ($1, $4, OR_OR); } + | list1 '&' newline_list list1 + { + if ($1->type == cm_connection) + $$ = connect_async_list ($1, $4, '&'); + else + $$ = command_connect ($1, $4, '&'); + } + | list1 ';' newline_list list1 + { $$ = command_connect ($1, $4, ';'); } + | list1 '\n' newline_list list1 + { $$ = command_connect ($1, $4, ';'); } + | pipeline_command + { $$ = $1; } + ; + +simple_list_terminator: '\n' + | yacc_EOF + ; + +list_terminator:'\n' + { $$ = '\n'; } + | ';' + { $$ = ';'; } + | yacc_EOF + { $$ = yacc_EOF; } + ; + +newline_list: + | newline_list '\n' + ; + +/* A simple_list is a list that contains no significant newlines + and no leading or trailing newlines. Newlines are allowed + only following operators, where they are not significant. + + This is what an inputunit consists of. */ + +simple_list: simple_list1 + { + $$ = $1; + if (need_here_doc) + gather_here_documents (); + if ((parser_state & PST_CMDSUBST) && current_token == shell_eof_token) + { + global_command = $1; + eof_encountered = 0; + rewind_input_string (); + YYACCEPT; + } + } + | simple_list1 '&' + { + if ($1->type == cm_connection) + $$ = connect_async_list ($1, (COMMAND *)NULL, '&'); + else + $$ = command_connect ($1, (COMMAND *)NULL, '&'); + if (need_here_doc) + gather_here_documents (); + if ((parser_state & PST_CMDSUBST) && current_token == shell_eof_token) + { + global_command = $1; + eof_encountered = 0; + rewind_input_string (); + YYACCEPT; + } + } + | simple_list1 ';' + { + $$ = $1; + if (need_here_doc) + gather_here_documents (); + if ((parser_state & PST_CMDSUBST) && current_token == shell_eof_token) + { + global_command = $1; + eof_encountered = 0; + rewind_input_string (); + YYACCEPT; + } + } + ; + +simple_list1: simple_list1 AND_AND newline_list simple_list1 + { $$ = command_connect ($1, $4, AND_AND); } + | simple_list1 OR_OR newline_list simple_list1 + { $$ = command_connect ($1, $4, OR_OR); } + | simple_list1 '&' simple_list1 + { + if ($1->type == cm_connection) + $$ = connect_async_list ($1, $3, '&'); + else + $$ = command_connect ($1, $3, '&'); + } + | simple_list1 ';' simple_list1 + { $$ = command_connect ($1, $3, ';'); } + + | pipeline_command + { $$ = $1; } + ; + +pipeline_command: pipeline + { $$ = $1; } + | BANG pipeline_command + { + if ($2) + $2->flags ^= CMD_INVERT_RETURN; /* toggle */ + $$ = $2; + } + | timespec pipeline_command + { + if ($2) + $2->flags |= $1; + $$ = $2; + } + | timespec list_terminator + { + ELEMENT x; + + /* Boy, this is unclean. `time' by itself can + time a null command. We cheat and push a + newline back if the list_terminator was a newline + to avoid the double-newline problem (one to + terminate this, one to terminate the command) */ + x.word = 0; + x.redirect = 0; + $$ = make_simple_command (x, (COMMAND *)NULL); + $$->flags |= $1; + /* XXX - let's cheat and push a newline back */ + if ($2 == '\n') + token_to_read = '\n'; + } + | BANG list_terminator + { + ELEMENT x; + + /* This is just as unclean. Posix says that `!' + by itself should be equivalent to `false'. + We cheat and push a + newline back if the list_terminator was a newline + to avoid the double-newline problem (one to + terminate this, one to terminate the command) */ + x.word = 0; + x.redirect = 0; + $$ = make_simple_command (x, (COMMAND *)NULL); + $$->flags |= CMD_INVERT_RETURN; + /* XXX - let's cheat and push a newline back */ + if ($2 == '\n') + token_to_read = '\n'; + } + ; + +pipeline: pipeline '|' newline_list pipeline + { $$ = command_connect ($1, $4, '|'); } + | pipeline BAR_AND newline_list pipeline + { + /* Make cmd1 |& cmd2 equivalent to cmd1 2>&1 | cmd2 */ + COMMAND *tc; + REDIRECTEE rd, sd; + REDIRECT *r; + + tc = $1->type == cm_simple ? (COMMAND *)$1->value.Simple : $1; + sd.dest = 2; + rd.dest = 1; + r = make_redirection (sd, r_duplicating_output, rd, 0); + if (tc->redirects) + { + register REDIRECT *t; + for (t = tc->redirects; t->next; t = t->next) + ; + t->next = r; + } + else + tc->redirects = r; + + $$ = command_connect ($1, $4, '|'); + } + | command + { $$ = $1; } + ; + +timespec: TIME + { $$ = CMD_TIME_PIPELINE; } + | TIME TIMEOPT + { $$ = CMD_TIME_PIPELINE|CMD_TIME_POSIX; } + | TIME TIMEOPT TIMEIGN + { $$ = CMD_TIME_PIPELINE|CMD_TIME_POSIX; } + ; +%% + +/* Initial size to allocate for tokens, and the + amount to grow them by. */ +#define TOKEN_DEFAULT_INITIAL_SIZE 496 +#define TOKEN_DEFAULT_GROW_SIZE 512 + +/* Should we call prompt_again? */ +#define SHOULD_PROMPT() \ + (interactive && (bash_input.type == st_stdin || bash_input.type == st_stream)) + +#if defined (ALIAS) +# define expanding_alias() (pushed_string_list && pushed_string_list->expander) +#else +# define expanding_alias() 0 +#endif + +/* Global var is non-zero when end of file has been reached. */ +int EOF_Reached = 0; + +#ifdef DEBUG +static void +debug_parser (i) + int i; +{ +#if YYDEBUG != 0 + yydebug = i; +#endif +} +#endif + +/* yy_getc () returns the next available character from input or EOF. + yy_ungetc (c) makes `c' the next character to read. + init_yy_io (get, unget, type, location) makes the function GET the + installed function for getting the next character, makes UNGET the + installed function for un-getting a character, sets the type of stream + (either string or file) from TYPE, and makes LOCATION point to where + the input is coming from. */ + +/* Unconditionally returns end-of-file. */ +int +return_EOF () +{ + return (EOF); +} + +/* Variable containing the current get and unget functions. + See ./input.h for a clearer description. */ +BASH_INPUT bash_input; + +/* Set all of the fields in BASH_INPUT to NULL. Free bash_input.name if it + is non-null, avoiding a memory leak. */ +void +initialize_bash_input () +{ + bash_input.type = st_none; + FREE (bash_input.name); + bash_input.name = (char *)NULL; + bash_input.location.file = (FILE *)NULL; + bash_input.location.string = (char *)NULL; + bash_input.getter = (sh_cget_func_t *)NULL; + bash_input.ungetter = (sh_cunget_func_t *)NULL; +} + +/* Set the contents of the current bash input stream from + GET, UNGET, TYPE, NAME, and LOCATION. */ +void +init_yy_io (get, unget, type, name, location) + sh_cget_func_t *get; + sh_cunget_func_t *unget; + enum stream_type type; + const char *name; + INPUT_STREAM location; +{ + bash_input.type = type; + FREE (bash_input.name); + bash_input.name = name ? savestring (name) : (char *)NULL; + + /* XXX */ +#if defined (CRAY) + memcpy((char *)&bash_input.location.string, (char *)&location.string, sizeof(location)); +#else + bash_input.location = location; +#endif + bash_input.getter = get; + bash_input.ungetter = unget; +} + +char * +yy_input_name () +{ + return (bash_input.name ? bash_input.name : "stdin"); +} + +/* Call this to get the next character of input. */ +static int +yy_getc () +{ + return (*(bash_input.getter)) (); +} + +/* Call this to unget C. That is, to make C the next character + to be read. */ +static int +yy_ungetc (c) + int c; +{ + return (*(bash_input.ungetter)) (c); +} + +#if defined (BUFFERED_INPUT) +#ifdef INCLUDE_UNUSED +int +input_file_descriptor () +{ + switch (bash_input.type) + { + case st_stream: + return (fileno (bash_input.location.file)); + case st_bstream: + return (bash_input.location.buffered_fd); + case st_stdin: + default: + return (fileno (stdin)); + } +} +#endif +#endif /* BUFFERED_INPUT */ + +/* **************************************************************** */ +/* */ +/* Let input be read from readline (). */ +/* */ +/* **************************************************************** */ + +#if defined (READLINE) +char *current_readline_prompt = (char *)NULL; +char *current_readline_line = (char *)NULL; +int current_readline_line_index = 0; + +static int +yy_readline_get () +{ + SigHandler *old_sigint; + int line_len; + unsigned char c; + + if (!current_readline_line) + { + if (!bash_readline_initialized) + initialize_readline (); + +#if defined (JOB_CONTROL) + if (job_control) + give_terminal_to (shell_pgrp, 0); +#endif /* JOB_CONTROL */ + + old_sigint = (SigHandler *)IMPOSSIBLE_TRAP_HANDLER; + if (signal_is_ignored (SIGINT) == 0) + { + interrupt_immediately++; + old_sigint = (SigHandler *)set_signal_handler (SIGINT, sigint_sighandler); + } + + current_readline_line = readline (current_readline_prompt ? + current_readline_prompt : ""); + + CHECK_TERMSIG; + if (signal_is_ignored (SIGINT) == 0) + { + interrupt_immediately--; + if (old_sigint != IMPOSSIBLE_TRAP_HANDLER) + set_signal_handler (SIGINT, old_sigint); + } + +#if 0 + /* Reset the prompt to the decoded value of prompt_string_pointer. */ + reset_readline_prompt (); +#endif + + if (current_readline_line == 0) + return (EOF); + + current_readline_line_index = 0; + line_len = strlen (current_readline_line); + + current_readline_line = (char *)xrealloc (current_readline_line, 2 + line_len); + current_readline_line[line_len++] = '\n'; + current_readline_line[line_len] = '\0'; + } + + if (current_readline_line[current_readline_line_index] == 0) + { + free (current_readline_line); + current_readline_line = (char *)NULL; + return (yy_readline_get ()); + } + else + { + c = current_readline_line[current_readline_line_index++]; + return (c); + } +} + +static int +yy_readline_unget (c) + int c; +{ + if (current_readline_line_index && current_readline_line) + current_readline_line[--current_readline_line_index] = c; + return (c); +} + +void +with_input_from_stdin () +{ + INPUT_STREAM location; + + if (bash_input.type != st_stdin && stream_on_stack (st_stdin) == 0) + { + location.string = current_readline_line; + init_yy_io (yy_readline_get, yy_readline_unget, + st_stdin, "readline stdin", location); + } +} + +#else /* !READLINE */ + +void +with_input_from_stdin () +{ + with_input_from_stream (stdin, "stdin"); +} +#endif /* !READLINE */ + +/* **************************************************************** */ +/* */ +/* Let input come from STRING. STRING is zero terminated. */ +/* */ +/* **************************************************************** */ + +static int +yy_string_get () +{ + register char *string; + register unsigned char c; + + string = bash_input.location.string; + + /* If the string doesn't exist, or is empty, EOF found. */ + if (string && *string) + { + c = *string++; + bash_input.location.string = string; + return (c); + } + else + return (EOF); +} + +static int +yy_string_unget (c) + int c; +{ + *(--bash_input.location.string) = c; + return (c); +} + +void +with_input_from_string (string, name) + char *string; + const char *name; +{ + INPUT_STREAM location; + + location.string = string; + init_yy_io (yy_string_get, yy_string_unget, st_string, name, location); +} + +/* Count the number of characters we've consumed from bash_input.location.string + and read into shell_input_line, but have not returned from shell_getc. + That is the true input location. Rewind bash_input.location.string by + that number of characters, so it points to the last character actually + consumed by the parser. */ +static void +rewind_input_string () +{ + int xchars; + + /* number of unconsumed characters in the input -- XXX need to take newlines + into account, e.g., $(...\n) */ + xchars = shell_input_line_len - shell_input_line_index; + if (bash_input.location.string[-1] == '\n') + xchars++; + + /* XXX - how to reflect bash_input.location.string back to string passed to + parse_and_execute or xparse_dolparen? xparse_dolparen needs to know how + far into the string we parsed. parse_and_execute knows where bash_input. + location.string is, and how far from orig_string that is -- that's the + number of characters the command consumed. */ + + /* bash_input.location.string - xchars should be where we parsed to */ + /* need to do more validation on xchars value for sanity -- test cases. */ + bash_input.location.string -= xchars; +} + +/* **************************************************************** */ +/* */ +/* Let input come from STREAM. */ +/* */ +/* **************************************************************** */ + +/* These two functions used to test the value of the HAVE_RESTARTABLE_SYSCALLS + define, and just use getc/ungetc if it was defined, but since bash + installs its signal handlers without the SA_RESTART flag, some signals + (like SIGCHLD, SIGWINCH, etc.) received during a read(2) will not cause + the read to be restarted. We need to restart it ourselves. */ + +static int +yy_stream_get () +{ + int result; + + result = EOF; + if (bash_input.location.file) + { + if (interactive) + interrupt_immediately++; + + /* XXX - don't need terminate_immediately; getc_with_restart checks + for terminating signals itself if read returns < 0 */ + result = getc_with_restart (bash_input.location.file); + + if (interactive) + interrupt_immediately--; + + } + return (result); +} + +static int +yy_stream_unget (c) + int c; +{ + return (ungetc_with_restart (c, bash_input.location.file)); +} + +void +with_input_from_stream (stream, name) + FILE *stream; + const char *name; +{ + INPUT_STREAM location; + + location.file = stream; + init_yy_io (yy_stream_get, yy_stream_unget, st_stream, name, location); +} + +typedef struct stream_saver { + struct stream_saver *next; + BASH_INPUT bash_input; + int line; +#if defined (BUFFERED_INPUT) + BUFFERED_STREAM *bstream; +#endif /* BUFFERED_INPUT */ +} STREAM_SAVER; + +/* The globally known line number. */ +int line_number = 0; + +/* The line number offset set by assigning to LINENO. Not currently used. */ +int line_number_base = 0; + +#if defined (COND_COMMAND) +static int cond_lineno; +static int cond_token; +#endif + +STREAM_SAVER *stream_list = (STREAM_SAVER *)NULL; + +void +push_stream (reset_lineno) + int reset_lineno; +{ + STREAM_SAVER *saver = (STREAM_SAVER *)xmalloc (sizeof (STREAM_SAVER)); + + xbcopy ((char *)&bash_input, (char *)&(saver->bash_input), sizeof (BASH_INPUT)); + +#if defined (BUFFERED_INPUT) + saver->bstream = (BUFFERED_STREAM *)NULL; + /* If we have a buffered stream, clear out buffers[fd]. */ + if (bash_input.type == st_bstream && bash_input.location.buffered_fd >= 0) + saver->bstream = set_buffered_stream (bash_input.location.buffered_fd, + (BUFFERED_STREAM *)NULL); +#endif /* BUFFERED_INPUT */ + + saver->line = line_number; + bash_input.name = (char *)NULL; + saver->next = stream_list; + stream_list = saver; + EOF_Reached = 0; + if (reset_lineno) + line_number = 0; +} + +void +pop_stream () +{ + if (!stream_list) + EOF_Reached = 1; + else + { + STREAM_SAVER *saver = stream_list; + + EOF_Reached = 0; + stream_list = stream_list->next; + + init_yy_io (saver->bash_input.getter, + saver->bash_input.ungetter, + saver->bash_input.type, + saver->bash_input.name, + saver->bash_input.location); + +#if defined (BUFFERED_INPUT) + /* If we have a buffered stream, restore buffers[fd]. */ + /* If the input file descriptor was changed while this was on the + save stack, update the buffered fd to the new file descriptor and + re-establish the buffer <-> bash_input fd correspondence. */ + if (bash_input.type == st_bstream && bash_input.location.buffered_fd >= 0) + { + if (bash_input_fd_changed) + { + bash_input_fd_changed = 0; + if (default_buffered_input >= 0) + { + bash_input.location.buffered_fd = default_buffered_input; + saver->bstream->b_fd = default_buffered_input; + SET_CLOSE_ON_EXEC (default_buffered_input); + } + } + /* XXX could free buffered stream returned as result here. */ + set_buffered_stream (bash_input.location.buffered_fd, saver->bstream); + } +#endif /* BUFFERED_INPUT */ + + line_number = saver->line; + + FREE (saver->bash_input.name); + free (saver); + } +} + +/* Return 1 if a stream of type TYPE is saved on the stack. */ +int +stream_on_stack (type) + enum stream_type type; +{ + register STREAM_SAVER *s; + + for (s = stream_list; s; s = s->next) + if (s->bash_input.type == type) + return 1; + return 0; +} + +/* Save the current token state and return it in a malloced array. */ +int * +save_token_state () +{ + int *ret; + + ret = (int *)xmalloc (4 * sizeof (int)); + ret[0] = last_read_token; + ret[1] = token_before_that; + ret[2] = two_tokens_ago; + ret[3] = current_token; + return ret; +} + +void +restore_token_state (ts) + int *ts; +{ + if (ts == 0) + return; + last_read_token = ts[0]; + token_before_that = ts[1]; + two_tokens_ago = ts[2]; + current_token = ts[3]; +} + +/* + * This is used to inhibit alias expansion and reserved word recognition + * inside case statement pattern lists. A `case statement pattern list' is: + * + * everything between the `in' in a `case word in' and the next ')' + * or `esac' + * everything between a `;;' and the next `)' or `esac' + */ + +#if defined (ALIAS) || defined (DPAREN_ARITHMETIC) + +#define END_OF_ALIAS 0 + +/* + * Pseudo-global variables used in implementing token-wise alias expansion. + */ + +/* + * Pushing and popping strings. This works together with shell_getc to + * implement alias expansion on a per-token basis. + */ + +typedef struct string_saver { + struct string_saver *next; + int expand_alias; /* Value to set expand_alias to when string is popped. */ + char *saved_line; +#if defined (ALIAS) + alias_t *expander; /* alias that caused this line to be pushed. */ +#endif + int saved_line_size, saved_line_index, saved_line_terminator; +} STRING_SAVER; + +STRING_SAVER *pushed_string_list = (STRING_SAVER *)NULL; + +/* + * Push the current shell_input_line onto a stack of such lines and make S + * the current input. Used when expanding aliases. EXPAND is used to set + * the value of expand_next_token when the string is popped, so that the + * word after the alias in the original line is handled correctly when the + * alias expands to multiple words. TOKEN is the token that was expanded + * into S; it is saved and used to prevent infinite recursive expansion. + */ +static void +push_string (s, expand, ap) + char *s; + int expand; + alias_t *ap; +{ + STRING_SAVER *temp = (STRING_SAVER *)xmalloc (sizeof (STRING_SAVER)); + + temp->expand_alias = expand; + temp->saved_line = shell_input_line; + temp->saved_line_size = shell_input_line_size; + temp->saved_line_index = shell_input_line_index; + temp->saved_line_terminator = shell_input_line_terminator; +#if defined (ALIAS) + temp->expander = ap; +#endif + temp->next = pushed_string_list; + pushed_string_list = temp; + +#if defined (ALIAS) + if (ap) + ap->flags |= AL_BEINGEXPANDED; +#endif + + shell_input_line = s; + shell_input_line_size = strlen (s); + shell_input_line_index = 0; + shell_input_line_terminator = '\0'; +#if 0 + parser_state &= ~PST_ALEXPNEXT; /* XXX */ +#endif + + set_line_mbstate (); +} + +/* + * Make the top of the pushed_string stack be the current shell input. + * Only called when there is something on the stack. Called from shell_getc + * when it thinks it has consumed the string generated by an alias expansion + * and needs to return to the original input line. + */ +static void +pop_string () +{ + STRING_SAVER *t; + + FREE (shell_input_line); + shell_input_line = pushed_string_list->saved_line; + shell_input_line_index = pushed_string_list->saved_line_index; + shell_input_line_size = pushed_string_list->saved_line_size; + shell_input_line_terminator = pushed_string_list->saved_line_terminator; + + if (pushed_string_list->expand_alias) + parser_state |= PST_ALEXPNEXT; + else + parser_state &= ~PST_ALEXPNEXT; + + t = pushed_string_list; + pushed_string_list = pushed_string_list->next; + +#if defined (ALIAS) + if (t->expander) + t->expander->flags &= ~AL_BEINGEXPANDED; +#endif + + free ((char *)t); + + set_line_mbstate (); +} + +static void +free_string_list () +{ + register STRING_SAVER *t, *t1; + + for (t = pushed_string_list; t; ) + { + t1 = t->next; + FREE (t->saved_line); +#if defined (ALIAS) + if (t->expander) + t->expander->flags &= ~AL_BEINGEXPANDED; +#endif + free ((char *)t); + t = t1; + } + pushed_string_list = (STRING_SAVER *)NULL; +} + +#endif /* ALIAS || DPAREN_ARITHMETIC */ + +void +free_pushed_string_input () +{ +#if defined (ALIAS) || defined (DPAREN_ARITHMETIC) + free_string_list (); +#endif +} + +/* Return a line of text, taken from wherever yylex () reads input. + If there is no more input, then we return NULL. If REMOVE_QUOTED_NEWLINE + is non-zero, we remove unquoted \ pairs. This is used by + read_secondary_line to read here documents. */ +static char * +read_a_line (remove_quoted_newline) + int remove_quoted_newline; +{ + static char *line_buffer = (char *)NULL; + static int buffer_size = 0; + int indx, c, peekc, pass_next; + +#if defined (READLINE) + if (no_line_editing && SHOULD_PROMPT ()) +#else + if (SHOULD_PROMPT ()) +#endif + print_prompt (); + + pass_next = indx = 0; + while (1) + { + /* Allow immediate exit if interrupted during input. */ + QUIT; + + c = yy_getc (); + + /* Ignore null bytes in input. */ + if (c == 0) + { +#if 0 + internal_warning ("read_a_line: ignored null byte in input"); +#endif + continue; + } + + /* If there is no more input, then we return NULL. */ + if (c == EOF) + { + if (interactive && bash_input.type == st_stream) + clearerr (stdin); + if (indx == 0) + return ((char *)NULL); + c = '\n'; + } + + /* `+2' in case the final character in the buffer is a newline. */ + RESIZE_MALLOCED_BUFFER (line_buffer, indx, 2, buffer_size, 128); + + /* IF REMOVE_QUOTED_NEWLINES is non-zero, we are reading a + here document with an unquoted delimiter. In this case, + the line will be expanded as if it were in double quotes. + We allow a backslash to escape the next character, but we + need to treat the backslash specially only if a backslash + quoting a backslash-newline pair appears in the line. */ + if (pass_next) + { + line_buffer[indx++] = c; + pass_next = 0; + } + else if (c == '\\' && remove_quoted_newline) + { + QUIT; + peekc = yy_getc (); + if (peekc == '\n') + { + line_number++; + continue; /* Make the unquoted \ pair disappear. */ + } + else + { + yy_ungetc (peekc); + pass_next = 1; + line_buffer[indx++] = c; /* Preserve the backslash. */ + } + } + else + line_buffer[indx++] = c; + + if (c == '\n') + { + line_buffer[indx] = '\0'; + return (line_buffer); + } + } +} + +/* Return a line as in read_a_line (), but insure that the prompt is + the secondary prompt. This is used to read the lines of a here + document. REMOVE_QUOTED_NEWLINE is non-zero if we should remove + newlines quoted with backslashes while reading the line. It is + non-zero unless the delimiter of the here document was quoted. */ +char * +read_secondary_line (remove_quoted_newline) + int remove_quoted_newline; +{ + char *ret; + int n, c; + + prompt_string_pointer = &ps2_prompt; + if (SHOULD_PROMPT()) + prompt_again (); + ret = read_a_line (remove_quoted_newline); +#if defined (HISTORY) + if (ret && remember_on_history && (parser_state & PST_HEREDOC)) + { + /* To make adding the the here-document body right, we need to rely + on history_delimiting_chars() returning \n for the first line of + the here-document body and the null string for the second and + subsequent lines, so we avoid double newlines. + current_command_line_count == 2 for the first line of the body. */ + + current_command_line_count++; + maybe_add_history (ret); + } +#endif /* HISTORY */ + return ret; +} + +/* **************************************************************** */ +/* */ +/* YYLEX () */ +/* */ +/* **************************************************************** */ + +/* Reserved words. These are only recognized as the first word of a + command. */ +STRING_INT_ALIST word_token_alist[] = { + { "if", IF }, + { "then", THEN }, + { "else", ELSE }, + { "elif", ELIF }, + { "fi", FI }, + { "case", CASE }, + { "esac", ESAC }, + { "for", FOR }, +#if defined (SELECT_COMMAND) + { "select", SELECT }, +#endif + { "while", WHILE }, + { "until", UNTIL }, + { "do", DO }, + { "done", DONE }, + { "in", IN }, + { "function", FUNCTION }, +#if defined (COMMAND_TIMING) + { "time", TIME }, +#endif + { "{", '{' }, + { "}", '}' }, + { "!", BANG }, +#if defined (COND_COMMAND) + { "[[", COND_START }, + { "]]", COND_END }, +#endif +#if defined (COPROCESS_SUPPORT) + { "coproc", COPROC }, +#endif + { (char *)NULL, 0} +}; + +/* other tokens that can be returned by read_token() */ +STRING_INT_ALIST other_token_alist[] = { + /* Multiple-character tokens with special values */ + { "--", TIMEIGN }, + { "-p", TIMEOPT }, + { "&&", AND_AND }, + { "||", OR_OR }, + { ">>", GREATER_GREATER }, + { "<<", LESS_LESS }, + { "<&", LESS_AND }, + { ">&", GREATER_AND }, + { ";;", SEMI_SEMI }, + { ";&", SEMI_AND }, + { ";;&", SEMI_SEMI_AND }, + { "<<-", LESS_LESS_MINUS }, + { "<<<", LESS_LESS_LESS }, + { "&>", AND_GREATER }, + { "&>>", AND_GREATER_GREATER }, + { "<>", LESS_GREATER }, + { ">|", GREATER_BAR }, + { "|&", BAR_AND }, + { "EOF", yacc_EOF }, + /* Tokens whose value is the character itself */ + { ">", '>' }, + { "<", '<' }, + { "-", '-' }, + { "{", '{' }, + { "}", '}' }, + { ";", ';' }, + { "(", '(' }, + { ")", ')' }, + { "|", '|' }, + { "&", '&' }, + { "newline", '\n' }, + { (char *)NULL, 0} +}; + +/* others not listed here: + WORD look at yylval.word + ASSIGNMENT_WORD look at yylval.word + NUMBER look at yylval.number + ARITH_CMD look at yylval.word_list + ARITH_FOR_EXPRS look at yylval.word_list + COND_CMD look at yylval.command +*/ + +/* These are used by read_token_word, but appear up here so that shell_getc + can use them to decide when to add otherwise blank lines to the history. */ + +/* The primary delimiter stack. */ +struct dstack dstack = { (char *)NULL, 0, 0 }; + +/* A temporary delimiter stack to be used when decoding prompt strings. + This is needed because command substitutions in prompt strings (e.g., PS2) + can screw up the parser's quoting state. */ +static struct dstack temp_dstack = { (char *)NULL, 0, 0 }; + +/* Macro for accessing the top delimiter on the stack. Returns the + delimiter or zero if none. */ +#define current_delimiter(ds) \ + (ds.delimiter_depth ? ds.delimiters[ds.delimiter_depth - 1] : 0) + +#define push_delimiter(ds, character) \ + do \ + { \ + if (ds.delimiter_depth + 2 > ds.delimiter_space) \ + ds.delimiters = (char *)xrealloc \ + (ds.delimiters, (ds.delimiter_space += 10) * sizeof (char)); \ + ds.delimiters[ds.delimiter_depth] = character; \ + ds.delimiter_depth++; \ + } \ + while (0) + +#define pop_delimiter(ds) ds.delimiter_depth-- + +/* Return the next shell input character. This always reads characters + from shell_input_line; when that line is exhausted, it is time to + read the next line. This is called by read_token when the shell is + processing normal command input. */ + +/* This implements one-character lookahead/lookbehind across physical input + lines, to avoid something being lost because it's pushed back with + shell_ungetc when we're at the start of a line. */ +static int eol_ungetc_lookahead = 0; + +static int +shell_getc (remove_quoted_newline) + int remove_quoted_newline; +{ + register int i; + int c; + unsigned char uc; + + QUIT; + + if (sigwinch_received) + { + sigwinch_received = 0; + get_new_window_size (0, (int *)0, (int *)0); + } + + if (eol_ungetc_lookahead) + { + c = eol_ungetc_lookahead; + eol_ungetc_lookahead = 0; + return (c); + } + +#if defined (ALIAS) || defined (DPAREN_ARITHMETIC) + /* If shell_input_line[shell_input_line_index] == 0, but there is + something on the pushed list of strings, then we don't want to go + off and get another line. We let the code down below handle it. */ + + if (!shell_input_line || ((!shell_input_line[shell_input_line_index]) && + (pushed_string_list == (STRING_SAVER *)NULL))) +#else /* !ALIAS && !DPAREN_ARITHMETIC */ + if (!shell_input_line || !shell_input_line[shell_input_line_index]) +#endif /* !ALIAS && !DPAREN_ARITHMETIC */ + { + line_number++; + + restart_read: + + /* Allow immediate exit if interrupted during input. */ + QUIT; + + i = 0; + shell_input_line_terminator = 0; + + /* If the shell is interatctive, but not currently printing a prompt + (interactive_shell && interactive == 0), we don't want to print + notifies or cleanup the jobs -- we want to defer it until we do + print the next prompt. */ + if (interactive_shell == 0 || SHOULD_PROMPT()) + { +#if defined (JOB_CONTROL) + /* This can cause a problem when reading a command as the result + of a trap, when the trap is called from flush_child. This call + had better not cause jobs to disappear from the job table in + that case, or we will have big trouble. */ + notify_and_cleanup (); +#else /* !JOB_CONTROL */ + cleanup_dead_jobs (); +#endif /* !JOB_CONTROL */ + } + +#if defined (READLINE) + if (no_line_editing && SHOULD_PROMPT()) +#else + if (SHOULD_PROMPT()) +#endif + print_prompt (); + + if (bash_input.type == st_stream) + clearerr (stdin); + + while (1) + { + c = yy_getc (); + + /* Allow immediate exit if interrupted during input. */ + QUIT; + + if (c == '\0') + { +#if 0 + internal_warning ("shell_getc: ignored null byte in input"); +#endif + continue; + } + + RESIZE_MALLOCED_BUFFER (shell_input_line, i, 2, shell_input_line_size, 256); + + if (c == EOF) + { + if (bash_input.type == st_stream) + clearerr (stdin); + + if (i == 0) + shell_input_line_terminator = EOF; + + shell_input_line[i] = '\0'; + break; + } + + shell_input_line[i++] = c; + + if (c == '\n') + { + shell_input_line[--i] = '\0'; + current_command_line_count++; + break; + } + } + + shell_input_line_index = 0; + shell_input_line_len = i; /* == strlen (shell_input_line) */ + + set_line_mbstate (); + +#if defined (HISTORY) + if (remember_on_history && shell_input_line && shell_input_line[0]) + { + char *expansions; +# if defined (BANG_HISTORY) + int old_hist; + + /* If the current delimiter is a single quote, we should not be + performing history expansion, even if we're on a different + line from the original single quote. */ + old_hist = history_expansion_inhibited; + if (current_delimiter (dstack) == '\'') + history_expansion_inhibited = 1; +# endif + expansions = pre_process_line (shell_input_line, 1, 1); +# if defined (BANG_HISTORY) + history_expansion_inhibited = old_hist; +# endif + if (expansions != shell_input_line) + { + free (shell_input_line); + shell_input_line = expansions; + shell_input_line_len = shell_input_line ? + strlen (shell_input_line) : 0; + if (shell_input_line_len == 0) + current_command_line_count--; + + /* We have to force the xrealloc below because we don't know + the true allocated size of shell_input_line anymore. */ + shell_input_line_size = shell_input_line_len; + + set_line_mbstate (); + } + } + /* Try to do something intelligent with blank lines encountered while + entering multi-line commands. XXX - this is grotesque */ + else if (remember_on_history && shell_input_line && + shell_input_line[0] == '\0' && + current_command_line_count > 1) + { + if (current_delimiter (dstack)) + /* We know shell_input_line[0] == 0 and we're reading some sort of + quoted string. This means we've got a line consisting of only + a newline in a quoted string. We want to make sure this line + gets added to the history. */ + maybe_add_history (shell_input_line); + else + { + char *hdcs; + hdcs = history_delimiting_chars (shell_input_line); + if (hdcs && hdcs[0] == ';') + maybe_add_history (shell_input_line); + } + } + +#endif /* HISTORY */ + + if (shell_input_line) + { + /* Lines that signify the end of the shell's input should not be + echoed. */ + if (echo_input_at_read && (shell_input_line[0] || + shell_input_line_terminator != EOF)) + fprintf (stderr, "%s\n", shell_input_line); + } + else + { + shell_input_line_size = 0; + prompt_string_pointer = ¤t_prompt_string; + if (SHOULD_PROMPT ()) + prompt_again (); + goto restart_read; + } + + /* Add the newline to the end of this string, iff the string does + not already end in an EOF character. */ + if (shell_input_line_terminator != EOF) + { + if (shell_input_line_len + 3 > shell_input_line_size) + shell_input_line = (char *)xrealloc (shell_input_line, + 1 + (shell_input_line_size += 2)); + + shell_input_line[shell_input_line_len] = '\n'; + shell_input_line[shell_input_line_len + 1] = '\0'; + + set_line_mbstate (); + } + } + +next_alias_char: + uc = shell_input_line[shell_input_line_index]; + + if (uc) + shell_input_line_index++; + +#if defined (ALIAS) || defined (DPAREN_ARITHMETIC) + /* If UC is NULL, we have reached the end of the current input string. If + pushed_string_list is non-empty, it's time to pop to the previous string + because we have fully consumed the result of the last alias expansion. + Do it transparently; just return the next character of the string popped + to. */ +pop_alias: + if (uc == 0 && (pushed_string_list != (STRING_SAVER *)NULL)) + { + pop_string (); + uc = shell_input_line[shell_input_line_index]; + if (uc) + shell_input_line_index++; + } +#endif /* ALIAS || DPAREN_ARITHMETIC */ + + if MBTEST(uc == '\\' && remove_quoted_newline && shell_input_line[shell_input_line_index] == '\n') + { + if (SHOULD_PROMPT ()) + prompt_again (); + line_number++; + /* What do we do here if we're expanding an alias whose definition + includes an escaped newline? If that's the last character in the + alias expansion, we just pop the pushed string list (recall that + we inhibit the appending of a space in mk_alexpansion() if newline + is the last character). If it's not the last character, we need + to consume the quoted newline and move to the next character in + the expansion. */ +#if defined (ALIAS) + if (expanding_alias () && shell_input_line[shell_input_line_index+1] == '\0') + { + uc = 0; + goto pop_alias; + } + else if (expanding_alias () && shell_input_line[shell_input_line_index+1] != '\0') + { + shell_input_line_index++; /* skip newline */ + goto next_alias_char; /* and get next character */ + } + else +#endif + goto restart_read; + } + + if (uc == 0 && shell_input_line_terminator == EOF) + return ((shell_input_line_index != 0) ? '\n' : EOF); + + return (uc); +} + +/* Put C back into the input for the shell. This might need changes for + HANDLE_MULTIBYTE around EOLs. Since we (currently) never push back a + character different than we read, shell_input_line_property doesn't need + to change when manipulating shell_input_line. The define for + last_shell_getc_is_singlebyte should take care of it, though. */ +static void +shell_ungetc (c) + int c; +{ + if (shell_input_line && shell_input_line_index) + shell_input_line[--shell_input_line_index] = c; + else + eol_ungetc_lookahead = c; +} + +#ifdef INCLUDE_UNUSED +/* Back the input pointer up by one, effectively `ungetting' a character. */ +static void +shell_ungetchar () +{ + if (shell_input_line && shell_input_line_index) + shell_input_line_index--; +} +#endif + +/* Discard input until CHARACTER is seen, then push that character back + onto the input stream. */ +static void +discard_until (character) + int character; +{ + int c; + + while ((c = shell_getc (0)) != EOF && c != character) + ; + + if (c != EOF) + shell_ungetc (c); +} + +void +execute_variable_command (command, vname) + char *command, *vname; +{ + char *last_lastarg; + sh_parser_state_t ps; + + save_parser_state (&ps); + last_lastarg = get_string_value ("_"); + if (last_lastarg) + last_lastarg = savestring (last_lastarg); + + parse_and_execute (savestring (command), vname, SEVAL_NONINT|SEVAL_NOHIST); + + restore_parser_state (&ps); + bind_variable ("_", last_lastarg, 0); + FREE (last_lastarg); + + if (token_to_read == '\n') /* reset_parser was called */ + token_to_read = 0; +} + +/* Place to remember the token. We try to keep the buffer + at a reasonable size, but it can grow. */ +static char *token = (char *)NULL; + +/* Current size of the token buffer. */ +static int token_buffer_size; + +/* Command to read_token () explaining what we want it to do. */ +#define READ 0 +#define RESET 1 +#define prompt_is_ps1 \ + (!prompt_string_pointer || prompt_string_pointer == &ps1_prompt) + +/* Function for yyparse to call. yylex keeps track of + the last two tokens read, and calls read_token. */ +static int +yylex () +{ + if (interactive && (current_token == 0 || current_token == '\n')) + { + /* Before we print a prompt, we might have to check mailboxes. + We do this only if it is time to do so. Notice that only here + is the mail alarm reset; nothing takes place in check_mail () + except the checking of mail. Please don't change this. */ + if (prompt_is_ps1 && parse_and_execute_level == 0 && time_to_check_mail ()) + { + check_mail (); + reset_mail_timer (); + } + + /* Avoid printing a prompt if we're not going to read anything, e.g. + after resetting the parser with read_token (RESET). */ + if (token_to_read == 0 && SHOULD_PROMPT ()) + prompt_again (); + } + + two_tokens_ago = token_before_that; + token_before_that = last_read_token; + last_read_token = current_token; + current_token = read_token (READ); + + if ((parser_state & PST_EOFTOKEN) && current_token == shell_eof_token) + { + current_token = yacc_EOF; + if (bash_input.type == st_string) + rewind_input_string (); + } + parser_state &= ~PST_EOFTOKEN; + + return (current_token); +} + +/* When non-zero, we have read the required tokens + which allow ESAC to be the next one read. */ +static int esacs_needed_count; + +void +gather_here_documents () +{ + int r; + + r = 0; + while (need_here_doc) + { + parser_state |= PST_HEREDOC; + make_here_document (redir_stack[r++], line_number); + parser_state &= ~PST_HEREDOC; + need_here_doc--; + } +} + +/* When non-zero, an open-brace used to create a group is awaiting a close + brace partner. */ +static int open_brace_count; + +#define command_token_position(token) \ + (((token) == ASSIGNMENT_WORD) || (parser_state&PST_REDIRLIST) || \ + ((token) != SEMI_SEMI && (token) != SEMI_AND && (token) != SEMI_SEMI_AND && reserved_word_acceptable(token))) + +#define assignment_acceptable(token) \ + (command_token_position(token) && ((parser_state & PST_CASEPAT) == 0)) + +/* Check to see if TOKEN is a reserved word and return the token + value if it is. */ +#define CHECK_FOR_RESERVED_WORD(tok) \ + do { \ + if (!dollar_present && !quoted && \ + reserved_word_acceptable (last_read_token)) \ + { \ + int i; \ + for (i = 0; word_token_alist[i].word != (char *)NULL; i++) \ + if (STREQ (tok, word_token_alist[i].word)) \ + { \ + if ((parser_state & PST_CASEPAT) && (word_token_alist[i].token != ESAC)) \ + break; \ + if (word_token_alist[i].token == TIME && time_command_acceptable () == 0) \ + break; \ + if (word_token_alist[i].token == ESAC) \ + parser_state &= ~(PST_CASEPAT|PST_CASESTMT); \ + else if (word_token_alist[i].token == CASE) \ + parser_state |= PST_CASESTMT; \ + else if (word_token_alist[i].token == COND_END) \ + parser_state &= ~(PST_CONDCMD|PST_CONDEXPR); \ + else if (word_token_alist[i].token == COND_START) \ + parser_state |= PST_CONDCMD; \ + else if (word_token_alist[i].token == '{') \ + open_brace_count++; \ + else if (word_token_alist[i].token == '}' && open_brace_count) \ + open_brace_count--; \ + return (word_token_alist[i].token); \ + } \ + } \ + } while (0) + +#if defined (ALIAS) + + /* OK, we have a token. Let's try to alias expand it, if (and only if) + it's eligible. + + It is eligible for expansion if EXPAND_ALIASES is set, and + the token is unquoted and the last token read was a command + separator (or expand_next_token is set), and we are currently + processing an alias (pushed_string_list is non-empty) and this + token is not the same as the current or any previously + processed alias. + + Special cases that disqualify: + In a pattern list in a case statement (parser_state & PST_CASEPAT). */ + +static char * +mk_alexpansion (s) + char *s; +{ + int l; + char *r; + + l = strlen (s); + r = xmalloc (l + 2); + strcpy (r, s); + /* If the last character in the alias is a newline, don't add a trailing + space to the expansion. Works with shell_getc above. */ + if (r[l - 1] != ' ' && r[l - 1] != '\n') + r[l++] = ' '; + r[l] = '\0'; + return r; +} + +static int +alias_expand_token (tokstr) + char *tokstr; +{ + char *expanded; + alias_t *ap; + + if (((parser_state & PST_ALEXPNEXT) || command_token_position (last_read_token)) && + (parser_state & PST_CASEPAT) == 0) + { + ap = find_alias (tokstr); + + /* Currently expanding this token. */ + if (ap && (ap->flags & AL_BEINGEXPANDED)) + return (NO_EXPANSION); + + /* mk_alexpansion puts an extra space on the end of the alias expansion, + so the lookahead by the parser works right. If this gets changed, + make sure the code in shell_getc that deals with reaching the end of + an expanded alias is changed with it. */ + expanded = ap ? mk_alexpansion (ap->value) : (char *)NULL; + + if (expanded) + { + push_string (expanded, ap->flags & AL_EXPANDNEXT, ap); + return (RE_READ_TOKEN); + } + else + /* This is an eligible token that does not have an expansion. */ + return (NO_EXPANSION); + } + return (NO_EXPANSION); +} +#endif /* ALIAS */ + +static int +time_command_acceptable () +{ +#if defined (COMMAND_TIMING) + int i; + + if (posixly_correct && shell_compatibility_level > 41) + { + /* Quick check of the rest of the line to find the next token. If it + begins with a `-', Posix says to not return `time' as the token. + This was interp 267. */ + i = shell_input_line_index; + while (i < shell_input_line_len && (shell_input_line[i] == ' ' || shell_input_line[i] == '\t')) + i++; + if (shell_input_line[i] == '-') + return 0; + } + + switch (last_read_token) + { + case 0: + case ';': + case '\n': + case AND_AND: + case OR_OR: + case '&': + case DO: + case THEN: + case ELSE: + case '{': /* } */ + case '(': /* ) */ + case BANG: /* ! time pipeline */ + case TIME: /* time time pipeline */ + case TIMEOPT: /* time -p time pipeline */ + case TIMEIGN: /* time -p -- ... */ + return 1; + default: + return 0; + } +#else + return 0; +#endif /* COMMAND_TIMING */ +} + +/* Handle special cases of token recognition: + IN is recognized if the last token was WORD and the token + before that was FOR or CASE or SELECT. + + DO is recognized if the last token was WORD and the token + before that was FOR or SELECT. + + ESAC is recognized if the last token caused `esacs_needed_count' + to be set + + `{' is recognized if the last token as WORD and the token + before that was FUNCTION, or if we just parsed an arithmetic + `for' command. + + `}' is recognized if there is an unclosed `{' present. + + `-p' is returned as TIMEOPT if the last read token was TIME. + `--' is returned as TIMEIGN if the last read token was TIMEOPT. + + ']]' is returned as COND_END if the parser is currently parsing + a conditional expression ((parser_state & PST_CONDEXPR) != 0) + + `time' is returned as TIME if and only if it is immediately + preceded by one of `;', `\n', `||', `&&', or `&'. +*/ + +static int +special_case_tokens (tokstr) + char *tokstr; +{ + if ((last_read_token == WORD) && +#if defined (SELECT_COMMAND) + ((token_before_that == FOR) || (token_before_that == CASE) || (token_before_that == SELECT)) && +#else + ((token_before_that == FOR) || (token_before_that == CASE)) && +#endif + (tokstr[0] == 'i' && tokstr[1] == 'n' && tokstr[2] == 0)) + { + if (token_before_that == CASE) + { + parser_state |= PST_CASEPAT; + esacs_needed_count++; + } + return (IN); + } + + if (last_read_token == WORD && +#if defined (SELECT_COMMAND) + (token_before_that == FOR || token_before_that == SELECT) && +#else + (token_before_that == FOR) && +#endif + (tokstr[0] == 'd' && tokstr[1] == 'o' && tokstr[2] == '\0')) + return (DO); + + /* Ditto for ESAC in the CASE case. + Specifically, this handles "case word in esac", which is a legal + construct, certainly because someone will pass an empty arg to the + case construct, and we don't want it to barf. Of course, we should + insist that the case construct has at least one pattern in it, but + the designers disagree. */ + if (esacs_needed_count) + { + esacs_needed_count--; + if (STREQ (tokstr, "esac")) + { + parser_state &= ~PST_CASEPAT; + return (ESAC); + } + } + + /* The start of a shell function definition. */ + if (parser_state & PST_ALLOWOPNBRC) + { + parser_state &= ~PST_ALLOWOPNBRC; + if (tokstr[0] == '{' && tokstr[1] == '\0') /* } */ + { + open_brace_count++; + function_bstart = line_number; + return ('{'); /* } */ + } + } + + /* We allow a `do' after a for ((...)) without an intervening + list_terminator */ + if (last_read_token == ARITH_FOR_EXPRS && tokstr[0] == 'd' && tokstr[1] == 'o' && !tokstr[2]) + return (DO); + if (last_read_token == ARITH_FOR_EXPRS && tokstr[0] == '{' && tokstr[1] == '\0') /* } */ + { + open_brace_count++; + return ('{'); /* } */ + } + + if (open_brace_count && reserved_word_acceptable (last_read_token) && tokstr[0] == '}' && !tokstr[1]) + { + open_brace_count--; /* { */ + return ('}'); + } + +#if defined (COMMAND_TIMING) + /* Handle -p after `time'. */ + if (last_read_token == TIME && tokstr[0] == '-' && tokstr[1] == 'p' && !tokstr[2]) + return (TIMEOPT); + /* Handle -- after `time -p'. */ + if (last_read_token == TIMEOPT && tokstr[0] == '-' && tokstr[1] == '-' && !tokstr[2]) + return (TIMEIGN); +#endif + +#if defined (COND_COMMAND) /* [[ */ + if ((parser_state & PST_CONDEXPR) && tokstr[0] == ']' && tokstr[1] == ']' && tokstr[2] == '\0') + return (COND_END); +#endif + + return (-1); +} + +/* Called from shell.c when Control-C is typed at top level. Or + by the error rule at top level. */ +void +reset_parser () +{ + dstack.delimiter_depth = 0; /* No delimiters found so far. */ + open_brace_count = 0; + +#if defined (EXTENDED_GLOB) + /* Reset to global value of extended glob */ + if (parser_state & PST_EXTPAT) + extended_glob = global_extglob; +#endif + + parser_state = 0; + +#if defined (ALIAS) || defined (DPAREN_ARITHMETIC) + if (pushed_string_list) + free_string_list (); +#endif /* ALIAS || DPAREN_ARITHMETIC */ + + if (shell_input_line) + { + free (shell_input_line); + shell_input_line = (char *)NULL; + shell_input_line_size = shell_input_line_index = 0; + } + + FREE (word_desc_to_read); + word_desc_to_read = (WORD_DESC *)NULL; + + current_token = '\n'; /* XXX */ + last_read_token = '\n'; + token_to_read = '\n'; +} + +/* Read the next token. Command can be READ (normal operation) or + RESET (to normalize state). */ +static int +read_token (command) + int command; +{ + int character; /* Current character. */ + int peek_char; /* Temporary look-ahead character. */ + int result; /* The thing to return. */ + + if (command == RESET) + { + reset_parser (); + return ('\n'); + } + + if (token_to_read) + { + result = token_to_read; + if (token_to_read == WORD || token_to_read == ASSIGNMENT_WORD) + { + yylval.word = word_desc_to_read; + word_desc_to_read = (WORD_DESC *)NULL; + } + token_to_read = 0; + return (result); + } + +#if defined (COND_COMMAND) + if ((parser_state & (PST_CONDCMD|PST_CONDEXPR)) == PST_CONDCMD) + { + cond_lineno = line_number; + parser_state |= PST_CONDEXPR; + yylval.command = parse_cond_command (); + if (cond_token != COND_END) + { + cond_error (); + return (-1); + } + token_to_read = COND_END; + parser_state &= ~(PST_CONDEXPR|PST_CONDCMD); + return (COND_CMD); + } +#endif + +#if defined (ALIAS) + /* This is a place to jump back to once we have successfully expanded a + token with an alias and pushed the string with push_string () */ + re_read_token: +#endif /* ALIAS */ + + /* Read a single word from input. Start by skipping blanks. */ + while ((character = shell_getc (1)) != EOF && shellblank (character)) + ; + + if (character == EOF) + { + EOF_Reached = 1; + return (yacc_EOF); + } + + if MBTEST(character == '#' && (!interactive || interactive_comments)) + { + /* A comment. Discard until EOL or EOF, and then return a newline. */ + discard_until ('\n'); + shell_getc (0); + character = '\n'; /* this will take the next if statement and return. */ + } + + if (character == '\n') + { + /* If we're about to return an unquoted newline, we can go and collect + the text of any pending here document. */ + if (need_here_doc) + gather_here_documents (); + +#if defined (ALIAS) + parser_state &= ~PST_ALEXPNEXT; +#endif /* ALIAS */ + + parser_state &= ~PST_ASSIGNOK; + + return (character); + } + + if (parser_state & PST_REGEXP) + goto tokword; + + /* Shell meta-characters. */ + if MBTEST(shellmeta (character) && ((parser_state & PST_DBLPAREN) == 0)) + { +#if defined (ALIAS) + /* Turn off alias tokenization iff this character sequence would + not leave us ready to read a command. */ + if (character == '<' || character == '>') + parser_state &= ~PST_ALEXPNEXT; +#endif /* ALIAS */ + + parser_state &= ~PST_ASSIGNOK; + + peek_char = shell_getc (1); + if (character == peek_char) + { + switch (character) + { + case '<': + /* If '<' then we could be at "<<" or at "<<-". We have to + look ahead one more character. */ + peek_char = shell_getc (1); + if MBTEST(peek_char == '-') + return (LESS_LESS_MINUS); + else if MBTEST(peek_char == '<') + return (LESS_LESS_LESS); + else + { + shell_ungetc (peek_char); + return (LESS_LESS); + } + + case '>': + return (GREATER_GREATER); + + case ';': + parser_state |= PST_CASEPAT; +#if defined (ALIAS) + parser_state &= ~PST_ALEXPNEXT; +#endif /* ALIAS */ + + peek_char = shell_getc (1); + if MBTEST(peek_char == '&') + return (SEMI_SEMI_AND); + else + { + shell_ungetc (peek_char); + return (SEMI_SEMI); + } + + case '&': + return (AND_AND); + + case '|': + return (OR_OR); + +#if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND) + case '(': /* ) */ + result = parse_dparen (character); + if (result == -2) + break; + else + return result; +#endif + } + } + else if MBTEST(character == '<' && peek_char == '&') + return (LESS_AND); + else if MBTEST(character == '>' && peek_char == '&') + return (GREATER_AND); + else if MBTEST(character == '<' && peek_char == '>') + return (LESS_GREATER); + else if MBTEST(character == '>' && peek_char == '|') + return (GREATER_BAR); + else if MBTEST(character == '&' && peek_char == '>') + { + peek_char = shell_getc (1); + if MBTEST(peek_char == '>') + return (AND_GREATER_GREATER); + else + { + shell_ungetc (peek_char); + return (AND_GREATER); + } + } + else if MBTEST(character == '|' && peek_char == '&') + return (BAR_AND); + else if MBTEST(character == ';' && peek_char == '&') + { + parser_state |= PST_CASEPAT; +#if defined (ALIAS) + parser_state &= ~PST_ALEXPNEXT; +#endif /* ALIAS */ + return (SEMI_AND); + } + + shell_ungetc (peek_char); + + /* If we look like we are reading the start of a function + definition, then let the reader know about it so that + we will do the right thing with `{'. */ + if MBTEST(character == ')' && last_read_token == '(' && token_before_that == WORD) + { + parser_state |= PST_ALLOWOPNBRC; +#if defined (ALIAS) + parser_state &= ~PST_ALEXPNEXT; +#endif /* ALIAS */ + function_dstart = line_number; + } + + /* case pattern lists may be preceded by an optional left paren. If + we're not trying to parse a case pattern list, the left paren + indicates a subshell. */ + if MBTEST(character == '(' && (parser_state & PST_CASEPAT) == 0) /* ) */ + parser_state |= PST_SUBSHELL; + /*(*/ + else if MBTEST((parser_state & PST_CASEPAT) && character == ')') + parser_state &= ~PST_CASEPAT; + /*(*/ + else if MBTEST((parser_state & PST_SUBSHELL) && character == ')') + parser_state &= ~PST_SUBSHELL; + +#if defined (PROCESS_SUBSTITUTION) + /* Check for the constructs which introduce process substitution. + Shells running in `posix mode' don't do process substitution. */ + if MBTEST(posixly_correct || ((character != '>' && character != '<') || peek_char != '(')) /*)*/ +#endif /* PROCESS_SUBSTITUTION */ + return (character); + } + + /* Hack <&- (close stdin) case. Also <&N- (dup and close). */ + if MBTEST(character == '-' && (last_read_token == LESS_AND || last_read_token == GREATER_AND)) + return (character); + +tokword: + /* Okay, if we got this far, we have to read a word. Read one, + and then check it against the known ones. */ + result = read_token_word (character); +#if defined (ALIAS) + if (result == RE_READ_TOKEN) + goto re_read_token; +#endif + return result; +} + +/* + * Match a $(...) or other grouping construct. This has to handle embedded + * quoted strings ('', ``, "") and nested constructs. It also must handle + * reprompting the user, if necessary, after reading a newline, and returning + * correct error values if it reads EOF. + */ +#define P_FIRSTCLOSE 0x0001 +#define P_ALLOWESC 0x0002 +#define P_DQUOTE 0x0004 +#define P_COMMAND 0x0008 /* parsing a command, so look for comments */ +#define P_BACKQUOTE 0x0010 /* parsing a backquoted command substitution */ +#define P_ARRAYSUB 0x0020 /* parsing a [...] array subscript for assignment */ +#define P_DOLBRACE 0x0040 /* parsing a ${...} construct */ + +/* Lexical state while parsing a grouping construct or $(...). */ +#define LEX_WASDOL 0x001 +#define LEX_CKCOMMENT 0x002 +#define LEX_INCOMMENT 0x004 +#define LEX_PASSNEXT 0x008 +#define LEX_RESWDOK 0x010 +#define LEX_CKCASE 0x020 +#define LEX_INCASE 0x040 +#define LEX_INHEREDOC 0x080 +#define LEX_HEREDELIM 0x100 /* reading here-doc delimiter */ +#define LEX_STRIPDOC 0x200 /* <<- strip tabs from here doc delim */ +#define LEX_INWORD 0x400 + +#define COMSUB_META(ch) ((ch) == ';' || (ch) == '&' || (ch) == '|') + +#define CHECK_NESTRET_ERROR() \ + do { \ + if (nestret == &matched_pair_error) \ + { \ + free (ret); \ + return &matched_pair_error; \ + } \ + } while (0) + +#define APPEND_NESTRET() \ + do { \ + if (nestlen) \ + { \ + RESIZE_MALLOCED_BUFFER (ret, retind, nestlen, retsize, 64); \ + strcpy (ret + retind, nestret); \ + retind += nestlen; \ + } \ + } while (0) + +static char matched_pair_error; + +static char * +parse_matched_pair (qc, open, close, lenp, flags) + int qc; /* `"' if this construct is within double quotes */ + int open, close; + int *lenp, flags; +{ + int count, ch, tflags; + int nestlen, ttranslen, start_lineno; + char *ret, *nestret, *ttrans; + int retind, retsize, rflags; + int dolbrace_state; + + dolbrace_state = (flags & P_DOLBRACE) ? DOLBRACE_PARAM : 0; + +/*itrace("parse_matched_pair[%d]: open = %c close = %c flags = %d", line_number, open, close, flags);*/ + count = 1; + tflags = 0; + + if ((flags & P_COMMAND) && qc != '`' && qc != '\'' && qc != '"' && (flags & P_DQUOTE) == 0) + tflags |= LEX_CKCOMMENT; + + /* RFLAGS is the set of flags we want to pass to recursive calls. */ + rflags = (qc == '"') ? P_DQUOTE : (flags & P_DQUOTE); + + ret = (char *)xmalloc (retsize = 64); + retind = 0; + + start_lineno = line_number; + while (count) + { + ch = shell_getc (qc != '\'' && (tflags & (LEX_PASSNEXT)) == 0); + + if (ch == EOF) + { + free (ret); + parser_error (start_lineno, _("unexpected EOF while looking for matching `%c'"), close); + EOF_Reached = 1; /* XXX */ + return (&matched_pair_error); + } + + /* Possible reprompting. */ + if (ch == '\n' && SHOULD_PROMPT ()) + prompt_again (); + + /* Don't bother counting parens or doing anything else if in a comment + or part of a case statement */ + if (tflags & LEX_INCOMMENT) + { + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + + if (ch == '\n') + tflags &= ~LEX_INCOMMENT; + + continue; + } + + /* Not exactly right yet, should handle shell metacharacters, too. If + any changes are made to this test, make analogous changes to subst.c: + extract_delimited_string(). */ + else if MBTEST((tflags & LEX_CKCOMMENT) && (tflags & LEX_INCOMMENT) == 0 && ch == '#' && (retind == 0 || ret[retind-1] == '\n' || shellblank (ret[retind - 1]))) + tflags |= LEX_INCOMMENT; + + if (tflags & LEX_PASSNEXT) /* last char was backslash */ + { + tflags &= ~LEX_PASSNEXT; + if (qc != '\'' && ch == '\n') /* double-quoted \ disappears. */ + { + if (retind > 0) + retind--; /* swallow previously-added backslash */ + continue; + } + + RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64); +#if 0 + if MBTEST(ch == CTLESC || ch == CTLNUL) +#else + if MBTEST(ch == CTLESC) +#endif + ret[retind++] = CTLESC; + ret[retind++] = ch; + continue; + } + /* If we're reparsing the input (e.g., from parse_string_to_word_list), + we've already prepended CTLESC to single-quoted results of $'...'. + We may want to do this for other CTLESC-quoted characters in + reparse, too. */ + else if MBTEST((parser_state & PST_REPARSE) && open == '\'' && (ch == CTLESC || ch == CTLNUL)) + { + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + continue; + } + else if MBTEST(ch == CTLESC || ch == CTLNUL) /* special shell escapes */ + { + RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64); + ret[retind++] = CTLESC; + ret[retind++] = ch; + continue; + } + else if MBTEST(ch == close) /* ending delimiter */ + count--; + /* handle nested ${...} specially. */ + else if MBTEST(open != close && (tflags & LEX_WASDOL) && open == '{' && ch == open) /* } */ + count++; + else if MBTEST(((flags & P_FIRSTCLOSE) == 0) && ch == open) /* nested begin */ + count++; + + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + + /* If we just read the ending character, don't bother continuing. */ + if (count == 0) + break; + + if (open == '\'') /* '' inside grouping construct */ + { + if MBTEST((flags & P_ALLOWESC) && ch == '\\') + tflags |= LEX_PASSNEXT; + continue; + } + + if MBTEST(ch == '\\') /* backslashes */ + tflags |= LEX_PASSNEXT; + + /* Based on which dolstate is currently in (param, op, or word), + decide what the op is. We're really only concerned if it's % or + #, so we can turn on a flag that says whether or not we should + treat single quotes as special when inside a double-quoted + ${...}. This logic must agree with subst.c:extract_dollar_brace_string + since they share the same defines. */ + if (flags & P_DOLBRACE) + { + /* ${param%[%]word} */ + if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '%' && retind > 1) + dolbrace_state = DOLBRACE_QUOTE; + /* ${param#[#]word} */ + else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '#' && retind > 1) + dolbrace_state = DOLBRACE_QUOTE; + /* ${param/[/]pat/rep} */ + else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '/' && retind > 1) + dolbrace_state = DOLBRACE_QUOTE; + /* ${param^[^]pat} */ + else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == '^' && retind > 1) + dolbrace_state = DOLBRACE_QUOTE; + /* ${param,[,]pat} */ + else if MBTEST(dolbrace_state == DOLBRACE_PARAM && ch == ',' && retind > 1) + dolbrace_state = DOLBRACE_QUOTE; + else if MBTEST(dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", ch) != 0) + dolbrace_state = DOLBRACE_OP; + else if MBTEST(dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", ch) == 0) + dolbrace_state = DOLBRACE_WORD; + } + + /* The big hammer. Single quotes aren't special in double quotes. The + problem is that Posix used to say the single quotes are semi-special: + within a double-quoted ${...} construct "an even number of + unescaped double-quotes or single-quotes, if any, shall occur." */ + /* This was changed in Austin Group Interp 221 */ + if MBTEST(posixly_correct && shell_compatibility_level > 41 && dolbrace_state != DOLBRACE_QUOTE && (flags & P_DQUOTE) && (flags & P_DOLBRACE) && ch == '\'') + continue; + + /* Could also check open == '`' if we want to parse grouping constructs + inside old-style command substitution. */ + if (open != close) /* a grouping construct */ + { + if MBTEST(shellquote (ch)) + { + /* '', ``, or "" inside $(...) or other grouping construct. */ + push_delimiter (dstack, ch); + if MBTEST((tflags & LEX_WASDOL) && ch == '\'') /* $'...' inside group */ + nestret = parse_matched_pair (ch, ch, ch, &nestlen, P_ALLOWESC|rflags); + else + nestret = parse_matched_pair (ch, ch, ch, &nestlen, rflags); + pop_delimiter (dstack); + CHECK_NESTRET_ERROR (); + + if MBTEST((tflags & LEX_WASDOL) && ch == '\'' && (extended_quote || (rflags & P_DQUOTE) == 0)) + { + /* Translate $'...' here. */ + ttrans = ansiexpand (nestret, 0, nestlen - 1, &ttranslen); + xfree (nestret); + + if ((rflags & P_DQUOTE) == 0) + { + nestret = sh_single_quote (ttrans); + free (ttrans); + nestlen = strlen (nestret); + } + else + { + nestret = ttrans; + nestlen = ttranslen; + } + retind -= 2; /* back up before the $' */ + } + else if MBTEST((tflags & LEX_WASDOL) && ch == '"' && (extended_quote || (rflags & P_DQUOTE) == 0)) + { + /* Locale expand $"..." here. */ + ttrans = localeexpand (nestret, 0, nestlen - 1, start_lineno, &ttranslen); + xfree (nestret); + + nestret = sh_mkdoublequoted (ttrans, ttranslen, 0); + free (ttrans); + nestlen = ttranslen + 2; + retind -= 2; /* back up before the $" */ + } + + APPEND_NESTRET (); + FREE (nestret); + } + else if ((flags & P_ARRAYSUB) && (tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */ + goto parse_dollar_word; + } + /* Parse an old-style command substitution within double quotes as a + single word. */ + /* XXX - sh and ksh93 don't do this - XXX */ + else if MBTEST(open == '"' && ch == '`') + { + nestret = parse_matched_pair (0, '`', '`', &nestlen, rflags); + + CHECK_NESTRET_ERROR (); + APPEND_NESTRET (); + + FREE (nestret); + } + else if MBTEST(open != '`' && (tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */ + /* check for $(), $[], or ${} inside quoted string. */ + { +parse_dollar_word: + if (open == ch) /* undo previous increment */ + count--; + if (ch == '(') /* ) */ + nestret = parse_comsub (0, '(', ')', &nestlen, (rflags|P_COMMAND) & ~P_DQUOTE); + else if (ch == '{') /* } */ + nestret = parse_matched_pair (0, '{', '}', &nestlen, P_FIRSTCLOSE|P_DOLBRACE|rflags); + else if (ch == '[') /* ] */ + nestret = parse_matched_pair (0, '[', ']', &nestlen, rflags); + + CHECK_NESTRET_ERROR (); + APPEND_NESTRET (); + + FREE (nestret); + } + if MBTEST(ch == '$') + tflags |= LEX_WASDOL; + else + tflags &= ~LEX_WASDOL; + } + + ret[retind] = '\0'; + if (lenp) + *lenp = retind; +/*itrace("parse_matched_pair[%d]: returning %s", line_number, ret);*/ + return ret; +} + +/* Parse a $(...) command substitution. This is messier than I'd like, and + reproduces a lot more of the token-reading code than I'd like. */ +static char * +parse_comsub (qc, open, close, lenp, flags) + int qc; /* `"' if this construct is within double quotes */ + int open, close; + int *lenp, flags; +{ + int count, ch, peekc, tflags, lex_rwlen, lex_wlen, lex_firstind; + int nestlen, ttranslen, start_lineno; + char *ret, *nestret, *ttrans, *heredelim; + int retind, retsize, rflags, hdlen; + + /* Posix interp 217 says arithmetic expressions have precedence, so + assume $(( introduces arithmetic expansion and parse accordingly. */ + peekc = shell_getc (0); + shell_ungetc (peekc); + if (peekc == '(') + return (parse_matched_pair (qc, open, close, lenp, 0)); + +/*itrace("parse_comsub: qc = `%c' open = %c close = %c", qc, open, close);*/ + count = 1; + tflags = LEX_RESWDOK; + + if ((flags & P_COMMAND) && qc != '\'' && qc != '"' && (flags & P_DQUOTE) == 0) + tflags |= LEX_CKCASE; + if ((tflags & LEX_CKCASE) && (interactive == 0 || interactive_comments)) + tflags |= LEX_CKCOMMENT; + + /* RFLAGS is the set of flags we want to pass to recursive calls. */ + rflags = (flags & P_DQUOTE); + + ret = (char *)xmalloc (retsize = 64); + retind = 0; + + start_lineno = line_number; + lex_rwlen = lex_wlen = 0; + + heredelim = 0; + lex_firstind = -1; + + while (count) + { +comsub_readchar: + ch = shell_getc (qc != '\'' && (tflags & (LEX_INCOMMENT|LEX_PASSNEXT)) == 0); + + if (ch == EOF) + { +eof_error: + free (ret); + FREE (heredelim); + parser_error (start_lineno, _("unexpected EOF while looking for matching `%c'"), close); + EOF_Reached = 1; /* XXX */ + return (&matched_pair_error); + } + + /* If we hit the end of a line and are reading the contents of a here + document, and it's not the same line that the document starts on, + check for this line being the here doc delimiter. Otherwise, if + we're in a here document, mark the next character as the beginning + of a line. */ + if (ch == '\n') + { + if ((tflags & LEX_HEREDELIM) && heredelim) + { + tflags &= ~LEX_HEREDELIM; + tflags |= LEX_INHEREDOC; + lex_firstind = retind + 1; + } + else if (tflags & LEX_INHEREDOC) + { + int tind; + tind = lex_firstind; + while ((tflags & LEX_STRIPDOC) && ret[tind] == '\t') + tind++; + if (STREQN (ret + tind, heredelim, hdlen)) + { + tflags &= ~(LEX_STRIPDOC|LEX_INHEREDOC); +/*itrace("parse_comsub:%d: found here doc end `%s'", line_number, ret + tind);*/ + free (heredelim); + heredelim = 0; + lex_firstind = -1; + } + else + lex_firstind = retind + 1; + } + } + + /* Possible reprompting. */ + if (ch == '\n' && SHOULD_PROMPT ()) + prompt_again (); + + /* XXX -- possibly allow here doc to be delimited by ending right + paren. */ + if ((tflags & LEX_INHEREDOC) && ch == close && count == 1) + { + int tind; +/*itrace("parse_comsub: in here doc, ch == close, retind - firstind = %d hdlen = %d retind = %d", retind-lex_firstind, hdlen, retind);*/ + tind = lex_firstind; + while ((tflags & LEX_STRIPDOC) && ret[tind] == '\t') + tind++; + if (retind-tind == hdlen && STREQN (ret + tind, heredelim, hdlen)) + { + tflags &= ~(LEX_STRIPDOC|LEX_INHEREDOC); +/*itrace("parse_comsub:%d: found here doc end `%s'", line_number, ret + tind);*/ + free (heredelim); + heredelim = 0; + lex_firstind = -1; + } + } + + /* Don't bother counting parens or doing anything else if in a comment */ + if (tflags & (LEX_INCOMMENT|LEX_INHEREDOC)) + { + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + + if ((tflags & LEX_INCOMMENT) && ch == '\n') +{ +/*itrace("parse_comsub:%d: lex_incomment -> 0 ch = `%c'", line_number, ch);*/ + tflags &= ~LEX_INCOMMENT; +} + + continue; + } + + if (tflags & LEX_PASSNEXT) /* last char was backslash */ + { +/*itrace("parse_comsub:%d: lex_passnext -> 0 ch = `%c' (%d)", line_number, ch, __LINE__);*/ + tflags &= ~LEX_PASSNEXT; + if (qc != '\'' && ch == '\n') /* double-quoted \ disappears. */ + { + if (retind > 0) + retind--; /* swallow previously-added backslash */ + continue; + } + + RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64); +#if 0 + if MBTEST(ch == CTLESC || ch == CTLNUL) +#else + if MBTEST(ch == CTLESC) +#endif + ret[retind++] = CTLESC; + ret[retind++] = ch; + continue; + } + + /* If this is a shell break character, we are not in a word. If not, + we either start or continue a word. */ + if MBTEST(shellbreak (ch)) + { + tflags &= ~LEX_INWORD; +/*itrace("parse_comsub:%d: lex_inword -> 0 ch = `%c' (%d)", line_number, ch, __LINE__);*/ + } + else + { + if (tflags & LEX_INWORD) + { + lex_wlen++; +/*itrace("parse_comsub:%d: lex_inword == 1 ch = `%c' lex_wlen = %d (%d)", line_number, ch, lex_wlen, __LINE__);*/ + } + else + { +/*itrace("parse_comsub:%d: lex_inword -> 1 ch = `%c' (%d)", line_number, ch, __LINE__);*/ + tflags |= LEX_INWORD; + lex_wlen = 0; + } + } + + /* Skip whitespace */ + if MBTEST(shellblank (ch) && (tflags & LEX_HEREDELIM) == 0 && lex_rwlen == 0) + { + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + continue; + } + + /* Either we are looking for the start of the here-doc delimiter + (lex_firstind == -1) or we are reading one (lex_firstind >= 0). + If this character is a shell break character and we are reading + the delimiter, save it and note that we are now reading a here + document. If we've found the start of the delimiter, note it by + setting lex_firstind. Backslashes can quote shell metacharacters + in here-doc delimiters. */ + if (tflags & LEX_HEREDELIM) + { + if (lex_firstind == -1 && shellbreak (ch) == 0) + lex_firstind = retind; +#if 0 + else if (heredelim && (tflags & LEX_PASSNEXT) == 0 && ch == '\n') + { + tflags |= LEX_INHEREDOC; + tflags &= ~LEX_HEREDELIM; + lex_firstind = retind + 1; + } +#endif + else if (lex_firstind >= 0 && (tflags & LEX_PASSNEXT) == 0 && shellbreak (ch)) + { + if (heredelim == 0) + { + nestret = substring (ret, lex_firstind, retind); + heredelim = string_quote_removal (nestret, 0); + free (nestret); + hdlen = STRLEN(heredelim); +/*itrace("parse_comsub:%d: found here doc delimiter `%s' (%d)", line_number, heredelim, hdlen);*/ + } + if (ch == '\n') + { + tflags |= LEX_INHEREDOC; + tflags &= ~LEX_HEREDELIM; + lex_firstind = retind + 1; + } + else + lex_firstind = -1; + } + } + + /* Meta-characters that can introduce a reserved word. Not perfect yet. */ + if MBTEST((tflags & LEX_RESWDOK) == 0 && (tflags & LEX_CKCASE) && (tflags & LEX_INCOMMENT) == 0 && (shellmeta(ch) || ch == '\n')) + { + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + peekc = shell_getc (1); + if (ch == peekc && (ch == '&' || ch == '|' || ch == ';')) /* two-character tokens */ + { + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = peekc; +/*itrace("parse_comsub:%d: set lex_reswordok = 1, ch = `%c'", line_number, ch);*/ + tflags |= LEX_RESWDOK; + lex_rwlen = 0; + continue; + } + else if (ch == '\n' || COMSUB_META(ch)) + { + shell_ungetc (peekc); +/*itrace("parse_comsub:%d: set lex_reswordok = 1, ch = `%c'", line_number, ch);*/ + tflags |= LEX_RESWDOK; + lex_rwlen = 0; + continue; + } + else if (ch == EOF) + goto eof_error; + else + { + /* `unget' the character we just added and fall through */ + retind--; + shell_ungetc (peekc); + } + } + + /* If we can read a reserved word, try to read one. */ + if (tflags & LEX_RESWDOK) + { + if MBTEST(islower (ch)) + { + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + lex_rwlen++; + continue; + } + else if MBTEST(lex_rwlen == 4 && shellbreak (ch)) + { + if (STREQN (ret + retind - 4, "case", 4)) +{ + tflags |= LEX_INCASE; +/*itrace("parse_comsub:%d: found `case', lex_incase -> 1 lex_reswdok -> 0", line_number);*/ +} + else if (STREQN (ret + retind - 4, "esac", 4)) +{ + tflags &= ~LEX_INCASE; +/*itrace("parse_comsub:%d: found `esac', lex_incase -> 0 lex_reswdok -> 0", line_number);*/ +} + tflags &= ~LEX_RESWDOK; + } + else if MBTEST((tflags & LEX_CKCOMMENT) && ch == '#' && (lex_rwlen == 0 || ((tflags & LEX_INWORD) && lex_wlen == 0))) + ; /* don't modify LEX_RESWDOK if we're starting a comment */ + else if MBTEST((tflags & LEX_INCASE) && ch != '\n') + /* If we can read a reserved word and we're in case, we're at the + point where we can read a new pattern list or an esac. We + handle the esac case above. If we read a newline, we want to + leave LEX_RESWDOK alone. If we read anything else, we want to + turn off LEX_RESWDOK, since we're going to read a pattern list. */ +{ + tflags &= ~LEX_RESWDOK; +/*itrace("parse_comsub:%d: lex_incase == 1 found `%c', lex_reswordok -> 0", line_number, ch);*/ +} + else if MBTEST(shellbreak (ch) == 0) +{ + tflags &= ~LEX_RESWDOK; +/*itrace("parse_comsub:%d: found `%c', lex_reswordok -> 0", line_number, ch);*/ +} + } + + /* Might be the start of a here-doc delimiter */ + if MBTEST((tflags & LEX_INCOMMENT) == 0 && (tflags & LEX_CKCASE) && ch == '<') + { + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + peekc = shell_getc (1); + if (peekc == EOF) + goto eof_error; + if (peekc == ch) + { + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = peekc; + peekc = shell_getc (1); + if (peekc == EOF) + goto eof_error; + if (peekc == '-') + { + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = peekc; + tflags |= LEX_STRIPDOC; + } + else + shell_ungetc (peekc); + if (peekc != '<') + { + tflags |= LEX_HEREDELIM; + lex_firstind = -1; + } + continue; + } + else + ch = peekc; /* fall through and continue XXX */ + } + else if MBTEST((tflags & LEX_CKCOMMENT) && (tflags & LEX_INCOMMENT) == 0 && ch == '#' && (((tflags & LEX_RESWDOK) && lex_rwlen == 0) || ((tflags & LEX_INWORD) && lex_wlen == 0))) +{ +/*itrace("parse_comsub:%d: lex_incomment -> 1 (%d)", line_number, __LINE__);*/ + tflags |= LEX_INCOMMENT; +} + + if MBTEST(ch == CTLESC || ch == CTLNUL) /* special shell escapes */ + { + RESIZE_MALLOCED_BUFFER (ret, retind, 2, retsize, 64); + ret[retind++] = CTLESC; + ret[retind++] = ch; + continue; + } +#if 0 + else if MBTEST((tflags & LEX_INCASE) && ch == close && close == ')') + tflags &= ~LEX_INCASE; /* XXX */ +#endif + else if MBTEST(ch == close && (tflags & LEX_INCASE) == 0) /* ending delimiter */ +{ + count--; +/*itrace("parse_comsub:%d: found close: count = %d", line_number, count);*/ +} + else if MBTEST(((flags & P_FIRSTCLOSE) == 0) && (tflags & LEX_INCASE) == 0 && ch == open) /* nested begin */ +{ + count++; +/*itrace("parse_comsub:%d: found open: count = %d", line_number, count);*/ +} + + /* Add this character. */ + RESIZE_MALLOCED_BUFFER (ret, retind, 1, retsize, 64); + ret[retind++] = ch; + + /* If we just read the ending character, don't bother continuing. */ + if (count == 0) + break; + + if MBTEST(ch == '\\') /* backslashes */ + tflags |= LEX_PASSNEXT; + + if MBTEST(shellquote (ch)) + { + /* '', ``, or "" inside $(...). */ + push_delimiter (dstack, ch); + if MBTEST((tflags & LEX_WASDOL) && ch == '\'') /* $'...' inside group */ + nestret = parse_matched_pair (ch, ch, ch, &nestlen, P_ALLOWESC|rflags); + else + nestret = parse_matched_pair (ch, ch, ch, &nestlen, rflags); + pop_delimiter (dstack); + CHECK_NESTRET_ERROR (); + + if MBTEST((tflags & LEX_WASDOL) && ch == '\'' && (extended_quote || (rflags & P_DQUOTE) == 0)) + { + /* Translate $'...' here. */ + ttrans = ansiexpand (nestret, 0, nestlen - 1, &ttranslen); + xfree (nestret); + + if ((rflags & P_DQUOTE) == 0) + { + nestret = sh_single_quote (ttrans); + free (ttrans); + nestlen = strlen (nestret); + } + else + { + nestret = ttrans; + nestlen = ttranslen; + } + retind -= 2; /* back up before the $' */ + } + else if MBTEST((tflags & LEX_WASDOL) && ch == '"' && (extended_quote || (rflags & P_DQUOTE) == 0)) + { + /* Locale expand $"..." here. */ + ttrans = localeexpand (nestret, 0, nestlen - 1, start_lineno, &ttranslen); + xfree (nestret); + + nestret = sh_mkdoublequoted (ttrans, ttranslen, 0); + free (ttrans); + nestlen = ttranslen + 2; + retind -= 2; /* back up before the $" */ + } + + APPEND_NESTRET (); + FREE (nestret); + } + else if MBTEST((tflags & LEX_WASDOL) && (ch == '(' || ch == '{' || ch == '[')) /* ) } ] */ + /* check for $(), $[], or ${} inside command substitution. */ + { + if ((tflags & LEX_INCASE) == 0 && open == ch) /* undo previous increment */ + count--; + if (ch == '(') /* ) */ + nestret = parse_comsub (0, '(', ')', &nestlen, (rflags|P_COMMAND) & ~P_DQUOTE); + else if (ch == '{') /* } */ + nestret = parse_matched_pair (0, '{', '}', &nestlen, P_FIRSTCLOSE|P_DOLBRACE|rflags); + else if (ch == '[') /* ] */ + nestret = parse_matched_pair (0, '[', ']', &nestlen, rflags); + + CHECK_NESTRET_ERROR (); + APPEND_NESTRET (); + + FREE (nestret); + } + if MBTEST(ch == '$') + tflags |= LEX_WASDOL; + else + tflags &= ~LEX_WASDOL; + } + + FREE (heredelim); + ret[retind] = '\0'; + if (lenp) + *lenp = retind; +/*itrace("parse_comsub:%d: returning `%s'", line_number, ret);*/ + return ret; +} + +/* Recursively call the parser to parse a $(...) command substitution. */ +char * +xparse_dolparen (base, string, indp, flags) + char *base; + char *string; + int *indp; + int flags; +{ + sh_parser_state_t ps; + sh_input_line_state_t ls; + int orig_ind, nc, sflags; + char *ret, *s, *ep, *ostring; + + /*yydebug = 1;*/ + orig_ind = *indp; + ostring = string; + +/*itrace("xparse_dolparen: size = %d shell_input_line = `%s'", shell_input_line_size, shell_input_line);*/ + sflags = SEVAL_NONINT|SEVAL_NOHIST|SEVAL_NOFREE; + if (flags & SX_NOLONGJMP) + sflags |= SEVAL_NOLONGJMP; + save_parser_state (&ps); + save_input_line_state (&ls); + + /*(*/ + parser_state |= PST_CMDSUBST|PST_EOFTOKEN; /* allow instant ')' */ /*(*/ + shell_eof_token = ')'; + parse_string (string, "command substitution", sflags, &ep); + + restore_parser_state (&ps); + reset_parser (); + /* reset_parser clears shell_input_line and associated variables */ + restore_input_line_state (&ls); + if (interactive) + token_to_read = 0; + + /* Need to find how many characters parse_and_execute consumed, update + *indp, if flags != 0, copy the portion of the string parsed into RET + and return it. If flags & 1 (EX_NOALLOC) we can return NULL. */ + + /*(*/ + if (ep[-1] != ')') + { +#if DEBUG + if (ep[-1] != '\n') + itrace("xparse_dolparen:%d: ep[-1] != RPAREN (%d), ep = `%s'", line_number, ep[-1], ep); +#endif + while (ep > ostring && ep[-1] == '\n') ep--; + } + + nc = ep - ostring; + *indp = ep - base - 1; + + /*(*/ +#if DEBUG + if (base[*indp] != ')') + itrace("xparse_dolparen:%d: base[%d] != RPAREN (%d), base = `%s'", line_number, *indp, base[*indp], base); +#endif + + if (flags & SX_NOALLOC) + return (char *)NULL; + + if (nc == 0) + { + ret = xmalloc (1); + ret[0] = '\0'; + } + else + ret = substring (ostring, 0, nc - 1); + + return ret; +} + +#if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND) +/* Parse a double-paren construct. It can be either an arithmetic + command, an arithmetic `for' command, or a nested subshell. Returns + the parsed token, -1 on error, or -2 if we didn't do anything and + should just go on. */ +static int +parse_dparen (c) + int c; +{ + int cmdtyp, sline; + char *wval; + WORD_DESC *wd; + +#if defined (ARITH_FOR_COMMAND) + if (last_read_token == FOR) + { + arith_for_lineno = line_number; + cmdtyp = parse_arith_cmd (&wval, 0); + if (cmdtyp == 1) + { + wd = alloc_word_desc (); + wd->word = wval; + yylval.word_list = make_word_list (wd, (WORD_LIST *)NULL); + return (ARITH_FOR_EXPRS); + } + else + return -1; /* ERROR */ + } +#endif + +#if defined (DPAREN_ARITHMETIC) + if (reserved_word_acceptable (last_read_token)) + { + sline = line_number; + + cmdtyp = parse_arith_cmd (&wval, 0); + if (cmdtyp == 1) /* arithmetic command */ + { + wd = alloc_word_desc (); + wd->word = wval; + wd->flags = W_QUOTED|W_NOSPLIT|W_NOGLOB|W_DQUOTE; + yylval.word_list = make_word_list (wd, (WORD_LIST *)NULL); + return (ARITH_CMD); + } + else if (cmdtyp == 0) /* nested subshell */ + { + push_string (wval, 0, (alias_t *)NULL); + if ((parser_state & PST_CASEPAT) == 0) + parser_state |= PST_SUBSHELL; + return (c); + } + else /* ERROR */ + return -1; + } +#endif + + return -2; /* XXX */ +} + +/* We've seen a `(('. Look for the matching `))'. If we get it, return 1. + If not, assume it's a nested subshell for backwards compatibility and + return 0. In any case, put the characters we've consumed into a locally- + allocated buffer and make *ep point to that buffer. Return -1 on an + error, for example EOF. */ +static int +parse_arith_cmd (ep, adddq) + char **ep; + int adddq; +{ + int exp_lineno, rval, c; + char *ttok, *tokstr; + int ttoklen; + + exp_lineno = line_number; + ttok = parse_matched_pair (0, '(', ')', &ttoklen, 0); + rval = 1; + if (ttok == &matched_pair_error) + return -1; + /* Check that the next character is the closing right paren. If + not, this is a syntax error. ( */ + c = shell_getc (0); + if MBTEST(c != ')') + rval = 0; + + tokstr = (char *)xmalloc (ttoklen + 4); + + /* if ADDDQ != 0 then (( ... )) -> "..." */ + if (rval == 1 && adddq) /* arith cmd, add double quotes */ + { + tokstr[0] = '"'; + strncpy (tokstr + 1, ttok, ttoklen - 1); + tokstr[ttoklen] = '"'; + tokstr[ttoklen+1] = '\0'; + } + else if (rval == 1) /* arith cmd, don't add double quotes */ + { + strncpy (tokstr, ttok, ttoklen - 1); + tokstr[ttoklen-1] = '\0'; + } + else /* nested subshell */ + { + tokstr[0] = '('; + strncpy (tokstr + 1, ttok, ttoklen - 1); + tokstr[ttoklen] = ')'; + tokstr[ttoklen+1] = c; + tokstr[ttoklen+2] = '\0'; + } + + *ep = tokstr; + FREE (ttok); + return rval; +} +#endif /* DPAREN_ARITHMETIC || ARITH_FOR_COMMAND */ + +#if defined (COND_COMMAND) +static void +cond_error () +{ + char *etext; + + if (EOF_Reached && cond_token != COND_ERROR) /* [[ */ + parser_error (cond_lineno, _("unexpected EOF while looking for `]]'")); + else if (cond_token != COND_ERROR) + { + if (etext = error_token_from_token (cond_token)) + { + parser_error (cond_lineno, _("syntax error in conditional expression: unexpected token `%s'"), etext); + free (etext); + } + else + parser_error (cond_lineno, _("syntax error in conditional expression")); + } +} + +static COND_COM * +cond_expr () +{ + return (cond_or ()); +} + +static COND_COM * +cond_or () +{ + COND_COM *l, *r; + + l = cond_and (); + if (cond_token == OR_OR) + { + r = cond_or (); + l = make_cond_node (COND_OR, (WORD_DESC *)NULL, l, r); + } + return l; +} + +static COND_COM * +cond_and () +{ + COND_COM *l, *r; + + l = cond_term (); + if (cond_token == AND_AND) + { + r = cond_and (); + l = make_cond_node (COND_AND, (WORD_DESC *)NULL, l, r); + } + return l; +} + +static int +cond_skip_newlines () +{ + while ((cond_token = read_token (READ)) == '\n') + { + if (SHOULD_PROMPT ()) + prompt_again (); + } + return (cond_token); +} + +#define COND_RETURN_ERROR() \ + do { cond_token = COND_ERROR; return ((COND_COM *)NULL); } while (0) + +static COND_COM * +cond_term () +{ + WORD_DESC *op; + COND_COM *term, *tleft, *tright; + int tok, lineno; + char *etext; + + /* Read a token. It can be a left paren, a `!', a unary operator, or a + word that should be the first argument of a binary operator. Start by + skipping newlines, since this is a compound command. */ + tok = cond_skip_newlines (); + lineno = line_number; + if (tok == COND_END) + { + COND_RETURN_ERROR (); + } + else if (tok == '(') + { + term = cond_expr (); + if (cond_token != ')') + { + if (term) + dispose_cond_node (term); /* ( */ + if (etext = error_token_from_token (cond_token)) + { + parser_error (lineno, _("unexpected token `%s', expected `)'"), etext); + free (etext); + } + else + parser_error (lineno, _("expected `)'")); + COND_RETURN_ERROR (); + } + term = make_cond_node (COND_EXPR, (WORD_DESC *)NULL, term, (COND_COM *)NULL); + (void)cond_skip_newlines (); + } + else if (tok == BANG || (tok == WORD && (yylval.word->word[0] == '!' && yylval.word->word[1] == '\0'))) + { + if (tok == WORD) + dispose_word (yylval.word); /* not needed */ + term = cond_term (); + if (term) + term->flags |= CMD_INVERT_RETURN; + } + else if (tok == WORD && yylval.word->word[0] == '-' && yylval.word->word[2] == 0 && test_unop (yylval.word->word)) + { + op = yylval.word; + tok = read_token (READ); + if (tok == WORD) + { + tleft = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL); + term = make_cond_node (COND_UNARY, op, tleft, (COND_COM *)NULL); + } + else + { + dispose_word (op); + if (etext = error_token_from_token (tok)) + { + parser_error (line_number, _("unexpected argument `%s' to conditional unary operator"), etext); + free (etext); + } + else + parser_error (line_number, _("unexpected argument to conditional unary operator")); + COND_RETURN_ERROR (); + } + + (void)cond_skip_newlines (); + } + else if (tok == WORD) /* left argument to binary operator */ + { + /* lhs */ + tleft = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL); + + /* binop */ + tok = read_token (READ); + if (tok == WORD && test_binop (yylval.word->word)) + { + op = yylval.word; + if (op->word[0] == '=' && (op->word[1] == '\0' || (op->word[1] == '=' && op->word[2] == '\0'))) + parser_state |= PST_EXTPAT; + else if (op->word[0] == '!' && op->word[1] == '=' && op->word[2] == '\0') + parser_state |= PST_EXTPAT; + } +#if defined (COND_REGEXP) + else if (tok == WORD && STREQ (yylval.word->word, "=~")) + { + op = yylval.word; + parser_state |= PST_REGEXP; + } +#endif + else if (tok == '<' || tok == '>') + op = make_word_from_token (tok); /* ( */ + /* There should be a check before blindly accepting the `)' that we have + seen the opening `('. */ + else if (tok == COND_END || tok == AND_AND || tok == OR_OR || tok == ')') + { + /* Special case. [[ x ]] is equivalent to [[ -n x ]], just like + the test command. Similarly for [[ x && expr ]] or + [[ x || expr ]] or [[ (x) ]]. */ + op = make_word ("-n"); + term = make_cond_node (COND_UNARY, op, tleft, (COND_COM *)NULL); + cond_token = tok; + return (term); + } + else + { + if (etext = error_token_from_token (tok)) + { + parser_error (line_number, _("unexpected token `%s', conditional binary operator expected"), etext); + free (etext); + } + else + parser_error (line_number, _("conditional binary operator expected")); + dispose_cond_node (tleft); + COND_RETURN_ERROR (); + } + + /* rhs */ + if (parser_state & PST_EXTPAT) + extended_glob = 1; + tok = read_token (READ); + if (parser_state & PST_EXTPAT) + extended_glob = global_extglob; + parser_state &= ~(PST_REGEXP|PST_EXTPAT); + + if (tok == WORD) + { + tright = make_cond_node (COND_TERM, yylval.word, (COND_COM *)NULL, (COND_COM *)NULL); + term = make_cond_node (COND_BINARY, op, tleft, tright); + } + else + { + if (etext = error_token_from_token (tok)) + { + parser_error (line_number, _("unexpected argument `%s' to conditional binary operator"), etext); + free (etext); + } + else + parser_error (line_number, _("unexpected argument to conditional binary operator")); + dispose_cond_node (tleft); + dispose_word (op); + COND_RETURN_ERROR (); + } + + (void)cond_skip_newlines (); + } + else + { + if (tok < 256) + parser_error (line_number, _("unexpected token `%c' in conditional command"), tok); + else if (etext = error_token_from_token (tok)) + { + parser_error (line_number, _("unexpected token `%s' in conditional command"), etext); + free (etext); + } + else + parser_error (line_number, _("unexpected token %d in conditional command"), tok); + COND_RETURN_ERROR (); + } + return (term); +} + +/* This is kind of bogus -- we slip a mini recursive-descent parser in + here to handle the conditional statement syntax. */ +static COMMAND * +parse_cond_command () +{ + COND_COM *cexp; + + global_extglob = extended_glob; + cexp = cond_expr (); + return (make_cond_command (cexp)); +} +#endif + +#if defined (ARRAY_VARS) +/* When this is called, it's guaranteed that we don't care about anything + in t beyond i. We do save and restore the chars, though. */ +static int +token_is_assignment (t, i) + char *t; + int i; +{ + unsigned char c, c1; + int r; + + c = t[i]; c1 = t[i+1]; + t[i] = '='; t[i+1] = '\0'; + r = assignment (t, (parser_state & PST_COMPASSIGN) != 0); + t[i] = c; t[i+1] = c1; + return r; +} + +/* XXX - possible changes here for `+=' */ +static int +token_is_ident (t, i) + char *t; + int i; +{ + unsigned char c; + int r; + + c = t[i]; + t[i] = '\0'; + r = legal_identifier (t); + t[i] = c; + return r; +} +#endif + +static int +read_token_word (character) + int character; +{ + /* The value for YYLVAL when a WORD is read. */ + WORD_DESC *the_word; + + /* Index into the token that we are building. */ + int token_index; + + /* ALL_DIGITS becomes zero when we see a non-digit. */ + int all_digit_token; + + /* DOLLAR_PRESENT becomes non-zero if we see a `$'. */ + int dollar_present; + + /* COMPOUND_ASSIGNMENT becomes non-zero if we are parsing a compound + assignment. */ + int compound_assignment; + + /* QUOTED becomes non-zero if we see one of ("), ('), (`), or (\). */ + int quoted; + + /* Non-zero means to ignore the value of the next character, and just + to add it no matter what. */ + int pass_next_character; + + /* The current delimiting character. */ + int cd; + int result, peek_char; + char *ttok, *ttrans; + int ttoklen, ttranslen; + intmax_t lvalue; + + if (token_buffer_size < TOKEN_DEFAULT_INITIAL_SIZE) + token = (char *)xrealloc (token, token_buffer_size = TOKEN_DEFAULT_INITIAL_SIZE); + + token_index = 0; + all_digit_token = DIGIT (character); + dollar_present = quoted = pass_next_character = compound_assignment = 0; + + for (;;) + { + if (character == EOF) + goto got_token; + + if (pass_next_character) + { + pass_next_character = 0; + goto got_escaped_character; + } + + cd = current_delimiter (dstack); + + /* Handle backslashes. Quote lots of things when not inside of + double-quotes, quote some things inside of double-quotes. */ + if MBTEST(character == '\\') + { + peek_char = shell_getc (0); + + /* Backslash-newline is ignored in all cases except + when quoted with single quotes. */ + if (peek_char == '\n') + { + character = '\n'; + goto next_character; + } + else + { + shell_ungetc (peek_char); + + /* If the next character is to be quoted, note it now. */ + if (cd == 0 || cd == '`' || + (cd == '"' && peek_char >= 0 && (sh_syntaxtab[peek_char] & CBSDQUOTE))) + pass_next_character++; + + quoted = 1; + goto got_character; + } + } + + /* Parse a matched pair of quote characters. */ + if MBTEST(shellquote (character)) + { + push_delimiter (dstack, character); + ttok = parse_matched_pair (character, character, character, &ttoklen, (character == '`') ? P_COMMAND : 0); + pop_delimiter (dstack); + if (ttok == &matched_pair_error) + return -1; /* Bail immediately. */ + RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2, + token_buffer_size, TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = character; + strcpy (token + token_index, ttok); + token_index += ttoklen; + all_digit_token = 0; + quoted = 1; + dollar_present |= (character == '"' && strchr (ttok, '$') != 0); + FREE (ttok); + goto next_character; + } + +#ifdef COND_REGEXP + /* When parsing a regexp as a single word inside a conditional command, + we need to special-case characters special to both the shell and + regular expressions. Right now, that is only '(' and '|'. */ /*)*/ + if MBTEST((parser_state & PST_REGEXP) && (character == '(' || character == '|')) /*)*/ + { + if (character == '|') + goto got_character; + + push_delimiter (dstack, character); + ttok = parse_matched_pair (cd, '(', ')', &ttoklen, 0); + pop_delimiter (dstack); + if (ttok == &matched_pair_error) + return -1; /* Bail immediately. */ + RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2, + token_buffer_size, TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = character; + strcpy (token + token_index, ttok); + token_index += ttoklen; + FREE (ttok); + dollar_present = all_digit_token = 0; + goto next_character; + } +#endif /* COND_REGEXP */ + +#ifdef EXTENDED_GLOB + /* Parse a ksh-style extended pattern matching specification. */ + if MBTEST(extended_glob && PATTERN_CHAR (character)) + { + peek_char = shell_getc (1); + if MBTEST(peek_char == '(') /* ) */ + { + push_delimiter (dstack, peek_char); + ttok = parse_matched_pair (cd, '(', ')', &ttoklen, 0); + pop_delimiter (dstack); + if (ttok == &matched_pair_error) + return -1; /* Bail immediately. */ + RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 3, + token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = character; + token[token_index++] = peek_char; + strcpy (token + token_index, ttok); + token_index += ttoklen; + FREE (ttok); + dollar_present = all_digit_token = 0; + goto next_character; + } + else + shell_ungetc (peek_char); + } +#endif /* EXTENDED_GLOB */ + + /* If the delimiter character is not single quote, parse some of + the shell expansions that must be read as a single word. */ + if (shellexp (character)) + { + peek_char = shell_getc (1); + /* $(...), <(...), >(...), $((...)), ${...}, and $[...] constructs */ + if MBTEST(peek_char == '(' || + ((peek_char == '{' || peek_char == '[') && character == '$')) /* ) ] } */ + { + if (peek_char == '{') /* } */ + ttok = parse_matched_pair (cd, '{', '}', &ttoklen, P_FIRSTCLOSE|P_DOLBRACE); + else if (peek_char == '(') /* ) */ + { + /* XXX - push and pop the `(' as a delimiter for use by + the command-oriented-history code. This way newlines + appearing in the $(...) string get added to the + history literally rather than causing a possibly- + incorrect `;' to be added. ) */ + push_delimiter (dstack, peek_char); + ttok = parse_comsub (cd, '(', ')', &ttoklen, P_COMMAND); + pop_delimiter (dstack); + } + else + ttok = parse_matched_pair (cd, '[', ']', &ttoklen, 0); + if (ttok == &matched_pair_error) + return -1; /* Bail immediately. */ + RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 3, + token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = character; + token[token_index++] = peek_char; + strcpy (token + token_index, ttok); + token_index += ttoklen; + FREE (ttok); + dollar_present = 1; + all_digit_token = 0; + goto next_character; + } + /* This handles $'...' and $"..." new-style quoted strings. */ + else if MBTEST(character == '$' && (peek_char == '\'' || peek_char == '"')) + { + int first_line; + + first_line = line_number; + push_delimiter (dstack, peek_char); + ttok = parse_matched_pair (peek_char, peek_char, peek_char, + &ttoklen, + (peek_char == '\'') ? P_ALLOWESC : 0); + pop_delimiter (dstack); + if (ttok == &matched_pair_error) + return -1; + if (peek_char == '\'') + { + ttrans = ansiexpand (ttok, 0, ttoklen - 1, &ttranslen); + free (ttok); + + /* Insert the single quotes and correctly quote any + embedded single quotes (allowed because P_ALLOWESC was + passed to parse_matched_pair). */ + ttok = sh_single_quote (ttrans); + free (ttrans); + ttranslen = strlen (ttok); + ttrans = ttok; + } + else + { + /* Try to locale-expand the converted string. */ + ttrans = localeexpand (ttok, 0, ttoklen - 1, first_line, &ttranslen); + free (ttok); + + /* Add the double quotes back */ + ttok = sh_mkdoublequoted (ttrans, ttranslen, 0); + free (ttrans); + ttranslen += 2; + ttrans = ttok; + } + + RESIZE_MALLOCED_BUFFER (token, token_index, ttranslen + 1, + token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + strcpy (token + token_index, ttrans); + token_index += ttranslen; + FREE (ttrans); + quoted = 1; + all_digit_token = 0; + goto next_character; + } + /* This could eventually be extended to recognize all of the + shell's single-character parameter expansions, and set flags.*/ + else if MBTEST(character == '$' && peek_char == '$') + { + RESIZE_MALLOCED_BUFFER (token, token_index, 3, + token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = '$'; + token[token_index++] = peek_char; + dollar_present = 1; + all_digit_token = 0; + goto next_character; + } + else + shell_ungetc (peek_char); + } + +#if defined (ARRAY_VARS) + /* Identify possible array subscript assignment; match [...]. If + parser_state&PST_COMPASSIGN, we need to parse [sub]=words treating + `sub' as if it were enclosed in double quotes. */ + else if MBTEST(character == '[' && /* ] */ + ((token_index > 0 && assignment_acceptable (last_read_token) && token_is_ident (token, token_index)) || + (token_index == 0 && (parser_state&PST_COMPASSIGN)))) + { + ttok = parse_matched_pair (cd, '[', ']', &ttoklen, P_ARRAYSUB); + if (ttok == &matched_pair_error) + return -1; /* Bail immediately. */ + RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 2, + token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = character; + strcpy (token + token_index, ttok); + token_index += ttoklen; + FREE (ttok); + all_digit_token = 0; + goto next_character; + } + /* Identify possible compound array variable assignment. */ + else if MBTEST(character == '=' && token_index > 0 && (assignment_acceptable (last_read_token) || (parser_state & PST_ASSIGNOK)) && token_is_assignment (token, token_index)) + { + peek_char = shell_getc (1); + if MBTEST(peek_char == '(') /* ) */ + { + ttok = parse_compound_assignment (&ttoklen); + + RESIZE_MALLOCED_BUFFER (token, token_index, ttoklen + 4, + token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + + token[token_index++] = '='; + token[token_index++] = '('; + if (ttok) + { + strcpy (token + token_index, ttok); + token_index += ttoklen; + } + token[token_index++] = ')'; + FREE (ttok); + all_digit_token = 0; + compound_assignment = 1; +#if 1 + goto next_character; +#else + goto got_token; /* ksh93 seems to do this */ +#endif + } + else + shell_ungetc (peek_char); + } +#endif + + /* When not parsing a multi-character word construct, shell meta- + characters break words. */ + if MBTEST(shellbreak (character)) + { + shell_ungetc (character); + goto got_token; + } + +got_character: + + if (character == CTLESC || character == CTLNUL) + { + RESIZE_MALLOCED_BUFFER (token, token_index, 2, token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + token[token_index++] = CTLESC; + } + else +got_escaped_character: + RESIZE_MALLOCED_BUFFER (token, token_index, 1, token_buffer_size, + TOKEN_DEFAULT_GROW_SIZE); + + token[token_index++] = character; + + all_digit_token &= DIGIT (character); + dollar_present |= character == '$'; + + next_character: + if (character == '\n' && SHOULD_PROMPT ()) + prompt_again (); + + /* We want to remove quoted newlines (that is, a \ pair) + unless we are within single quotes or pass_next_character is + set (the shell equivalent of literal-next). */ + cd = current_delimiter (dstack); + character = shell_getc (cd != '\'' && pass_next_character == 0); + } /* end for (;;) */ + +got_token: + + /* Calls to RESIZE_MALLOCED_BUFFER ensure there is sufficient room. */ + token[token_index] = '\0'; + + /* Check to see what thing we should return. If the last_read_token + is a `<', or a `&', or the character which ended this token is + a '>' or '<', then, and ONLY then, is this input token a NUMBER. + Otherwise, it is just a word, and should be returned as such. */ + if MBTEST(all_digit_token && (character == '<' || character == '>' || + last_read_token == LESS_AND || + last_read_token == GREATER_AND)) + { + if (legal_number (token, &lvalue) && (int)lvalue == lvalue) +#if 0 + yylval.number = lvalue; + else + yylval.number = -1; + return (NUMBER); +#else + { + yylval.number = lvalue; + return (NUMBER); + } +#endif + } + + /* Check for special case tokens. */ + result = (last_shell_getc_is_singlebyte) ? special_case_tokens (token) : -1; + if (result >= 0) + return result; + +#if defined (ALIAS) + /* Posix.2 does not allow reserved words to be aliased, so check for all + of them, including special cases, before expanding the current token + as an alias. */ + if MBTEST(posixly_correct) + CHECK_FOR_RESERVED_WORD (token); + + /* Aliases are expanded iff EXPAND_ALIASES is non-zero, and quoting + inhibits alias expansion. */ + if (expand_aliases && quoted == 0) + { + result = alias_expand_token (token); + if (result == RE_READ_TOKEN) + return (RE_READ_TOKEN); + else if (result == NO_EXPANSION) + parser_state &= ~PST_ALEXPNEXT; + } + + /* If not in Posix.2 mode, check for reserved words after alias + expansion. */ + if MBTEST(posixly_correct == 0) +#endif + CHECK_FOR_RESERVED_WORD (token); + + the_word = (WORD_DESC *)xmalloc (sizeof (WORD_DESC)); + the_word->word = (char *)xmalloc (1 + token_index); + the_word->flags = 0; + strcpy (the_word->word, token); + if (dollar_present) + the_word->flags |= W_HASDOLLAR; + if (quoted) + the_word->flags |= W_QUOTED; /*(*/ + if (compound_assignment && token[token_index-1] == ')') + the_word->flags |= W_COMPASSIGN; + /* A word is an assignment if it appears at the beginning of a + simple command, or after another assignment word. This is + context-dependent, so it cannot be handled in the grammar. */ + if (assignment (token, (parser_state & PST_COMPASSIGN) != 0)) + { + the_word->flags |= W_ASSIGNMENT; + /* Don't perform word splitting on assignment statements. */ + if (assignment_acceptable (last_read_token) || (parser_state & PST_COMPASSIGN) != 0) + the_word->flags |= W_NOSPLIT; + } + + if (command_token_position (last_read_token)) + { + struct builtin *b; + b = builtin_address_internal (token, 0); + if (b && (b->flags & ASSIGNMENT_BUILTIN)) + parser_state |= PST_ASSIGNOK; + else if (STREQ (token, "eval") || STREQ (token, "let")) + parser_state |= PST_ASSIGNOK; + } + + yylval.word = the_word; + + if (token[0] == '{' && token[token_index-1] == '}' && + (character == '<' || character == '>')) + { + /* can use token; already copied to the_word */ + token[token_index-1] = '\0'; + if (legal_identifier (token+1)) + { + strcpy (the_word->word, token+1); +/*itrace("read_token_word: returning REDIR_WORD for %s", the_word->word);*/ + return (REDIR_WORD); + } + } + + result = ((the_word->flags & (W_ASSIGNMENT|W_NOSPLIT)) == (W_ASSIGNMENT|W_NOSPLIT)) + ? ASSIGNMENT_WORD : WORD; + + switch (last_read_token) + { + case FUNCTION: + parser_state |= PST_ALLOWOPNBRC; + function_dstart = line_number; + break; + case CASE: + case SELECT: + case FOR: + if (word_top < MAX_CASE_NEST) + word_top++; + word_lineno[word_top] = line_number; + break; + } + + return (result); +} + +/* Return 1 if TOKSYM is a token that after being read would allow + a reserved word to be seen, else 0. */ +static int +reserved_word_acceptable (toksym) + int toksym; +{ + switch (toksym) + { + case '\n': + case ';': + case '(': + case ')': + case '|': + case '&': + case '{': + case '}': /* XXX */ + case AND_AND: + case BANG: + case BAR_AND: + case DO: + case DONE: + case ELIF: + case ELSE: + case ESAC: + case FI: + case IF: + case OR_OR: + case SEMI_SEMI: + case SEMI_AND: + case SEMI_SEMI_AND: + case THEN: + case TIME: + case TIMEOPT: + case TIMEIGN: + case COPROC: + case UNTIL: + case WHILE: + case 0: + return 1; + default: +#if defined (COPROCESS_SUPPORT) + if (last_read_token == WORD && token_before_that == COPROC) + return 1; +#endif + if (last_read_token == WORD && token_before_that == FUNCTION) + return 1; + return 0; + } +} + +/* Return the index of TOKEN in the alist of reserved words, or -1 if + TOKEN is not a shell reserved word. */ +int +find_reserved_word (tokstr) + char *tokstr; +{ + int i; + for (i = 0; word_token_alist[i].word; i++) + if (STREQ (tokstr, word_token_alist[i].word)) + return i; + return -1; +} + +/* An interface to let the rest of the shell (primarily the completion + system) know what the parser is expecting. */ +int +parser_in_command_position () +{ + return (command_token_position (last_read_token)); +} + +#if 0 +#if defined (READLINE) +/* Called after each time readline is called. This insures that whatever + the new prompt string is gets propagated to readline's local prompt + variable. */ +static void +reset_readline_prompt () +{ + char *temp_prompt; + + if (prompt_string_pointer) + { + temp_prompt = (*prompt_string_pointer) + ? decode_prompt_string (*prompt_string_pointer) + : (char *)NULL; + + if (temp_prompt == 0) + { + temp_prompt = (char *)xmalloc (1); + temp_prompt[0] = '\0'; + } + + FREE (current_readline_prompt); + current_readline_prompt = temp_prompt; + } +} +#endif /* READLINE */ +#endif /* 0 */ + +#if defined (HISTORY) +/* A list of tokens which can be followed by newlines, but not by + semi-colons. When concatenating multiple lines of history, the + newline separator for such tokens is replaced with a space. */ +static const int no_semi_successors[] = { + '\n', '{', '(', ')', ';', '&', '|', + CASE, DO, ELSE, IF, SEMI_SEMI, SEMI_AND, SEMI_SEMI_AND, THEN, UNTIL, + WHILE, AND_AND, OR_OR, IN, + 0 +}; + +/* If we are not within a delimited expression, try to be smart + about which separators can be semi-colons and which must be + newlines. Returns the string that should be added into the + history entry. LINE is the line we're about to add; it helps + make some more intelligent decisions in certain cases. */ +char * +history_delimiting_chars (line) + const char *line; +{ + static int last_was_heredoc = 0; /* was the last entry the start of a here document? */ + register int i; + + if ((parser_state & PST_HEREDOC) == 0) + last_was_heredoc = 0; + + if (dstack.delimiter_depth != 0) + return ("\n"); + + /* We look for current_command_line_count == 2 because we are looking to + add the first line of the body of the here document (the second line + of the command). We also keep LAST_WAS_HEREDOC as a private sentinel + variable to note when we think we added the first line of a here doc + (the one with a "<<" somewhere in it) */ + if (parser_state & PST_HEREDOC) + { + if (last_was_heredoc) + { + last_was_heredoc = 0; + return "\n"; + } + return (current_command_line_count == 2 ? "\n" : ""); + } + + /* First, handle some special cases. */ + /*(*/ + /* If we just read `()', assume it's a function definition, and don't + add a semicolon. If the token before the `)' was not `(', and we're + not in the midst of parsing a case statement, assume it's a + parenthesized command and add the semicolon. */ + /*)(*/ + if (token_before_that == ')') + { + if (two_tokens_ago == '(') /*)*/ /* function def */ + return " "; + /* This does not work for subshells inside case statement + command lists. It's a suboptimal solution. */ + else if (parser_state & PST_CASESTMT) /* case statement pattern */ + return " "; + else + return "; "; /* (...) subshell */ + } + else if (token_before_that == WORD && two_tokens_ago == FUNCTION) + return " "; /* function def using `function name' without `()' */ + + /* If we're not in a here document, but we think we're about to parse one, + and we would otherwise return a `;', return a newline to delimit the + line with the here-doc delimiter */ + else if ((parser_state & PST_HEREDOC) == 0 && current_command_line_count > 1 && last_read_token == '\n' && strstr (line, "<<")) + { + last_was_heredoc = 1; + return "\n"; + } + + else if (token_before_that == WORD && two_tokens_ago == FOR) + { + /* Tricky. `for i\nin ...' should not have a semicolon, but + `for i\ndo ...' should. We do what we can. */ + for (i = shell_input_line_index; whitespace (shell_input_line[i]); i++) + ; + if (shell_input_line[i] && shell_input_line[i] == 'i' && shell_input_line[i+1] == 'n') + return " "; + return ";"; + } + else if (two_tokens_ago == CASE && token_before_that == WORD && (parser_state & PST_CASESTMT)) + return " "; + + for (i = 0; no_semi_successors[i]; i++) + { + if (token_before_that == no_semi_successors[i]) + return (" "); + } + + return ("; "); +} +#endif /* HISTORY */ + +/* Issue a prompt, or prepare to issue a prompt when the next character + is read. */ +static void +prompt_again () +{ + char *temp_prompt; + + if (interactive == 0 || expanding_alias ()) /* XXX */ + return; + + ps1_prompt = get_string_value ("PS1"); + ps2_prompt = get_string_value ("PS2"); + + if (!prompt_string_pointer) + prompt_string_pointer = &ps1_prompt; + + temp_prompt = *prompt_string_pointer + ? decode_prompt_string (*prompt_string_pointer) + : (char *)NULL; + + if (temp_prompt == 0) + { + temp_prompt = (char *)xmalloc (1); + temp_prompt[0] = '\0'; + } + + current_prompt_string = *prompt_string_pointer; + prompt_string_pointer = &ps2_prompt; + +#if defined (READLINE) + if (!no_line_editing) + { + FREE (current_readline_prompt); + current_readline_prompt = temp_prompt; + } + else +#endif /* READLINE */ + { + FREE (current_decoded_prompt); + current_decoded_prompt = temp_prompt; + } +} + +int +get_current_prompt_level () +{ + return ((current_prompt_string && current_prompt_string == ps2_prompt) ? 2 : 1); +} + +void +set_current_prompt_level (x) + int x; +{ + prompt_string_pointer = (x == 2) ? &ps2_prompt : &ps1_prompt; + current_prompt_string = *prompt_string_pointer; +} + +static void +print_prompt () +{ + fprintf (stderr, "%s", current_decoded_prompt); + fflush (stderr); +} + +/* Return a string which will be printed as a prompt. The string + may contain special characters which are decoded as follows: + + \a bell (ascii 07) + \d the date in Day Mon Date format + \e escape (ascii 033) + \h the hostname up to the first `.' + \H the hostname + \j the number of active jobs + \l the basename of the shell's tty device name + \n CRLF + \r CR + \s the name of the shell + \t the time in 24-hour hh:mm:ss format + \T the time in 12-hour hh:mm:ss format + \@ the time in 12-hour hh:mm am/pm format + \A the time in 24-hour hh:mm format + \D{fmt} the result of passing FMT to strftime(3) + \u your username + \v the version of bash (e.g., 2.00) + \V the release of bash, version + patchlevel (e.g., 2.00.0) + \w the current working directory + \W the last element of $PWD + \! the history number of this command + \# the command number of this command + \$ a $ or a # if you are root + \nnn character code nnn in octal + \\ a backslash + \[ begin a sequence of non-printing chars + \] end a sequence of non-printing chars +*/ +#define PROMPT_GROWTH 48 +char * +decode_prompt_string (string) + char *string; +{ + WORD_LIST *list; + char *result, *t; + struct dstack save_dstack; + int last_exit_value, last_comsub_pid; +#if defined (PROMPT_STRING_DECODE) + int result_size, result_index; + int c, n, i; + char *temp, octal_string[4]; + struct tm *tm; + time_t the_time; + char timebuf[128]; + char *timefmt; + + result = (char *)xmalloc (result_size = PROMPT_GROWTH); + result[result_index = 0] = 0; + temp = (char *)NULL; + + while (c = *string++) + { + if (posixly_correct && c == '!') + { + if (*string == '!') + { + temp = savestring ("!"); + goto add_string; + } + else + { +#if !defined (HISTORY) + temp = savestring ("1"); +#else /* HISTORY */ + temp = itos (history_number ()); +#endif /* HISTORY */ + string--; /* add_string increments string again. */ + goto add_string; + } + } + if (c == '\\') + { + c = *string; + + switch (c) + { + case '0': + case '1': + case '2': + case '3': + case '4': + case '5': + case '6': + case '7': + strncpy (octal_string, string, 3); + octal_string[3] = '\0'; + + n = read_octal (octal_string); + temp = (char *)xmalloc (3); + + if (n == CTLESC || n == CTLNUL) + { + temp[0] = CTLESC; + temp[1] = n; + temp[2] = '\0'; + } + else if (n == -1) + { + temp[0] = '\\'; + temp[1] = '\0'; + } + else + { + temp[0] = n; + temp[1] = '\0'; + } + + for (c = 0; n != -1 && c < 3 && ISOCTAL (*string); c++) + string++; + + c = 0; /* tested at add_string: */ + goto add_string; + + case 'd': + case 't': + case 'T': + case '@': + case 'A': + /* Make the current time/date into a string. */ + (void) time (&the_time); +#if defined (HAVE_TZSET) + sv_tz ("TZ"); /* XXX -- just make sure */ +#endif + tm = localtime (&the_time); + + if (c == 'd') + n = strftime (timebuf, sizeof (timebuf), "%a %b %d", tm); + else if (c == 't') + n = strftime (timebuf, sizeof (timebuf), "%H:%M:%S", tm); + else if (c == 'T') + n = strftime (timebuf, sizeof (timebuf), "%I:%M:%S", tm); + else if (c == '@') + n = strftime (timebuf, sizeof (timebuf), "%I:%M %p", tm); + else if (c == 'A') + n = strftime (timebuf, sizeof (timebuf), "%H:%M", tm); + + if (n == 0) + timebuf[0] = '\0'; + else + timebuf[sizeof(timebuf) - 1] = '\0'; + + temp = savestring (timebuf); + goto add_string; + + case 'D': /* strftime format */ + if (string[1] != '{') /* } */ + goto not_escape; + + (void) time (&the_time); + tm = localtime (&the_time); + string += 2; /* skip { */ + timefmt = xmalloc (strlen (string) + 3); + for (t = timefmt; *string && *string != '}'; ) + *t++ = *string++; + *t = '\0'; + c = *string; /* tested at add_string */ + if (timefmt[0] == '\0') + { + timefmt[0] = '%'; + timefmt[1] = 'X'; /* locale-specific current time */ + timefmt[2] = '\0'; + } + n = strftime (timebuf, sizeof (timebuf), timefmt, tm); + free (timefmt); + + if (n == 0) + timebuf[0] = '\0'; + else + timebuf[sizeof(timebuf) - 1] = '\0'; + + if (promptvars || posixly_correct) + /* Make sure that expand_prompt_string is called with a + second argument of Q_DOUBLE_QUOTES if we use this + function here. */ + temp = sh_backslash_quote_for_double_quotes (timebuf); + else + temp = savestring (timebuf); + goto add_string; + + case 'n': + temp = (char *)xmalloc (3); + temp[0] = no_line_editing ? '\n' : '\r'; + temp[1] = no_line_editing ? '\0' : '\n'; + temp[2] = '\0'; + goto add_string; + + case 's': + temp = base_pathname (shell_name); + temp = savestring (temp); + goto add_string; + + case 'v': + case 'V': + temp = (char *)xmalloc (16); + if (c == 'v') + strcpy (temp, dist_version); + else + sprintf (temp, "%s.%d", dist_version, patch_level); + goto add_string; + + case 'w': + case 'W': + { + /* Use the value of PWD because it is much more efficient. */ + char t_string[PATH_MAX]; + int tlen; + + temp = get_string_value ("PWD"); + + if (temp == 0) + { + if (getcwd (t_string, sizeof(t_string)) == 0) + { + t_string[0] = '.'; + tlen = 1; + } + else + tlen = strlen (t_string); + } + else + { + tlen = sizeof (t_string) - 1; + strncpy (t_string, temp, tlen); + } + t_string[tlen] = '\0'; + +#if defined (MACOSX) + /* Convert from "fs" format to "input" format */ + temp = fnx_fromfs (t_string, strlen (t_string)); + if (temp != t_string) + strcpy (t_string, temp); +#endif + +#define ROOT_PATH(x) ((x)[0] == '/' && (x)[1] == 0) +#define DOUBLE_SLASH_ROOT(x) ((x)[0] == '/' && (x)[1] == '/' && (x)[2] == 0) + /* Abbreviate \W as ~ if $PWD == $HOME */ + if (c == 'W' && (((t = get_string_value ("HOME")) == 0) || STREQ (t, t_string) == 0)) + { + if (ROOT_PATH (t_string) == 0 && DOUBLE_SLASH_ROOT (t_string) == 0) + { + t = strrchr (t_string, '/'); + if (t) + memmove (t_string, t + 1, strlen (t)); /* strlen(t) to copy NULL */ + } + } +#undef ROOT_PATH +#undef DOUBLE_SLASH_ROOT + else + /* polite_directory_format is guaranteed to return a string + no longer than PATH_MAX - 1 characters. */ + strcpy (t_string, polite_directory_format (t_string)); + + temp = trim_pathname (t_string, PATH_MAX - 1); + /* If we're going to be expanding the prompt string later, + quote the directory name. */ + if (promptvars || posixly_correct) + /* Make sure that expand_prompt_string is called with a + second argument of Q_DOUBLE_QUOTES if we use this + function here. */ + temp = sh_backslash_quote_for_double_quotes (t_string); + else + temp = savestring (t_string); + + goto add_string; + } + + case 'u': + if (current_user.user_name == 0) + get_current_user_info (); + temp = savestring (current_user.user_name); + goto add_string; + + case 'h': + case 'H': + temp = savestring (current_host_name); + if (c == 'h' && (t = (char *)strchr (temp, '.'))) + *t = '\0'; + goto add_string; + + case '#': + temp = itos (current_command_number); + goto add_string; + + case '!': +#if !defined (HISTORY) + temp = savestring ("1"); +#else /* HISTORY */ + temp = itos (history_number ()); +#endif /* HISTORY */ + goto add_string; + + case '$': + t = temp = (char *)xmalloc (3); + if ((promptvars || posixly_correct) && (current_user.euid != 0)) + *t++ = '\\'; + *t++ = current_user.euid == 0 ? '#' : '$'; + *t = '\0'; + goto add_string; + + case 'j': + temp = itos (count_all_jobs ()); + goto add_string; + + case 'l': +#if defined (HAVE_TTYNAME) + temp = (char *)ttyname (fileno (stdin)); + t = temp ? base_pathname (temp) : "tty"; + temp = savestring (t); +#else + temp = savestring ("tty"); +#endif /* !HAVE_TTYNAME */ + goto add_string; + +#if defined (READLINE) + case '[': + case ']': + if (no_line_editing) + { + string++; + break; + } + temp = (char *)xmalloc (3); + n = (c == '[') ? RL_PROMPT_START_IGNORE : RL_PROMPT_END_IGNORE; + i = 0; + if (n == CTLESC || n == CTLNUL) + temp[i++] = CTLESC; + temp[i++] = n; + temp[i] = '\0'; + goto add_string; +#endif /* READLINE */ + + case '\\': + case 'a': + case 'e': + case 'r': + temp = (char *)xmalloc (2); + if (c == 'a') + temp[0] = '\07'; + else if (c == 'e') + temp[0] = '\033'; + else if (c == 'r') + temp[0] = '\r'; + else /* (c == '\\') */ + temp[0] = c; + temp[1] = '\0'; + goto add_string; + + default: +not_escape: + temp = (char *)xmalloc (3); + temp[0] = '\\'; + temp[1] = c; + temp[2] = '\0'; + + add_string: + if (c) + string++; + result = + sub_append_string (temp, result, &result_index, &result_size); + temp = (char *)NULL; /* Freed in sub_append_string (). */ + result[result_index] = '\0'; + break; + } + } + else + { + RESIZE_MALLOCED_BUFFER (result, result_index, 3, result_size, PROMPT_GROWTH); + result[result_index++] = c; + result[result_index] = '\0'; + } + } +#else /* !PROMPT_STRING_DECODE */ + result = savestring (string); +#endif /* !PROMPT_STRING_DECODE */ + + /* Save the delimiter stack and point `dstack' to temp space so any + command substitutions in the prompt string won't result in screwing + up the parser's quoting state. */ + save_dstack = dstack; + dstack = temp_dstack; + dstack.delimiter_depth = 0; + + /* Perform variable and parameter expansion and command substitution on + the prompt string. */ + if (promptvars || posixly_correct) + { + last_exit_value = last_command_exit_value; + last_comsub_pid = last_command_subst_pid; + list = expand_prompt_string (result, Q_DOUBLE_QUOTES, 0); + free (result); + result = string_list (list); + dispose_words (list); + last_command_exit_value = last_exit_value; + last_command_subst_pid = last_comsub_pid; + } + else + { + t = dequote_string (result); + free (result); + result = t; + } + + dstack = save_dstack; + + return (result); +} + +/************************************************ + * * + * ERROR HANDLING * + * * + ************************************************/ + +/* Report a syntax error, and restart the parser. Call here for fatal + errors. */ +int +yyerror (msg) + const char *msg; +{ + report_syntax_error ((char *)NULL); + reset_parser (); + return (0); +} + +static char * +error_token_from_token (tok) + int tok; +{ + char *t; + + if (t = find_token_in_alist (tok, word_token_alist, 0)) + return t; + + if (t = find_token_in_alist (tok, other_token_alist, 0)) + return t; + + t = (char *)NULL; + /* This stuff is dicy and needs closer inspection */ + switch (current_token) + { + case WORD: + case ASSIGNMENT_WORD: + if (yylval.word) + t = savestring (yylval.word->word); + break; + case NUMBER: + t = itos (yylval.number); + break; + case ARITH_CMD: + if (yylval.word_list) + t = string_list (yylval.word_list); + break; + case ARITH_FOR_EXPRS: + if (yylval.word_list) + t = string_list_internal (yylval.word_list, " ; "); + break; + case COND_CMD: + t = (char *)NULL; /* punt */ + break; + } + + return t; +} + +static char * +error_token_from_text () +{ + char *msg, *t; + int token_end, i; + + t = shell_input_line; + i = shell_input_line_index; + token_end = 0; + msg = (char *)NULL; + + if (i && t[i] == '\0') + i--; + + while (i && (whitespace (t[i]) || t[i] == '\n')) + i--; + + if (i) + token_end = i + 1; + + while (i && (member (t[i], " \n\t;|&") == 0)) + i--; + + while (i != token_end && (whitespace (t[i]) || t[i] == '\n')) + i++; + + /* Return our idea of the offending token. */ + if (token_end || (i == 0 && token_end == 0)) + { + if (token_end) + msg = substring (t, i, token_end); + else /* one-character token */ + { + msg = (char *)xmalloc (2); + msg[0] = t[i]; + msg[1] = '\0'; + } + } + + return (msg); +} + +static void +print_offending_line () +{ + char *msg; + int token_end; + + msg = savestring (shell_input_line); + token_end = strlen (msg); + while (token_end && msg[token_end - 1] == '\n') + msg[--token_end] = '\0'; + + parser_error (line_number, "`%s'", msg); + free (msg); +} + +/* Report a syntax error with line numbers, etc. + Call here for recoverable errors. If you have a message to print, + then place it in MESSAGE, otherwise pass NULL and this will figure + out an appropriate message for you. */ +static void +report_syntax_error (message) + char *message; +{ + char *msg, *p; + + if (message) + { + parser_error (line_number, "%s", message); + if (interactive && EOF_Reached) + EOF_Reached = 0; + last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE; + return; + } + + /* If the line of input we're reading is not null, try to find the + objectionable token. First, try to figure out what token the + parser's complaining about by looking at current_token. */ + if (current_token != 0 && EOF_Reached == 0 && (msg = error_token_from_token (current_token))) + { + if (ansic_shouldquote (msg)) + { + p = ansic_quote (msg, 0, NULL); + free (msg); + msg = p; + } + parser_error (line_number, _("syntax error near unexpected token `%s'"), msg); + free (msg); + + if (interactive == 0) + print_offending_line (); + + last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE; + return; + } + + /* If looking at the current token doesn't prove fruitful, try to find the + offending token by analyzing the text of the input line near the current + input line index and report what we find. */ + if (shell_input_line && *shell_input_line) + { + msg = error_token_from_text (); + if (msg) + { + parser_error (line_number, _("syntax error near `%s'"), msg); + free (msg); + } + + /* If not interactive, print the line containing the error. */ + if (interactive == 0) + print_offending_line (); + } + else + { + msg = EOF_Reached ? _("syntax error: unexpected end of file") : _("syntax error"); + parser_error (line_number, "%s", msg); + /* When the shell is interactive, this file uses EOF_Reached + only for error reporting. Other mechanisms are used to + decide whether or not to exit. */ + if (interactive && EOF_Reached) + EOF_Reached = 0; + } + + last_command_exit_value = parse_and_execute_level ? EX_BADSYNTAX : EX_BADUSAGE; +} + +/* ??? Needed function. ??? We have to be able to discard the constructs + created during parsing. In the case of error, we want to return + allocated objects to the memory pool. In the case of no error, we want + to throw away the information about where the allocated objects live. + (dispose_command () will actually free the command.) */ +static void +discard_parser_constructs (error_p) + int error_p; +{ +} + +/************************************************ + * * + * EOF HANDLING * + * * + ************************************************/ + +/* Do that silly `type "bye" to exit' stuff. You know, "ignoreeof". */ + +/* A flag denoting whether or not ignoreeof is set. */ +int ignoreeof = 0; + +/* The number of times that we have encountered an EOF character without + another character intervening. When this gets above the limit, the + shell terminates. */ +int eof_encountered = 0; + +/* The limit for eof_encountered. */ +int eof_encountered_limit = 10; + +/* If we have EOF as the only input unit, this user wants to leave + the shell. If the shell is not interactive, then just leave. + Otherwise, if ignoreeof is set, and we haven't done this the + required number of times in a row, print a message. */ +static void +handle_eof_input_unit () +{ + if (interactive) + { + /* shell.c may use this to decide whether or not to write out the + history, among other things. We use it only for error reporting + in this file. */ + if (EOF_Reached) + EOF_Reached = 0; + + /* If the user wants to "ignore" eof, then let her do so, kind of. */ + if (ignoreeof) + { + if (eof_encountered < eof_encountered_limit) + { + fprintf (stderr, _("Use \"%s\" to leave the shell.\n"), + login_shell ? "logout" : "exit"); + eof_encountered++; + /* Reset the parsing state. */ + last_read_token = current_token = '\n'; + /* Reset the prompt string to be $PS1. */ + prompt_string_pointer = (char **)NULL; + prompt_again (); + return; + } + } + + /* In this case EOF should exit the shell. Do it now. */ + reset_parser (); + exit_builtin ((WORD_LIST *)NULL); + } + else + { + /* We don't write history files, etc., for non-interactive shells. */ + EOF_Reached = 1; + } +} + +/************************************************ + * * + * STRING PARSING FUNCTIONS * + * * + ************************************************/ + +/* It's very important that these two functions treat the characters + between ( and ) identically. */ + +static WORD_LIST parse_string_error; + +/* Take a string and run it through the shell parser, returning the + resultant word list. Used by compound array assignment. */ +WORD_LIST * +parse_string_to_word_list (s, flags, whom) + char *s; + int flags; + const char *whom; +{ + WORD_LIST *wl; + int tok, orig_current_token, orig_line_number, orig_input_terminator; + int orig_line_count; + int old_echo_input, old_expand_aliases; +#if defined (HISTORY) + int old_remember_on_history, old_history_expansion_inhibited; +#endif + +#if defined (HISTORY) + old_remember_on_history = remember_on_history; +# if defined (BANG_HISTORY) + old_history_expansion_inhibited = history_expansion_inhibited; +# endif + bash_history_disable (); +#endif + + orig_line_number = line_number; + orig_line_count = current_command_line_count; + orig_input_terminator = shell_input_line_terminator; + old_echo_input = echo_input_at_read; + old_expand_aliases = expand_aliases; + + push_stream (1); + last_read_token = WORD; /* WORD to allow reserved words here */ + current_command_line_count = 0; + echo_input_at_read = expand_aliases = 0; + + with_input_from_string (s, whom); + wl = (WORD_LIST *)NULL; + + if (flags & 1) + parser_state |= PST_COMPASSIGN|PST_REPARSE; + + while ((tok = read_token (READ)) != yacc_EOF) + { + if (tok == '\n' && *bash_input.location.string == '\0') + break; + if (tok == '\n') /* Allow newlines in compound assignments */ + continue; + if (tok != WORD && tok != ASSIGNMENT_WORD) + { + line_number = orig_line_number + line_number - 1; + orig_current_token = current_token; + current_token = tok; + yyerror (NULL); /* does the right thing */ + current_token = orig_current_token; + if (wl) + dispose_words (wl); + wl = &parse_string_error; + break; + } + wl = make_word_list (yylval.word, wl); + } + + last_read_token = '\n'; + pop_stream (); + +#if defined (HISTORY) + remember_on_history = old_remember_on_history; +# if defined (BANG_HISTORY) + history_expansion_inhibited = old_history_expansion_inhibited; +# endif /* BANG_HISTORY */ +#endif /* HISTORY */ + + echo_input_at_read = old_echo_input; + expand_aliases = old_expand_aliases; + + current_command_line_count = orig_line_count; + shell_input_line_terminator = orig_input_terminator; + + if (flags & 1) + parser_state &= ~(PST_COMPASSIGN|PST_REPARSE); + + if (wl == &parse_string_error) + { + last_command_exit_value = EXECUTION_FAILURE; + if (interactive_shell == 0 && posixly_correct) + jump_to_top_level (FORCE_EOF); + else + jump_to_top_level (DISCARD); + } + + return (REVERSE_LIST (wl, WORD_LIST *)); +} + +static char * +parse_compound_assignment (retlenp) + int *retlenp; +{ + WORD_LIST *wl, *rl; + int tok, orig_line_number, orig_token_size, orig_last_token, assignok; + char *saved_token, *ret; + + saved_token = token; + orig_token_size = token_buffer_size; + orig_line_number = line_number; + orig_last_token = last_read_token; + + last_read_token = WORD; /* WORD to allow reserved words here */ + + token = (char *)NULL; + token_buffer_size = 0; + + assignok = parser_state&PST_ASSIGNOK; /* XXX */ + + wl = (WORD_LIST *)NULL; /* ( */ + parser_state |= PST_COMPASSIGN; + + while ((tok = read_token (READ)) != ')') + { + if (tok == '\n') /* Allow newlines in compound assignments */ + { + if (SHOULD_PROMPT ()) + prompt_again (); + continue; + } + if (tok != WORD && tok != ASSIGNMENT_WORD) + { + current_token = tok; /* for error reporting */ + if (tok == yacc_EOF) /* ( */ + parser_error (orig_line_number, _("unexpected EOF while looking for matching `)'")); + else + yyerror(NULL); /* does the right thing */ + if (wl) + dispose_words (wl); + wl = &parse_string_error; + break; + } + wl = make_word_list (yylval.word, wl); + } + + FREE (token); + token = saved_token; + token_buffer_size = orig_token_size; + + parser_state &= ~PST_COMPASSIGN; + + if (wl == &parse_string_error) + { + last_command_exit_value = EXECUTION_FAILURE; + last_read_token = '\n'; /* XXX */ + if (interactive_shell == 0 && posixly_correct) + jump_to_top_level (FORCE_EOF); + else + jump_to_top_level (DISCARD); + } + + last_read_token = orig_last_token; /* XXX - was WORD? */ + + if (wl) + { + rl = REVERSE_LIST (wl, WORD_LIST *); + ret = string_list (rl); + dispose_words (rl); + } + else + ret = (char *)NULL; + + if (retlenp) + *retlenp = (ret && *ret) ? strlen (ret) : 0; + + if (assignok) + parser_state |= PST_ASSIGNOK; + + return ret; +} + +/************************************************ + * * + * SAVING AND RESTORING PARTIAL PARSE STATE * + * * + ************************************************/ + +sh_parser_state_t * +save_parser_state (ps) + sh_parser_state_t *ps; +{ + if (ps == 0) + ps = (sh_parser_state_t *)xmalloc (sizeof (sh_parser_state_t)); + if (ps == 0) + return ((sh_parser_state_t *)NULL); + + ps->parser_state = parser_state; + ps->token_state = save_token_state (); + + ps->input_line_terminator = shell_input_line_terminator; + ps->eof_encountered = eof_encountered; + + ps->prompt_string_pointer = prompt_string_pointer; + + ps->current_command_line_count = current_command_line_count; + +#if defined (HISTORY) + ps->remember_on_history = remember_on_history; +# if defined (BANG_HISTORY) + ps->history_expansion_inhibited = history_expansion_inhibited; +# endif +#endif + + ps->last_command_exit_value = last_command_exit_value; +#if defined (ARRAY_VARS) + ps->pipestatus = save_pipestatus_array (); +#endif + + ps->last_shell_builtin = last_shell_builtin; + ps->this_shell_builtin = this_shell_builtin; + + ps->expand_aliases = expand_aliases; + ps->echo_input_at_read = echo_input_at_read; + + ps->token = token; + ps->token_buffer_size = token_buffer_size; + /* Force reallocation on next call to read_token_word */ + token = 0; + token_buffer_size = 0; + + return (ps); +} + +void +restore_parser_state (ps) + sh_parser_state_t *ps; +{ + if (ps == 0) + return; + + parser_state = ps->parser_state; + if (ps->token_state) + { + restore_token_state (ps->token_state); + free (ps->token_state); + } + + shell_input_line_terminator = ps->input_line_terminator; + eof_encountered = ps->eof_encountered; + + prompt_string_pointer = ps->prompt_string_pointer; + + current_command_line_count = ps->current_command_line_count; + +#if defined (HISTORY) + remember_on_history = ps->remember_on_history; +# if defined (BANG_HISTORY) + history_expansion_inhibited = ps->history_expansion_inhibited; +# endif +#endif + + last_command_exit_value = ps->last_command_exit_value; +#if defined (ARRAY_VARS) + restore_pipestatus_array (ps->pipestatus); +#endif + + last_shell_builtin = ps->last_shell_builtin; + this_shell_builtin = ps->this_shell_builtin; + + expand_aliases = ps->expand_aliases; + echo_input_at_read = ps->echo_input_at_read; + + FREE (token); + token = ps->token; + token_buffer_size = ps->token_buffer_size; +} + +sh_input_line_state_t * +save_input_line_state (ls) + sh_input_line_state_t *ls; +{ + if (ls == 0) + ls = (sh_input_line_state_t *)xmalloc (sizeof (sh_input_line_state_t)); + if (ls == 0) + return ((sh_input_line_state_t *)NULL); + + ls->input_line = shell_input_line; + ls->input_line_size = shell_input_line_size; + ls->input_line_len = shell_input_line_len; + ls->input_line_index = shell_input_line_index; + + /* force reallocation */ + shell_input_line = 0; + shell_input_line_size = shell_input_line_len = shell_input_line_index = 0; + + return ls; +} + +void +restore_input_line_state (ls) + sh_input_line_state_t *ls; +{ + FREE (shell_input_line); + shell_input_line = ls->input_line; + shell_input_line_size = ls->input_line_size; + shell_input_line_len = ls->input_line_len; + shell_input_line_index = ls->input_line_index; + + set_line_mbstate (); +} + +/************************************************ + * * + * MULTIBYTE CHARACTER HANDLING * + * * + ************************************************/ + +#if defined (HANDLE_MULTIBYTE) +static void +set_line_mbstate () +{ + int i, previ, len, c; + mbstate_t mbs, prevs; + size_t mbclen; + + if (shell_input_line == NULL) + return; + len = strlen (shell_input_line); /* XXX - shell_input_line_len ? */ + FREE (shell_input_line_property); + shell_input_line_property = (char *)xmalloc (len + 1); + + memset (&prevs, '\0', sizeof (mbstate_t)); + for (i = previ = 0; i < len; i++) + { + mbs = prevs; + + c = shell_input_line[i]; + if (c == EOF) + { + int j; + for (j = i; j < len; j++) + shell_input_line_property[j] = 1; + break; + } + + mbclen = mbrlen (shell_input_line + previ, i - previ + 1, &mbs); + if (mbclen == 1 || mbclen == (size_t)-1) + { + mbclen = 1; + previ = i + 1; + } + else if (mbclen == (size_t)-2) + mbclen = 0; + else if (mbclen > 1) + { + mbclen = 0; + previ = i + 1; + prevs = mbs; + } + else + { + /* XXX - what to do if mbrlen returns 0? (null wide character) */ + int j; + for (j = i; j < len; j++) + shell_input_line_property[j] = 1; + break; + } + + shell_input_line_property[i] = mbclen; + } +} +#endif /* HANDLE_MULTIBYTE */ diff --git a/print_cmd.c b/print_cmd.c index 34cec3f33..ac5773a3a 100644 --- a/print_cmd.c +++ b/print_cmd.c @@ -1416,7 +1416,7 @@ indent (amount) for (i = 0; amount > 0; amount--) indentation_string[i++] = ' '; indentation_string[i] = '\0'; - cprintf (indentation_string); + cprintf ("%s", indentation_string); } static void diff --git a/print_cmd.c~ b/print_cmd.c~ new file mode 100644 index 000000000..34cec3f33 --- /dev/null +++ b/print_cmd.c~ @@ -0,0 +1,1576 @@ +/* print_command -- A way to make readable commands from a command tree. */ + +/* Copyright (C) 1989-2011 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include "config.h" + +#include + +#if defined (HAVE_UNISTD_H) +# ifdef _MINIX +# include +# endif +# include +#endif + +#if defined (PREFER_STDARG) +# include +#else +# include +#endif + +#include "bashansi.h" +#include "bashintl.h" + +#include "shell.h" +#include "flags.h" +#include /* use <...> so we pick it up from the build directory */ + +#include "shmbutil.h" + +#include "builtins/common.h" + +#if !HAVE_DECL_PRINTF +extern int printf __P((const char *, ...)); /* Yuck. Double yuck. */ +#endif + +extern int indirection_level; + +static int indentation; +static int indentation_amount = 4; + +#if defined (PREFER_STDARG) +typedef void PFUNC __P((const char *, ...)); + +static void cprintf __P((const char *, ...)) __attribute__((__format__ (printf, 1, 2))); +static void xprintf __P((const char *, ...)) __attribute__((__format__ (printf, 1, 2))); +#else +#define PFUNC VFunction +static void cprintf (); +static void xprintf (); +#endif + +static void reset_locals __P((void)); +static void newline __P((char *)); +static void indent __P((int)); +static void semicolon __P((void)); +static void the_printed_command_resize __P((int)); + +static void make_command_string_internal __P((COMMAND *)); +static void _print_word_list __P((WORD_LIST *, char *, PFUNC *)); +static void command_print_word_list __P((WORD_LIST *, char *)); +static void print_case_clauses __P((PATTERN_LIST *)); +static void print_redirection_list __P((REDIRECT *)); +static void print_redirection __P((REDIRECT *)); +static void print_heredoc_header __P((REDIRECT *)); +static void print_heredoc_body __P((REDIRECT *)); +static void print_heredocs __P((REDIRECT *)); +static void print_deferred_heredocs __P((const char *)); + +static void print_for_command __P((FOR_COM *)); +#if defined (ARITH_FOR_COMMAND) +static void print_arith_for_command __P((ARITH_FOR_COM *)); +#endif +#if defined (SELECT_COMMAND) +static void print_select_command __P((SELECT_COM *)); +#endif +static void print_group_command __P((GROUP_COM *)); +static void print_case_command __P((CASE_COM *)); +static void print_while_command __P((WHILE_COM *)); +static void print_until_command __P((WHILE_COM *)); +static void print_until_or_while __P((WHILE_COM *, char *)); +static void print_if_command __P((IF_COM *)); +#if defined (COND_COMMAND) +static void print_cond_node __P((COND_COM *)); +#endif +static void print_function_def __P((FUNCTION_DEF *)); + +#define PRINTED_COMMAND_INITIAL_SIZE 64 +#define PRINTED_COMMAND_GROW_SIZE 128 + +char *the_printed_command = (char *)NULL; +int the_printed_command_size = 0; +int command_string_index = 0; + +int xtrace_fd = -1; +FILE *xtrace_fp = 0; + +#define CHECK_XTRACE_FP xtrace_fp = (xtrace_fp ? xtrace_fp : stderr) + +/* shell expansion characters: used in print_redirection_list */ +#define EXPCHAR(c) ((c) == '{' || (c) == '~' || (c) == '$' || (c) == '`') + +#define PRINT_DEFERRED_HEREDOCS(x) \ + do { \ + if (deferred_heredocs) \ + print_deferred_heredocs (x); \ + } while (0) + +/* Non-zero means the stuff being printed is inside of a function def. */ +static int inside_function_def; +static int skip_this_indent; +static int was_heredoc; +static int printing_connection; +static REDIRECT *deferred_heredocs; + +/* The depth of the group commands that we are currently printing. This + includes the group command that is a function body. */ +static int group_command_nesting; + +/* A buffer to indicate the indirection level (PS4) when set -x is enabled. */ +static char *indirection_string = 0; +static int indirection_stringsiz = 0; + +/* Print COMMAND (a command tree) on standard output. */ +void +print_command (command) + COMMAND *command; +{ + command_string_index = 0; + printf ("%s", make_command_string (command)); +} + +/* Make a string which is the printed representation of the command + tree in COMMAND. We return this string. However, the string is + not consed, so you have to do that yourself if you want it to + remain around. */ +char * +make_command_string (command) + COMMAND *command; +{ + command_string_index = was_heredoc = 0; + deferred_heredocs = 0; + make_command_string_internal (command); + return (the_printed_command); +} + +/* The internal function. This is the real workhorse. */ +static void +make_command_string_internal (command) + COMMAND *command; +{ + char s[3]; + + if (command == 0) + cprintf (""); + else + { + if (skip_this_indent) + skip_this_indent--; + else + indent (indentation); + + if (command->flags & CMD_TIME_PIPELINE) + { + cprintf ("time "); + if (command->flags & CMD_TIME_POSIX) + cprintf ("-p "); + } + + if (command->flags & CMD_INVERT_RETURN) + cprintf ("! "); + + switch (command->type) + { + case cm_for: + print_for_command (command->value.For); + break; + +#if defined (ARITH_FOR_COMMAND) + case cm_arith_for: + print_arith_for_command (command->value.ArithFor); + break; +#endif + +#if defined (SELECT_COMMAND) + case cm_select: + print_select_command (command->value.Select); + break; +#endif + + case cm_case: + print_case_command (command->value.Case); + break; + + case cm_while: + print_while_command (command->value.While); + break; + + case cm_until: + print_until_command (command->value.While); + break; + + case cm_if: + print_if_command (command->value.If); + break; + +#if defined (DPAREN_ARITHMETIC) + case cm_arith: + print_arith_command (command->value.Arith->exp); + break; +#endif + +#if defined (COND_COMMAND) + case cm_cond: + print_cond_command (command->value.Cond); + break; +#endif + + case cm_simple: + print_simple_command (command->value.Simple); + break; + + case cm_connection: + + skip_this_indent++; + printing_connection++; + make_command_string_internal (command->value.Connection->first); + + switch (command->value.Connection->connector) + { + case '&': + case '|': + { + char c = command->value.Connection->connector; + + s[0] = ' '; + s[1] = c; + s[2] = '\0'; + + print_deferred_heredocs (s); + + if (c != '&' || command->value.Connection->second) + { + cprintf (" "); + skip_this_indent++; + } + } + break; + + case AND_AND: + print_deferred_heredocs (" && "); + if (command->value.Connection->second) + skip_this_indent++; + break; + + case OR_OR: + print_deferred_heredocs (" || "); + if (command->value.Connection->second) + skip_this_indent++; + break; + + case ';': + if (deferred_heredocs == 0) + { + if (was_heredoc == 0) + cprintf (";"); + else + was_heredoc = 0; + } + else + print_deferred_heredocs (inside_function_def ? "" : ";"); + + if (inside_function_def) + cprintf ("\n"); + else + { + cprintf (" "); + if (command->value.Connection->second) + skip_this_indent++; + } + break; + + default: + cprintf (_("print_command: bad connector `%d'"), + command->value.Connection->connector); + break; + } + + make_command_string_internal (command->value.Connection->second); + PRINT_DEFERRED_HEREDOCS (""); + printing_connection--; + break; + + case cm_function_def: + print_function_def (command->value.Function_def); + break; + + case cm_group: + print_group_command (command->value.Group); + break; + + case cm_subshell: + cprintf ("( "); + skip_this_indent++; + make_command_string_internal (command->value.Subshell->command); + PRINT_DEFERRED_HEREDOCS (""); + cprintf (" )"); + break; + + case cm_coproc: + cprintf ("coproc %s ", command->value.Coproc->name); + skip_this_indent++; + make_command_string_internal (command->value.Coproc->command); + break; + + default: + command_error ("print_command", CMDERR_BADTYPE, command->type, 0); + break; + } + + + if (command->redirects) + { + cprintf (" "); + print_redirection_list (command->redirects); + } + } +} + +static void +_print_word_list (list, separator, pfunc) + WORD_LIST *list; + char *separator; + PFUNC *pfunc; +{ + WORD_LIST *w; + + for (w = list; w; w = w->next) + (*pfunc) ("%s%s", w->word->word, w->next ? separator : ""); +} + +void +print_word_list (list, separator) + WORD_LIST *list; + char *separator; +{ + _print_word_list (list, separator, xprintf); +} + +void +xtrace_set (fd, fp) + int fd; + FILE *fp; +{ + if (fd >= 0 && sh_validfd (fd) == 0) + { + internal_error (_("xtrace_set: %d: invalid file descriptor"), fd); + return; + } + if (fp == 0) + { + internal_error (_("xtrace_set: NULL file pointer")); + return; + } + if (fd >= 0 && fileno (fp) != fd) + internal_warning (_("xtrace fd (%d) != fileno xtrace fp (%d)"), fd, fileno (fp)); + + xtrace_fd = fd; + xtrace_fp = fp; +} + +void +xtrace_init () +{ + xtrace_set (-1, stderr); +} + +void +xtrace_reset () +{ + if (xtrace_fd >= 0 && xtrace_fp) + { + fflush (xtrace_fp); + fclose (xtrace_fp); + } + else if (xtrace_fd >= 0) + close (xtrace_fd); + + xtrace_fd = -1; + xtrace_fp = stderr; +} + +void +xtrace_fdchk (fd) + int fd; +{ + if (fd == xtrace_fd) + xtrace_reset (); +} + +/* Return a string denoting what our indirection level is. */ + +char * +indirection_level_string () +{ + register int i, j; + char *ps4; + char ps4_firstc[MB_LEN_MAX+1]; + int ps4_firstc_len, ps4_len, ineed, old; + + ps4 = get_string_value ("PS4"); + if (indirection_string == 0) + indirection_string = xmalloc (indirection_stringsiz = 100); + indirection_string[0] = '\0'; + + if (ps4 == 0 || *ps4 == '\0') + return (indirection_string); + + old = change_flag ('x', FLAG_OFF); + ps4 = decode_prompt_string (ps4); + if (old) + change_flag ('x', FLAG_ON); + + if (ps4 == 0 || *ps4 == '\0') + return (indirection_string); + +#if defined (HANDLE_MULTIBYTE) + ps4_len = strnlen (ps4, MB_CUR_MAX); + ps4_firstc_len = MBLEN (ps4, ps4_len); + if (ps4_firstc_len == 1 || ps4_firstc_len == 0 || ps4_firstc_len < 0) + { + ps4_firstc[0] = ps4[0]; + ps4_firstc[ps4_firstc_len = 1] = '\0'; + } + else + memcpy (ps4_firstc, ps4, ps4_firstc_len); +#else + ps4_firstc[0] = ps4[0]; + ps4_firstc[ps4_firstc_len = 1] = '\0'; +#endif + + /* Dynamically resize indirection_string so we have room for everything + and we don't have to truncate ps4 */ + ineed = (ps4_firstc_len * indirection_level) + strlen (ps4); + if (ineed > indirection_stringsiz - 1) + { + indirection_stringsiz = ineed + 1; + indirection_string = xrealloc (indirection_string, indirection_stringsiz); + } + + for (i = j = 0; ps4_firstc[0] && j < indirection_level && i < indirection_stringsiz - 1; i += ps4_firstc_len, j++) + { + if (ps4_firstc_len == 1) + indirection_string[i] = ps4_firstc[0]; + else + memcpy (indirection_string+i, ps4_firstc, ps4_firstc_len); + } + + for (j = ps4_firstc_len; *ps4 && ps4[j] && i < indirection_stringsiz - 1; i++, j++) + indirection_string[i] = ps4[j]; + + indirection_string[i] = '\0'; + free (ps4); + return (indirection_string); +} + +void +xtrace_print_assignment (name, value, assign_list, xflags) + char *name, *value; + int assign_list, xflags; +{ + char *nval; + + CHECK_XTRACE_FP; + + if (xflags) + fprintf (xtrace_fp, "%s", indirection_level_string ()); + + /* VALUE should not be NULL when this is called. */ + if (*value == '\0' || assign_list) + nval = value; + else if (sh_contains_shell_metas (value)) + nval = sh_single_quote (value); + else if (ansic_shouldquote (value)) + nval = ansic_quote (value, 0, (int *)0); + else + nval = value; + + if (assign_list) + fprintf (xtrace_fp, "%s=(%s)\n", name, nval); + else + fprintf (xtrace_fp, "%s=%s\n", name, nval); + + if (nval != value) + FREE (nval); + + fflush (xtrace_fp); +} + +/* A function to print the words of a simple command when set -x is on. */ +void +xtrace_print_word_list (list, xtflags) + WORD_LIST *list; + int xtflags; +{ + WORD_LIST *w; + char *t, *x; + + CHECK_XTRACE_FP; + + if (xtflags) + fprintf (xtrace_fp, "%s", indirection_level_string ()); + + for (w = list; w; w = w->next) + { + t = w->word->word; + if (t == 0 || *t == '\0') + fprintf (xtrace_fp, "''%s", w->next ? " " : ""); + else if (sh_contains_shell_metas (t)) + { + x = sh_single_quote (t); + fprintf (xtrace_fp, "%s%s", x, w->next ? " " : ""); + free (x); + } + else if (ansic_shouldquote (t)) + { + x = ansic_quote (t, 0, (int *)0); + fprintf (xtrace_fp, "%s%s", x, w->next ? " " : ""); + free (x); + } + else + fprintf (xtrace_fp, "%s%s", t, w->next ? " " : ""); + } + fprintf (xtrace_fp, "\n"); + fflush (xtrace_fp); +} + +static void +command_print_word_list (list, separator) + WORD_LIST *list; + char *separator; +{ + _print_word_list (list, separator, cprintf); +} + +void +print_for_command_head (for_command) + FOR_COM *for_command; +{ + cprintf ("for %s in ", for_command->name->word); + command_print_word_list (for_command->map_list, " "); +} + +void +xtrace_print_for_command_head (for_command) + FOR_COM *for_command; +{ + CHECK_XTRACE_FP; + fprintf (xtrace_fp, "%s", indirection_level_string ()); + fprintf (xtrace_fp, "for %s in ", for_command->name->word); + xtrace_print_word_list (for_command->map_list, 0); +} + +static void +print_for_command (for_command) + FOR_COM *for_command; +{ + print_for_command_head (for_command); + cprintf (";"); + newline ("do\n"); + + indentation += indentation_amount; + make_command_string_internal (for_command->action); + PRINT_DEFERRED_HEREDOCS (""); + semicolon (); + indentation -= indentation_amount; + + newline ("done"); +} + +#if defined (ARITH_FOR_COMMAND) +static void +print_arith_for_command (arith_for_command) + ARITH_FOR_COM *arith_for_command; +{ + cprintf ("for (("); + command_print_word_list (arith_for_command->init, " "); + cprintf ("; "); + command_print_word_list (arith_for_command->test, " "); + cprintf ("; "); + command_print_word_list (arith_for_command->step, " "); + cprintf ("))"); + newline ("do\n"); + indentation += indentation_amount; + make_command_string_internal (arith_for_command->action); + PRINT_DEFERRED_HEREDOCS (""); + semicolon (); + indentation -= indentation_amount; + newline ("done"); +} +#endif /* ARITH_FOR_COMMAND */ + +#if defined (SELECT_COMMAND) +void +print_select_command_head (select_command) + SELECT_COM *select_command; +{ + cprintf ("select %s in ", select_command->name->word); + command_print_word_list (select_command->map_list, " "); +} + +void +xtrace_print_select_command_head (select_command) + SELECT_COM *select_command; +{ + CHECK_XTRACE_FP; + fprintf (xtrace_fp, "%s", indirection_level_string ()); + fprintf (xtrace_fp, "select %s in ", select_command->name->word); + xtrace_print_word_list (select_command->map_list, 0); +} + +static void +print_select_command (select_command) + SELECT_COM *select_command; +{ + print_select_command_head (select_command); + + cprintf (";"); + newline ("do\n"); + indentation += indentation_amount; + make_command_string_internal (select_command->action); + PRINT_DEFERRED_HEREDOCS (""); + semicolon (); + indentation -= indentation_amount; + newline ("done"); +} +#endif /* SELECT_COMMAND */ + +static void +print_group_command (group_command) + GROUP_COM *group_command; +{ + group_command_nesting++; + cprintf ("{ "); + + if (inside_function_def == 0) + skip_this_indent++; + else + { + /* This is a group command { ... } inside of a function + definition, and should be printed as a multiline group + command, using the current indentation. */ + cprintf ("\n"); + indentation += indentation_amount; + } + + make_command_string_internal (group_command->command); + PRINT_DEFERRED_HEREDOCS (""); + + if (inside_function_def) + { + cprintf ("\n"); + indentation -= indentation_amount; + indent (indentation); + } + else + { + semicolon (); + cprintf (" "); + } + + cprintf ("}"); + + group_command_nesting--; +} + +void +print_case_command_head (case_command) + CASE_COM *case_command; +{ + cprintf ("case %s in ", case_command->word->word); +} + +void +xtrace_print_case_command_head (case_command) + CASE_COM *case_command; +{ + CHECK_XTRACE_FP; + fprintf (xtrace_fp, "%s", indirection_level_string ()); + fprintf (xtrace_fp, "case %s in\n", case_command->word->word); +} + +static void +print_case_command (case_command) + CASE_COM *case_command; +{ + print_case_command_head (case_command); + + if (case_command->clauses) + print_case_clauses (case_command->clauses); + newline ("esac"); +} + +static void +print_case_clauses (clauses) + PATTERN_LIST *clauses; +{ + indentation += indentation_amount; + while (clauses) + { + newline (""); + command_print_word_list (clauses->patterns, " | "); + cprintf (")\n"); + indentation += indentation_amount; + make_command_string_internal (clauses->action); + indentation -= indentation_amount; + PRINT_DEFERRED_HEREDOCS (""); + if (clauses->flags & CASEPAT_FALLTHROUGH) + newline (";&"); + else if (clauses->flags & CASEPAT_TESTNEXT) + newline (";;&"); + else + newline (";;"); + clauses = clauses->next; + } + indentation -= indentation_amount; +} + +static void +print_while_command (while_command) + WHILE_COM *while_command; +{ + print_until_or_while (while_command, "while"); +} + +static void +print_until_command (while_command) + WHILE_COM *while_command; +{ + print_until_or_while (while_command, "until"); +} + +static void +print_until_or_while (while_command, which) + WHILE_COM *while_command; + char *which; +{ + cprintf ("%s ", which); + skip_this_indent++; + make_command_string_internal (while_command->test); + PRINT_DEFERRED_HEREDOCS (""); + semicolon (); + cprintf (" do\n"); /* was newline ("do\n"); */ + indentation += indentation_amount; + make_command_string_internal (while_command->action); + PRINT_DEFERRED_HEREDOCS (""); + indentation -= indentation_amount; + semicolon (); + newline ("done"); +} + +static void +print_if_command (if_command) + IF_COM *if_command; +{ + cprintf ("if "); + skip_this_indent++; + make_command_string_internal (if_command->test); + semicolon (); + cprintf (" then\n"); + indentation += indentation_amount; + make_command_string_internal (if_command->true_case); + PRINT_DEFERRED_HEREDOCS (""); + indentation -= indentation_amount; + + if (if_command->false_case) + { + semicolon (); + newline ("else\n"); + indentation += indentation_amount; + make_command_string_internal (if_command->false_case); + PRINT_DEFERRED_HEREDOCS (""); + indentation -= indentation_amount; + } + semicolon (); + newline ("fi"); +} + +#if defined (DPAREN_ARITHMETIC) +void +print_arith_command (arith_cmd_list) + WORD_LIST *arith_cmd_list; +{ + cprintf ("(("); + command_print_word_list (arith_cmd_list, " "); + cprintf ("))"); +} +#endif + +#if defined (COND_COMMAND) +static void +print_cond_node (cond) + COND_COM *cond; +{ + if (cond->flags & CMD_INVERT_RETURN) + cprintf ("! "); + + if (cond->type == COND_EXPR) + { + cprintf ("( "); + print_cond_node (cond->left); + cprintf (" )"); + } + else if (cond->type == COND_AND) + { + print_cond_node (cond->left); + cprintf (" && "); + print_cond_node (cond->right); + } + else if (cond->type == COND_OR) + { + print_cond_node (cond->left); + cprintf (" || "); + print_cond_node (cond->right); + } + else if (cond->type == COND_UNARY) + { + cprintf ("%s", cond->op->word); + cprintf (" "); + print_cond_node (cond->left); + } + else if (cond->type == COND_BINARY) + { + print_cond_node (cond->left); + cprintf (" "); + cprintf ("%s", cond->op->word); + cprintf (" "); + print_cond_node (cond->right); + } + else if (cond->type == COND_TERM) + { + cprintf ("%s", cond->op->word); /* need to add quoting here */ + } +} + +void +print_cond_command (cond) + COND_COM *cond; +{ + cprintf ("[[ "); + print_cond_node (cond); + cprintf (" ]]"); +} + +#ifdef DEBUG +void +debug_print_cond_command (cond) + COND_COM *cond; +{ + fprintf (stderr, "DEBUG: "); + command_string_index = 0; + print_cond_command (cond); + fprintf (stderr, "%s\n", the_printed_command); +} +#endif + +void +xtrace_print_cond_term (type, invert, op, arg1, arg2) + int type, invert; + WORD_DESC *op; + char *arg1, *arg2; +{ + CHECK_XTRACE_FP; + command_string_index = 0; + fprintf (xtrace_fp, "%s", indirection_level_string ()); + fprintf (xtrace_fp, "[[ "); + if (invert) + fprintf (xtrace_fp, "! "); + + if (type == COND_UNARY) + { + fprintf (xtrace_fp, "%s ", op->word); + fprintf (xtrace_fp, "%s", (arg1 && *arg1) ? arg1 : "''"); + } + else if (type == COND_BINARY) + { + fprintf (xtrace_fp, "%s", (arg1 && *arg1) ? arg1 : "''"); + fprintf (xtrace_fp, " %s ", op->word); + fprintf (xtrace_fp, "%s", (arg2 && *arg2) ? arg2 : "''"); + } + + fprintf (xtrace_fp, " ]]\n"); + + fflush (xtrace_fp); +} +#endif /* COND_COMMAND */ + +#if defined (DPAREN_ARITHMETIC) || defined (ARITH_FOR_COMMAND) +/* A function to print the words of an arithmetic command when set -x is on. */ +void +xtrace_print_arith_cmd (list) + WORD_LIST *list; +{ + WORD_LIST *w; + + CHECK_XTRACE_FP; + fprintf (xtrace_fp, "%s", indirection_level_string ()); + fprintf (xtrace_fp, "(( "); + for (w = list; w; w = w->next) + fprintf (xtrace_fp, "%s%s", w->word->word, w->next ? " " : ""); + fprintf (xtrace_fp, " ))\n"); + + fflush (xtrace_fp); +} +#endif + +void +print_simple_command (simple_command) + SIMPLE_COM *simple_command; +{ + command_print_word_list (simple_command->words, " "); + + if (simple_command->redirects) + { + cprintf (" "); + print_redirection_list (simple_command->redirects); + } +} + +static void +print_heredocs (heredocs) + REDIRECT *heredocs; +{ + REDIRECT *hdtail; + + cprintf (" "); + for (hdtail = heredocs; hdtail; hdtail = hdtail->next) + { + print_redirection (hdtail); + cprintf ("\n"); + } + was_heredoc = 1; +} + +/* Print heredocs that are attached to the command before the connector + represented by CSTRING. The parsing semantics require us to print the + here-doc delimiters, then the connector (CSTRING), then the here-doc + bodies. We don't print the connector if it's a `;', but we use it to + note not to print an extra space after the last heredoc body and + newline. */ +static void +print_deferred_heredocs (cstring) + const char *cstring; +{ + REDIRECT *hdtail; + + for (hdtail = deferred_heredocs; hdtail; hdtail = hdtail->next) + { + cprintf (" "); + print_heredoc_header (hdtail); + } + if (cstring && cstring[0] && (cstring[0] != ';' || cstring[1])) + cprintf ("%s", cstring); + if (deferred_heredocs) + cprintf ("\n"); + for (hdtail = deferred_heredocs; hdtail; hdtail = hdtail->next) + { + print_heredoc_body (hdtail); + cprintf ("\n"); + } + if (deferred_heredocs) + { + if (cstring && cstring[0] && (cstring[0] != ';' || cstring[1])) + cprintf (" "); /* make sure there's at least one space */ + dispose_redirects (deferred_heredocs); + was_heredoc = 1; + } + deferred_heredocs = (REDIRECT *)NULL; +} + +static void +print_redirection_list (redirects) + REDIRECT *redirects; +{ + REDIRECT *heredocs, *hdtail, *newredir; + char *rw; + + heredocs = (REDIRECT *)NULL; + hdtail = heredocs; + + was_heredoc = 0; + while (redirects) + { + /* Defer printing the here documents until we've printed the + rest of the redirections. */ + if (redirects->instruction == r_reading_until || redirects->instruction == r_deblank_reading_until) + { + newredir = copy_redirect (redirects); + newredir->next = (REDIRECT *)NULL; + if (heredocs) + { + hdtail->next = newredir; + hdtail = newredir; + } + else + hdtail = heredocs = newredir; + } + else if (redirects->instruction == r_duplicating_output_word && (redirects->flags & REDIR_VARASSIGN) == 0 && redirects->redirector.dest == 1) + { + /* Temporarily translate it as the execution code does. */ + rw = redirects->redirectee.filename->word; + if (rw && *rw != '-' && DIGIT (*rw) == 0 && EXPCHAR (*rw) == 0) + redirects->instruction = r_err_and_out; + print_redirection (redirects); + redirects->instruction = r_duplicating_output_word; + } + else + print_redirection (redirects); + + redirects = redirects->next; + if (redirects) + cprintf (" "); + } + + /* Now that we've printed all the other redirections (on one line), + print the here documents. */ + if (heredocs && printing_connection) + deferred_heredocs = heredocs; + else if (heredocs) + { + print_heredocs (heredocs); + dispose_redirects (heredocs); + } +} + +static void +print_heredoc_header (redirect) + REDIRECT *redirect; +{ + int kill_leading; + char *x; + + kill_leading = redirect->instruction == r_deblank_reading_until; + + /* Here doc header */ + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redirect->redirector.filename->word); + else if (redirect->redirector.dest != 0) + cprintf ("%d", redirect->redirector.dest); + + /* If the here document delimiter is quoted, single-quote it. */ + if (redirect->redirectee.filename->flags & W_QUOTED) + { + x = sh_single_quote (redirect->here_doc_eof); + cprintf ("<<%s%s", kill_leading ? "-" : "", x); + free (x); + } + else + cprintf ("<<%s%s", kill_leading ? "-" : "", redirect->here_doc_eof); +} + +static void +print_heredoc_body (redirect) + REDIRECT *redirect; +{ + /* Here doc body */ + cprintf ("%s%s", redirect->redirectee.filename->word, redirect->here_doc_eof); +} + +static void +print_redirection (redirect) + REDIRECT *redirect; +{ + int redirector, redir_fd; + WORD_DESC *redirectee, *redir_word; + + redirectee = redirect->redirectee.filename; + redir_fd = redirect->redirectee.dest; + + redir_word = redirect->redirector.filename; + redirector = redirect->redirector.dest; + + switch (redirect->instruction) + { + case r_input_direction: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redir_word->word); + else if (redirector != 0) + cprintf ("%d", redirector); + cprintf ("< %s", redirectee->word); + break; + + case r_output_direction: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redir_word->word); + else if (redirector != 1) + cprintf ("%d", redirector); + cprintf ("> %s", redirectee->word); + break; + + case r_inputa_direction: /* Redirection created by the shell. */ + cprintf ("&"); + break; + + case r_output_force: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redir_word->word); + else if (redirector != 1) + cprintf ("%d", redirector); + cprintf (">|%s", redirectee->word); + break; + + case r_appending_to: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redir_word->word); + else if (redirector != 1) + cprintf ("%d", redirector); + cprintf (">> %s", redirectee->word); + break; + + case r_input_output: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redir_word->word); + else if (redirector != 1) + cprintf ("%d", redirector); + cprintf ("<> %s", redirectee->word); + break; + + case r_deblank_reading_until: + case r_reading_until: + print_heredoc_header (redirect); + cprintf ("\n"); + print_heredoc_body (redirect); + break; + + case r_reading_string: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}", redir_word->word); + else if (redirector != 0) + cprintf ("%d", redirector); +#if 0 + /* Don't need to check whether or not to requote, since original quotes + are still intact. The only thing that has happened is that $'...' + has been replaced with 'expanded ...'. */ + if (ansic_shouldquote (redirect->redirectee.filename->word)) + { + char *x; + x = ansic_quote (redirect->redirectee.filename->word, 0, (int *)0); + cprintf ("<<< %s", x); + free (x); + } + else +#endif + cprintf ("<<< %s", redirect->redirectee.filename->word); + break; + + case r_duplicating_input: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}<&%d", redir_word->word, redir_fd); + else + cprintf ("%d<&%d", redirector, redir_fd); + break; + + case r_duplicating_output: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}>&%d", redir_word->word, redir_fd); + else + cprintf ("%d>&%d", redirector, redir_fd); + break; + + case r_duplicating_input_word: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}<&%s", redir_word->word, redirectee->word); + else + cprintf ("%d<&%s", redirector, redirectee->word); + break; + + case r_duplicating_output_word: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}>&%s", redir_word->word, redirectee->word); + else + cprintf ("%d>&%s", redirector, redirectee->word); + break; + + case r_move_input: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}<&%d-", redir_word->word, redir_fd); + else + cprintf ("%d<&%d-", redirector, redir_fd); + break; + + case r_move_output: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}>&%d-", redir_word->word, redir_fd); + else + cprintf ("%d>&%d-", redirector, redir_fd); + break; + + case r_move_input_word: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}<&%s-", redir_word->word, redirectee->word); + else + cprintf ("%d<&%s-", redirector, redirectee->word); + break; + + case r_move_output_word: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}>&%s-", redir_word->word, redirectee->word); + else + cprintf ("%d>&%s-", redirector, redirectee->word); + break; + + case r_close_this: + if (redirect->rflags & REDIR_VARASSIGN) + cprintf ("{%s}>&-", redir_word->word); + else + cprintf ("%d>&-", redirector); + break; + + case r_err_and_out: + cprintf ("&>%s", redirectee->word); + break; + + case r_append_err_and_out: + cprintf ("&>>%s", redirectee->word); + break; + } +} + +static void +reset_locals () +{ + inside_function_def = 0; + indentation = 0; + printing_connection = 0; + deferred_heredocs = 0; +} + +static void +print_function_def (func) + FUNCTION_DEF *func; +{ + COMMAND *cmdcopy; + REDIRECT *func_redirects; + + func_redirects = NULL; + cprintf ("function %s () \n", func->name->word); + add_unwind_protect (reset_locals, 0); + + indent (indentation); + cprintf ("{ \n"); + + inside_function_def++; + indentation += indentation_amount; + + cmdcopy = copy_command (func->command); + if (cmdcopy->type == cm_group) + { + func_redirects = cmdcopy->redirects; + cmdcopy->redirects = (REDIRECT *)NULL; + } + make_command_string_internal (cmdcopy->type == cm_group + ? cmdcopy->value.Group->command + : cmdcopy); + /* XXX - PRINT_DEFERRED_HEREDOCS (""); ? */ + + remove_unwind_protect (); + indentation -= indentation_amount; + inside_function_def--; + + if (func_redirects) + { /* { */ + newline ("} "); + print_redirection_list (func_redirects); + cmdcopy->redirects = func_redirects; + } + else + newline ("}"); + + dispose_command (cmdcopy); +} + +/* Return the string representation of the named function. + NAME is the name of the function. + COMMAND is the function body. It should be a GROUP_COM. + flags&FUNC_MULTILINE is non-zero to pretty-print, or zero for all on one line. + flags&FUNC_EXTERNAL means convert from internal to external form + */ +char * +named_function_string (name, command, flags) + char *name; + COMMAND *command; + int flags; +{ + char *result; + int old_indent, old_amount; + COMMAND *cmdcopy; + REDIRECT *func_redirects; + + old_indent = indentation; + old_amount = indentation_amount; + command_string_index = was_heredoc = 0; + deferred_heredocs = 0; + + if (name && *name) + cprintf ("%s ", name); + + cprintf ("() "); + + if ((flags & FUNC_MULTILINE) == 0) + { + indentation = 1; + indentation_amount = 0; + } + else + { + cprintf ("\n"); + indentation += indentation_amount; + } + + inside_function_def++; + + cprintf ((flags & FUNC_MULTILINE) ? "{ \n" : "{ "); + + cmdcopy = copy_command (command); + /* Take any redirections specified in the function definition (which should + apply to the function as a whole) and save them for printing later. */ + func_redirects = (REDIRECT *)NULL; + if (cmdcopy->type == cm_group) + { + func_redirects = cmdcopy->redirects; + cmdcopy->redirects = (REDIRECT *)NULL; + } + make_command_string_internal (cmdcopy->type == cm_group + ? cmdcopy->value.Group->command + : cmdcopy); + /* XXX - PRINT_DEFERRED_HEREDOCS (""); ? */ + + indentation = old_indent; + indentation_amount = old_amount; + inside_function_def--; + + if (func_redirects) + { /* { */ + newline ("} "); + print_redirection_list (func_redirects); + cmdcopy->redirects = func_redirects; + } + else + newline ("}"); + + result = the_printed_command; + + if ((flags & FUNC_MULTILINE) == 0) + { +#if 0 + register int i; + for (i = 0; result[i]; i++) + if (result[i] == '\n') + { + strcpy (result + i, result + i + 1); + --i; + } +#else + if (result[2] == '\n') /* XXX -- experimental */ + memmove (result + 2, result + 3, strlen (result) - 2); + +#endif + } + + dispose_command (cmdcopy); + + if (flags & FUNC_EXTERNAL) + result = remove_quoted_escapes (result); + + return (result); +} + +static void +newline (string) + char *string; +{ + cprintf ("\n"); + indent (indentation); + if (string && *string) + cprintf ("%s", string); +} + +static char *indentation_string; +static int indentation_size; + +static void +indent (amount) + int amount; +{ + register int i; + + RESIZE_MALLOCED_BUFFER (indentation_string, 0, amount, indentation_size, 16); + + for (i = 0; amount > 0; amount--) + indentation_string[i++] = ' '; + indentation_string[i] = '\0'; + cprintf (indentation_string); +} + +static void +semicolon () +{ + if (command_string_index > 0 && + (the_printed_command[command_string_index - 1] == '&' || + the_printed_command[command_string_index - 1] == '\n')) + return; + cprintf (";"); +} + +/* How to make the string. */ +static void +#if defined (PREFER_STDARG) +cprintf (const char *control, ...) +#else +cprintf (control, va_alist) + const char *control; + va_dcl +#endif +{ + register const char *s; + char char_arg[2], *argp, intbuf[INT_STRLEN_BOUND (int) + 1]; + int digit_arg, arg_len, c; + va_list args; + + SH_VA_START (args, control); + + arg_len = strlen (control); + the_printed_command_resize (arg_len + 1); + + char_arg[1] = '\0'; + s = control; + while (s && *s) + { + c = *s++; + argp = (char *)NULL; + if (c != '%' || !*s) + { + char_arg[0] = c; + argp = char_arg; + arg_len = 1; + } + else + { + c = *s++; + switch (c) + { + case '%': + char_arg[0] = c; + argp = char_arg; + arg_len = 1; + break; + + case 's': + argp = va_arg (args, char *); + arg_len = strlen (argp); + break; + + case 'd': + /* Represent an out-of-range file descriptor with an out-of-range + integer value. We can do this because the only use of `%d' in + the calls to cprintf is to output a file descriptor number for + a redirection. */ + digit_arg = va_arg (args, int); + if (digit_arg < 0) + { + sprintf (intbuf, "%u", (unsigned)-1); + argp = intbuf; + } + else + argp = inttostr (digit_arg, intbuf, sizeof (intbuf)); + arg_len = strlen (argp); + break; + + case 'c': + char_arg[0] = va_arg (args, int); + argp = char_arg; + arg_len = 1; + break; + + default: + programming_error (_("cprintf: `%c': invalid format character"), c); + /*NOTREACHED*/ + } + } + + if (argp && arg_len) + { + the_printed_command_resize (arg_len + 1); + FASTCOPY (argp, the_printed_command + command_string_index, arg_len); + command_string_index += arg_len; + } + } + + va_end (args); + + the_printed_command[command_string_index] = '\0'; +} + +/* Ensure that there is enough space to stuff LENGTH characters into + THE_PRINTED_COMMAND. */ +static void +the_printed_command_resize (length) + int length; +{ + if (the_printed_command == 0) + { + the_printed_command_size = (length + PRINTED_COMMAND_INITIAL_SIZE - 1) & ~(PRINTED_COMMAND_INITIAL_SIZE - 1); + the_printed_command = (char *)xmalloc (the_printed_command_size); + command_string_index = 0; + } + else if ((command_string_index + length) >= the_printed_command_size) + { + int new; + new = command_string_index + length + 1; + + /* Round up to the next multiple of PRINTED_COMMAND_GROW_SIZE. */ + new = (new + PRINTED_COMMAND_GROW_SIZE - 1) & ~(PRINTED_COMMAND_GROW_SIZE - 1); + the_printed_command_size = new; + + the_printed_command = (char *)xrealloc (the_printed_command, the_printed_command_size); + } +} + +#if defined (HAVE_VPRINTF) +/* ``If vprintf is available, you may assume that vfprintf and vsprintf are + also available.'' */ + +static void +#if defined (PREFER_STDARG) +xprintf (const char *format, ...) +#else +xprintf (format, va_alist) + const char *format; + va_dcl +#endif +{ + va_list args; + + SH_VA_START (args, format); + + vfprintf (stdout, format, args); + va_end (args); +} + +#else + +static void +xprintf (format, arg1, arg2, arg3, arg4, arg5) + const char *format; +{ + printf (format, arg1, arg2, arg3, arg4, arg5); +} + +#endif /* !HAVE_VPRINTF */ diff --git a/redir.c b/redir.c index 8b61a71ac..027e6c910 100644 --- a/redir.c +++ b/redir.c @@ -1,6 +1,6 @@ /* redir.c -- Functions to perform input and output redirection. */ -/* Copyright (C) 1997-2009 Free Software Foundation, Inc. +/* Copyright (C) 1997-2012 Free Software Foundation, Inc. This file is part of GNU Bash, the Bourne Again SHell. diff --git a/redir.c~ b/redir.c~ new file mode 100644 index 000000000..8b61a71ac --- /dev/null +++ b/redir.c~ @@ -0,0 +1,1322 @@ +/* redir.c -- Functions to perform input and output redirection. */ + +/* Copyright (C) 1997-2009 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include "config.h" + +#if !defined (__GNUC__) && !defined (HAVE_ALLOCA_H) && defined (_AIX) + #pragma alloca +#endif /* _AIX && RISC6000 && !__GNUC__ */ + +#include +#include "bashtypes.h" +#if !defined (_MINIX) && defined (HAVE_SYS_FILE_H) +# include +#endif +#include "filecntl.h" +#include "posixstat.h" + +#if defined (HAVE_UNISTD_H) +# include +#endif + +#include + +#if !defined (errno) +extern int errno; +#endif + +#include "bashansi.h" +#include "bashintl.h" +#include "memalloc.h" + +#define NEED_FPURGE_DECL + +#include "shell.h" +#include "flags.h" +#include "execute_cmd.h" +#include "redir.h" + +#if defined (BUFFERED_INPUT) +# include "input.h" +#endif + +#define SHELL_FD_BASE 10 + +int expanding_redir; + +extern int posixly_correct; +extern int last_command_exit_value; +extern REDIRECT *redirection_undo_list; +extern REDIRECT *exec_redirection_undo_list; + +/* Static functions defined and used in this file. */ +static void add_exec_redirect __P((REDIRECT *)); +static int add_undo_redirect __P((int, enum r_instruction, int)); +static int add_undo_close_redirect __P((int)); +static int expandable_redirection_filename __P((REDIRECT *)); +static int stdin_redirection __P((enum r_instruction, int)); +static int undoablefd __P((int)); +static int do_redirection_internal __P((REDIRECT *, int)); + +static int write_here_document __P((int, WORD_DESC *)); +static int write_here_string __P((int, WORD_DESC *)); +static int here_document_to_fd __P((WORD_DESC *, enum r_instruction)); + +static int redir_special_open __P((int, char *, int, int, enum r_instruction)); +static int noclobber_open __P((char *, int, int, enum r_instruction)); +static int redir_open __P((char *, int, int, enum r_instruction)); + +static int redir_varassign __P((REDIRECT *, int)); +static int redir_varvalue __P((REDIRECT *)); + +/* Spare redirector used when translating [N]>&WORD[-] or [N]<&WORD[-] to + a new redirection and when creating the redirection undo list. */ +static REDIRECTEE rd; + +/* Set to errno when a here document cannot be created for some reason. + Used to print a reasonable error message. */ +static int heredoc_errno; + +#define REDIRECTION_ERROR(r, e, fd) \ +do { \ + if ((r) < 0) \ + { \ + if (fd >= 0) \ + close (fd); \ + last_command_exit_value = EXECUTION_FAILURE;\ + return ((e) == 0 ? EINVAL : (e));\ + } \ +} while (0) + +void +redirection_error (temp, error) + REDIRECT *temp; + int error; +{ + char *filename, *allocname; + int oflags; + + allocname = 0; + if (temp->rflags & REDIR_VARASSIGN) + filename = allocname = savestring (temp->redirector.filename->word); + else if (temp->redirector.dest < 0) + /* This can happen when read_token_word encounters overflow, like in + exec 4294967297>x */ + filename = _("file descriptor out of range"); +#ifdef EBADF + /* This error can never involve NOCLOBBER */ + else if (error != NOCLOBBER_REDIRECT && temp->redirector.dest >= 0 && error == EBADF) + { + /* If we're dealing with two file descriptors, we have to guess about + which one is invalid; in the cases of r_{duplicating,move}_input and + r_{duplicating,move}_output we're here because dup2() failed. */ + switch (temp->instruction) + { + case r_duplicating_input: + case r_duplicating_output: + case r_move_input: + case r_move_output: + filename = allocname = itos (temp->redirectee.dest); + break; + case r_duplicating_input_word: + if (temp->redirector.dest == 0) /* Guess */ + filename = temp->redirectee.filename->word; /* XXX */ + else + filename = allocname = itos (temp->redirector.dest); + break; + case r_duplicating_output_word: + if (temp->redirector.dest == 1) /* Guess */ + filename = temp->redirectee.filename->word; /* XXX */ + else + filename = allocname = itos (temp->redirector.dest); + break; + default: + filename = allocname = itos (temp->redirector.dest); + break; + } + } +#endif + else if (expandable_redirection_filename (temp)) + { +expandable_filename: + if (posixly_correct && interactive_shell == 0) + { + oflags = temp->redirectee.filename->flags; + temp->redirectee.filename->flags |= W_NOGLOB; + } + filename = allocname = redirection_expand (temp->redirectee.filename); + if (posixly_correct && interactive_shell == 0) + temp->redirectee.filename->flags = oflags; + if (filename == 0) + filename = temp->redirectee.filename->word; + } + else if (temp->redirectee.dest < 0) + filename = "file descriptor out of range"; + else + filename = allocname = itos (temp->redirectee.dest); + + switch (error) + { + case AMBIGUOUS_REDIRECT: + internal_error (_("%s: ambiguous redirect"), filename); + break; + + case NOCLOBBER_REDIRECT: + internal_error (_("%s: cannot overwrite existing file"), filename); + break; + +#if defined (RESTRICTED_SHELL) + case RESTRICTED_REDIRECT: + internal_error (_("%s: restricted: cannot redirect output"), filename); + break; +#endif /* RESTRICTED_SHELL */ + + case HEREDOC_REDIRECT: + internal_error (_("cannot create temp file for here-document: %s"), strerror (heredoc_errno)); + break; + + case BADVAR_REDIRECT: + internal_error (_("%s: cannot assign fd to variable"), filename); + break; + + default: + internal_error ("%s: %s", filename, strerror (error)); + break; + } + + FREE (allocname); +} + +/* Perform the redirections on LIST. If flags & RX_ACTIVE, then actually + make input and output file descriptors, otherwise just do whatever is + neccessary for side effecting. flags & RX_UNDOABLE says to remember + how to undo the redirections later, if non-zero. If flags & RX_CLEXEC + is non-zero, file descriptors opened in do_redirection () have their + close-on-exec flag set. */ +int +do_redirections (list, flags) + REDIRECT *list; + int flags; +{ + int error; + REDIRECT *temp; + + if (flags & RX_UNDOABLE) + { + if (redirection_undo_list) + { + dispose_redirects (redirection_undo_list); + redirection_undo_list = (REDIRECT *)NULL; + } + if (exec_redirection_undo_list) + dispose_exec_redirects (); + } + + for (temp = list; temp; temp = temp->next) + { + error = do_redirection_internal (temp, flags); + if (error) + { + redirection_error (temp, error); + return (error); + } + } + return (0); +} + +/* Return non-zero if the redirection pointed to by REDIRECT has a + redirectee.filename that can be expanded. */ +static int +expandable_redirection_filename (redirect) + REDIRECT *redirect; +{ + switch (redirect->instruction) + { + case r_output_direction: + case r_appending_to: + case r_input_direction: + case r_inputa_direction: + case r_err_and_out: + case r_append_err_and_out: + case r_input_output: + case r_output_force: + case r_duplicating_input_word: + case r_duplicating_output_word: + case r_move_input_word: + case r_move_output_word: + return 1; + + default: + return 0; + } +} + +/* Expand the word in WORD returning a string. If WORD expands to + multiple words (or no words), then return NULL. */ +char * +redirection_expand (word) + WORD_DESC *word; +{ + char *result; + WORD_LIST *tlist1, *tlist2; + WORD_DESC *w; + + w = copy_word (word); + if (posixly_correct) + w->flags |= W_NOSPLIT; + + tlist1 = make_word_list (w, (WORD_LIST *)NULL); + expanding_redir = 1; + tlist2 = expand_words_no_vars (tlist1); + expanding_redir = 0; + dispose_words (tlist1); + + if (!tlist2 || tlist2->next) + { + /* We expanded to no words, or to more than a single word. + Dispose of the word list and return NULL. */ + if (tlist2) + dispose_words (tlist2); + return ((char *)NULL); + } + result = string_list (tlist2); /* XXX savestring (tlist2->word->word)? */ + dispose_words (tlist2); + return (result); +} + +static int +write_here_string (fd, redirectee) + int fd; + WORD_DESC *redirectee; +{ + char *herestr; + int herelen, n, e; + + expanding_redir = 1; + herestr = expand_string_to_string (redirectee->word, 0); + expanding_redir = 0; + herelen = STRLEN (herestr); + + n = write (fd, herestr, herelen); + if (n == herelen) + { + n = write (fd, "\n", 1); + herelen = 1; + } + e = errno; + FREE (herestr); + if (n != herelen) + { + if (e == 0) + e = ENOSPC; + return e; + } + return 0; +} + +/* Write the text of the here document pointed to by REDIRECTEE to the file + descriptor FD, which is already open to a temp file. Return 0 if the + write is successful, otherwise return errno. */ +static int +write_here_document (fd, redirectee) + int fd; + WORD_DESC *redirectee; +{ + char *document; + int document_len, fd2; + FILE *fp; + register WORD_LIST *t, *tlist; + + /* Expand the text if the word that was specified had + no quoting. The text that we expand is treated + exactly as if it were surrounded by double quotes. */ + + if (redirectee->flags & W_QUOTED) + { + document = redirectee->word; + document_len = strlen (document); + /* Set errno to something reasonable if the write fails. */ + if (write (fd, document, document_len) < document_len) + { + if (errno == 0) + errno = ENOSPC; + return (errno); + } + else + return 0; + } + + expanding_redir = 1; + tlist = expand_string (redirectee->word, Q_HERE_DOCUMENT); + expanding_redir = 0; + + if (tlist) + { + /* Try using buffered I/O (stdio) and writing a word + at a time, letting stdio do the work of buffering + for us rather than managing our own strings. Most + stdios are not particularly fast, however -- this + may need to be reconsidered later. */ + if ((fd2 = dup (fd)) < 0 || (fp = fdopen (fd2, "w")) == NULL) + { + if (fd2 >= 0) + close (fd2); + return (errno); + } + errno = 0; + for (t = tlist; t; t = t->next) + { + /* This is essentially the body of + string_list_internal expanded inline. */ + document = t->word->word; + document_len = strlen (document); + if (t != tlist) + putc (' ', fp); /* separator */ + fwrite (document, document_len, 1, fp); + if (ferror (fp)) + { + if (errno == 0) + errno = ENOSPC; + fd2 = errno; + fclose(fp); + dispose_words (tlist); + return (fd2); + } + } + dispose_words (tlist); + if (fclose (fp) != 0) + { + if (errno == 0) + errno = ENOSPC; + return (errno); + } + } + return 0; +} + +/* Create a temporary file holding the text of the here document pointed to + by REDIRECTEE, and return a file descriptor open for reading to the temp + file. Return -1 on any error, and make sure errno is set appropriately. */ +static int +here_document_to_fd (redirectee, ri) + WORD_DESC *redirectee; + enum r_instruction ri; +{ + char *filename; + int r, fd, fd2; + + fd = sh_mktmpfd ("sh-thd", MT_USERANDOM|MT_USETMPDIR, &filename); + + /* If we failed for some reason other than the file existing, abort */ + if (fd < 0) + { + FREE (filename); + return (fd); + } + + errno = r = 0; /* XXX */ + /* write_here_document returns 0 on success, errno on failure. */ + if (redirectee->word) + r = (ri != r_reading_string) ? write_here_document (fd, redirectee) + : write_here_string (fd, redirectee); + + if (r) + { + close (fd); + unlink (filename); + free (filename); + errno = r; + return (-1); + } + + /* In an attempt to avoid races, we close the first fd only after opening + the second. */ + /* Make the document really temporary. Also make it the input. */ + fd2 = open (filename, O_RDONLY|O_BINARY, 0600); + + if (fd2 < 0) + { + r = errno; + unlink (filename); + free (filename); + close (fd); + errno = r; + return -1; + } + + close (fd); + if (unlink (filename) < 0) + { + r = errno; + close (fd2); + free (filename); + errno = r; + return (-1); + } + + free (filename); + return (fd2); +} + +#define RF_DEVFD 1 +#define RF_DEVSTDERR 2 +#define RF_DEVSTDIN 3 +#define RF_DEVSTDOUT 4 +#define RF_DEVTCP 5 +#define RF_DEVUDP 6 + +/* A list of pattern/value pairs for filenames that the redirection + code handles specially. */ +static STRING_INT_ALIST _redir_special_filenames[] = { +#if !defined (HAVE_DEV_FD) + { "/dev/fd/[0-9]*", RF_DEVFD }, +#endif +#if !defined (HAVE_DEV_STDIN) + { "/dev/stderr", RF_DEVSTDERR }, + { "/dev/stdin", RF_DEVSTDIN }, + { "/dev/stdout", RF_DEVSTDOUT }, +#endif +#if defined (NETWORK_REDIRECTIONS) + { "/dev/tcp/*/*", RF_DEVTCP }, + { "/dev/udp/*/*", RF_DEVUDP }, +#endif + { (char *)NULL, -1 } +}; + +static int +redir_special_open (spec, filename, flags, mode, ri) + int spec; + char *filename; + int flags, mode; + enum r_instruction ri; +{ + int fd; +#if !defined (HAVE_DEV_FD) + intmax_t lfd; +#endif + + fd = -1; + switch (spec) + { +#if !defined (HAVE_DEV_FD) + case RF_DEVFD: + if (all_digits (filename+8) && legal_number (filename+8, &lfd) && lfd == (int)lfd) + { + fd = lfd; + fd = fcntl (fd, F_DUPFD, SHELL_FD_BASE); + } + else + fd = AMBIGUOUS_REDIRECT; + break; +#endif + +#if !defined (HAVE_DEV_STDIN) + case RF_DEVSTDIN: + fd = fcntl (0, F_DUPFD, SHELL_FD_BASE); + break; + case RF_DEVSTDOUT: + fd = fcntl (1, F_DUPFD, SHELL_FD_BASE); + break; + case RF_DEVSTDERR: + fd = fcntl (2, F_DUPFD, SHELL_FD_BASE); + break; +#endif + +#if defined (NETWORK_REDIRECTIONS) + case RF_DEVTCP: + case RF_DEVUDP: +#if defined (HAVE_NETWORK) + fd = netopen (filename); +#else + internal_warning (_("/dev/(tcp|udp)/host/port not supported without networking")); + fd = open (filename, flags, mode); +#endif + break; +#endif /* NETWORK_REDIRECTIONS */ + } + + return fd; +} + +/* Open FILENAME with FLAGS in noclobber mode, hopefully avoiding most + race conditions and avoiding the problem where the file is replaced + between the stat(2) and open(2). */ +static int +noclobber_open (filename, flags, mode, ri) + char *filename; + int flags, mode; + enum r_instruction ri; +{ + int r, fd; + struct stat finfo, finfo2; + + /* If the file exists and is a regular file, return an error + immediately. */ + r = stat (filename, &finfo); + if (r == 0 && (S_ISREG (finfo.st_mode))) + return (NOCLOBBER_REDIRECT); + + /* If the file was not present (r != 0), make sure we open it + exclusively so that if it is created before we open it, our open + will fail. Make sure that we do not truncate an existing file. + Note that we don't turn on O_EXCL unless the stat failed -- if + the file was not a regular file, we leave O_EXCL off. */ + flags &= ~O_TRUNC; + if (r != 0) + { + fd = open (filename, flags|O_EXCL, mode); + return ((fd < 0 && errno == EEXIST) ? NOCLOBBER_REDIRECT : fd); + } + fd = open (filename, flags, mode); + + /* If the open failed, return the file descriptor right away. */ + if (fd < 0) + return (errno == EEXIST ? NOCLOBBER_REDIRECT : fd); + + /* OK, the open succeeded, but the file may have been changed from a + non-regular file to a regular file between the stat and the open. + We are assuming that the O_EXCL open handles the case where FILENAME + did not exist and is symlinked to an existing file between the stat + and open. */ + + /* If we can open it and fstat the file descriptor, and neither check + revealed that it was a regular file, and the file has not been replaced, + return the file descriptor. */ + if ((fstat (fd, &finfo2) == 0) && (S_ISREG (finfo2.st_mode) == 0) && + r == 0 && (S_ISREG (finfo.st_mode) == 0) && + same_file (filename, filename, &finfo, &finfo2)) + return fd; + + /* The file has been replaced. badness. */ + close (fd); + errno = EEXIST; + return (NOCLOBBER_REDIRECT); +} + +static int +redir_open (filename, flags, mode, ri) + char *filename; + int flags, mode; + enum r_instruction ri; +{ + int fd, r; + + r = find_string_in_alist (filename, _redir_special_filenames, 1); + if (r >= 0) + return (redir_special_open (r, filename, flags, mode, ri)); + + /* If we are in noclobber mode, you are not allowed to overwrite + existing files. Check before opening. */ + if (noclobber && CLOBBERING_REDIRECT (ri)) + { + fd = noclobber_open (filename, flags, mode, ri); + if (fd == NOCLOBBER_REDIRECT) + return (NOCLOBBER_REDIRECT); + } + else + { + fd = open (filename, flags, mode); +#if defined (AFS) + if ((fd < 0) && (errno == EACCES)) + { + fd = open (filename, flags & ~O_CREAT, mode); + errno = EACCES; /* restore errno */ + } +#endif /* AFS */ + } + + return fd; +} + +static int +undoablefd (fd) + int fd; +{ + int clexec; + + clexec = fcntl (fd, F_GETFD, 0); + if (clexec == -1 || (fd >= SHELL_FD_BASE && clexec == 1)) + return 0; + return 1; +} + +/* Do the specific redirection requested. Returns errno or one of the + special redirection errors (*_REDIRECT) in case of error, 0 on success. + If flags & RX_ACTIVE is zero, then just do whatever is neccessary to + produce the appropriate side effects. flags & RX_UNDOABLE, if non-zero, + says to remember how to undo each redirection. If flags & RX_CLEXEC is + non-zero, then we set all file descriptors > 2 that we open to be + close-on-exec. */ +static int +do_redirection_internal (redirect, flags) + REDIRECT *redirect; + int flags; +{ + WORD_DESC *redirectee; + int redir_fd, fd, redirector, r, oflags; + intmax_t lfd; + char *redirectee_word; + enum r_instruction ri; + REDIRECT *new_redirect; + REDIRECTEE sd; + + redirectee = redirect->redirectee.filename; + redir_fd = redirect->redirectee.dest; + redirector = redirect->redirector.dest; + ri = redirect->instruction; + + if (redirect->flags & RX_INTERNAL) + flags |= RX_INTERNAL; + + if (TRANSLATE_REDIRECT (ri)) + { + /* We have [N]>&WORD[-] or [N]<&WORD[-] (or {V}>&WORD[-] or {V}<&WORD-). + and WORD, then translate the redirection into a new one and + continue. */ + redirectee_word = redirection_expand (redirectee); + + /* XXX - what to do with [N]<&$w- where w is unset or null? ksh93 + closes N. */ + if (redirectee_word == 0) + return (AMBIGUOUS_REDIRECT); + else if (redirectee_word[0] == '-' && redirectee_word[1] == '\0') + { + sd = redirect->redirector; + rd.dest = 0; + new_redirect = make_redirection (sd, r_close_this, rd, 0); + } + else if (all_digits (redirectee_word)) + { + sd = redirect->redirector; + if (legal_number (redirectee_word, &lfd) && (int)lfd == lfd) + rd.dest = lfd; + else + rd.dest = -1; /* XXX */ + switch (ri) + { + case r_duplicating_input_word: + new_redirect = make_redirection (sd, r_duplicating_input, rd, 0); + break; + case r_duplicating_output_word: + new_redirect = make_redirection (sd, r_duplicating_output, rd, 0); + break; + case r_move_input_word: + new_redirect = make_redirection (sd, r_move_input, rd, 0); + break; + case r_move_output_word: + new_redirect = make_redirection (sd, r_move_output, rd, 0); + break; + } + } + else if (ri == r_duplicating_output_word && (redirect->rflags & REDIR_VARASSIGN) == 0 && redirector == 1) + { + sd = redirect->redirector; + rd.filename = make_bare_word (redirectee_word); + new_redirect = make_redirection (sd, r_err_and_out, rd, 0); + } + else + { + free (redirectee_word); + return (AMBIGUOUS_REDIRECT); + } + + free (redirectee_word); + + /* Set up the variables needed by the rest of the function from the + new redirection. */ + if (new_redirect->instruction == r_err_and_out) + { + char *alloca_hack; + + /* Copy the word without allocating any memory that must be + explicitly freed. */ + redirectee = (WORD_DESC *)alloca (sizeof (WORD_DESC)); + xbcopy ((char *)new_redirect->redirectee.filename, + (char *)redirectee, sizeof (WORD_DESC)); + + alloca_hack = (char *) + alloca (1 + strlen (new_redirect->redirectee.filename->word)); + redirectee->word = alloca_hack; + strcpy (redirectee->word, new_redirect->redirectee.filename->word); + } + else + /* It's guaranteed to be an integer, and shouldn't be freed. */ + redirectee = new_redirect->redirectee.filename; + + redir_fd = new_redirect->redirectee.dest; + redirector = new_redirect->redirector.dest; + ri = new_redirect->instruction; + + /* Overwrite the flags element of the old redirect with the new value. */ + redirect->flags = new_redirect->flags; + dispose_redirects (new_redirect); + } + + switch (ri) + { + case r_output_direction: + case r_appending_to: + case r_input_direction: + case r_inputa_direction: + case r_err_and_out: /* command &>filename */ + case r_append_err_and_out: /* command &>> filename */ + case r_input_output: + case r_output_force: + if (posixly_correct && interactive_shell == 0) + { + oflags = redirectee->flags; + redirectee->flags |= W_NOGLOB; + } + redirectee_word = redirection_expand (redirectee); + if (posixly_correct && interactive_shell == 0) + redirectee->flags = oflags; + + if (redirectee_word == 0) + return (AMBIGUOUS_REDIRECT); + +#if defined (RESTRICTED_SHELL) + if (restricted && (WRITE_REDIRECT (ri))) + { + free (redirectee_word); + return (RESTRICTED_REDIRECT); + } +#endif /* RESTRICTED_SHELL */ + + fd = redir_open (redirectee_word, redirect->flags, 0666, ri); + free (redirectee_word); + + if (fd == NOCLOBBER_REDIRECT) + return (fd); + + if (fd < 0) + return (errno); + + if (flags & RX_ACTIVE) + { + if (redirect->rflags & REDIR_VARASSIGN) + { + redirector = fcntl (fd, F_DUPFD, SHELL_FD_BASE); /* XXX try this for now */ + r = errno; + if (redirector < 0) + sys_error (_("redirection error: cannot duplicate fd")); + REDIRECTION_ERROR (redirector, r, fd); + } + + if (flags & RX_UNDOABLE) + { + /* Only setup to undo it if the thing to undo is active. */ + if ((fd != redirector) && (fcntl (redirector, F_GETFD, 0) != -1)) + r = add_undo_redirect (redirector, ri, -1); + else + r = add_undo_close_redirect (redirector); + if (r < 0 && (redirect->rflags & REDIR_VARASSIGN)) + close (redirector); + REDIRECTION_ERROR (r, errno, fd); + } + +#if defined (BUFFERED_INPUT) + check_bash_input (redirector); +#endif + + /* Make sure there is no pending output before we change the state + of the underlying file descriptor, since the builtins use stdio + for output. */ + if (redirector == 1 && fileno (stdout) == redirector) + { + fflush (stdout); + fpurge (stdout); + } + else if (redirector == 2 && fileno (stderr) == redirector) + { + fflush (stderr); + fpurge (stderr); + } + + if (redirect->rflags & REDIR_VARASSIGN) + { + if ((r = redir_varassign (redirect, redirector)) < 0) + { + close (redirector); + close (fd); + return (r); /* XXX */ + } + } + else if ((fd != redirector) && (dup2 (fd, redirector) < 0)) + return (errno); + +#if defined (BUFFERED_INPUT) + /* Do not change the buffered stream for an implicit redirection + of /dev/null to fd 0 for asynchronous commands without job + control (r_inputa_direction). */ + if (ri == r_input_direction || ri == r_input_output) + duplicate_buffered_stream (fd, redirector); +#endif /* BUFFERED_INPUT */ + + /* + * If we're remembering, then this is the result of a while, for + * or until loop with a loop redirection, or a function/builtin + * executing in the parent shell with a redirection. In the + * function/builtin case, we want to set all file descriptors > 2 + * to be close-on-exec to duplicate the effect of the old + * for i = 3 to NOFILE close(i) loop. In the case of the loops, + * both sh and ksh leave the file descriptors open across execs. + * The Posix standard mentions only the exec builtin. + */ + if ((flags & RX_CLEXEC) && (redirector > 2)) + SET_CLOSE_ON_EXEC (redirector); + } + + if (fd != redirector) + { +#if defined (BUFFERED_INPUT) + if (INPUT_REDIRECT (ri)) + close_buffered_fd (fd); + else +#endif /* !BUFFERED_INPUT */ + close (fd); /* Don't close what we just opened! */ + } + + /* If we are hacking both stdout and stderr, do the stderr + redirection here. XXX - handle {var} here? */ + if (ri == r_err_and_out || ri == r_append_err_and_out) + { + if (flags & RX_ACTIVE) + { + if (flags & RX_UNDOABLE) + add_undo_redirect (2, ri, -1); + if (dup2 (1, 2) < 0) + return (errno); + } + } + break; + + case r_reading_until: + case r_deblank_reading_until: + case r_reading_string: + /* REDIRECTEE is a pointer to a WORD_DESC containing the text of + the new input. Place it in a temporary file. */ + if (redirectee) + { + fd = here_document_to_fd (redirectee, ri); + + if (fd < 0) + { + heredoc_errno = errno; + return (HEREDOC_REDIRECT); + } + + if (redirect->rflags & REDIR_VARASSIGN) + { + redirector = fcntl (fd, F_DUPFD, SHELL_FD_BASE); /* XXX try this for now */ + r = errno; + if (redirector < 0) + sys_error (_("redirection error: cannot duplicate fd")); + REDIRECTION_ERROR (redirector, r, fd); + } + + if (flags & RX_ACTIVE) + { + if (flags & RX_UNDOABLE) + { + /* Only setup to undo it if the thing to undo is active. */ + if ((fd != redirector) && (fcntl (redirector, F_GETFD, 0) != -1)) + r = add_undo_redirect (redirector, ri, -1); + else + r = add_undo_close_redirect (redirector); + if (r < 0 && (redirect->rflags & REDIR_VARASSIGN)) + close (redirector); + REDIRECTION_ERROR (r, errno, fd); + } + +#if defined (BUFFERED_INPUT) + check_bash_input (redirector); +#endif + if (redirect->rflags & REDIR_VARASSIGN) + { + if ((r = redir_varassign (redirect, redirector)) < 0) + { + close (redirector); + close (fd); + return (r); /* XXX */ + } + } + else if (fd != redirector && dup2 (fd, redirector) < 0) + { + r = errno; + close (fd); + return (r); + } + +#if defined (BUFFERED_INPUT) + duplicate_buffered_stream (fd, redirector); +#endif + + if ((flags & RX_CLEXEC) && (redirector > 2)) + SET_CLOSE_ON_EXEC (redirector); + } + + if (fd != redirector) +#if defined (BUFFERED_INPUT) + close_buffered_fd (fd); +#else + close (fd); +#endif + } + break; + + case r_duplicating_input: + case r_duplicating_output: + case r_move_input: + case r_move_output: + if ((flags & RX_ACTIVE) && (redirect->rflags & REDIR_VARASSIGN)) + { + redirector = fcntl (redir_fd, F_DUPFD, SHELL_FD_BASE); /* XXX try this for now */ + r = errno; + if (redirector < 0) + sys_error (_("redirection error: cannot duplicate fd")); + REDIRECTION_ERROR (redirector, r, -1); + } + + if ((flags & RX_ACTIVE) && (redir_fd != redirector)) + { + if (flags & RX_UNDOABLE) + { + /* Only setup to undo it if the thing to undo is active. */ + if (fcntl (redirector, F_GETFD, 0) != -1) + r = add_undo_redirect (redirector, ri, redir_fd); + else + r = add_undo_close_redirect (redirector); + if (r < 0 && (redirect->rflags & REDIR_VARASSIGN)) + close (redirector); + REDIRECTION_ERROR (r, errno, -1); + } +#if defined (BUFFERED_INPUT) + check_bash_input (redirector); +#endif + if (redirect->rflags & REDIR_VARASSIGN) + { + if ((r = redir_varassign (redirect, redirector)) < 0) + { + close (redirector); + return (r); /* XXX */ + } + } + /* This is correct. 2>&1 means dup2 (1, 2); */ + else if (dup2 (redir_fd, redirector) < 0) + return (errno); + +#if defined (BUFFERED_INPUT) + if (ri == r_duplicating_input || ri == r_move_input) + duplicate_buffered_stream (redir_fd, redirector); +#endif /* BUFFERED_INPUT */ + + /* First duplicate the close-on-exec state of redirectee. dup2 + leaves the flag unset on the new descriptor, which means it + stays open. Only set the close-on-exec bit for file descriptors + greater than 2 in any case, since 0-2 should always be open + unless closed by something like `exec 2<&-'. It should always + be safe to set fds > 2 to close-on-exec if they're being used to + save file descriptors < 2, since we don't need to preserve the + state of the close-on-exec flag for those fds -- they should + always be open. */ + /* if ((already_set || set_unconditionally) && (ok_to_set)) + set_it () */ +#if 0 + if (((fcntl (redir_fd, F_GETFD, 0) == 1) || redir_fd < 2 || (flags & RX_CLEXEC)) && + (redirector > 2)) +#else + if (((fcntl (redir_fd, F_GETFD, 0) == 1) || (redir_fd < 2 && (flags & RX_INTERNAL)) || (flags & RX_CLEXEC)) && + (redirector > 2)) +#endif + SET_CLOSE_ON_EXEC (redirector); + + /* When undoing saving of non-standard file descriptors (>=3) using + file descriptors >= SHELL_FD_BASE, we set the saving fd to be + close-on-exec and use a flag to decide how to set close-on-exec + when the fd is restored. */ + if ((redirect->flags & RX_INTERNAL) && (redirect->flags & RX_SAVCLEXEC) && redirector >= 3 && (redir_fd >= SHELL_FD_BASE || (redirect->flags & RX_SAVEFD))) + SET_OPEN_ON_EXEC (redirector); + + /* dup-and-close redirection */ + if (ri == r_move_input || ri == r_move_output) + { + xtrace_fdchk (redir_fd); + + close (redir_fd); +#if defined (COPROCESS_SUPPORT) + coproc_fdchk (redir_fd); /* XXX - loses coproc fds */ +#endif + } + } + break; + + case r_close_this: + if (flags & RX_ACTIVE) + { + if (redirect->rflags & REDIR_VARASSIGN) + { + redirector = redir_varvalue (redirect); + if (redirector < 0) + return AMBIGUOUS_REDIRECT; + } + + r = 0; + if ((flags & RX_UNDOABLE) && (fcntl (redirector, F_GETFD, 0) != -1)) + { + r = add_undo_redirect (redirector, ri, -1); + REDIRECTION_ERROR (r, errno, redirector); + } + +#if defined (COPROCESS_SUPPORT) + coproc_fdchk (redirector); +#endif + xtrace_fdchk (redirector); + +#if defined (BUFFERED_INPUT) + check_bash_input (redirector); + close_buffered_fd (redirector); +#else /* !BUFFERED_INPUT */ + close (redirector); +#endif /* !BUFFERED_INPUT */ + } + break; + + case r_duplicating_input_word: + case r_duplicating_output_word: + break; + } + return (0); +} + +/* Remember the file descriptor associated with the slot FD, + on REDIRECTION_UNDO_LIST. Note that the list will be reversed + before it is executed. Any redirections that need to be undone + even if REDIRECTION_UNDO_LIST is discarded by the exec builtin + are also saved on EXEC_REDIRECTION_UNDO_LIST. FDBASE says where to + start the duplicating. If it's less than SHELL_FD_BASE, we're ok, + and can use SHELL_FD_BASE (-1 == don't care). If it's >= SHELL_FD_BASE, + we have to make sure we don't use fdbase to save a file descriptor, + since we're going to use it later (e.g., make sure we don't save fd 0 + to fd 10 if we have a redirection like 0<&10). If the value of fdbase + puts the process over its fd limit, causing fcntl to fail, we try + again with SHELL_FD_BASE. Return 0 on success, -1 on error. */ +static int +add_undo_redirect (fd, ri, fdbase) + int fd; + enum r_instruction ri; + int fdbase; +{ + int new_fd, clexec_flag, savefd_flag; + REDIRECT *new_redirect, *closer, *dummy_redirect; + REDIRECTEE sd; + + savefd_flag = 0; + new_fd = fcntl (fd, F_DUPFD, (fdbase < SHELL_FD_BASE) ? SHELL_FD_BASE : fdbase+1); + if (new_fd < 0) + new_fd = fcntl (fd, F_DUPFD, SHELL_FD_BASE); + if (new_fd < 0) + { + new_fd = fcntl (fd, F_DUPFD, 0); + savefd_flag = 1; + } + + if (new_fd < 0) + { + sys_error (_("redirection error: cannot duplicate fd")); + return (-1); + } + + clexec_flag = fcntl (fd, F_GETFD, 0); + + sd.dest = new_fd; + rd.dest = 0; + closer = make_redirection (sd, r_close_this, rd, 0); + closer->flags |= RX_INTERNAL; + dummy_redirect = copy_redirects (closer); + + sd.dest = fd; + rd.dest = new_fd; + if (fd == 0) + new_redirect = make_redirection (sd, r_duplicating_input, rd, 0); + else + new_redirect = make_redirection (sd, r_duplicating_output, rd, 0); + new_redirect->flags |= RX_INTERNAL; + if (savefd_flag) + new_redirect->flags |= RX_SAVEFD; + if (clexec_flag == 0 && fd >= 3 && (new_fd >= SHELL_FD_BASE || savefd_flag)) + new_redirect->flags |= RX_SAVCLEXEC; + new_redirect->next = closer; + + closer->next = redirection_undo_list; + redirection_undo_list = new_redirect; + + /* Save redirections that need to be undone even if the undo list + is thrown away by the `exec' builtin. */ + add_exec_redirect (dummy_redirect); + + /* experimental: if we're saving a redirection to undo for a file descriptor + above SHELL_FD_BASE, add a redirection to be undone if the exec builtin + causes redirections to be discarded. There needs to be a difference + between fds that are used to save other fds and then are the target of + user redirections and fds that are just the target of user redirections. + We use the close-on-exec flag to tell the difference; fds > SHELL_FD_BASE + that have the close-on-exec flag set are assumed to be fds used internally + to save others. */ + if (fd >= SHELL_FD_BASE && ri != r_close_this && clexec_flag) + { + sd.dest = fd; + rd.dest = new_fd; + new_redirect = make_redirection (sd, r_duplicating_output, rd, 0); + new_redirect->flags |= RX_INTERNAL; + + add_exec_redirect (new_redirect); + } + + /* File descriptors used only for saving others should always be + marked close-on-exec. Unfortunately, we have to preserve the + close-on-exec state of the file descriptor we are saving, since + fcntl (F_DUPFD) sets the new file descriptor to remain open + across execs. If, however, the file descriptor whose state we + are saving is <= 2, we can just set the close-on-exec flag, + because file descriptors 0-2 should always be open-on-exec, + and the restore above in do_redirection() will take care of it. */ + if (clexec_flag || fd < 3) + SET_CLOSE_ON_EXEC (new_fd); + else if (redirection_undo_list->flags & RX_SAVCLEXEC) + SET_CLOSE_ON_EXEC (new_fd); + + return (0); +} + +/* Set up to close FD when we are finished with the current command + and its redirections. Return 0 on success, -1 on error. */ +static int +add_undo_close_redirect (fd) + int fd; +{ + REDIRECT *closer; + REDIRECTEE sd; + + sd.dest = fd; + rd.dest = 0; + closer = make_redirection (sd, r_close_this, rd, 0); + closer->flags |= RX_INTERNAL; + closer->next = redirection_undo_list; + redirection_undo_list = closer; + + return 0; +} + +static void +add_exec_redirect (dummy_redirect) + REDIRECT *dummy_redirect; +{ + dummy_redirect->next = exec_redirection_undo_list; + exec_redirection_undo_list = dummy_redirect; +} + +/* Return 1 if the redirection specified by RI and REDIRECTOR alters the + standard input. */ +static int +stdin_redirection (ri, redirector) + enum r_instruction ri; + int redirector; +{ + switch (ri) + { + case r_input_direction: + case r_inputa_direction: + case r_input_output: + case r_reading_until: + case r_deblank_reading_until: + case r_reading_string: + return (1); + case r_duplicating_input: + case r_duplicating_input_word: + case r_close_this: + return (redirector == 0); + case r_output_direction: + case r_appending_to: + case r_duplicating_output: + case r_err_and_out: + case r_append_err_and_out: + case r_output_force: + case r_duplicating_output_word: + return (0); + } + return (0); +} + +/* Return non-zero if any of the redirections in REDIRS alter the standard + input. */ +int +stdin_redirects (redirs) + REDIRECT *redirs; +{ + REDIRECT *rp; + int n; + + for (n = 0, rp = redirs; rp; rp = rp->next) + if ((rp->rflags & REDIR_VARASSIGN) == 0) + n += stdin_redirection (rp->instruction, rp->redirector.dest); + return n; +} + +/* These don't yet handle array references */ +static int +redir_varassign (redir, fd) + REDIRECT *redir; + int fd; +{ + WORD_DESC *w; + SHELL_VAR *v; + + w = redir->redirector.filename; + v = bind_var_to_int (w->word, fd); + if (v == 0 || readonly_p (v) || noassign_p (v)) + return BADVAR_REDIRECT; + + return 0; +} + +static int +redir_varvalue (redir) + REDIRECT *redir; +{ + SHELL_VAR *v; + char *val; + intmax_t vmax; + int i; + + /* XXX - handle set -u here? */ + v = find_variable (redir->redirector.filename->word); + if (v == 0 || invisible_p (v)) + return -1; + + val = get_variable_value (v); + if (val == 0 || *val == 0) + return -1; + + if (legal_number (val, &vmax) < 0) + return -1; + + i = vmax; /* integer truncation */ + return i; +} diff --git a/subst.c b/subst.c index f9c23ab22..ce6df0ab5 100644 --- a/subst.c +++ b/subst.c @@ -4,7 +4,7 @@ /* ``Have a little faith, there's magic in the night. You ain't a beauty, but, hey, you're alright.'' */ -/* Copyright (C) 1987-2010 Free Software Foundation, Inc. +/* Copyright (C) 1987-2012 Free Software Foundation, Inc. This file is part of GNU Bash, the Bourne Again SHell. diff --git a/subst.c~ b/subst.c~ new file mode 100644 index 000000000..c6d673797 --- /dev/null +++ b/subst.c~ @@ -0,0 +1,9417 @@ +/* subst.c -- The part of the shell that does parameter, command, arithmetic, + and globbing substitutions. */ + +/* ``Have a little faith, there's magic in the night. You ain't a + beauty, but, hey, you're alright.'' */ + +/* Copyright (C) 1987-2012 Free Software Foundation, Inc. + + This file is part of GNU Bash, the Bourne Again SHell. + + Bash is free software: you can redistribute it and/or modify + it under the terms of the GNU General Public License as published by + the Free Software Foundation, either version 3 of the License, or + (at your option) any later version. + + Bash is distributed in the hope that it will be useful, + but WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + GNU General Public License for more details. + + You should have received a copy of the GNU General Public License + along with Bash. If not, see . +*/ + +#include "config.h" + +#include "bashtypes.h" +#include +#include "chartypes.h" +#if defined (HAVE_PWD_H) +# include +#endif +#include +#include + +#if defined (HAVE_UNISTD_H) +# include +#endif + +#include "bashansi.h" +#include "posixstat.h" +#include "bashintl.h" + +#include "shell.h" +#include "parser.h" +#include "flags.h" +#include "jobs.h" +#include "execute_cmd.h" +#include "filecntl.h" +#include "trap.h" +#include "pathexp.h" +#include "mailcheck.h" + +#include "shmbutil.h" +#include "typemax.h" + +#include "builtins/getopt.h" +#include "builtins/common.h" + +#include "builtins/builtext.h" + +#include +#include + +#if !defined (errno) +extern int errno; +#endif /* !errno */ + +/* The size that strings change by. */ +#define DEFAULT_INITIAL_ARRAY_SIZE 112 +#define DEFAULT_ARRAY_SIZE 128 + +/* Variable types. */ +#define VT_VARIABLE 0 +#define VT_POSPARMS 1 +#define VT_ARRAYVAR 2 +#define VT_ARRAYMEMBER 3 +#define VT_ASSOCVAR 4 + +#define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */ + +/* Flags for quoted_strchr */ +#define ST_BACKSL 0x01 +#define ST_CTLESC 0x02 +#define ST_SQUOTE 0x04 /* unused yet */ +#define ST_DQUOTE 0x08 /* unused yet */ + +/* Flags for the `pflags' argument to param_expand() */ +#define PF_NOCOMSUB 0x01 /* Do not perform command substitution */ +#define PF_IGNUNBOUND 0x02 /* ignore unbound vars even if -u set */ +#define PF_NOSPLIT2 0x04 /* same as W_NOSPLIT2 */ +#define PF_ASSIGNRHS 0x08 /* same as W_ASSIGNRHS */ + +/* These defs make it easier to use the editor. */ +#define LBRACE '{' +#define RBRACE '}' +#define LPAREN '(' +#define RPAREN ')' + +#if defined (HANDLE_MULTIBYTE) +#define WLPAREN L'(' +#define WRPAREN L')' +#endif + +/* Evaluates to 1 if C is one of the shell's special parameters whose length + can be taken, but is also one of the special expansion characters. */ +#define VALID_SPECIAL_LENGTH_PARAM(c) \ + ((c) == '-' || (c) == '?' || (c) == '#') + +/* Evaluates to 1 if C is one of the shell's special parameters for which an + indirect variable reference may be made. */ +#define VALID_INDIR_PARAM(c) \ + ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*') + +/* Evaluates to 1 if C is one of the OP characters that follows the parameter + in ${parameter[:]OPword}. */ +#define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP) + +/* Evaluates to 1 if this is one of the shell's special variables. */ +#define SPECIAL_VAR(name, wi) \ + ((DIGIT (*name) && all_digits (name)) || \ + (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \ + (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1]))) + +/* An expansion function that takes a string and a quoted flag and returns + a WORD_LIST *. Used as the type of the third argument to + expand_string_if_necessary(). */ +typedef WORD_LIST *EXPFUNC __P((char *, int)); + +/* Process ID of the last command executed within command substitution. */ +pid_t last_command_subst_pid = NO_PID; +pid_t current_command_subst_pid = NO_PID; + +/* Variables used to keep track of the characters in IFS. */ +SHELL_VAR *ifs_var; +char *ifs_value; +unsigned char ifs_cmap[UCHAR_MAX + 1]; + +#if defined (HANDLE_MULTIBYTE) +unsigned char ifs_firstc[MB_LEN_MAX]; +size_t ifs_firstc_len; +#else +unsigned char ifs_firstc; +#endif + +/* Sentinel to tell when we are performing variable assignments preceding a + command name and putting them into the environment. Used to make sure + we use the temporary environment when looking up variable values. */ +int assigning_in_environment; + +/* Used to hold a list of variable assignments preceding a command. Global + so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a + SIGCHLD trap and so it can be saved and restored by the trap handlers. */ +WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL; + +/* Extern functions and variables from different files. */ +extern int last_command_exit_value, last_command_exit_signal; +extern int subshell_environment, line_number; +extern int subshell_level, parse_and_execute_level, sourcelevel; +extern int eof_encountered; +extern int return_catch_flag, return_catch_value; +extern pid_t dollar_dollar_pid; +extern int posixly_correct; +extern char *this_command_name; +extern struct fd_bitmap *current_fds_to_close; +extern int wordexp_only; +extern int expanding_redir; +extern int tempenv_assign_error; + +#if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE) +extern wchar_t *wcsdup __P((const wchar_t *)); +#endif + +/* Non-zero means to allow unmatched globbed filenames to expand to + a null file. */ +int allow_null_glob_expansion; + +/* Non-zero means to throw an error when globbing fails to match anything. */ +int fail_glob_expansion; + +#if 0 +/* Variables to keep track of which words in an expanded word list (the + output of expand_word_list_internal) are the result of globbing + expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c. + (CURRENTLY UNUSED). */ +char *glob_argv_flags; +static int glob_argv_flags_size; +#endif + +static WORD_LIST expand_word_error, expand_word_fatal; +static WORD_DESC expand_wdesc_error, expand_wdesc_fatal; +static char expand_param_error, expand_param_fatal; +static char extract_string_error, extract_string_fatal; + +/* Tell the expansion functions to not longjmp back to top_level on fatal + errors. Enabled when doing completion and prompt string expansion. */ +static int no_longjmp_on_fatal_error = 0; + +/* Set by expand_word_unsplit; used to inhibit splitting and re-joining + $* on $IFS, primarily when doing assignment statements. */ +static int expand_no_split_dollar_star = 0; + +/* A WORD_LIST of words to be expanded by expand_word_list_internal, + without any leading variable assignments. */ +static WORD_LIST *garglist = (WORD_LIST *)NULL; + +static char *quoted_substring __P((char *, int, int)); +static int quoted_strlen __P((char *)); +static char *quoted_strchr __P((char *, int, int)); + +static char *expand_string_if_necessary __P((char *, int, EXPFUNC *)); +static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *)); +static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *)); +static WORD_LIST *expand_string_internal __P((char *, int)); +static WORD_LIST *expand_string_leave_quoted __P((char *, int)); +static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *)); + +static WORD_LIST *list_quote_escapes __P((WORD_LIST *)); +static char *make_quoted_char __P((int)); +static WORD_LIST *quote_list __P((WORD_LIST *)); + +static int unquoted_substring __P((char *, char *)); +static int unquoted_member __P((int, char *)); + +#if defined (ARRAY_VARS) +static SHELL_VAR *do_compound_assignment __P((char *, char *, int)); +#endif +static int do_assignment_internal __P((const WORD_DESC *, int)); + +static char *string_extract_verbatim __P((char *, size_t, int *, char *, int)); +static char *string_extract __P((char *, int *, char *, int)); +static char *string_extract_double_quoted __P((char *, int *, int)); +static inline char *string_extract_single_quoted __P((char *, int *)); +static inline int skip_single_quoted __P((const char *, size_t, int)); +static int skip_double_quoted __P((char *, size_t, int)); +static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int)); +static char *extract_dollar_brace_string __P((char *, int *, int, int)); +static int skip_matched_pair __P((const char *, int, int, int, int)); + +static char *pos_params __P((char *, int, int, int)); + +static unsigned char *mb_getcharlens __P((char *, int)); + +static char *remove_upattern __P((char *, char *, int)); +#if defined (HANDLE_MULTIBYTE) +static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int)); +#endif +static char *remove_pattern __P((char *, char *, int)); + +static int match_upattern __P((char *, char *, int, char **, char **)); +#if defined (HANDLE_MULTIBYTE) +static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **)); +#endif +static int match_pattern __P((char *, char *, int, char **, char **)); +static int getpatspec __P((int, char *)); +static char *getpattern __P((char *, int, int)); +static char *variable_remove_pattern __P((char *, char *, int, int)); +static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int)); +static char *parameter_list_remove_pattern __P((int, char *, int, int)); +#ifdef ARRAY_VARS +static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int)); +#endif +static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int)); + +static char *process_substitute __P((char *, int)); + +static char *read_comsub __P((int, int, int *)); + +#ifdef ARRAY_VARS +static arrayind_t array_length_reference __P((char *)); +#endif + +static int valid_brace_expansion_word __P((char *, int)); +static int chk_atstar __P((char *, int, int *, int *)); +static int chk_arithsub __P((const char *, int)); + +static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *)); +static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *)); +static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *)); +static void parameter_brace_expand_error __P((char *, char *)); + +static int valid_length_expression __P((char *)); +static intmax_t parameter_brace_expand_length __P((char *)); + +static char *skiparith __P((char *, int)); +static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *)); +static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **)); +static char *mb_substring __P((char *, int, int)); +static char *parameter_brace_substring __P((char *, char *, int, char *, int, int)); + +static int shouldexp_replacement __P((char *)); + +static char *pos_params_pat_subst __P((char *, char *, char *, int)); + +static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int)); + +static char *pos_params_casemod __P((char *, char *, int, int)); +static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int)); + +static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *)); +static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int)); + +static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *)); + +static WORD_LIST *word_list_split __P((WORD_LIST *)); + +static void exp_jump_to_top_level __P((int)); + +static WORD_LIST *separate_out_assignments __P((WORD_LIST *)); +static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int)); +#ifdef BRACE_EXPANSION +static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int)); +#endif +#if defined (ARRAY_VARS) +static int make_internal_declare __P((char *, char *)); +#endif +static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int)); +static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int)); + +/* **************************************************************** */ +/* */ +/* Utility Functions */ +/* */ +/* **************************************************************** */ + +#if defined (DEBUG) +void +dump_word_flags (flags) + int flags; +{ + int f; + + f = flags; + fprintf (stderr, "%d -> ", f); + if (f & W_ASSIGNASSOC) + { + f &= ~W_ASSIGNASSOC; + fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : ""); + } + if (f & W_HASCTLESC) + { + f &= ~W_HASCTLESC; + fprintf (stderr, "W_HASCTLESC%s", f ? "|" : ""); + } + if (f & W_NOPROCSUB) + { + f &= ~W_NOPROCSUB; + fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : ""); + } + if (f & W_DQUOTE) + { + f &= ~W_DQUOTE; + fprintf (stderr, "W_DQUOTE%s", f ? "|" : ""); + } + if (f & W_HASQUOTEDNULL) + { + f &= ~W_HASQUOTEDNULL; + fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : ""); + } + if (f & W_ASSIGNARG) + { + f &= ~W_ASSIGNARG; + fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : ""); + } + if (f & W_ASSNBLTIN) + { + f &= ~W_ASSNBLTIN; + fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : ""); + } + if (f & W_COMPASSIGN) + { + f &= ~W_COMPASSIGN; + fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : ""); + } + if (f & W_NOEXPAND) + { + f &= ~W_NOEXPAND; + fprintf (stderr, "W_NOEXPAND%s", f ? "|" : ""); + } + if (f & W_ITILDE) + { + f &= ~W_ITILDE; + fprintf (stderr, "W_ITILDE%s", f ? "|" : ""); + } + if (f & W_NOTILDE) + { + f &= ~W_NOTILDE; + fprintf (stderr, "W_NOTILDE%s", f ? "|" : ""); + } + if (f & W_ASSIGNRHS) + { + f &= ~W_ASSIGNRHS; + fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : ""); + } + if (f & W_NOCOMSUB) + { + f &= ~W_NOCOMSUB; + fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : ""); + } + if (f & W_DOLLARSTAR) + { + f &= ~W_DOLLARSTAR; + fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : ""); + } + if (f & W_DOLLARAT) + { + f &= ~W_DOLLARAT; + fprintf (stderr, "W_DOLLARAT%s", f ? "|" : ""); + } + if (f & W_TILDEEXP) + { + f &= ~W_TILDEEXP; + fprintf (stderr, "W_TILDEEXP%s", f ? "|" : ""); + } + if (f & W_NOSPLIT2) + { + f &= ~W_NOSPLIT2; + fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : ""); + } + if (f & W_NOGLOB) + { + f &= ~W_NOGLOB; + fprintf (stderr, "W_NOGLOB%s", f ? "|" : ""); + } + if (f & W_NOSPLIT) + { + f &= ~W_NOSPLIT; + fprintf (stderr, "W_NOSPLIT%s", f ? "|" : ""); + } + if (f & W_GLOBEXP) + { + f &= ~W_GLOBEXP; + fprintf (stderr, "W_GLOBEXP%s", f ? "|" : ""); + } + if (f & W_ASSIGNMENT) + { + f &= ~W_ASSIGNMENT; + fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : ""); + } + if (f & W_QUOTED) + { + f &= ~W_QUOTED; + fprintf (stderr, "W_QUOTED%s", f ? "|" : ""); + } + if (f & W_HASDOLLAR) + { + f &= ~W_HASDOLLAR; + fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : ""); + } + fprintf (stderr, "\n"); + fflush (stderr); +} +#endif + +#ifdef INCLUDE_UNUSED +static char * +quoted_substring (string, start, end) + char *string; + int start, end; +{ + register int len, l; + register char *result, *s, *r; + + len = end - start; + + /* Move to string[start], skipping quoted characters. */ + for (s = string, l = 0; *s && l < start; ) + { + if (*s == CTLESC) + { + s++; + continue; + } + l++; + if (*s == 0) + break; + } + + r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */ + + /* Copy LEN characters, including quote characters. */ + s = string + l; + for (l = 0; l < len; s++) + { + if (*s == CTLESC) + *r++ = *s++; + *r++ = *s; + l++; + if (*s == 0) + break; + } + *r = '\0'; + return result; +} +#endif + +#ifdef INCLUDE_UNUSED +/* Return the length of S, skipping over quoted characters */ +static int +quoted_strlen (s) + char *s; +{ + register char *p; + int i; + + i = 0; + for (p = s; *p; p++) + { + if (*p == CTLESC) + { + p++; + if (*p == 0) + return (i + 1); + } + i++; + } + + return i; +} +#endif + +/* Find the first occurrence of character C in string S, obeying shell + quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped + characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters + escaped with CTLESC are skipped. */ +static char * +quoted_strchr (s, c, flags) + char *s; + int c, flags; +{ + register char *p; + + for (p = s; *p; p++) + { + if (((flags & ST_BACKSL) && *p == '\\') + || ((flags & ST_CTLESC) && *p == CTLESC)) + { + p++; + if (*p == '\0') + return ((char *)NULL); + continue; + } + else if (*p == c) + return p; + } + return ((char *)NULL); +} + +/* Return 1 if CHARACTER appears in an unquoted portion of + STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */ +static int +unquoted_member (character, string) + int character; + char *string; +{ + size_t slen; + int sindex, c; + DECLARE_MBSTATE; + + slen = strlen (string); + sindex = 0; + while (c = string[sindex]) + { + if (c == character) + return (1); + + switch (c) + { + default: + ADVANCE_CHAR (string, slen, sindex); + break; + + case '\\': + sindex++; + if (string[sindex]) + ADVANCE_CHAR (string, slen, sindex); + break; + + case '\'': + sindex = skip_single_quoted (string, slen, ++sindex); + break; + + case '"': + sindex = skip_double_quoted (string, slen, ++sindex); + break; + } + } + return (0); +} + +/* Return 1 if SUBSTR appears in an unquoted portion of STRING. */ +static int +unquoted_substring (substr, string) + char *substr, *string; +{ + size_t slen; + int sindex, c, sublen; + DECLARE_MBSTATE; + + if (substr == 0 || *substr == '\0') + return (0); + + slen = strlen (string); + sublen = strlen (substr); + for (sindex = 0; c = string[sindex]; ) + { + if (STREQN (string + sindex, substr, sublen)) + return (1); + + switch (c) + { + case '\\': + sindex++; + if (string[sindex]) + ADVANCE_CHAR (string, slen, sindex); + break; + + case '\'': + sindex = skip_single_quoted (string, slen, ++sindex); + break; + + case '"': + sindex = skip_double_quoted (string, slen, ++sindex); + break; + + default: + ADVANCE_CHAR (string, slen, sindex); + break; + } + } + return (0); +} + +/* Most of the substitutions must be done in parallel. In order + to avoid using tons of unclear goto's, I have some functions + for manipulating malloc'ed strings. They all take INDX, a + pointer to an integer which is the offset into the string + where manipulation is taking place. They also take SIZE, a + pointer to an integer which is the current length of the + character array for this string. */ + +/* Append SOURCE to TARGET at INDEX. SIZE is the current amount + of space allocated to TARGET. SOURCE can be NULL, in which + case nothing happens. Gets rid of SOURCE by freeing it. + Returns TARGET in case the location has changed. */ +INLINE char * +sub_append_string (source, target, indx, size) + char *source, *target; + int *indx, *size; +{ + if (source) + { + int srclen, n; + + srclen = STRLEN (source); + if (srclen >= (int)(*size - *indx)) + { + n = srclen + *indx; + n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE); + target = (char *)xrealloc (target, (*size = n)); + } + + FASTCOPY (source, target + *indx, srclen); + *indx += srclen; + target[*indx] = '\0'; + + free (source); + } + return (target); +} + +#if 0 +/* UNUSED */ +/* Append the textual representation of NUMBER to TARGET. + INDX and SIZE are as in SUB_APPEND_STRING. */ +char * +sub_append_number (number, target, indx, size) + intmax_t number; + int *indx, *size; + char *target; +{ + char *temp; + + temp = itos (number); + return (sub_append_string (temp, target, indx, size)); +} +#endif + +/* Extract a substring from STRING, starting at SINDEX and ending with + one of the characters in CHARLIST. Don't make the ending character + part of the string. Leave SINDEX pointing at the ending character. + Understand about backslashes in the string. If (flags & SX_VARNAME) + is non-zero, and array variables have been compiled into the shell, + everything between a `[' and a corresponding `]' is skipped over. + If (flags & SX_NOALLOC) is non-zero, don't return the substring, just + update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must + contain a closing character from CHARLIST. */ +static char * +string_extract (string, sindex, charlist, flags) + char *string; + int *sindex; + char *charlist; + int flags; +{ + register int c, i; + int found; + size_t slen; + char *temp; + DECLARE_MBSTATE; + + slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0; + i = *sindex; + found = 0; + while (c = string[i]) + { + if (c == '\\') + { + if (string[i + 1]) + i++; + else + break; + } +#if defined (ARRAY_VARS) + else if ((flags & SX_VARNAME) && c == '[') + { + int ni; + /* If this is an array subscript, skip over it and continue. */ + ni = skipsubscript (string, i, 0); + if (string[ni] == ']') + i = ni; + } +#endif + else if (MEMBER (c, charlist)) + { + found = 1; + break; + } + + ADVANCE_CHAR (string, slen, i); + } + + /* If we had to have a matching delimiter and didn't find one, return an + error and let the caller deal with it. */ + if ((flags & SX_REQMATCH) && found == 0) + { + *sindex = i; + return (&extract_string_error); + } + + temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i); + *sindex = i; + + return (temp); +} + +/* Extract the contents of STRING as if it is enclosed in double quotes. + SINDEX, when passed in, is the offset of the character immediately + following the opening double quote; on exit, SINDEX is left pointing after + the closing double quote. If STRIPDQ is non-zero, unquoted double + quotes are stripped and the string is terminated by a null byte. + Backslashes between the embedded double quotes are processed. If STRIPDQ + is zero, an unquoted `"' terminates the string. */ +static char * +string_extract_double_quoted (string, sindex, stripdq) + char *string; + int *sindex, stripdq; +{ + size_t slen; + char *send; + int j, i, t; + unsigned char c; + char *temp, *ret; /* The new string we return. */ + int pass_next, backquote, si; /* State variables for the machine. */ + int dquote; + DECLARE_MBSTATE; + + slen = strlen (string + *sindex) + *sindex; + send = string + slen; + + pass_next = backquote = dquote = 0; + temp = (char *)xmalloc (1 + slen - *sindex); + + j = 0; + i = *sindex; + while (c = string[i]) + { + /* Process a character that was quoted by a backslash. */ + if (pass_next) + { + /* XXX - take another look at this in light of Interp 221 */ + /* Posix.2 sez: + + ``The backslash shall retain its special meaning as an escape + character only when followed by one of the characters: + $ ` " \ ''. + + If STRIPDQ is zero, we handle the double quotes here and let + expand_word_internal handle the rest. If STRIPDQ is non-zero, + we have already been through one round of backslash stripping, + and want to strip these backslashes only if DQUOTE is non-zero, + indicating that we are inside an embedded double-quoted string. */ + + /* If we are in an embedded quoted string, then don't strip + backslashes before characters for which the backslash + retains its special meaning, but remove backslashes in + front of other characters. If we are not in an + embedded quoted string, don't strip backslashes at all. + This mess is necessary because the string was already + surrounded by double quotes (and sh has some really weird + quoting rules). + The returned string will be run through expansion as if + it were double-quoted. */ + if ((stripdq == 0 && c != '"') || + (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0))) + temp[j++] = '\\'; + pass_next = 0; + +add_one_character: + COPY_CHAR_I (temp, j, string, send, i); + continue; + } + + /* A backslash protects the next character. The code just above + handles preserving the backslash in front of any character but + a double quote. */ + if (c == '\\') + { + pass_next++; + i++; + continue; + } + + /* Inside backquotes, ``the portion of the quoted string from the + initial backquote and the characters up to the next backquote + that is not preceded by a backslash, having escape characters + removed, defines that command''. */ + if (backquote) + { + if (c == '`') + backquote = 0; + temp[j++] = c; + i++; + continue; + } + + if (c == '`') + { + temp[j++] = c; + backquote++; + i++; + continue; + } + + /* Pass everything between `$(' and the matching `)' or a quoted + ${ ... } pair through according to the Posix.2 specification. */ + if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE))) + { + int free_ret = 1; + + si = i + 2; + if (string[i + 1] == LPAREN) + ret = extract_command_subst (string, &si, 0); + else + ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0); + + temp[j++] = '$'; + temp[j++] = string[i + 1]; + + /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error + is set. */ + if (ret == 0 && no_longjmp_on_fatal_error) + { + free_ret = 0; + ret = string + i + 2; + } + + for (t = 0; ret[t]; t++, j++) + temp[j] = ret[t]; + temp[j] = string[si]; + + if (string[si]) + { + j++; + i = si + 1; + } + else + i = si; + + if (free_ret) + free (ret); + continue; + } + + /* Add any character but a double quote to the quoted string we're + accumulating. */ + if (c != '"') + goto add_one_character; + + /* c == '"' */ + if (stripdq) + { + dquote ^= 1; + i++; + continue; + } + + break; + } + temp[j] = '\0'; + + /* Point to after the closing quote. */ + if (c) + i++; + *sindex = i; + + return (temp); +} + +/* This should really be another option to string_extract_double_quoted. */ +static int +skip_double_quoted (string, slen, sind) + char *string; + size_t slen; + int sind; +{ + int c, i; + char *ret; + int pass_next, backquote, si; + DECLARE_MBSTATE; + + pass_next = backquote = 0; + i = sind; + while (c = string[i]) + { + if (pass_next) + { + pass_next = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (c == '\\') + { + pass_next++; + i++; + continue; + } + else if (backquote) + { + if (c == '`') + backquote = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (c == '`') + { + backquote++; + i++; + continue; + } + else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE))) + { + si = i + 2; + if (string[i + 1] == LPAREN) + ret = extract_command_subst (string, &si, SX_NOALLOC); + else + ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC); + + i = si + 1; + continue; + } + else if (c != '"') + { + ADVANCE_CHAR (string, slen, i); + continue; + } + else + break; + } + + if (c) + i++; + + return (i); +} + +/* Extract the contents of STRING as if it is enclosed in single quotes. + SINDEX, when passed in, is the offset of the character immediately + following the opening single quote; on exit, SINDEX is left pointing after + the closing single quote. */ +static inline char * +string_extract_single_quoted (string, sindex) + char *string; + int *sindex; +{ + register int i; + size_t slen; + char *t; + DECLARE_MBSTATE; + + /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */ + slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0; + i = *sindex; + while (string[i] && string[i] != '\'') + ADVANCE_CHAR (string, slen, i); + + t = substring (string, *sindex, i); + + if (string[i]) + i++; + *sindex = i; + + return (t); +} + +static inline int +skip_single_quoted (string, slen, sind) + const char *string; + size_t slen; + int sind; +{ + register int c; + DECLARE_MBSTATE; + + c = sind; + while (string[c] && string[c] != '\'') + ADVANCE_CHAR (string, slen, c); + + if (string[c]) + c++; + return c; +} + +/* Just like string_extract, but doesn't hack backslashes or any of + that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */ +static char * +string_extract_verbatim (string, slen, sindex, charlist, flags) + char *string; + size_t slen; + int *sindex; + char *charlist; + int flags; +{ + register int i; +#if defined (HANDLE_MULTIBYTE) + size_t clen; + wchar_t *wcharlist; +#endif + int c; + char *temp; + DECLARE_MBSTATE; + + if (charlist[0] == '\'' && charlist[1] == '\0') + { + temp = string_extract_single_quoted (string, sindex); + --*sindex; /* leave *sindex at separator character */ + return temp; + } + + i = *sindex; +#if 0 + /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need + this only if MB_CUR_MAX > 1. */ + slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1; +#endif +#if defined (HANDLE_MULTIBYTE) + clen = strlen (charlist); + wcharlist = 0; +#endif + while (c = string[i]) + { +#if defined (HANDLE_MULTIBYTE) + size_t mblength; +#endif + if ((flags & SX_NOCTLESC) == 0 && c == CTLESC) + { + i += 2; + continue; + } + /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL + through, to protect the CTLNULs from later calls to + remove_quoted_nulls. */ + else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL) + { + i += 2; + continue; + } + +#if defined (HANDLE_MULTIBYTE) + mblength = MBLEN (string + i, slen - i); + if (mblength > 1) + { + wchar_t wc; + mblength = mbtowc (&wc, string + i, slen - i); + if (MB_INVALIDCH (mblength)) + { + if (MEMBER (c, charlist)) + break; + } + else + { + if (wcharlist == 0) + { + size_t len; + len = mbstowcs (wcharlist, charlist, 0); + if (len == -1) + len = 0; + wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1)); + mbstowcs (wcharlist, charlist, len + 1); + } + + if (wcschr (wcharlist, wc)) + break; + } + } + else +#endif + if (MEMBER (c, charlist)) + break; + + ADVANCE_CHAR (string, slen, i); + } + +#if defined (HANDLE_MULTIBYTE) + FREE (wcharlist); +#endif + + temp = substring (string, *sindex, i); + *sindex = i; + + return (temp); +} + +/* Extract the $( construct in STRING, and return a new string. + Start extracting at (SINDEX) as if we had just seen "$(". + Make (SINDEX) get the position of the matching ")". ) + XFLAGS is additional flags to pass to other extraction functions. */ +char * +extract_command_subst (string, sindex, xflags) + char *string; + int *sindex; + int xflags; +{ + if (string[*sindex] == LPAREN) + return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/ + else + { + xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0); + return (xparse_dolparen (string, string+*sindex, sindex, xflags)); + } +} + +/* Extract the $[ construct in STRING, and return a new string. (]) + Start extracting at (SINDEX) as if we had just seen "$[". + Make (SINDEX) get the position of the matching "]". */ +char * +extract_arithmetic_subst (string, sindex) + char *string; + int *sindex; +{ + return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/ +} + +#if defined (PROCESS_SUBSTITUTION) +/* Extract the <( or >( construct in STRING, and return a new string. + Start extracting at (SINDEX) as if we had just seen "<(". + Make (SINDEX) get the position of the matching ")". */ /*))*/ +char * +extract_process_subst (string, starter, sindex) + char *string; + char *starter; + int *sindex; +{ + return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND)); +} +#endif /* PROCESS_SUBSTITUTION */ + +#if defined (ARRAY_VARS) +/* This can be fooled by unquoted right parens in the passed string. If + each caller verifies that the last character in STRING is a right paren, + we don't even need to call extract_delimited_string. */ +char * +extract_array_assignment_list (string, sindex) + char *string; + int *sindex; +{ + int slen; + char *ret; + + slen = strlen (string); /* ( */ + if (string[slen - 1] == ')') + { + ret = substring (string, *sindex, slen - 1); + *sindex = slen - 1; + return ret; + } + return 0; +} +#endif + +/* Extract and create a new string from the contents of STRING, a + character string delimited with OPENER and CLOSER. SINDEX is + the address of an int describing the current offset in STRING; + it should point to just after the first OPENER found. On exit, + SINDEX gets the position of the last character of the matching CLOSER. + If OPENER is more than a single character, ALT_OPENER, if non-null, + contains a character string that can also match CLOSER and thus + needs to be skipped. */ +static char * +extract_delimited_string (string, sindex, opener, alt_opener, closer, flags) + char *string; + int *sindex; + char *opener, *alt_opener, *closer; + int flags; +{ + int i, c, si; + size_t slen; + char *t, *result; + int pass_character, nesting_level, in_comment; + int len_closer, len_opener, len_alt_opener; + DECLARE_MBSTATE; + + slen = strlen (string + *sindex) + *sindex; + len_opener = STRLEN (opener); + len_alt_opener = STRLEN (alt_opener); + len_closer = STRLEN (closer); + + pass_character = in_comment = 0; + + nesting_level = 1; + i = *sindex; + + while (nesting_level) + { + c = string[i]; + + if (c == 0) + break; + + if (in_comment) + { + if (c == '\n') + in_comment = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + + if (pass_character) /* previous char was backslash */ + { + pass_character = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + + /* Not exactly right yet; should handle shell metacharacters and + multibyte characters, too. See COMMENT_BEGIN define in parse.y */ + if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1]))) + { + in_comment = 1; + ADVANCE_CHAR (string, slen, i); + continue; + } + + if (c == CTLESC || c == '\\') + { + pass_character++; + i++; + continue; + } + + /* Process a nested command substitution, but only if we're parsing an + arithmetic substitution. */ + if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN) + { + si = i + 2; + t = extract_command_subst (string, &si, flags|SX_NOALLOC); + i = si + 1; + continue; + } + + /* Process a nested OPENER. */ + if (STREQN (string + i, opener, len_opener)) + { + si = i + len_opener; + t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC); + i = si + 1; + continue; + } + + /* Process a nested ALT_OPENER */ + if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener)) + { + si = i + len_alt_opener; + t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC); + i = si + 1; + continue; + } + + /* If the current substring terminates the delimited string, decrement + the nesting level. */ + if (STREQN (string + i, closer, len_closer)) + { + i += len_closer - 1; /* move to last byte of the closer */ + nesting_level--; + if (nesting_level == 0) + break; + } + + /* Pass old-style command substitution through verbatim. */ + if (c == '`') + { + si = i + 1; + t = string_extract (string, &si, "`", flags|SX_NOALLOC); + i = si + 1; + continue; + } + + /* Pass single-quoted and double-quoted strings through verbatim. */ + if (c == '\'' || c == '"') + { + si = i + 1; + i = (c == '\'') ? skip_single_quoted (string, slen, si) + : skip_double_quoted (string, slen, si); + continue; + } + + /* move past this character, which was not special. */ + ADVANCE_CHAR (string, slen, i); + } + + if (c == 0 && nesting_level) + { + if (no_longjmp_on_fatal_error == 0) + { + last_command_exit_value = EXECUTION_FAILURE; + report_error (_("bad substitution: no closing `%s' in %s"), closer, string); + exp_jump_to_top_level (DISCARD); + } + else + { + *sindex = i; + return (char *)NULL; + } + } + + si = i - *sindex - len_closer + 1; + if (flags & SX_NOALLOC) + result = (char *)NULL; + else + { + result = (char *)xmalloc (1 + si); + strncpy (result, string + *sindex, si); + result[si] = '\0'; + } + *sindex = i; + + return (result); +} + +/* Extract a parameter expansion expression within ${ and } from STRING. + Obey the Posix.2 rules for finding the ending `}': count braces while + skipping over enclosed quoted strings and command substitutions. + SINDEX is the address of an int describing the current offset in STRING; + it should point to just after the first `{' found. On exit, SINDEX + gets the position of the matching `}'. QUOTED is non-zero if this + occurs inside double quotes. */ +/* XXX -- this is very similar to extract_delimited_string -- XXX */ +static char * +extract_dollar_brace_string (string, sindex, quoted, flags) + char *string; + int *sindex, quoted, flags; +{ + register int i, c; + size_t slen; + int pass_character, nesting_level, si, dolbrace_state; + char *result, *t; + DECLARE_MBSTATE; + + pass_character = 0; + nesting_level = 1; + slen = strlen (string + *sindex) + *sindex; + + /* The handling of dolbrace_state needs to agree with the code in parse.y: + parse_matched_pair(). The different initial value is to handle the + case where this function is called to parse the word in + ${param op word} (SX_WORD). */ + dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM; + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP)) + dolbrace_state = DOLBRACE_QUOTE; + + i = *sindex; + while (c = string[i]) + { + if (pass_character) + { + pass_character = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + + /* CTLESCs and backslashes quote the next character. */ + if (c == CTLESC || c == '\\') + { + pass_character++; + i++; + continue; + } + + if (string[i] == '$' && string[i+1] == LBRACE) + { + nesting_level++; + i += 2; + continue; + } + + if (c == RBRACE) + { + nesting_level--; + if (nesting_level == 0) + break; + i++; + continue; + } + + /* Pass the contents of old-style command substitutions through + verbatim. */ + if (c == '`') + { + si = i + 1; + t = string_extract (string, &si, "`", flags|SX_NOALLOC); + i = si + 1; + continue; + } + + /* Pass the contents of new-style command substitutions and + arithmetic substitutions through verbatim. */ + if (string[i] == '$' && string[i+1] == LPAREN) + { + si = i + 2; + t = extract_command_subst (string, &si, flags|SX_NOALLOC); + i = si + 1; + continue; + } + +#if 0 + /* Pass the contents of single-quoted and double-quoted strings + through verbatim. */ + if (c == '\'' || c == '"') + { + si = i + 1; + i = (c == '\'') ? skip_single_quoted (string, slen, si) + : skip_double_quoted (string, slen, si); + /* skip_XXX_quoted leaves index one past close quote */ + continue; + } +#else /* XXX - bash-4.2 */ + /* Pass the contents of double-quoted strings through verbatim. */ + if (c == '"') + { + si = i + 1; + i = skip_double_quoted (string, slen, si); + /* skip_XXX_quoted leaves index one past close quote */ + continue; + } + + if (c == '\'') + { +/*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/ + if (posixly_correct && shell_compatibility_level > 41 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ADVANCE_CHAR (string, slen, i); + else + { + si = i + 1; + i = skip_single_quoted (string, slen, si); + } + + continue; + } +#endif + + /* move past this character, which was not special. */ + ADVANCE_CHAR (string, slen, i); + + /* This logic must agree with parse.y:parse_matched_pair, since they + share the same defines. */ + if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1) + dolbrace_state = DOLBRACE_QUOTE; + else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1) + dolbrace_state = DOLBRACE_QUOTE; + else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1) + dolbrace_state = DOLBRACE_QUOTE; + else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1) + dolbrace_state = DOLBRACE_QUOTE; + else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1) + dolbrace_state = DOLBRACE_QUOTE; + else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0) + dolbrace_state = DOLBRACE_OP; + else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0) + dolbrace_state = DOLBRACE_WORD; + } + + if (c == 0 && nesting_level) + { + if (no_longjmp_on_fatal_error == 0) + { /* { */ + last_command_exit_value = EXECUTION_FAILURE; + report_error (_("bad substitution: no closing `%s' in %s"), "}", string); + exp_jump_to_top_level (DISCARD); + } + else + { + *sindex = i; + return ((char *)NULL); + } + } + + result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i); + *sindex = i; + + return (result); +} + +/* Remove backslashes which are quoting backquotes from STRING. Modifies + STRING, and returns a pointer to it. */ +char * +de_backslash (string) + char *string; +{ + register size_t slen; + register int i, j, prev_i; + DECLARE_MBSTATE; + + slen = strlen (string); + i = j = 0; + + /* Loop copying string[i] to string[j], i >= j. */ + while (i < slen) + { + if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' || + string[i + 1] == '$')) + i++; + prev_i = i; + ADVANCE_CHAR (string, slen, i); + if (j < prev_i) + do string[j++] = string[prev_i++]; while (prev_i < i); + else + j = i; + } + string[j] = '\0'; + + return (string); +} + +#if 0 +/*UNUSED*/ +/* Replace instances of \! in a string with !. */ +void +unquote_bang (string) + char *string; +{ + register int i, j; + register char *temp; + + temp = (char *)xmalloc (1 + strlen (string)); + + for (i = 0, j = 0; (temp[j] = string[i]); i++, j++) + { + if (string[i] == '\\' && string[i + 1] == '!') + { + temp[j] = '!'; + i++; + } + } + strcpy (string, temp); + free (temp); +} +#endif + +#define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0) + +/* This function assumes s[i] == open; returns with s[ret] == close; used to + parse array subscripts. FLAGS & 1 means to not attempt to skip over + matched pairs of quotes or backquotes, or skip word expansions; it is + intended to be used after expansion has been performed and during final + assignment parsing (see arrayfunc.c:assign_compound_array_list()). */ +static int +skip_matched_pair (string, start, open, close, flags) + const char *string; + int start, open, close, flags; +{ + int i, pass_next, backq, si, c, count; + size_t slen; + char *temp, *ss; + DECLARE_MBSTATE; + + slen = strlen (string + start) + start; + no_longjmp_on_fatal_error = 1; + + i = start + 1; /* skip over leading bracket */ + count = 1; + pass_next = backq = 0; + ss = (char *)string; + while (c = string[i]) + { + if (pass_next) + { + pass_next = 0; + if (c == 0) + CQ_RETURN(i); + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (c == '\\') + { + pass_next = 1; + i++; + continue; + } + else if (backq) + { + if (c == '`') + backq = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + else if ((flags & 1) == 0 && c == '`') + { + backq = 1; + i++; + continue; + } + else if ((flags & 1) == 0 && c == open) + { + count++; + i++; + continue; + } + else if (c == close) + { + count--; + if (count == 0) + break; + i++; + continue; + } + else if ((flags & 1) == 0 && (c == '\'' || c == '"')) + { + i = (c == '\'') ? skip_single_quoted (ss, slen, ++i) + : skip_double_quoted (ss, slen, ++i); + /* no increment, the skip functions increment past the closing quote. */ + } + else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE)) + { + si = i + 2; + if (string[si] == '\0') + CQ_RETURN(si); + + if (string[i+1] == LPAREN) + temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */ + else + temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC); + i = si; + if (string[i] == '\0') /* don't increment i past EOS in loop */ + break; + i++; + continue; + } + else + ADVANCE_CHAR (string, slen, i); + } + + CQ_RETURN(i); +} + +#if defined (ARRAY_VARS) +int +skipsubscript (string, start, flags) + const char *string; + int start, flags; +{ + return (skip_matched_pair (string, start, '[', ']', flags)); +} +#endif + +/* Skip characters in STRING until we find a character in DELIMS, and return + the index of that character. START is the index into string at which we + begin. This is similar in spirit to strpbrk, but it returns an index into + STRING and takes a starting index. This little piece of code knows quite + a lot of shell syntax. It's very similar to skip_double_quoted and other + functions of that ilk. */ +int +skip_to_delim (string, start, delims, flags) + char *string; + int start; + char *delims; + int flags; +{ + int i, pass_next, backq, si, c, invert, skipquote, skipcmd; + size_t slen; + char *temp, open[3]; + DECLARE_MBSTATE; + + slen = strlen (string + start) + start; + if (flags & SD_NOJMP) + no_longjmp_on_fatal_error = 1; + invert = (flags & SD_INVERT); + skipcmd = (flags & SD_NOSKIPCMD) == 0; + + i = start; + pass_next = backq = 0; + while (c = string[i]) + { + /* If this is non-zero, we should not let quote characters be delimiters + and the current character is a single or double quote. We should not + test whether or not it's a delimiter until after we skip single- or + double-quoted strings. */ + skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"')); + if (pass_next) + { + pass_next = 0; + if (c == 0) + CQ_RETURN(i); + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (c == '\\') + { + pass_next = 1; + i++; + continue; + } + else if (backq) + { + if (c == '`') + backq = 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (c == '`') + { + backq = 1; + i++; + continue; + } + else if (skipquote == 0 && invert == 0 && member (c, delims)) + break; + else if (c == '\'' || c == '"') + { + i = (c == '\'') ? skip_single_quoted (string, slen, ++i) + : skip_double_quoted (string, slen, ++i); + /* no increment, the skip functions increment past the closing quote. */ + } + else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE)) + { + si = i + 2; + if (string[si] == '\0') + CQ_RETURN(si); + + if (string[i+1] == LPAREN) + temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */ + else + temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC); + i = si; + if (string[i] == '\0') /* don't increment i past EOS in loop */ + break; + i++; + continue; + } +#if defined (PROCESS_SUBSTITUTION) + else if (skipcmd && (c == '<' || c == '>') && string[i+1] == LPAREN) + { + si = i + 2; + if (string[si] == '\0') + CQ_RETURN(si); + temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si); + free (temp); /* no SX_ALLOC here */ + i = si; + if (string[i] == '\0') + break; + i++; + continue; + } +#endif /* PROCESS_SUBSTITUTION */ +#if defined (EXTENDED_GLOB) + else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@")) + { + si = i + 2; + if (string[si] == '\0') + CQ_RETURN(si); + + open[0] = c; + open[1] = LPAREN; + open[2] = '\0'; + temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */ + + i = si; + if (string[i] == '\0') /* don't increment i past EOS in loop */ + break; + i++; + continue; + } +#endif + else if ((skipquote || invert) && (member (c, delims) == 0)) + break; + else + ADVANCE_CHAR (string, slen, i); + } + + CQ_RETURN(i); +} + +#if defined (READLINE) +/* Return 1 if the portion of STRING ending at EINDEX is quoted (there is + an unclosed quoted string), or if the character at EINDEX is quoted + by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various + single and double-quoted string parsing functions should not return an + error if there are unclosed quotes or braces. The characters that this + recognizes need to be the same as the contents of + rl_completer_quote_characters. */ + +int +char_is_quoted (string, eindex) + char *string; + int eindex; +{ + int i, pass_next, c; + size_t slen; + DECLARE_MBSTATE; + + slen = strlen (string); + no_longjmp_on_fatal_error = 1; + i = pass_next = 0; + while (i <= eindex) + { + c = string[i]; + + if (pass_next) + { + pass_next = 0; + if (i >= eindex) /* XXX was if (i >= eindex - 1) */ + CQ_RETURN(1); + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (c == '\\') + { + pass_next = 1; + i++; + continue; + } + else if (c == '\'' || c == '"') + { + i = (c == '\'') ? skip_single_quoted (string, slen, ++i) + : skip_double_quoted (string, slen, ++i); + if (i > eindex) + CQ_RETURN(1); + /* no increment, the skip_xxx functions go one past end */ + } + else + ADVANCE_CHAR (string, slen, i); + } + + CQ_RETURN(0); +} + +int +unclosed_pair (string, eindex, openstr) + char *string; + int eindex; + char *openstr; +{ + int i, pass_next, openc, olen; + size_t slen; + DECLARE_MBSTATE; + + slen = strlen (string); + olen = strlen (openstr); + i = pass_next = openc = 0; + while (i <= eindex) + { + if (pass_next) + { + pass_next = 0; + if (i >= eindex) /* XXX was if (i >= eindex - 1) */ + return 0; + ADVANCE_CHAR (string, slen, i); + continue; + } + else if (string[i] == '\\') + { + pass_next = 1; + i++; + continue; + } + else if (STREQN (string + i, openstr, olen)) + { + openc = 1 - openc; + i += olen; + } + else if (string[i] == '\'' || string[i] == '"') + { + i = (string[i] == '\'') ? skip_single_quoted (string, slen, i) + : skip_double_quoted (string, slen, i); + if (i > eindex) + return 0; + } + else + ADVANCE_CHAR (string, slen, i); + } + return (openc); +} + +/* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the + individual words. If DELIMS is NULL, the current value of $IFS is used + to split the string, and the function follows the shell field splitting + rules. SENTINEL is an index to look for. NWP, if non-NULL, + gets the number of words in the returned list. CWP, if non-NULL, gets + the index of the word containing SENTINEL. Non-whitespace chars in + DELIMS delimit separate fields. */ +WORD_LIST * +split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp) + char *string; + int slen; + char *delims; + int sentinel, flags; + int *nwp, *cwp; +{ + int ts, te, i, nw, cw, ifs_split, dflags; + char *token, *d, *d2; + WORD_LIST *ret, *tl; + + if (string == 0 || *string == '\0') + { + if (nwp) + *nwp = 0; + if (cwp) + *cwp = 0; + return ((WORD_LIST *)NULL); + } + + d = (delims == 0) ? ifs_value : delims; + ifs_split = delims == 0; + + /* Make d2 the non-whitespace characters in delims */ + d2 = 0; + if (delims) + { + size_t slength; +#if defined (HANDLE_MULTIBYTE) + size_t mblength = 1; +#endif + DECLARE_MBSTATE; + + slength = strlen (delims); + d2 = (char *)xmalloc (slength + 1); + i = ts = 0; + while (delims[i]) + { +#if defined (HANDLE_MULTIBYTE) + mbstate_t state_bak; + state_bak = state; + mblength = MBRLEN (delims + i, slength, &state); + if (MB_INVALIDCH (mblength)) + state = state_bak; + else if (mblength > 1) + { + memcpy (d2 + ts, delims + i, mblength); + ts += mblength; + i += mblength; + slength -= mblength; + continue; + } +#endif + if (whitespace (delims[i]) == 0) + d2[ts++] = delims[i]; + + i++; + slength--; + } + d2[ts] = '\0'; + } + + ret = (WORD_LIST *)NULL; + + /* Remove sequences of whitespace characters at the start of the string, as + long as those characters are delimiters. */ + for (i = 0; member (string[i], d) && spctabnl (string[i]); i++) + ; + if (string[i] == '\0') + return (ret); + + ts = i; + nw = 0; + cw = -1; + dflags = flags|SD_NOJMP; + while (1) + { + te = skip_to_delim (string, ts, d, dflags); + + /* If we have a non-whitespace delimiter character, use it to make a + separate field. This is just about what $IFS splitting does and + is closer to the behavior of the shell parser. */ + if (ts == te && d2 && member (string[ts], d2)) + { + te = ts + 1; + /* If we're using IFS splitting, the non-whitespace delimiter char + and any additional IFS whitespace delimits a field. */ + if (ifs_split) + while (member (string[te], d) && spctabnl (string[te])) + te++; + else + while (member (string[te], d2)) + te++; + } + + token = substring (string, ts, te); + + ret = add_string_to_list (token, ret); + free (token); + nw++; + + if (sentinel >= ts && sentinel <= te) + cw = nw; + + /* If the cursor is at whitespace just before word start, set the + sentinel word to the current word. */ + if (cwp && cw == -1 && sentinel == ts-1) + cw = nw; + + /* If the cursor is at whitespace between two words, make a new, empty + word, add it before (well, after, since the list is in reverse order) + the word we just added, and set the current word to that one. */ + if (cwp && cw == -1 && sentinel < ts) + { + tl = make_word_list (make_word (""), ret->next); + ret->next = tl; + cw = nw; + nw++; + } + + if (string[te] == 0) + break; + + i = te; + while (member (string[i], d) && (ifs_split || spctabnl(string[i]))) + i++; + + if (string[i]) + ts = i; + else + break; + } + + /* Special case for SENTINEL at the end of STRING. If we haven't found + the word containing SENTINEL yet, and the index we're looking for is at + the end of STRING (or past the end of the previously-found token, + possible if the end of the line is composed solely of IFS whitespace) + add an additional null argument and set the current word pointer to that. */ + if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te)) + { + if (whitespace (string[sentinel - 1])) + { + token = ""; + ret = add_string_to_list (token, ret); + nw++; + } + cw = nw; + } + + if (nwp) + *nwp = nw; + if (cwp) + *cwp = cw; + + FREE (d2); + + return (REVERSE_LIST (ret, WORD_LIST *)); +} +#endif /* READLINE */ + +#if 0 +/* UNUSED */ +/* Extract the name of the variable to bind to from the assignment string. */ +char * +assignment_name (string) + char *string; +{ + int offset; + char *temp; + + offset = assignment (string, 0); + if (offset == 0) + return (char *)NULL; + temp = substring (string, 0, offset); + return (temp); +} +#endif + +/* **************************************************************** */ +/* */ +/* Functions to convert strings to WORD_LISTs and vice versa */ +/* */ +/* **************************************************************** */ + +/* Return a single string of all the words in LIST. SEP is the separator + to put between individual elements of LIST in the output string. */ +char * +string_list_internal (list, sep) + WORD_LIST *list; + char *sep; +{ + register WORD_LIST *t; + char *result, *r; + int word_len, sep_len, result_size; + + if (list == 0) + return ((char *)NULL); + + /* Short-circuit quickly if we don't need to separate anything. */ + if (list->next == 0) + return (savestring (list->word->word)); + + /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */ + sep_len = STRLEN (sep); + result_size = 0; + + for (t = list; t; t = t->next) + { + if (t != list) + result_size += sep_len; + result_size += strlen (t->word->word); + } + + r = result = (char *)xmalloc (result_size + 1); + + for (t = list; t; t = t->next) + { + if (t != list && sep_len) + { + if (sep_len > 1) + { + FASTCOPY (sep, r, sep_len); + r += sep_len; + } + else + *r++ = sep[0]; + } + + word_len = strlen (t->word->word); + FASTCOPY (t->word->word, r, word_len); + r += word_len; + } + + *r = '\0'; + return (result); +} + +/* Return a single string of all the words present in LIST, separating + each word with a space. */ +char * +string_list (list) + WORD_LIST *list; +{ + return (string_list_internal (list, " ")); +} + +/* An external interface that can be used by the rest of the shell to + obtain a string containing the first character in $IFS. Handles all + the multibyte complications. If LENP is non-null, it is set to the + length of the returned string. */ +char * +ifs_firstchar (lenp) + int *lenp; +{ + char *ret; + int len; + + ret = xmalloc (MB_LEN_MAX + 1); +#if defined (HANDLE_MULTIBYTE) + if (ifs_firstc_len == 1) + { + ret[0] = ifs_firstc[0]; + ret[1] = '\0'; + len = ret[0] ? 1 : 0; + } + else + { + memcpy (ret, ifs_firstc, ifs_firstc_len); + ret[len = ifs_firstc_len] = '\0'; + } +#else + ret[0] = ifs_firstc; + ret[1] = '\0'; + len = ret[0] ? 0 : 1; +#endif + + if (lenp) + *lenp = len; + + return ret; +} + +/* Return a single string of all the words present in LIST, obeying the + quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the + expansion [of $*] appears within a double quoted string, it expands + to a single field with the value of each parameter separated by the + first character of the IFS variable, or by a if IFS is unset." */ +char * +string_list_dollar_star (list) + WORD_LIST *list; +{ + char *ret; +#if defined (HANDLE_MULTIBYTE) +# if defined (__GNUC__) + char sep[MB_CUR_MAX + 1]; +# else + char *sep = 0; +# endif +#else + char sep[2]; +#endif + +#if defined (HANDLE_MULTIBYTE) +# if !defined (__GNUC__) + sep = (char *)xmalloc (MB_CUR_MAX + 1); +# endif /* !__GNUC__ */ + if (ifs_firstc_len == 1) + { + sep[0] = ifs_firstc[0]; + sep[1] = '\0'; + } + else + { + memcpy (sep, ifs_firstc, ifs_firstc_len); + sep[ifs_firstc_len] = '\0'; + } +#else + sep[0] = ifs_firstc; + sep[1] = '\0'; +#endif + + ret = string_list_internal (list, sep); +#if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__) + free (sep); +#endif + return ret; +} + +/* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + is non-zero, the $@ appears within double quotes, and we should quote + the list before converting it into a string. If IFS is unset, and the + word is not quoted, we just need to quote CTLESC and CTLNUL characters + in the words in the list, because the default value of $IFS is + , IFS characters in the words in the list should + also be split. If IFS is null, and the word is not quoted, we need + to quote the words in the list to preserve the positional parameters + exactly. */ +char * +string_list_dollar_at (list, quoted) + WORD_LIST *list; + int quoted; +{ + char *ifs, *ret; +#if defined (HANDLE_MULTIBYTE) +# if defined (__GNUC__) + char sep[MB_CUR_MAX + 1]; +# else + char *sep = 0; +# endif /* !__GNUC__ */ +#else + char sep[2]; +#endif + WORD_LIST *tlist; + + /* XXX this could just be ifs = ifs_value; */ + ifs = ifs_var ? value_cell (ifs_var) : (char *)0; + +#if defined (HANDLE_MULTIBYTE) +# if !defined (__GNUC__) + sep = (char *)xmalloc (MB_CUR_MAX + 1); +# endif /* !__GNUC__ */ + if (ifs && *ifs) + { + if (ifs_firstc_len == 1) + { + sep[0] = ifs_firstc[0]; + sep[1] = '\0'; + } + else + { + memcpy (sep, ifs_firstc, ifs_firstc_len); + sep[ifs_firstc_len] = '\0'; + } + } + else + { + sep[0] = ' '; + sep[1] = '\0'; + } +#else + sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs; + sep[1] = '\0'; +#endif + + /* XXX -- why call quote_list if ifs == 0? we can get away without doing + it now that quote_escapes quotes spaces */ + tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE)) + ? quote_list (list) + : list_quote_escapes (list); + + ret = string_list_internal (tlist, sep); +#if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__) + free (sep); +#endif + return ret; +} + +/* Turn the positional paramters into a string, understanding quoting and + the various subtleties of using the first character of $IFS as the + separator. Calls string_list_dollar_at, string_list_dollar_star, and + string_list as appropriate. */ +char * +string_list_pos_params (pchar, list, quoted) + int pchar; + WORD_LIST *list; + int quoted; +{ + char *ret; + WORD_LIST *tlist; + + if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES)) + { + tlist = quote_list (list); + word_list_remove_quoted_nulls (tlist); + ret = string_list_dollar_star (tlist); + } + else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT)) + { + tlist = quote_list (list); + word_list_remove_quoted_nulls (tlist); + ret = string_list (tlist); + } + else if (pchar == '*') + { + /* Even when unquoted, string_list_dollar_star does the right thing + making sure that the first character of $IFS is used as the + separator. */ + ret = string_list_dollar_star (list); + } + else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + /* We use string_list_dollar_at, but only if the string is quoted, since + that quotes the escapes if it's not, which we don't want. We could + use string_list (the old code did), but that doesn't do the right + thing if the first character of $IFS is not a space. We use + string_list_dollar_star if the string is unquoted so we make sure that + the elements of $@ are separated by the first character of $IFS for + later splitting. */ + ret = string_list_dollar_at (list, quoted); + else if (pchar == '@') + ret = string_list_dollar_star (list); + else + ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list); + + return ret; +} + +/* Return the list of words present in STRING. Separate the string into + words at any of the characters found in SEPARATORS. If QUOTED is + non-zero then word in the list will have its quoted flag set, otherwise + the quoted flag is left as make_word () deemed fit. + + This obeys the P1003.2 word splitting semantics. If `separators' is + exactly , then the splitting algorithm is that of + the Bourne shell, which treats any sequence of characters from `separators' + as a delimiter. If IFS is unset, which results in `separators' being set + to "", no splitting occurs. If separators has some other value, the + following rules are applied (`IFS white space' means zero or more + occurrences of , , or , as long as those characters + are in `separators'): + + 1) IFS white space is ignored at the start and the end of the + string. + 2) Each occurrence of a character in `separators' that is not + IFS white space, along with any adjacent occurrences of + IFS white space delimits a field. + 3) Any nonzero-length sequence of IFS white space delimits a field. + */ + +/* BEWARE! list_string strips null arguments. Don't call it twice and + expect to have "" preserved! */ + +/* This performs word splitting and quoted null character removal on + STRING. */ +#define issep(c) \ + (((separators)[0]) ? ((separators)[1] ? isifs(c) \ + : (c) == (separators)[0]) \ + : 0) + +WORD_LIST * +list_string (string, separators, quoted) + register char *string, *separators; + int quoted; +{ + WORD_LIST *result; + WORD_DESC *t; + char *current_word, *s; + int sindex, sh_style_split, whitesep, xflags; + size_t slen; + + if (!string || !*string) + return ((WORD_LIST *)NULL); + + sh_style_split = separators && separators[0] == ' ' && + separators[1] == '\t' && + separators[2] == '\n' && + separators[3] == '\0'; + for (xflags = 0, s = ifs_value; s && *s; s++) + { + if (*s == CTLESC) xflags |= SX_NOCTLESC; + else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL; + } + + slen = 0; + /* Remove sequences of whitespace at the beginning of STRING, as + long as those characters appear in IFS. Do not do this if + STRING is quoted or if there are no separator characters. */ + if (!quoted || !separators || !*separators) + { + for (s = string; *s && spctabnl (*s) && issep (*s); s++); + + if (!*s) + return ((WORD_LIST *)NULL); + + string = s; + } + + /* OK, now STRING points to a word that does not begin with white space. + The splitting algorithm is: + extract a word, stopping at a separator + skip sequences of spc, tab, or nl as long as they are separators + This obeys the field splitting rules in Posix.2. */ + slen = (MB_CUR_MAX > 1) ? strlen (string) : 1; + for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; ) + { + /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim + unless multibyte chars are possible. */ + current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags); + if (current_word == 0) + break; + + /* If we have a quoted empty string, add a quoted null argument. We + want to preserve the quoted null character iff this is a quoted + empty string; otherwise the quoted null characters are removed + below. */ + if (QUOTED_NULL (current_word)) + { + t = alloc_word_desc (); + t->word = make_quoted_char ('\0'); + t->flags |= W_QUOTED|W_HASQUOTEDNULL; + result = make_word_list (t, result); + } + else if (current_word[0] != '\0') + { + /* If we have something, then add it regardless. However, + perform quoted null character removal on the current word. */ + remove_quoted_nulls (current_word); + result = add_string_to_list (current_word, result); + result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */ + if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) + result->word->flags |= W_QUOTED; + } + + /* If we're not doing sequences of separators in the traditional + Bourne shell style, then add a quoted null argument. */ + else if (!sh_style_split && !spctabnl (string[sindex])) + { + t = alloc_word_desc (); + t->word = make_quoted_char ('\0'); + t->flags |= W_QUOTED|W_HASQUOTEDNULL; + result = make_word_list (t, result); + } + + free (current_word); + + /* Note whether or not the separator is IFS whitespace, used later. */ + whitesep = string[sindex] && spctabnl (string[sindex]); + + /* Move past the current separator character. */ + if (string[sindex]) + { + DECLARE_MBSTATE; + ADVANCE_CHAR (string, slen, sindex); + } + + /* Now skip sequences of space, tab, or newline characters if they are + in the list of separators. */ + while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex])) + sindex++; + + /* If the first separator was IFS whitespace and the current character + is a non-whitespace IFS character, it should be part of the current + field delimiter, not a separate delimiter that would result in an + empty field. Look at POSIX.2, 3.6.5, (3)(b). */ + if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex])) + { + sindex++; + /* An IFS character that is not IFS white space, along with any + adjacent IFS white space, shall delimit a field. (SUSv3) */ + while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex])) + sindex++; + } + } + return (REVERSE_LIST (result, WORD_LIST *)); +} + +/* Parse a single word from STRING, using SEPARATORS to separate fields. + ENDPTR is set to the first character after the word. This is used by + the `read' builtin. This is never called with SEPARATORS != $IFS; + it should be simplified. + + XXX - this function is very similar to list_string; they should be + combined - XXX */ +char * +get_word_from_string (stringp, separators, endptr) + char **stringp, *separators, **endptr; +{ + register char *s; + char *current_word; + int sindex, sh_style_split, whitesep, xflags; + size_t slen; + + if (!stringp || !*stringp || !**stringp) + return ((char *)NULL); + + sh_style_split = separators && separators[0] == ' ' && + separators[1] == '\t' && + separators[2] == '\n' && + separators[3] == '\0'; + for (xflags = 0, s = ifs_value; s && *s; s++) + { + if (*s == CTLESC) xflags |= SX_NOCTLESC; + if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL; + } + + s = *stringp; + slen = 0; + + /* Remove sequences of whitespace at the beginning of STRING, as + long as those characters appear in IFS. */ + if (sh_style_split || !separators || !*separators) + { + for (; *s && spctabnl (*s) && isifs (*s); s++); + + /* If the string is nothing but whitespace, update it and return. */ + if (!*s) + { + *stringp = s; + if (endptr) + *endptr = s; + return ((char *)NULL); + } + } + + /* OK, S points to a word that does not begin with white space. + Now extract a word, stopping at a separator, save a pointer to + the first character after the word, then skip sequences of spc, + tab, or nl as long as they are separators. + + This obeys the field splitting rules in Posix.2. */ + sindex = 0; + /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim + unless multibyte chars are possible. */ + slen = (MB_CUR_MAX > 1) ? strlen (s) : 1; + current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags); + + /* Set ENDPTR to the first character after the end of the word. */ + if (endptr) + *endptr = s + sindex; + + /* Note whether or not the separator is IFS whitespace, used later. */ + whitesep = s[sindex] && spctabnl (s[sindex]); + + /* Move past the current separator character. */ + if (s[sindex]) + { + DECLARE_MBSTATE; + ADVANCE_CHAR (s, slen, sindex); + } + + /* Now skip sequences of space, tab, or newline characters if they are + in the list of separators. */ + while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex])) + sindex++; + + /* If the first separator was IFS whitespace and the current character is + a non-whitespace IFS character, it should be part of the current field + delimiter, not a separate delimiter that would result in an empty field. + Look at POSIX.2, 3.6.5, (3)(b). */ + if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex])) + { + sindex++; + /* An IFS character that is not IFS white space, along with any adjacent + IFS white space, shall delimit a field. */ + while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex])) + sindex++; + } + + /* Update STRING to point to the next field. */ + *stringp = s + sindex; + return (current_word); +} + +/* Remove IFS white space at the end of STRING. Start at the end + of the string and walk backwards until the beginning of the string + or we find a character that's not IFS white space and not CTLESC. + Only let CTLESC escape a white space character if SAW_ESCAPE is + non-zero. */ +char * +strip_trailing_ifs_whitespace (string, separators, saw_escape) + char *string, *separators; + int saw_escape; +{ + char *s; + + s = string + STRLEN (string) - 1; + while (s > string && ((spctabnl (*s) && isifs (*s)) || + (saw_escape && *s == CTLESC && spctabnl (s[1])))) + s--; + *++s = '\0'; + return string; +} + +#if 0 +/* UNUSED */ +/* Split STRING into words at whitespace. Obeys shell-style quoting with + backslashes, single and double quotes. */ +WORD_LIST * +list_string_with_quotes (string) + char *string; +{ + WORD_LIST *list; + char *token, *s; + size_t s_len; + int c, i, tokstart, len; + + for (s = string; s && *s && spctabnl (*s); s++) + ; + if (s == 0 || *s == 0) + return ((WORD_LIST *)NULL); + + s_len = strlen (s); + tokstart = i = 0; + list = (WORD_LIST *)NULL; + while (1) + { + c = s[i]; + if (c == '\\') + { + i++; + if (s[i]) + i++; + } + else if (c == '\'') + i = skip_single_quoted (s, s_len, ++i); + else if (c == '"') + i = skip_double_quoted (s, s_len, ++i); + else if (c == 0 || spctabnl (c)) + { + /* We have found the end of a token. Make a word out of it and + add it to the word list. */ + token = substring (s, tokstart, i); + list = add_string_to_list (token, list); + free (token); + while (spctabnl (s[i])) + i++; + if (s[i]) + tokstart = i; + else + break; + } + else + i++; /* normal character */ + } + return (REVERSE_LIST (list, WORD_LIST *)); +} +#endif + +/********************************************************/ +/* */ +/* Functions to perform assignment statements */ +/* */ +/********************************************************/ + +#if defined (ARRAY_VARS) +static SHELL_VAR * +do_compound_assignment (name, value, flags) + char *name, *value; + int flags; +{ + SHELL_VAR *v; + int mklocal, mkassoc; + WORD_LIST *list; + + mklocal = flags & ASS_MKLOCAL; + mkassoc = flags & ASS_MKASSOC; + + if (mklocal && variable_context) + { + v = find_variable (name); + list = expand_compound_array_assignment (v, value, flags); + if (mkassoc) + v = make_local_assoc_variable (name); + else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context) + v = make_local_array_variable (name, 0); + assign_compound_array_list (v, list, flags); + } + else + v = assign_array_from_string (name, value, flags); + + return (v); +} +#endif + +/* Given STRING, an assignment string, get the value of the right side + of the `=', and bind it to the left side. If EXPAND is true, then + perform parameter expansion, command substitution, and arithmetic + expansion on the right-hand side. Perform tilde expansion in any + case. Do not perform word splitting on the result of expansion. */ +static int +do_assignment_internal (word, expand) + const WORD_DESC *word; + int expand; +{ + int offset, appendop, assign_list, aflags, retval; + char *name, *value, *temp; + SHELL_VAR *entry; +#if defined (ARRAY_VARS) + char *t; + int ni; +#endif + const char *string; + + if (word == 0 || word->word == 0) + return 0; + + appendop = assign_list = aflags = 0; + string = word->word; + offset = assignment (string, 0); + name = savestring (string); + value = (char *)NULL; + + if (name[offset] == '=') + { + if (name[offset - 1] == '+') + { + appendop = 1; + name[offset - 1] = '\0'; + } + + name[offset] = 0; /* might need this set later */ + temp = name + offset + 1; + +#if defined (ARRAY_VARS) + if (expand && (word->flags & W_COMPASSIGN)) + { + assign_list = ni = 1; + value = extract_array_assignment_list (temp, &ni); + } + else +#endif + if (expand && temp[0]) + value = expand_string_if_necessary (temp, 0, expand_string_assignment); + else + value = savestring (temp); + } + + if (value == 0) + { + value = (char *)xmalloc (1); + value[0] = '\0'; + } + + if (echo_command_at_execute) + { + if (appendop) + name[offset - 1] = '+'; + xtrace_print_assignment (name, value, assign_list, 1); + if (appendop) + name[offset - 1] = '\0'; + } + +#define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0) + + if (appendop) + aflags |= ASS_APPEND; + +#if defined (ARRAY_VARS) + if (t = mbschr (name, '[')) /*]*/ + { + if (assign_list) + { + report_error (_("%s: cannot assign list to array member"), name); + ASSIGN_RETURN (0); + } + entry = assign_array_element (name, value, aflags); + if (entry == 0) + ASSIGN_RETURN (0); + } + else if (assign_list) + { + if (word->flags & W_ASSIGNARG) + aflags |= ASS_MKLOCAL; + if (word->flags & W_ASSIGNASSOC) + aflags |= ASS_MKASSOC; + entry = do_compound_assignment (name, value, aflags); + } + else +#endif /* ARRAY_VARS */ + entry = bind_variable (name, value, aflags); + + stupidly_hack_special_variables (name); + +#if 1 + /* Return 1 if the assignment seems to have been performed correctly. */ + if (entry == 0 || readonly_p (entry)) + retval = 0; /* assignment failure */ + else if (noassign_p (entry)) + { + last_command_exit_value = EXECUTION_FAILURE; + retval = 1; /* error status, but not assignment failure */ + } + else + retval = 1; + + if (entry && retval != 0 && noassign_p (entry) == 0) + VUNSETATTR (entry, att_invisible); + + ASSIGN_RETURN (retval); +#else + if (entry) + VUNSETATTR (entry, att_invisible); + + ASSIGN_RETURN (entry ? ((readonly_p (entry) == 0) && noassign_p (entry) == 0) : 0); +#endif +} + +/* Perform the assignment statement in STRING, and expand the + right side by doing tilde, command and parameter expansion. */ +int +do_assignment (string) + char *string; +{ + WORD_DESC td; + + td.flags = W_ASSIGNMENT; + td.word = string; + + return do_assignment_internal (&td, 1); +} + +int +do_word_assignment (word, flags) + WORD_DESC *word; + int flags; +{ + return do_assignment_internal (word, 1); +} + +/* Given STRING, an assignment string, get the value of the right side + of the `=', and bind it to the left side. Do not perform any word + expansions on the right hand side. */ +int +do_assignment_no_expand (string) + char *string; +{ + WORD_DESC td; + + td.flags = W_ASSIGNMENT; + td.word = string; + + return (do_assignment_internal (&td, 0)); +} + +/*************************************************** + * * + * Functions to manage the positional parameters * + * * + ***************************************************/ + +/* Return the word list that corresponds to `$*'. */ +WORD_LIST * +list_rest_of_args () +{ + register WORD_LIST *list, *args; + int i; + + /* Break out of the loop as soon as one of the dollar variables is null. */ + for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++) + list = make_word_list (make_bare_word (dollar_vars[i]), list); + + for (args = rest_of_args; args; args = args->next) + list = make_word_list (make_bare_word (args->word->word), list); + + return (REVERSE_LIST (list, WORD_LIST *)); +} + +int +number_of_args () +{ + register WORD_LIST *list; + int n; + + for (n = 0; n < 9 && dollar_vars[n+1]; n++) + ; + for (list = rest_of_args; list; list = list->next) + n++; + return n; +} + +/* Return the value of a positional parameter. This handles values > 10. */ +char * +get_dollar_var_value (ind) + intmax_t ind; +{ + char *temp; + WORD_LIST *p; + + if (ind < 10) + temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL; + else /* We want something like ${11} */ + { + ind -= 10; + for (p = rest_of_args; p && ind--; p = p->next) + ; + temp = p ? savestring (p->word->word) : (char *)NULL; + } + return (temp); +} + +/* Make a single large string out of the dollar digit variables, + and the rest_of_args. If DOLLAR_STAR is 1, then obey the special + case of "$*" with respect to IFS. */ +char * +string_rest_of_args (dollar_star) + int dollar_star; +{ + register WORD_LIST *list; + char *string; + + list = list_rest_of_args (); + string = dollar_star ? string_list_dollar_star (list) : string_list (list); + dispose_words (list); + return (string); +} + +/* Return a string containing the positional parameters from START to + END, inclusive. If STRING[0] == '*', we obey the rules for $*, + which only makes a difference if QUOTED is non-zero. If QUOTED includes + Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise + no quoting chars are added. */ +static char * +pos_params (string, start, end, quoted) + char *string; + int start, end, quoted; +{ + WORD_LIST *save, *params, *h, *t; + char *ret; + int i; + + /* see if we can short-circuit. if start == end, we want 0 parameters. */ + if (start == end) + return ((char *)NULL); + + save = params = list_rest_of_args (); + if (save == 0) + return ((char *)NULL); + + if (start == 0) /* handle ${@:0[:x]} specially */ + { + t = make_word_list (make_word (dollar_vars[0]), params); + save = params = t; + } + + for (i = start ? 1 : 0; params && i < start; i++) + params = params->next; + if (params == 0) + return ((char *)NULL); + for (h = t = params; params && i < end; i++) + { + t = params; + params = params->next; + } + + t->next = (WORD_LIST *)NULL; + + ret = string_list_pos_params (string[0], h, quoted); + + if (t != params) + t->next = params; + + dispose_words (save); + return (ret); +} + +/******************************************************************/ +/* */ +/* Functions to expand strings to strings or WORD_LISTs */ +/* */ +/******************************************************************/ + +#if defined (PROCESS_SUBSTITUTION) +#define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~') +#else +#define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~') +#endif + +/* If there are any characters in STRING that require full expansion, + then call FUNC to expand STRING; otherwise just perform quote + removal if necessary. This returns a new string. */ +static char * +expand_string_if_necessary (string, quoted, func) + char *string; + int quoted; + EXPFUNC *func; +{ + WORD_LIST *list; + size_t slen; + int i, saw_quote; + char *ret; + DECLARE_MBSTATE; + + /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */ + slen = (MB_CUR_MAX > 1) ? strlen (string) : 0; + i = saw_quote = 0; + while (string[i]) + { + if (EXP_CHAR (string[i])) + break; + else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"') + saw_quote = 1; + ADVANCE_CHAR (string, slen, i); + } + + if (string[i]) + { + list = (*func) (string, quoted); + if (list) + { + ret = string_list (list); + dispose_words (list); + } + else + ret = (char *)NULL; + } + else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) + ret = string_quote_removal (string, quoted); + else + ret = savestring (string); + + return ret; +} + +static inline char * +expand_string_to_string_internal (string, quoted, func) + char *string; + int quoted; + EXPFUNC *func; +{ + WORD_LIST *list; + char *ret; + + if (string == 0 || *string == '\0') + return ((char *)NULL); + + list = (*func) (string, quoted); + if (list) + { + ret = string_list (list); + dispose_words (list); + } + else + ret = (char *)NULL; + + return (ret); +} + +char * +expand_string_to_string (string, quoted) + char *string; + int quoted; +{ + return (expand_string_to_string_internal (string, quoted, expand_string)); +} + +char * +expand_string_unsplit_to_string (string, quoted) + char *string; + int quoted; +{ + return (expand_string_to_string_internal (string, quoted, expand_string_unsplit)); +} + +char * +expand_assignment_string_to_string (string, quoted) + char *string; + int quoted; +{ + return (expand_string_to_string_internal (string, quoted, expand_string_assignment)); +} + +char * +expand_arith_string (string, quoted) + char *string; + int quoted; +{ + return (expand_string_if_necessary (string, quoted, expand_string)); +} + +#if defined (COND_COMMAND) +/* Just remove backslashes in STRING. Returns a new string. */ +char * +remove_backslashes (string) + char *string; +{ + char *r, *ret, *s; + + r = ret = (char *)xmalloc (strlen (string) + 1); + for (s = string; s && *s; ) + { + if (*s == '\\') + s++; + if (*s == 0) + break; + *r++ = *s++; + } + *r = '\0'; + return ret; +} + +/* This needs better error handling. */ +/* Expand W for use as an argument to a unary or binary operator in a + [[...]] expression. If SPECIAL is 1, this is the rhs argument + to the != or == operator, and should be treated as a pattern. In + this case, we quote the string specially for the globbing code. If + SPECIAL is 2, this is an rhs argument for the =~ operator, and should + be quoted appropriately for regcomp/regexec. The caller is responsible + for removing the backslashes if the unquoted word is needed later. */ +char * +cond_expand_word (w, special) + WORD_DESC *w; + int special; +{ + char *r, *p; + WORD_LIST *l; + int qflags; + + if (w->word == 0 || w->word[0] == '\0') + return ((char *)NULL); + + w->flags |= W_NOSPLIT2; + l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0); + if (l) + { + if (special == 0) + { + dequote_list (l); + r = string_list (l); + } + else + { + qflags = QGLOB_CVTNULL; + if (special == 2) + qflags |= QGLOB_REGEXP; + p = string_list (l); + r = quote_string_for_globbing (p, qflags); + free (p); + } + dispose_words (l); + } + else + r = (char *)NULL; + + return r; +} +#endif + +/* Call expand_word_internal to expand W and handle error returns. + A convenience function for functions that don't want to handle + any errors or free any memory before aborting. */ +static WORD_LIST * +call_expand_word_internal (w, q, i, c, e) + WORD_DESC *w; + int q, i, *c, *e; +{ + WORD_LIST *result; + + result = expand_word_internal (w, q, i, c, e); + if (result == &expand_word_error || result == &expand_word_fatal) + { + /* By convention, each time this error is returned, w->word has + already been freed (it sometimes may not be in the fatal case, + but that doesn't result in a memory leak because we're going + to exit in most cases). */ + w->word = (char *)NULL; + last_command_exit_value = EXECUTION_FAILURE; + exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF); + /* NOTREACHED */ + } + else + return (result); +} + +/* Perform parameter expansion, command substitution, and arithmetic + expansion on STRING, as if it were a word. Leave the result quoted. + Since this does not perform word splitting, it leaves quoted nulls + in the result. */ +static WORD_LIST * +expand_string_internal (string, quoted) + char *string; + int quoted; +{ + WORD_DESC td; + WORD_LIST *tresult; + + if (string == 0 || *string == 0) + return ((WORD_LIST *)NULL); + + td.flags = 0; + td.word = savestring (string); + + tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); + + FREE (td.word); + return (tresult); +} + +/* Expand STRING by performing parameter expansion, command substitution, + and arithmetic expansion. Dequote the resulting WORD_LIST before + returning it, but do not perform word splitting. The call to + remove_quoted_nulls () is in here because word splitting normally + takes care of quote removal. */ +WORD_LIST * +expand_string_unsplit (string, quoted) + char *string; + int quoted; +{ + WORD_LIST *value; + + if (string == 0 || *string == '\0') + return ((WORD_LIST *)NULL); + + expand_no_split_dollar_star = 1; + value = expand_string_internal (string, quoted); + expand_no_split_dollar_star = 0; + + if (value) + { + if (value->word) + { + remove_quoted_nulls (value->word->word); + value->word->flags &= ~W_HASQUOTEDNULL; + } + dequote_list (value); + } + return (value); +} + +/* Expand the rhs of an assignment statement */ +WORD_LIST * +expand_string_assignment (string, quoted) + char *string; + int quoted; +{ + WORD_DESC td; + WORD_LIST *value; + + if (string == 0 || *string == '\0') + return ((WORD_LIST *)NULL); + + expand_no_split_dollar_star = 1; + + td.flags = W_ASSIGNRHS; + td.word = savestring (string); + value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); + FREE (td.word); + + expand_no_split_dollar_star = 0; + + if (value) + { + if (value->word) + { + remove_quoted_nulls (value->word->word); + value->word->flags &= ~W_HASQUOTEDNULL; + } + dequote_list (value); + } + return (value); +} + + +/* Expand one of the PS? prompt strings. This is a sort of combination of + expand_string_unsplit and expand_string_internal, but returns the + passed string when an error occurs. Might want to trap other calls + to jump_to_top_level here so we don't endlessly loop. */ +WORD_LIST * +expand_prompt_string (string, quoted, wflags) + char *string; + int quoted; + int wflags; +{ + WORD_LIST *value; + WORD_DESC td; + + if (string == 0 || *string == 0) + return ((WORD_LIST *)NULL); + + td.flags = wflags; + td.word = savestring (string); + + no_longjmp_on_fatal_error = 1; + value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); + no_longjmp_on_fatal_error = 0; + + if (value == &expand_word_error || value == &expand_word_fatal) + { + value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL); + return value; + } + FREE (td.word); + if (value) + { + if (value->word) + { + remove_quoted_nulls (value->word->word); + value->word->flags &= ~W_HASQUOTEDNULL; + } + dequote_list (value); + } + return (value); +} + +/* Expand STRING just as if you were expanding a word, but do not dequote + the resultant WORD_LIST. This is called only from within this file, + and is used to correctly preserve quoted characters when expanding + things like ${1+"$@"}. This does parameter expansion, command + substitution, arithmetic expansion, and word splitting. */ +static WORD_LIST * +expand_string_leave_quoted (string, quoted) + char *string; + int quoted; +{ + WORD_LIST *tlist; + WORD_LIST *tresult; + + if (string == 0 || *string == '\0') + return ((WORD_LIST *)NULL); + + tlist = expand_string_internal (string, quoted); + + if (tlist) + { + tresult = word_list_split (tlist); + dispose_words (tlist); + return (tresult); + } + return ((WORD_LIST *)NULL); +} + +/* This does not perform word splitting or dequote the WORD_LIST + it returns. */ +static WORD_LIST * +expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at) + char *string; + int quoted, *dollar_at_p, *has_dollar_at; +{ + WORD_DESC td; + WORD_LIST *tresult; + + if (string == 0 || *string == '\0') + return (WORD_LIST *)NULL; + + td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */ + td.word = string; + tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at); + return (tresult); +} + +/* Expand STRING just as if you were expanding a word. This also returns + a list of words. Note that filename globbing is *NOT* done for word + or string expansion, just when the shell is expanding a command. This + does parameter expansion, command substitution, arithmetic expansion, + and word splitting. Dequote the resultant WORD_LIST before returning. */ +WORD_LIST * +expand_string (string, quoted) + char *string; + int quoted; +{ + WORD_LIST *result; + + if (string == 0 || *string == '\0') + return ((WORD_LIST *)NULL); + + result = expand_string_leave_quoted (string, quoted); + return (result ? dequote_list (result) : result); +} + +/*************************************************** + * * + * Functions to handle quoting chars * + * * + ***************************************************/ + +/* Conventions: + + A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string. + The parser passes CTLNUL as CTLESC CTLNUL. */ + +/* Quote escape characters in string s, but no other characters. This is + used to protect CTLESC and CTLNUL in variable values from the rest of + the word expansion process after the variable is expanded (word splitting + and filename generation). If IFS is null, we quote spaces as well, just + in case we split on spaces later (in the case of unquoted $@, we will + eventually attempt to split the entire word on spaces). Corresponding + code exists in dequote_escapes. Even if we don't end up splitting on + spaces, quoting spaces is not a problem. This should never be called on + a string that is quoted with single or double quotes or part of a here + document (effectively double-quoted). */ +char * +quote_escapes (string) + char *string; +{ + register char *s, *t; + size_t slen; + char *result, *send; + int quote_spaces, skip_ctlesc, skip_ctlnul; + DECLARE_MBSTATE; + + slen = strlen (string); + send = string + slen; + + quote_spaces = (ifs_value && *ifs_value == 0); + + for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++) + skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL; + + t = result = (char *)xmalloc ((slen * 2) + 1); + s = string; + + while (*s) + { + if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' ')) + *t++ = CTLESC; + COPY_CHAR_P (t, s, send); + } + *t = '\0'; + return (result); +} + +static WORD_LIST * +list_quote_escapes (list) + WORD_LIST *list; +{ + register WORD_LIST *w; + char *t; + + for (w = list; w; w = w->next) + { + t = w->word->word; + w->word->word = quote_escapes (t); + free (t); + } + return list; +} + +/* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL. + + The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL. + This is necessary to make unquoted CTLESC and CTLNUL characters in the + data stream pass through properly. + + We need to remove doubled CTLESC characters inside quoted strings before + quoting the entire string, so we do not double the number of CTLESC + characters. + + Also used by parts of the pattern substitution code. */ +char * +dequote_escapes (string) + char *string; +{ + register char *s, *t, *s1; + size_t slen; + char *result, *send; + int quote_spaces; + DECLARE_MBSTATE; + + if (string == 0) + return string; + + slen = strlen (string); + send = string + slen; + + t = result = (char *)xmalloc (slen + 1); + + if (strchr (string, CTLESC) == 0) + return (strcpy (result, string)); + + quote_spaces = (ifs_value && *ifs_value == 0); + + s = string; + while (*s) + { + if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' '))) + { + s++; + if (*s == '\0') + break; + } + COPY_CHAR_P (t, s, send); + } + *t = '\0'; + return result; +} + +/* Return a new string with the quoted representation of character C. + This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be + set in any resultant WORD_DESC where this value is the word. */ +static char * +make_quoted_char (c) + int c; +{ + char *temp; + + temp = (char *)xmalloc (3); + if (c == 0) + { + temp[0] = CTLNUL; + temp[1] = '\0'; + } + else + { + temp[0] = CTLESC; + temp[1] = c; + temp[2] = '\0'; + } + return (temp); +} + +/* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so + the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where + this value is the word. */ +char * +quote_string (string) + char *string; +{ + register char *t; + size_t slen; + char *result, *send; + + if (*string == 0) + { + result = (char *)xmalloc (2); + result[0] = CTLNUL; + result[1] = '\0'; + } + else + { + DECLARE_MBSTATE; + + slen = strlen (string); + send = string + slen; + + result = (char *)xmalloc ((slen * 2) + 1); + + for (t = result; string < send; ) + { + *t++ = CTLESC; + COPY_CHAR_P (t, string, send); + } + *t = '\0'; + } + return (result); +} + +/* De-quote quoted characters in STRING. */ +char * +dequote_string (string) + char *string; +{ + register char *s, *t; + size_t slen; + char *result, *send; + DECLARE_MBSTATE; + + slen = strlen (string); + + t = result = (char *)xmalloc (slen + 1); + + if (QUOTED_NULL (string)) + { + result[0] = '\0'; + return (result); + } + + /* If no character in the string can be quoted, don't bother examining + each character. Just return a copy of the string passed to us. */ + if (strchr (string, CTLESC) == NULL) + return (strcpy (result, string)); + + send = string + slen; + s = string; + while (*s) + { + if (*s == CTLESC) + { + s++; + if (*s == '\0') + break; + } + COPY_CHAR_P (t, s, send); + } + + *t = '\0'; + return (result); +} + +/* Quote the entire WORD_LIST list. */ +static WORD_LIST * +quote_list (list) + WORD_LIST *list; +{ + register WORD_LIST *w; + char *t; + + for (w = list; w; w = w->next) + { + t = w->word->word; + w->word->word = quote_string (t); + if (*t == 0) + w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */ + w->word->flags |= W_QUOTED; + free (t); + } + return list; +} + +/* De-quote quoted characters in each word in LIST. */ +WORD_LIST * +dequote_list (list) + WORD_LIST *list; +{ + register char *s; + register WORD_LIST *tlist; + + for (tlist = list; tlist; tlist = tlist->next) + { + s = dequote_string (tlist->word->word); + if (QUOTED_NULL (tlist->word->word)) + tlist->word->flags &= ~W_HASQUOTEDNULL; + free (tlist->word->word); + tlist->word->word = s; + } + return list; +} + +/* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed + string. */ +char * +remove_quoted_escapes (string) + char *string; +{ + char *t; + + if (string) + { + t = dequote_escapes (string); + strcpy (string, t); + free (t); + } + + return (string); +} + +/* Perform quoted null character removal on STRING. We don't allow any + quoted null characters in the middle or at the ends of strings because + of how expand_word_internal works. remove_quoted_nulls () turns + STRING into an empty string iff it only consists of a quoted null, + and removes all unquoted CTLNUL characters. */ +char * +remove_quoted_nulls (string) + char *string; +{ + register size_t slen; + register int i, j, prev_i; + DECLARE_MBSTATE; + + if (strchr (string, CTLNUL) == 0) /* XXX */ + return string; /* XXX */ + + slen = strlen (string); + i = j = 0; + + while (i < slen) + { + if (string[i] == CTLESC) + { + /* Old code had j++, but we cannot assume that i == j at this + point -- what if a CTLNUL has already been removed from the + string? We don't want to drop the CTLESC or recopy characters + that we've already copied down. */ + i++; string[j++] = CTLESC; + if (i == slen) + break; + } + else if (string[i] == CTLNUL) + { + i++; + continue; + } + + prev_i = i; + ADVANCE_CHAR (string, slen, i); + if (j < prev_i) + { + do string[j++] = string[prev_i++]; while (prev_i < i); + } + else + j = i; + } + string[j] = '\0'; + + return (string); +} + +/* Perform quoted null character removal on each element of LIST. + This modifies LIST. */ +void +word_list_remove_quoted_nulls (list) + WORD_LIST *list; +{ + register WORD_LIST *t; + + for (t = list; t; t = t->next) + { + remove_quoted_nulls (t->word->word); + t->word->flags &= ~W_HASQUOTEDNULL; + } +} + +/* **************************************************************** */ +/* */ +/* Functions for Matching and Removing Patterns */ +/* */ +/* **************************************************************** */ + +#if defined (HANDLE_MULTIBYTE) +#if 0 /* Currently unused */ +static unsigned char * +mb_getcharlens (string, len) + char *string; + int len; +{ + int i, offset, last; + unsigned char *ret; + char *p; + DECLARE_MBSTATE; + + i = offset = 0; + last = 0; + ret = (unsigned char *)xmalloc (len); + memset (ret, 0, len); + while (string[last]) + { + ADVANCE_CHAR (string, len, offset); + ret[last] = offset - last; + last = offset; + } + return ret; +} +#endif +#endif + +/* Remove the portion of PARAM matched by PATTERN according to OP, where OP + can have one of 4 values: + RP_LONG_LEFT remove longest matching portion at start of PARAM + RP_SHORT_LEFT remove shortest matching portion at start of PARAM + RP_LONG_RIGHT remove longest matching portion at end of PARAM + RP_SHORT_RIGHT remove shortest matching portion at end of PARAM +*/ + +#define RP_LONG_LEFT 1 +#define RP_SHORT_LEFT 2 +#define RP_LONG_RIGHT 3 +#define RP_SHORT_RIGHT 4 + +/* Returns its first argument if nothing matched; new memory otherwise */ +static char * +remove_upattern (param, pattern, op) + char *param, *pattern; + int op; +{ + register int len; + register char *end; + register char *p, *ret, c; + + len = STRLEN (param); + end = param + len; + + switch (op) + { + case RP_LONG_LEFT: /* remove longest match at start */ + for (p = end; p >= param; p--) + { + c = *p; *p = '\0'; + if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + *p = c; + return (savestring (p)); + } + *p = c; + + } + break; + + case RP_SHORT_LEFT: /* remove shortest match at start */ + for (p = param; p <= end; p++) + { + c = *p; *p = '\0'; + if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + *p = c; + return (savestring (p)); + } + *p = c; + } + break; + + case RP_LONG_RIGHT: /* remove longest match at end */ + for (p = param; p <= end; p++) + { + if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + c = *p; *p = '\0'; + ret = savestring (param); + *p = c; + return (ret); + } + } + break; + + case RP_SHORT_RIGHT: /* remove shortest match at end */ + for (p = end; p >= param; p--) + { + if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + c = *p; *p = '\0'; + ret = savestring (param); + *p = c; + return (ret); + } + } + break; + } + + return (param); /* no match, return original string */ +} + +#if defined (HANDLE_MULTIBYTE) +/* Returns its first argument if nothing matched; new memory otherwise */ +static wchar_t * +remove_wpattern (wparam, wstrlen, wpattern, op) + wchar_t *wparam; + size_t wstrlen; + wchar_t *wpattern; + int op; +{ + wchar_t wc, *ret; + int n; + + switch (op) + { + case RP_LONG_LEFT: /* remove longest match at start */ + for (n = wstrlen; n >= 0; n--) + { + wc = wparam[n]; wparam[n] = L'\0'; + if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + wparam[n] = wc; + return (wcsdup (wparam + n)); + } + wparam[n] = wc; + } + break; + + case RP_SHORT_LEFT: /* remove shortest match at start */ + for (n = 0; n <= wstrlen; n++) + { + wc = wparam[n]; wparam[n] = L'\0'; + if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + wparam[n] = wc; + return (wcsdup (wparam + n)); + } + wparam[n] = wc; + } + break; + + case RP_LONG_RIGHT: /* remove longest match at end */ + for (n = 0; n <= wstrlen; n++) + { + if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + wc = wparam[n]; wparam[n] = L'\0'; + ret = wcsdup (wparam); + wparam[n] = wc; + return (ret); + } + } + break; + + case RP_SHORT_RIGHT: /* remove shortest match at end */ + for (n = wstrlen; n >= 0; n--) + { + if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH) + { + wc = wparam[n]; wparam[n] = L'\0'; + ret = wcsdup (wparam); + wparam[n] = wc; + return (ret); + } + } + break; + } + + return (wparam); /* no match, return original string */ +} +#endif /* HANDLE_MULTIBYTE */ + +static char * +remove_pattern (param, pattern, op) + char *param, *pattern; + int op; +{ + char *xret; + + if (param == NULL) + return (param); + if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */ + return (savestring (param)); + +#if defined (HANDLE_MULTIBYTE) + if (MB_CUR_MAX > 1) + { + wchar_t *ret, *oret; + size_t n; + wchar_t *wparam, *wpattern; + mbstate_t ps; + + n = xdupmbstowcs (&wpattern, NULL, pattern); + if (n == (size_t)-1) + { + xret = remove_upattern (param, pattern, op); + return ((xret == param) ? savestring (param) : xret); + } + n = xdupmbstowcs (&wparam, NULL, param); + if (n == (size_t)-1) + { + free (wpattern); + xret = remove_upattern (param, pattern, op); + return ((xret == param) ? savestring (param) : xret); + } + oret = ret = remove_wpattern (wparam, n, wpattern, op); + /* Don't bother to convert wparam back to multibyte string if nothing + matched; just return copy of original string */ + if (ret == wparam) + { + free (wparam); + free (wpattern); + return (savestring (param)); + } + + free (wparam); + free (wpattern); + + n = strlen (param); + xret = (char *)xmalloc (n + 1); + memset (&ps, '\0', sizeof (mbstate_t)); + n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps); + xret[n] = '\0'; /* just to make sure */ + free (oret); + return xret; + } + else +#endif + { + xret = remove_upattern (param, pattern, op); + return ((xret == param) ? savestring (param) : xret); + } +} + +/* Match PAT anywhere in STRING and return the match boundaries. + This returns 1 in case of a successful match, 0 otherwise. SP + and EP are pointers into the string where the match begins and + ends, respectively. MTYPE controls what kind of match is attempted. + MATCH_BEG and MATCH_END anchor the match at the beginning and end + of the string, respectively. The longest match is returned. */ +static int +match_upattern (string, pat, mtype, sp, ep) + char *string, *pat; + int mtype; + char **sp, **ep; +{ + int c, len, mlen; + register char *p, *p1, *npat; + char *end; + int n1; + + /* If the pattern doesn't match anywhere in the string, go ahead and + short-circuit right away. A minor optimization, saves a bunch of + unnecessary calls to strmatch (up to N calls for a string of N + characters) if the match is unsuccessful. To preserve the semantics + of the substring matches below, we make sure that the pattern has + `*' as first and last character, making a new pattern if necessary. */ + /* XXX - check this later if I ever implement `**' with special meaning, + since this will potentially result in `**' at the beginning or end */ + len = STRLEN (pat); + if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*') + { + p = npat = (char *)xmalloc (len + 3); + p1 = pat; + if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob)) + *p++ = '*'; + while (*p1) + *p++ = *p1++; + if (p1[-1] != '*' || p[-2] == '\\') + *p++ = '*'; + *p = '\0'; + } + else + npat = pat; + c = strmatch (npat, string, FNMATCH_EXTFLAG); + if (npat != pat) + free (npat); + if (c == FNM_NOMATCH) + return (0); + + len = STRLEN (string); + end = string + len; + + mlen = umatchlen (pat, len); + + switch (mtype) + { + case MATCH_ANY: + for (p = string; p <= end; p++) + { + if (match_pattern_char (pat, p)) + { +#if 0 + for (p1 = end; p1 >= p; p1--) +#else + p1 = (mlen == -1) ? end : p + mlen; + /* p1 - p = length of portion of string to be considered + p = current position in string + mlen = number of characters consumed by match (-1 for entire string) + end = end of string + we want to break immediately if the potential match len + is greater than the number of characters remaining in the + string + */ + if (p1 > end) + break; + for ( ; p1 >= p; p1--) +#endif + { + c = *p1; *p1 = '\0'; + if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0) + { + *p1 = c; + *sp = p; + *ep = p1; + return 1; + } + *p1 = c; +#if 1 + /* If MLEN != -1, we have a fixed length pattern. */ + if (mlen != -1) + break; +#endif + } + } + } + + return (0); + + case MATCH_BEG: + if (match_pattern_char (pat, string) == 0) + return (0); + +#if 0 + for (p = end; p >= string; p--) +#else + for (p = (mlen == -1) ? end : string + mlen; p >= string; p--) +#endif + { + c = *p; *p = '\0'; + if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0) + { + *p = c; + *sp = string; + *ep = p; + return 1; + } + *p = c; +#if 1 + /* If MLEN != -1, we have a fixed length pattern. */ + if (mlen != -1) + break; +#endif + } + + return (0); + + case MATCH_END: +#if 0 + for (p = string; p <= end; p++) +#else + for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++) +#endif + { + if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0) + { + *sp = p; + *ep = end; + return 1; + } +#if 1 + /* If MLEN != -1, we have a fixed length pattern. */ + if (mlen != -1) + break; +#endif + } + + return (0); + } + + return (0); +} + +#if defined (HANDLE_MULTIBYTE) +/* Match WPAT anywhere in WSTRING and return the match boundaries. + This returns 1 in case of a successful match, 0 otherwise. Wide + character version. */ +static int +match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep) + wchar_t *wstring; + char **indices; + size_t wstrlen; + wchar_t *wpat; + int mtype; + char **sp, **ep; +{ + wchar_t wc, *wp, *nwpat, *wp1; + size_t len; + int mlen; + int n, n1, n2, simple; + + simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'['); +#if defined (EXTENDED_GLOB) + if (extended_glob) + simple |= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/ +#endif + + /* If the pattern doesn't match anywhere in the string, go ahead and + short-circuit right away. A minor optimization, saves a bunch of + unnecessary calls to strmatch (up to N calls for a string of N + characters) if the match is unsuccessful. To preserve the semantics + of the substring matches below, we make sure that the pattern has + `*' as first and last character, making a new pattern if necessary. */ + len = wcslen (wpat); + if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*') + { + wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t)); + wp1 = wpat; + if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob)) + *wp++ = L'*'; + while (*wp1 != L'\0') + *wp++ = *wp1++; + if (wp1[-1] != L'*' || wp1[-2] == L'\\') + *wp++ = L'*'; + *wp = '\0'; + } + else + nwpat = wpat; + len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG); + if (nwpat != wpat) + free (nwpat); + if (len == FNM_NOMATCH) + return (0); + + mlen = wmatchlen (wpat, wstrlen); + +/* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */ + switch (mtype) + { + case MATCH_ANY: + for (n = 0; n <= wstrlen; n++) + { +#if 1 + n2 = simple ? (*wpat == wstring[n]) : match_pattern_wchar (wpat, wstring + n); +#else + n2 = match_pattern_wchar (wpat, wstring + n); +#endif + if (n2) + { +#if 0 + for (n1 = wstrlen; n1 >= n; n1--) +#else + n1 = (mlen == -1) ? wstrlen : n + mlen; + if (n1 > wstrlen) + break; + + for ( ; n1 >= n; n1--) +#endif + { + wc = wstring[n1]; wstring[n1] = L'\0'; + if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0) + { + wstring[n1] = wc; + *sp = indices[n]; + *ep = indices[n1]; + return 1; + } + wstring[n1] = wc; +#if 1 + /* If MLEN != -1, we have a fixed length pattern. */ + if (mlen != -1) + break; +#endif + } + } + } + + return (0); + + case MATCH_BEG: + if (match_pattern_wchar (wpat, wstring) == 0) + return (0); + +#if 0 + for (n = wstrlen; n >= 0; n--) +#else + for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--) +#endif + { + wc = wstring[n]; wstring[n] = L'\0'; + if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0) + { + wstring[n] = wc; + *sp = indices[0]; + *ep = indices[n]; + return 1; + } + wstring[n] = wc; +#if 1 + /* If MLEN != -1, we have a fixed length pattern. */ + if (mlen != -1) + break; +#endif + } + + return (0); + + case MATCH_END: +#if 0 + for (n = 0; n <= wstrlen; n++) +#else + for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++) +#endif + { + if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0) + { + *sp = indices[n]; + *ep = indices[wstrlen]; + return 1; + } +#if 1 + /* If MLEN != -1, we have a fixed length pattern. */ + if (mlen != -1) + break; +#endif + } + + return (0); + } + + return (0); +} +#endif /* HANDLE_MULTIBYTE */ + +static int +match_pattern (string, pat, mtype, sp, ep) + char *string, *pat; + int mtype; + char **sp, **ep; +{ +#if defined (HANDLE_MULTIBYTE) + int ret; + size_t n; + wchar_t *wstring, *wpat; + char **indices; + size_t slen, plen, mslen, mplen; +#endif + + if (string == 0 || *string == 0 || pat == 0 || *pat == 0) + return (0); + +#if defined (HANDLE_MULTIBYTE) + if (MB_CUR_MAX > 1) + { +#if 0 + slen = STRLEN (string); + mslen = MBSLEN (string); + plen = STRLEN (pat); + mplen = MBSLEN (pat); + if (slen == mslen && plen == mplen) +#else + if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0) +#endif + return (match_upattern (string, pat, mtype, sp, ep)); + + n = xdupmbstowcs (&wpat, NULL, pat); + if (n == (size_t)-1) + return (match_upattern (string, pat, mtype, sp, ep)); + n = xdupmbstowcs (&wstring, &indices, string); + if (n == (size_t)-1) + { + free (wpat); + return (match_upattern (string, pat, mtype, sp, ep)); + } + ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep); + + free (wpat); + free (wstring); + free (indices); + + return (ret); + } + else +#endif + return (match_upattern (string, pat, mtype, sp, ep)); +} + +static int +getpatspec (c, value) + int c; + char *value; +{ + if (c == '#') + return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT); + else /* c == '%' */ + return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT); +} + +/* Posix.2 says that the WORD should be run through tilde expansion, + parameter expansion, command substitution and arithmetic expansion. + This leaves the result quoted, so quote_string_for_globbing () has + to be called to fix it up for strmatch (). If QUOTED is non-zero, + it means that the entire expression was enclosed in double quotes. + This means that quoting characters in the pattern do not make any + special pattern characters quoted. For example, the `*' in the + following retains its special meaning: "${foo#'*'}". */ +static char * +getpattern (value, quoted, expandpat) + char *value; + int quoted, expandpat; +{ + char *pat, *tword; + WORD_LIST *l; +#if 0 + int i; +#endif + /* There is a problem here: how to handle single or double quotes in the + pattern string when the whole expression is between double quotes? + POSIX.2 says that enclosing double quotes do not cause the pattern to + be quoted, but does that leave us a problem with @ and array[@] and their + expansions inside a pattern? */ +#if 0 + if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword) + { + i = 0; + pat = string_extract_double_quoted (tword, &i, 1); + free (tword); + tword = pat; + } +#endif + + /* expand_string_for_rhs () leaves WORD quoted and does not perform + word splitting. */ + l = *value ? expand_string_for_rhs (value, + (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted, + (int *)NULL, (int *)NULL) + : (WORD_LIST *)0; + pat = string_list (l); + dispose_words (l); + if (pat) + { + tword = quote_string_for_globbing (pat, QGLOB_CVTNULL); + free (pat); + pat = tword; + } + return (pat); +} + +#if 0 +/* Handle removing a pattern from a string as a result of ${name%[%]value} + or ${name#[#]value}. */ +static char * +variable_remove_pattern (value, pattern, patspec, quoted) + char *value, *pattern; + int patspec, quoted; +{ + char *tword; + + tword = remove_pattern (value, pattern, patspec); + + return (tword); +} +#endif + +static char * +list_remove_pattern (list, pattern, patspec, itype, quoted) + WORD_LIST *list; + char *pattern; + int patspec, itype, quoted; +{ + WORD_LIST *new, *l; + WORD_DESC *w; + char *tword; + + for (new = (WORD_LIST *)NULL, l = list; l; l = l->next) + { + tword = remove_pattern (l->word->word, pattern, patspec); + w = alloc_word_desc (); + w->word = tword ? tword : savestring (""); + new = make_word_list (w, new); + } + + l = REVERSE_LIST (new, WORD_LIST *); + tword = string_list_pos_params (itype, l, quoted); + dispose_words (l); + + return (tword); +} + +static char * +parameter_list_remove_pattern (itype, pattern, patspec, quoted) + int itype; + char *pattern; + int patspec, quoted; +{ + char *ret; + WORD_LIST *list; + + list = list_rest_of_args (); + if (list == 0) + return ((char *)NULL); + ret = list_remove_pattern (list, pattern, patspec, itype, quoted); + dispose_words (list); + return (ret); +} + +#if defined (ARRAY_VARS) +static char * +array_remove_pattern (var, pattern, patspec, varname, quoted) + SHELL_VAR *var; + char *pattern; + int patspec; + char *varname; /* so we can figure out how it's indexed */ + int quoted; +{ + ARRAY *a; + HASH_TABLE *h; + int itype; + char *ret; + WORD_LIST *list; + SHELL_VAR *v; + + /* compute itype from varname here */ + v = array_variable_part (varname, &ret, 0); + itype = ret[0]; + + a = (v && array_p (v)) ? array_cell (v) : 0; + h = (v && assoc_p (v)) ? assoc_cell (v) : 0; + + list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0); + if (list == 0) + return ((char *)NULL); + ret = list_remove_pattern (list, pattern, patspec, itype, quoted); + dispose_words (list); + + return ret; +} +#endif /* ARRAY_VARS */ + +static char * +parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags) + char *varname, *value; + int ind; + char *patstr; + int rtype, quoted, flags; +{ + int vtype, patspec, starsub; + char *temp1, *val, *pattern; + SHELL_VAR *v; + + if (value == 0) + return ((char *)NULL); + + this_command_name = varname; + + vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); + if (vtype == -1) + return ((char *)NULL); + + starsub = vtype & VT_STARSUB; + vtype &= ~VT_STARSUB; + + patspec = getpatspec (rtype, patstr); + if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT) + patstr++; + + /* Need to pass getpattern newly-allocated memory in case of expansion -- + the expansion code will free the passed string on an error. */ + temp1 = savestring (patstr); + pattern = getpattern (temp1, quoted, 1); + free (temp1); + + temp1 = (char *)NULL; /* shut up gcc */ + switch (vtype) + { + case VT_VARIABLE: + case VT_ARRAYMEMBER: + temp1 = remove_pattern (val, pattern, patspec); + if (vtype == VT_VARIABLE) + FREE (val); + if (temp1) + { + val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + ? quote_string (temp1) + : quote_escapes (temp1); + free (temp1); + temp1 = val; + } + break; +#if defined (ARRAY_VARS) + case VT_ARRAYVAR: + temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted); + if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) + { + val = quote_escapes (temp1); + free (temp1); + temp1 = val; + } + break; +#endif + case VT_POSPARMS: + temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted); + if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) + { + val = quote_escapes (temp1); + free (temp1); + temp1 = val; + } + break; + } + + FREE (pattern); + return temp1; +} + +/******************************************* + * * + * Functions to expand WORD_DESCs * + * * + *******************************************/ + +/* Expand WORD, performing word splitting on the result. This does + parameter expansion, command substitution, arithmetic expansion, + word splitting, and quote removal. */ + +WORD_LIST * +expand_word (word, quoted) + WORD_DESC *word; + int quoted; +{ + WORD_LIST *result, *tresult; + + tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); + result = word_list_split (tresult); + dispose_words (tresult); + return (result ? dequote_list (result) : result); +} + +/* Expand WORD, but do not perform word splitting on the result. This + does parameter expansion, command substitution, arithmetic expansion, + and quote removal. */ +WORD_LIST * +expand_word_unsplit (word, quoted) + WORD_DESC *word; + int quoted; +{ + WORD_LIST *result; + + expand_no_split_dollar_star = 1; +#if defined (HANDLE_MULTIBYTE) + if (ifs_firstc[0] == 0) +#else + if (ifs_firstc == 0) +#endif + word->flags |= W_NOSPLIT; + word->flags |= W_NOSPLIT2; + result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); + expand_no_split_dollar_star = 0; + + return (result ? dequote_list (result) : result); +} + +/* Perform shell expansions on WORD, but do not perform word splitting or + quote removal on the result. Virtually identical to expand_word_unsplit; + could be combined if implementations don't diverge. */ +WORD_LIST * +expand_word_leave_quoted (word, quoted) + WORD_DESC *word; + int quoted; +{ + WORD_LIST *result; + + expand_no_split_dollar_star = 1; +#if defined (HANDLE_MULTIBYTE) + if (ifs_firstc[0] == 0) +#else + if (ifs_firstc == 0) +#endif + word->flags |= W_NOSPLIT; + word->flags |= W_NOSPLIT2; + result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); + expand_no_split_dollar_star = 0; + + return result; +} + +#if defined (PROCESS_SUBSTITUTION) + +/*****************************************************************/ +/* */ +/* Hacking Process Substitution */ +/* */ +/*****************************************************************/ + +#if !defined (HAVE_DEV_FD) +/* Named pipes must be removed explicitly with `unlink'. This keeps a list + of FIFOs the shell has open. unlink_fifo_list will walk the list and + unlink all of them. add_fifo_list adds the name of an open FIFO to the + list. NFIFO is a count of the number of FIFOs in the list. */ +#define FIFO_INCR 20 + +struct temp_fifo { + char *file; + pid_t proc; +}; + +static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL; +static int nfifo; +static int fifo_list_size; + +char * +copy_fifo_list (sizep) + int *sizep; +{ + if (sizep) + *sizep = 0; + return (char *)NULL; +} + +static void +add_fifo_list (pathname) + char *pathname; +{ + if (nfifo >= fifo_list_size - 1) + { + fifo_list_size += FIFO_INCR; + fifo_list = (struct temp_fifo *)xrealloc (fifo_list, + fifo_list_size * sizeof (struct temp_fifo)); + } + + fifo_list[nfifo].file = savestring (pathname); + nfifo++; +} + +void +unlink_fifo (i) + int i; +{ + if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1)) + { + unlink (fifo_list[i].file); + free (fifo_list[i].file); + fifo_list[i].file = (char *)NULL; + fifo_list[i].proc = -1; + } +} + +void +unlink_fifo_list () +{ + int saved, i, j; + + if (nfifo == 0) + return; + + for (i = saved = 0; i < nfifo; i++) + { + if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1)) + { + unlink (fifo_list[i].file); + free (fifo_list[i].file); + fifo_list[i].file = (char *)NULL; + fifo_list[i].proc = -1; + } + else + saved++; + } + + /* If we didn't remove some of the FIFOs, compact the list. */ + if (saved) + { + for (i = j = 0; i < nfifo; i++) + if (fifo_list[i].file) + { + fifo_list[j].file = fifo_list[i].file; + fifo_list[j].proc = fifo_list[i].proc; + j++; + } + nfifo = j; + } + else + nfifo = 0; +} + +/* Take LIST, which is a bitmap denoting active FIFOs in fifo_list + from some point in the past, and close all open FIFOs in fifo_list + that are not marked as active in LIST. If LIST is NULL, close + everything in fifo_list. LSIZE is the number of elements in LIST, in + case it's larger than fifo_list_size (size of fifo_list). */ +void +close_new_fifos (list, lsize) + char *list; + int lsize; +{ + int i; + + if (list == 0) + { + unlink_fifo_list (); + return; + } + + for (i = 0; i < lsize; i++) + if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1) + unlink_fifo (i); + + for (i = lsize; i < fifo_list_size; i++) + unlink_fifo (i); +} + +int +fifos_pending () +{ + return nfifo; +} + +int +num_fifos () +{ + return nfifo; +} + +static char * +make_named_pipe () +{ + char *tname; + + tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR); + if (mkfifo (tname, 0600) < 0) + { + free (tname); + return ((char *)NULL); + } + + add_fifo_list (tname); + return (tname); +} + +#else /* HAVE_DEV_FD */ + +/* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell + has open to children. NFDS is a count of the number of bits currently + set in DEV_FD_LIST. TOTFDS is a count of the highest possible number + of open files. */ +static char *dev_fd_list = (char *)NULL; +static int nfds; +static int totfds; /* The highest possible number of open files. */ + +char * +copy_fifo_list (sizep) + int *sizep; +{ + char *ret; + + if (nfds == 0 || totfds == 0) + { + if (sizep) + *sizep = 0; + return (char *)NULL; + } + + if (sizep) + *sizep = totfds; + ret = (char *)xmalloc (totfds); + return (memcpy (ret, dev_fd_list, totfds)); +} + +static void +add_fifo_list (fd) + int fd; +{ + if (dev_fd_list == 0 || fd >= totfds) + { + int ofds; + + ofds = totfds; + totfds = getdtablesize (); + if (totfds < 0 || totfds > 256) + totfds = 256; + if (fd >= totfds) + totfds = fd + 2; + + dev_fd_list = (char *)xrealloc (dev_fd_list, totfds); + memset (dev_fd_list + ofds, '\0', totfds - ofds); + } + + dev_fd_list[fd] = 1; + nfds++; +} + +int +fifos_pending () +{ + return 0; /* used for cleanup; not needed with /dev/fd */ +} + +int +num_fifos () +{ + return nfds; +} + +void +unlink_fifo (fd) + int fd; +{ + if (dev_fd_list[fd]) + { + close (fd); + dev_fd_list[fd] = 0; + nfds--; + } +} + +void +unlink_fifo_list () +{ + register int i; + + if (nfds == 0) + return; + + for (i = 0; nfds && i < totfds; i++) + unlink_fifo (i); + + nfds = 0; +} + +/* Take LIST, which is a snapshot copy of dev_fd_list from some point in + the past, and close all open fds in dev_fd_list that are not marked + as open in LIST. If LIST is NULL, close everything in dev_fd_list. + LSIZE is the number of elements in LIST, in case it's larger than + totfds (size of dev_fd_list). */ +void +close_new_fifos (list, lsize) + char *list; + int lsize; +{ + int i; + + if (list == 0) + { + unlink_fifo_list (); + return; + } + + for (i = 0; i < lsize; i++) + if (list[i] == 0 && i < totfds && dev_fd_list[i]) + unlink_fifo (i); + + for (i = lsize; i < totfds; i++) + unlink_fifo (i); +} + +#if defined (NOTDEF) +print_dev_fd_list () +{ + register int i; + + fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ()); + fflush (stderr); + + for (i = 0; i < totfds; i++) + { + if (dev_fd_list[i]) + fprintf (stderr, " %d", i); + } + fprintf (stderr, "\n"); +} +#endif /* NOTDEF */ + +static char * +make_dev_fd_filename (fd) + int fd; +{ + char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p; + + ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8); + + strcpy (ret, DEV_FD_PREFIX); + p = inttostr (fd, intbuf, sizeof (intbuf)); + strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p); + + add_fifo_list (fd); + return (ret); +} + +#endif /* HAVE_DEV_FD */ + +/* Return a filename that will open a connection to the process defined by + executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return + a filename in /dev/fd corresponding to a descriptor that is one of the + ends of the pipe. If not defined, we use named pipes on systems that have + them. Systems without /dev/fd and named pipes are out of luck. + + OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or + use the read end of the pipe and dup that file descriptor to fd 0 in + the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for + writing or use the write end of the pipe in the child, and dup that + file descriptor to fd 1 in the child. The parent does the opposite. */ + +static char * +process_substitute (string, open_for_read_in_child) + char *string; + int open_for_read_in_child; +{ + char *pathname; + int fd, result; + pid_t old_pid, pid; +#if defined (HAVE_DEV_FD) + int parent_pipe_fd, child_pipe_fd; + int fildes[2]; +#endif /* HAVE_DEV_FD */ +#if defined (JOB_CONTROL) + pid_t old_pipeline_pgrp; +#endif + + if (!string || !*string || wordexp_only) + return ((char *)NULL); + +#if !defined (HAVE_DEV_FD) + pathname = make_named_pipe (); +#else /* HAVE_DEV_FD */ + if (pipe (fildes) < 0) + { + sys_error (_("cannot make pipe for process substitution")); + return ((char *)NULL); + } + /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of + the pipe in the parent, otherwise the read end. */ + parent_pipe_fd = fildes[open_for_read_in_child]; + child_pipe_fd = fildes[1 - open_for_read_in_child]; + /* Move the parent end of the pipe to some high file descriptor, to + avoid clashes with FDs used by the script. */ + parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64); + + pathname = make_dev_fd_filename (parent_pipe_fd); +#endif /* HAVE_DEV_FD */ + + if (pathname == 0) + { + sys_error (_("cannot make pipe for process substitution")); + return ((char *)NULL); + } + + old_pid = last_made_pid; + +#if defined (JOB_CONTROL) + old_pipeline_pgrp = pipeline_pgrp; + pipeline_pgrp = shell_pgrp; + save_pipeline (1); +#endif /* JOB_CONTROL */ + + pid = make_child ((char *)NULL, 1); + if (pid == 0) + { + reset_terminating_signals (); /* XXX */ + free_pushed_string_input (); + /* Cancel traps, in trap.c. */ + restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */ + setup_async_signals (); + subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB; + } + +#if defined (JOB_CONTROL) + set_sigchld_handler (); + stop_making_children (); + /* XXX - should we only do this in the parent? (as in command subst) */ + pipeline_pgrp = old_pipeline_pgrp; +#endif /* JOB_CONTROL */ + + if (pid < 0) + { + sys_error (_("cannot make child for process substitution")); + free (pathname); +#if defined (HAVE_DEV_FD) + close (parent_pipe_fd); + close (child_pipe_fd); +#endif /* HAVE_DEV_FD */ + return ((char *)NULL); + } + + if (pid > 0) + { +#if defined (JOB_CONTROL) + restore_pipeline (1); +#endif + +#if !defined (HAVE_DEV_FD) + fifo_list[nfifo-1].proc = pid; +#endif + + last_made_pid = old_pid; + +#if defined (JOB_CONTROL) && defined (PGRP_PIPE) + close_pgrp_pipe (); +#endif /* JOB_CONTROL && PGRP_PIPE */ + +#if defined (HAVE_DEV_FD) + close (child_pipe_fd); +#endif /* HAVE_DEV_FD */ + + return (pathname); + } + + set_sigint_handler (); + +#if defined (JOB_CONTROL) + set_job_control (0); +#endif /* JOB_CONTROL */ + +#if !defined (HAVE_DEV_FD) + /* Open the named pipe in the child. */ + fd = open (pathname, open_for_read_in_child ? O_RDONLY|O_NONBLOCK : O_WRONLY); + if (fd < 0) + { + /* Two separate strings for ease of translation. */ + if (open_for_read_in_child) + sys_error (_("cannot open named pipe %s for reading"), pathname); + else + sys_error (_("cannot open named pipe %s for writing"), pathname); + + exit (127); + } + if (open_for_read_in_child) + { + if (sh_unset_nodelay_mode (fd) < 0) + { + sys_error (_("cannot reset nodelay mode for fd %d"), fd); + exit (127); + } + } +#else /* HAVE_DEV_FD */ + fd = child_pipe_fd; +#endif /* HAVE_DEV_FD */ + + if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0) + { + sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname, + open_for_read_in_child ? 0 : 1); + exit (127); + } + + if (fd != (open_for_read_in_child ? 0 : 1)) + close (fd); + + /* Need to close any files that this process has open to pipes inherited + from its parent. */ + if (current_fds_to_close) + { + close_fd_bitmap (current_fds_to_close); + current_fds_to_close = (struct fd_bitmap *)NULL; + } + +#if defined (HAVE_DEV_FD) + /* Make sure we close the parent's end of the pipe and clear the slot + in the fd list so it is not closed later, if reallocated by, for + instance, pipe(2). */ + close (parent_pipe_fd); + dev_fd_list[parent_pipe_fd] = 0; +#endif /* HAVE_DEV_FD */ + + result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST)); + +#if !defined (HAVE_DEV_FD) + /* Make sure we close the named pipe in the child before we exit. */ + close (open_for_read_in_child ? 0 : 1); +#endif /* !HAVE_DEV_FD */ + + exit (result); + /*NOTREACHED*/ +} +#endif /* PROCESS_SUBSTITUTION */ + +/***********************************/ +/* */ +/* Command Substitution */ +/* */ +/***********************************/ + +static char * +read_comsub (fd, quoted, rflag) + int fd, quoted; + int *rflag; +{ + char *istring, buf[128], *bufp, *s; + int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul; + ssize_t bufn; + + istring = (char *)NULL; + istring_index = istring_size = bufn = tflag = 0; + + for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++) + skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL; + + /* Read the output of the command through the pipe. This may need to be + changed to understand multibyte characters in the future. */ + while (1) + { + if (fd < 0) + break; + if (--bufn <= 0) + { + bufn = zread (fd, buf, sizeof (buf)); + if (bufn <= 0) + break; + bufp = buf; + } + c = *bufp++; + + if (c == 0) + { +#if 0 + internal_warning ("read_comsub: ignored null byte in input"); +#endif + continue; + } + + /* Add the character to ISTRING, possibly after resizing it. */ + RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE); + + /* This is essentially quote_string inline */ + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */) + istring[istring_index++] = CTLESC; + /* Escape CTLESC and CTLNUL in the output to protect those characters + from the rest of the word expansions (word splitting and globbing.) + This is essentially quote_escapes inline. */ + else if (skip_ctlesc == 0 && c == CTLESC) + { + tflag |= W_HASCTLESC; + istring[istring_index++] = CTLESC; + } + else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0))) + istring[istring_index++] = CTLESC; + + istring[istring_index++] = c; + +#if 0 +#if defined (__CYGWIN__) + if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r') + { + istring_index--; + istring[istring_index - 1] = '\n'; + } +#endif +#endif + } + + if (istring) + istring[istring_index] = '\0'; + + /* If we read no output, just return now and save ourselves some + trouble. */ + if (istring_index == 0) + { + FREE (istring); + if (rflag) + *rflag = tflag; + return (char *)NULL; + } + + /* Strip trailing newlines from the output of the command. */ + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + { + while (istring_index > 0) + { + if (istring[istring_index - 1] == '\n') + { + --istring_index; + + /* If the newline was quoted, remove the quoting char. */ + if (istring[istring_index - 1] == CTLESC) + --istring_index; + } + else + break; + } + istring[istring_index] = '\0'; + } + else + strip_trailing (istring, istring_index - 1, 1); + + if (rflag) + *rflag = tflag; + return istring; +} + +/* Perform command substitution on STRING. This returns a WORD_DESC * with the + contained string possibly quoted. */ +WORD_DESC * +command_substitute (string, quoted) + char *string; + int quoted; +{ + pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid; + char *istring; + int result, fildes[2], function_value, pflags, rc, tflag; + WORD_DESC *ret; + + istring = (char *)NULL; + + /* Don't fork () if there is no need to. In the case of no command to + run, just return NULL. */ + if (!string || !*string || (string[0] == '\n' && !string[1])) + return ((WORD_DESC *)NULL); + + if (wordexp_only && read_but_dont_execute) + { + last_command_exit_value = EX_WEXPCOMSUB; + jump_to_top_level (EXITPROG); + } + + /* We're making the assumption here that the command substitution will + eventually run a command from the file system. Since we'll run + maybe_make_export_env in this subshell before executing that command, + the parent shell and any other shells it starts will have to remake + the environment. If we make it before we fork, other shells won't + have to. Don't bother if we have any temporary variable assignments, + though, because the export environment will be remade after this + command completes anyway, but do it if all the words to be expanded + are variable assignments. */ + if (subst_assign_varlist == 0 || garglist == 0) + maybe_make_export_env (); /* XXX */ + + /* Flags to pass to parse_and_execute() */ + pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0; + + /* Pipe the output of executing STRING into the current shell. */ + if (pipe (fildes) < 0) + { + sys_error (_("cannot make pipe for command substitution")); + goto error_exit; + } + + old_pid = last_made_pid; +#if defined (JOB_CONTROL) + old_pipeline_pgrp = pipeline_pgrp; + /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */ + if ((subshell_environment & SUBSHELL_PIPE) == 0) + pipeline_pgrp = shell_pgrp; + cleanup_the_pipeline (); +#endif /* JOB_CONTROL */ + + old_async_pid = last_asynchronous_pid; + pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC); + last_asynchronous_pid = old_async_pid; + + if (pid == 0) + { + /* Reset the signal handlers in the child, but don't free the + trap strings. Set a flag noting that we have to free the + trap strings if we run trap to change a signal disposition. */ + reset_signal_handlers (); + subshell_environment |= SUBSHELL_RESETTRAP; + } + +#if defined (JOB_CONTROL) + /* XXX DO THIS ONLY IN PARENT ? XXX */ + set_sigchld_handler (); + stop_making_children (); + if (pid != 0) + pipeline_pgrp = old_pipeline_pgrp; +#else + stop_making_children (); +#endif /* JOB_CONTROL */ + + if (pid < 0) + { + sys_error (_("cannot make child for command substitution")); + error_exit: + + FREE (istring); + close (fildes[0]); + close (fildes[1]); + return ((WORD_DESC *)NULL); + } + + if (pid == 0) + { + set_sigint_handler (); /* XXX */ + + free_pushed_string_input (); + + if (dup2 (fildes[1], 1) < 0) + { + sys_error (_("command_substitute: cannot duplicate pipe as fd 1")); + exit (EXECUTION_FAILURE); + } + + /* If standard output is closed in the parent shell + (such as after `exec >&-'), file descriptor 1 will be + the lowest available file descriptor, and end up in + fildes[0]. This can happen for stdin and stderr as well, + but stdout is more important -- it will cause no output + to be generated from this command. */ + if ((fildes[1] != fileno (stdin)) && + (fildes[1] != fileno (stdout)) && + (fildes[1] != fileno (stderr))) + close (fildes[1]); + + if ((fildes[0] != fileno (stdin)) && + (fildes[0] != fileno (stdout)) && + (fildes[0] != fileno (stderr))) + close (fildes[0]); + +#ifdef __CYGWIN__ + /* Let stdio know the fd may have changed from text to binary mode, and + make sure to preserve stdout line buffering. */ + freopen (NULL, "w", stdout); + sh_setlinebuf (stdout); +#endif /* __CYGWIN__ */ + + /* The currently executing shell is not interactive. */ + interactive = 0; + + /* This is a subshell environment. */ + subshell_environment |= SUBSHELL_COMSUB; + + /* When not in POSIX mode, command substitution does not inherit + the -e flag. */ + if (posixly_correct == 0) + exit_immediately_on_error = 0; + + remove_quoted_escapes (string); + + startup_state = 2; /* see if we can avoid a fork */ + /* Give command substitution a place to jump back to on failure, + so we don't go back up to main (). */ + result = setjmp (top_level); +itrace("command_substitute: %s", string); + /* If we're running a command substitution inside a shell function, + trap `return' so we don't return from the function in the subshell + and go off to never-never land. */ + if (result == 0 && return_catch_flag) + function_value = setjmp (return_catch); + else + function_value = 0; + + if (result == ERREXIT) + rc = last_command_exit_value; + else if (result == EXITPROG) + rc = last_command_exit_value; + else if (result) + rc = EXECUTION_FAILURE; + else if (function_value) + rc = return_catch_value; + else + { + subshell_level++; + rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST); + subshell_level--; + } + + last_command_exit_value = rc; + rc = run_exit_trap (); +#if defined (PROCESS_SUBSTITUTION) + unlink_fifo_list (); +#endif + exit (rc); + } + else + { +#if defined (JOB_CONTROL) && defined (PGRP_PIPE) + close_pgrp_pipe (); +#endif /* JOB_CONTROL && PGRP_PIPE */ + + close (fildes[1]); + + tflag = 0; + istring = read_comsub (fildes[0], quoted, &tflag); + + close (fildes[0]); + + current_command_subst_pid = pid; + last_command_exit_value = wait_for (pid); + last_command_subst_pid = pid; + last_made_pid = old_pid; + +#if defined (JOB_CONTROL) + /* If last_command_exit_value > 128, then the substituted command + was terminated by a signal. If that signal was SIGINT, then send + SIGINT to ourselves. This will break out of loops, for instance. */ + if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT) + kill (getpid (), SIGINT); + + /* wait_for gives the terminal back to shell_pgrp. If some other + process group should have it, give it away to that group here. + pipeline_pgrp is non-zero only while we are constructing a + pipline, so what we are concerned about is whether or not that + pipeline was started in the background. A pipeline started in + the background should never get the tty back here. */ + if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0) + give_terminal_to (pipeline_pgrp, 0); +#endif /* JOB_CONTROL */ + + ret = alloc_word_desc (); + ret->word = istring; + ret->flags = tflag; + + return ret; + } +} + +/******************************************************** + * * + * Utility functions for parameter expansion * + * * + ********************************************************/ + +#if defined (ARRAY_VARS) + +static arrayind_t +array_length_reference (s) + char *s; +{ + int len; + arrayind_t ind; + char *akey; + char *t, c; + ARRAY *array; + HASH_TABLE *h; + SHELL_VAR *var; + + var = array_variable_part (s, &t, &len); + + /* If unbound variables should generate an error, report one and return + failure. */ + if ((var == 0 || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error) + { + c = *--t; + *t = '\0'; + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (s); + *t = c; + return (-1); + } + else if (var == 0) + return 0; + + /* We support a couple of expansions for variables that are not arrays. + We'll return the length of the value for v[0], and 1 for v[@] or + v[*]. Return 0 for everything else. */ + + array = array_p (var) ? array_cell (var) : (ARRAY *)NULL; + h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL; + + if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']') + { + if (assoc_p (var)) + return (h ? assoc_num_elements (h) : 0); + else if (array_p (var)) + return (array ? array_num_elements (array) : 0); + else + return (var_isset (var) ? 1 : 0); + } + + if (assoc_p (var)) + { + t[len - 1] = '\0'; + akey = expand_assignment_string_to_string (t, 0); /* [ */ + t[len - 1] = ']'; + if (akey == 0 || *akey == 0) + { + err_badarraysub (t); + FREE (akey); + return (-1); + } + t = assoc_reference (assoc_cell (var), akey); + free (akey); + } + else + { + ind = array_expand_index (var, t, len); + if (ind < 0) + { + err_badarraysub (t); + return (-1); + } + if (array_p (var)) + t = array_reference (array, ind); + else + t = (ind == 0) ? value_cell (var) : (char *)NULL; + } + + len = MB_STRLEN (t); + return (len); +} +#endif /* ARRAY_VARS */ + +static int +valid_brace_expansion_word (name, var_is_special) + char *name; + int var_is_special; +{ + if (DIGIT (*name) && all_digits (name)) + return 1; + else if (var_is_special) + return 1; +#if defined (ARRAY_VARS) + else if (valid_array_reference (name)) + return 1; +#endif /* ARRAY_VARS */ + else if (legal_identifier (name)) + return 1; + else + return 0; +} + +static int +chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at) + char *name; + int quoted; + int *quoted_dollar_atp, *contains_dollar_at; +{ + char *temp1; + + if (name == 0) + { + if (quoted_dollar_atp) + *quoted_dollar_atp = 0; + if (contains_dollar_at) + *contains_dollar_at = 0; + return 0; + } + + /* check for $@ and $* */ + if (name[0] == '@' && name[1] == 0) + { + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 1; + if (contains_dollar_at) + *contains_dollar_at = 1; + return 1; + } + else if (name[0] == '*' && name[1] == '\0' && quoted == 0) + { + if (contains_dollar_at) + *contains_dollar_at = 1; + return 1; + } + + /* Now check for ${array[@]} and ${array[*]} */ +#if defined (ARRAY_VARS) + else if (valid_array_reference (name)) + { + temp1 = mbschr (name, '['); + if (temp1 && temp1[1] == '@' && temp1[2] == ']') + { + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 1; + if (contains_dollar_at) + *contains_dollar_at = 1; + return 1; + } /* [ */ + /* ${array[*]}, when unquoted, should be treated like ${array[@]}, + which should result in separate words even when IFS is unset. */ + if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0) + { + if (contains_dollar_at) + *contains_dollar_at = 1; + return 1; + } + } +#endif + return 0; +} + +/* Parameter expand NAME, and return a new string which is the expansion, + or NULL if there was no expansion. + VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in + the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that + NAME was found inside of a double-quoted expression. */ +static WORD_DESC * +parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp) + char *name; + int var_is_special, quoted, pflags; + arrayind_t *indp; +{ + WORD_DESC *ret; + char *temp, *tt; + intmax_t arg_index; + SHELL_VAR *var; + int atype, rflags; + arrayind_t ind; + + ret = 0; + temp = 0; + rflags = 0; + + if (indp) + *indp = INTMAX_MIN; + + /* Handle multiple digit arguments, as in ${11}. */ + if (legal_number (name, &arg_index)) + { + tt = get_dollar_var_value (arg_index); + if (tt) + temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) + ? quote_string (tt) + : quote_escapes (tt); + else + temp = (char *)NULL; + FREE (tt); + } + else if (var_is_special) /* ${@} */ + { + int sindex; + tt = (char *)xmalloc (2 + strlen (name)); + tt[sindex = 0] = '$'; + strcpy (tt + 1, name); + + ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL, + (int *)NULL, (int *)NULL, pflags); + free (tt); + } +#if defined (ARRAY_VARS) + else if (valid_array_reference (name)) + { + /* XXX - does this leak if name[@] or name[*]? */ + temp = array_value (name, quoted, 0, &atype, &ind); + if (atype == 0 && temp) + { + temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) + ? quote_string (temp) + : quote_escapes (temp); + rflags |= W_ARRAYIND; + if (indp) + *indp = ind; + } + else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) + rflags |= W_HASQUOTEDNULL; + } +#endif + else if (var = find_variable (name)) + { + if (var_isset (var) && invisible_p (var) == 0) + { +#if defined (ARRAY_VARS) + if (assoc_p (var)) + temp = assoc_reference (assoc_cell (var), "0"); + else if (array_p (var)) + temp = array_reference (array_cell (var), 0); + else + temp = value_cell (var); +#else + temp = value_cell (var); +#endif + + if (temp) + temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) + ? quote_string (temp) + : quote_escapes (temp); + } + else + temp = (char *)NULL; + } + else + temp = (char *)NULL; + + if (ret == 0) + { + ret = alloc_word_desc (); + ret->word = temp; + ret->flags |= rflags; + } + return ret; +} + +/* Expand an indirect reference to a variable: ${!NAME} expands to the + value of the variable whose name is the value of NAME. */ +static WORD_DESC * +parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at) + char *name; + int var_is_special, quoted; + int *quoted_dollar_atp, *contains_dollar_at; +{ + char *temp, *t; + WORD_DESC *w; + + w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0); + t = w->word; + /* Have to dequote here if necessary */ + if (t) + { + temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) + ? dequote_string (t) + : dequote_escapes (t); + free (t); + t = temp; + } + dispose_word_desc (w); + + chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at); + if (t == 0) + return (WORD_DESC *)NULL; + + w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0); + free (t); + + return w; +} + +/* Expand the right side of a parameter expansion of the form ${NAMEcVALUE}, + depending on the value of C, the separating character. C can be one of + "-", "+", or "=". QUOTED is true if the entire brace expression occurs + between double quotes. */ +static WORD_DESC * +parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat) + char *name, *value; + int c, quoted, *qdollaratp, *hasdollarat; +{ + WORD_DESC *w; + WORD_LIST *l; + char *t, *t1, *temp; + int hasdol; + + /* If the entire expression is between double quotes, we want to treat + the value as a double-quoted string, with the exception that we strip + embedded unescaped double quotes (for sh backwards compatibility). */ + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value) + { + hasdol = 0; + temp = string_extract_double_quoted (value, &hasdol, 1); + } + else + temp = value; + + w = alloc_word_desc (); + hasdol = 0; + /* XXX was 0 not quoted */ + l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL) + : (WORD_LIST *)0; + if (hasdollarat) + *hasdollarat = hasdol || (l && l->next); + if (temp != value) + free (temp); + if (l) + { + /* The expansion of TEMP returned something. We need to treat things + slightly differently if HASDOL is non-zero. If we have "$@", the + individual words have already been quoted. We need to turn them + into a string with the words separated by the first character of + $IFS without any additional quoting, so string_list_dollar_at won't + do the right thing. We use string_list_dollar_star instead. */ + temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l); + + /* If l->next is not null, we know that TEMP contained "$@", since that + is the only expansion that creates more than one word. */ + if (qdollaratp && ((hasdol && quoted) || l->next)) + *qdollaratp = 1; + dispose_words (l); + } + else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol) + { + /* The brace expansion occurred between double quotes and there was + a $@ in TEMP. It does not matter if the $@ is quoted, as long as + it does not expand to anything. In this case, we want to return + a quoted empty string. */ + temp = make_quoted_char ('\0'); + w->flags |= W_HASQUOTEDNULL; + } + else + temp = (char *)NULL; + + if (c == '-' || c == '+') + { + w->word = temp; + return w; + } + + /* c == '=' */ + t = temp ? savestring (temp) : savestring (""); + t1 = dequote_string (t); + free (t); +#if defined (ARRAY_VARS) + if (valid_array_reference (name)) + assign_array_element (name, t1, 0); + else +#endif /* ARRAY_VARS */ + bind_variable (name, t1, 0); + + /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */ + free (temp); + + w->word = t1; + return w; +} + +/* Deal with the right hand side of a ${name:?value} expansion in the case + that NAME is null or not set. If VALUE is non-null it is expanded and + used as the error message to print, otherwise a standard message is + printed. */ +static void +parameter_brace_expand_error (name, value) + char *name, *value; +{ + WORD_LIST *l; + char *temp; + + last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */ + if (value && *value) + { + l = expand_string (value, 0); + temp = string_list (l); + report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */ + FREE (temp); + dispose_words (l); + } + else + report_error (_("%s: parameter null or not set"), name); + + /* Free the data we have allocated during this expansion, since we + are about to longjmp out. */ + free (name); + FREE (value); +} + +/* Return 1 if NAME is something for which parameter_brace_expand_length is + OK to do. */ +static int +valid_length_expression (name) + char *name; +{ + return (name[1] == '\0' || /* ${#} */ + ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */ + (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */ +#if defined (ARRAY_VARS) + valid_array_reference (name + 1) || /* ${#a[7]} */ +#endif + legal_identifier (name + 1)); /* ${#PS1} */ +} + +/* Handle the parameter brace expansion that requires us to return the + length of a parameter. */ +static intmax_t +parameter_brace_expand_length (name) + char *name; +{ + char *t, *newname; + intmax_t number, arg_index; + WORD_LIST *list; +#if defined (ARRAY_VARS) + SHELL_VAR *var; +#endif + + if (name[1] == '\0') /* ${#} */ + number = number_of_args (); + else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */ + number = number_of_args (); + else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') + { + /* Take the lengths of some of the shell's special parameters. */ + switch (name[1]) + { + case '-': + t = which_set_flags (); + break; + case '?': + t = itos (last_command_exit_value); + break; + case '$': + t = itos (dollar_dollar_pid); + break; + case '!': + if (last_asynchronous_pid == NO_PID) + t = (char *)NULL; /* XXX - error if set -u set? */ + else + t = itos (last_asynchronous_pid); + break; + case '#': + t = itos (number_of_args ()); + break; + } + number = STRLEN (t); + FREE (t); + } +#if defined (ARRAY_VARS) + else if (valid_array_reference (name + 1)) + number = array_length_reference (name + 1); +#endif /* ARRAY_VARS */ + else + { + number = 0; + + if (legal_number (name + 1, &arg_index)) /* ${#1} */ + { + t = get_dollar_var_value (arg_index); + if (t == 0 && unbound_vars_is_error) + return INTMAX_MIN; + number = MB_STRLEN (t); + FREE (t); + } +#if defined (ARRAY_VARS) + else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var))) + { + if (assoc_p (var)) + t = assoc_reference (assoc_cell (var), "0"); + else + t = array_reference (array_cell (var), 0); + if (t == 0 && unbound_vars_is_error) + return INTMAX_MIN; + number = MB_STRLEN (t); + } +#endif + else /* ${#PS1} */ + { + newname = savestring (name); + newname[0] = '$'; + list = expand_string (newname, Q_DOUBLE_QUOTES); + t = list ? string_list (list) : (char *)NULL; + free (newname); + if (list) + dispose_words (list); + + number = t ? MB_STRLEN (t) : 0; + FREE (t); + } + } + + return (number); +} + +/* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression, + so we do some ad-hoc parsing of an arithmetic expression to find + the first DELIM, instead of using strchr(3). Two rules: + 1. If the substring contains a `(', read until closing `)'. + 2. If the substring contains a `?', read past one `:' for each `?'. +*/ + +static char * +skiparith (substr, delim) + char *substr; + int delim; +{ + size_t sublen; + int skipcol, pcount, i; + DECLARE_MBSTATE; + + sublen = strlen (substr); + i = skipcol = pcount = 0; + while (substr[i]) + { + /* Balance parens */ + if (substr[i] == LPAREN) + { + pcount++; + i++; + continue; + } + if (substr[i] == RPAREN && pcount) + { + pcount--; + i++; + continue; + } + if (pcount) + { + ADVANCE_CHAR (substr, sublen, i); + continue; + } + + /* Skip one `:' for each `?' */ + if (substr[i] == ':' && skipcol) + { + skipcol--; + i++; + continue; + } + if (substr[i] == delim) + break; + if (substr[i] == '?') + { + skipcol++; + i++; + continue; + } + ADVANCE_CHAR (substr, sublen, i); + } + + return (substr + i); +} + +/* Verify and limit the start and end of the desired substring. If + VTYPE == 0, a regular shell variable is being used; if it is 1, + then the positional parameters are being used; if it is 2, then + VALUE is really a pointer to an array variable that should be used. + Return value is 1 if both values were OK, 0 if there was a problem + with an invalid expression, or -1 if the values were out of range. */ +static int +verify_substring_values (v, value, substr, vtype, e1p, e2p) + SHELL_VAR *v; + char *value, *substr; + int vtype; + intmax_t *e1p, *e2p; +{ + char *t, *temp1, *temp2; + arrayind_t len; + int expok; +#if defined (ARRAY_VARS) + ARRAY *a; + HASH_TABLE *h; +#endif + + /* duplicate behavior of strchr(3) */ + t = skiparith (substr, ':'); + if (*t && *t == ':') + *t = '\0'; + else + t = (char *)0; + + temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES); + *e1p = evalexp (temp1, &expok); + free (temp1); + if (expok == 0) + return (0); + + len = -1; /* paranoia */ + switch (vtype) + { + case VT_VARIABLE: + case VT_ARRAYMEMBER: + len = MB_STRLEN (value); + break; + case VT_POSPARMS: + len = number_of_args () + 1; + if (*e1p == 0) + len++; /* add one arg if counting from $0 */ + break; +#if defined (ARRAY_VARS) + case VT_ARRAYVAR: + /* For arrays, the first value deals with array indices. Negative + offsets count from one past the array's maximum index. Associative + arrays treat the number of elements as the maximum index. */ + if (assoc_p (v)) + { + h = assoc_cell (v); + len = assoc_num_elements (h) + (*e1p < 0); + } + else + { + a = (ARRAY *)value; + len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */ + } + break; +#endif + } + + if (len == -1) /* paranoia */ + return -1; + + if (*e1p < 0) /* negative offsets count from end */ + *e1p += len; + + if (*e1p > len || *e1p < 0) + return (-1); + +#if defined (ARRAY_VARS) + /* For arrays, the second offset deals with the number of elements. */ + if (vtype == VT_ARRAYVAR) + len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a); +#endif + + if (t) + { + t++; + temp2 = savestring (t); + temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES); + free (temp2); + t[-1] = ':'; + *e2p = evalexp (temp1, &expok); + free (temp1); + if (expok == 0) + return (0); + if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0) + { + internal_error (_("%s: substring expression < 0"), t); + return (0); + } +#if defined (ARRAY_VARS) + /* In order to deal with sparse arrays, push the intelligence about how + to deal with the number of elements desired down to the array- + specific functions. */ + if (vtype != VT_ARRAYVAR) +#endif + { + if (*e2p < 0) + { + *e2p += len; + if (*e2p < 0 || *e2p < *e1p) + { + internal_error (_("%s: substring expression < 0"), t); + return (0); + } + } + else + *e2p += *e1p; /* want E2 chars starting at E1 */ + if (*e2p > len) + *e2p = len; + } + } + else + *e2p = len; + + return (1); +} + +/* Return the type of variable specified by VARNAME (simple variable, + positional param, or array variable). Also return the value specified + by VARNAME (value of a variable or a reference to an array element). + QUOTED is the standard description of quoting state, using Q_* defines. + FLAGS is currently a set of flags to pass to array_value. If IND is + non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is + passed to array_value so the array index is not computed again. + If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL + characters in the value are quoted with CTLESC and takes appropriate + steps. For convenience, *VALP is set to the dequoted VALUE. */ +static int +get_var_and_type (varname, value, ind, quoted, flags, varp, valp) + char *varname, *value; + arrayind_t ind; + int quoted, flags; + SHELL_VAR **varp; + char **valp; +{ + int vtype; + char *temp; +#if defined (ARRAY_VARS) + SHELL_VAR *v; +#endif + arrayind_t lind; + + /* This sets vtype to VT_VARIABLE or VT_POSPARMS */ + vtype = (varname[0] == '@' || varname[0] == '*') && varname[1] == '\0'; + if (vtype == VT_POSPARMS && varname[0] == '*') + vtype |= VT_STARSUB; + *varp = (SHELL_VAR *)NULL; + +#if defined (ARRAY_VARS) + if (valid_array_reference (varname)) + { + v = array_variable_part (varname, &temp, (int *)0); + /* If we want to signal array_value to use an already-computed index, + set LIND to that index */ + lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0; + if (v && (array_p (v) || assoc_p (v))) + { /* [ */ + if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']') + { + /* Callers have to differentiate betwen indexed and associative */ + vtype = VT_ARRAYVAR; + if (temp[0] == '*') + vtype |= VT_STARSUB; + *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v); + } + else + { + vtype = VT_ARRAYMEMBER; + *valp = array_value (varname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind); + } + *varp = v; + } + else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')) + { + vtype = VT_VARIABLE; + *varp = v; + if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) + *valp = dequote_string (value); + else + *valp = dequote_escapes (value); + } + else + { + vtype = VT_ARRAYMEMBER; + *varp = v; + *valp = array_value (varname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind); + } + } + else if ((v = find_variable (varname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v))) + { + vtype = VT_ARRAYMEMBER; + *varp = v; + *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0); + } + else +#endif + { + if (value && vtype == VT_VARIABLE) + { + if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) + *valp = dequote_string (value); + else + *valp = dequote_escapes (value); + } + else + *valp = value; + } + + return vtype; +} + +/******************************************************/ +/* */ +/* Functions to extract substrings of variable values */ +/* */ +/******************************************************/ + +#if defined (HANDLE_MULTIBYTE) +/* Character-oriented rather than strictly byte-oriented substrings. S and + E, rather being strict indices into STRING, indicate character (possibly + multibyte character) positions that require calculation. + Used by the ${param:offset[:length]} expansion. */ +static char * +mb_substring (string, s, e) + char *string; + int s, e; +{ + char *tt; + int start, stop, i, slen; + DECLARE_MBSTATE; + + start = 0; + /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */ + slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0; + + i = s; + while (string[start] && i--) + ADVANCE_CHAR (string, slen, start); + stop = start; + i = e - s; + while (string[stop] && i--) + ADVANCE_CHAR (string, slen, stop); + tt = substring (string, start, stop); + return tt; +} +#endif + +/* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME + is `@', use the positional parameters; otherwise, use the value of + VARNAME. If VARNAME is an array variable, use the array elements. */ + +static char * +parameter_brace_substring (varname, value, ind, substr, quoted, flags) + char *varname, *value; + int ind; + char *substr; + int quoted, flags; +{ + intmax_t e1, e2; + int vtype, r, starsub; + char *temp, *val, *tt, *oname; + SHELL_VAR *v; + + if (value == 0) + return ((char *)NULL); + + oname = this_command_name; + this_command_name = varname; + + vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); + if (vtype == -1) + { + this_command_name = oname; + return ((char *)NULL); + } + + starsub = vtype & VT_STARSUB; + vtype &= ~VT_STARSUB; + + r = verify_substring_values (v, val, substr, vtype, &e1, &e2); + this_command_name = oname; + if (r <= 0) + { + if (vtype == VT_VARIABLE) + FREE (val); + return ((r == 0) ? &expand_param_error : (char *)NULL); + } + + switch (vtype) + { + case VT_VARIABLE: + case VT_ARRAYMEMBER: +#if defined (HANDLE_MULTIBYTE) + if (MB_CUR_MAX > 1) + tt = mb_substring (val, e1, e2); + else +#endif + tt = substring (val, e1, e2); + + if (vtype == VT_VARIABLE) + FREE (val); + if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) + temp = quote_string (tt); + else + temp = tt ? quote_escapes (tt) : (char *)NULL; + FREE (tt); + break; + case VT_POSPARMS: + tt = pos_params (varname, e1, e2, quoted); + if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0) + { + temp = tt ? quote_escapes (tt) : (char *)NULL; + FREE (tt); + } + else + temp = tt; + break; +#if defined (ARRAY_VARS) + case VT_ARRAYVAR: + if (assoc_p (v)) + /* we convert to list and take first e2 elements starting at e1th + element -- officially undefined for now */ + temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted); + else + /* We want E2 to be the number of elements desired (arrays can be sparse, + so verify_substring_values just returns the numbers specified and we + rely on array_subrange to understand how to deal with them). */ + temp = array_subrange (array_cell (v), e1, e2, starsub, quoted); + /* array_subrange now calls array_quote_escapes as appropriate, so the + caller no longer needs to. */ + break; +#endif + default: + temp = (char *)NULL; + } + + return temp; +} + +/****************************************************************/ +/* */ +/* Functions to perform pattern substitution on variable values */ +/* */ +/****************************************************************/ + +static int +shouldexp_replacement (s) + char *s; +{ + register char *p; + + for (p = s; p && *p; p++) + { + if (*p == '\\') + p++; + else if (*p == '&') + return 1; + } + return 0; +} + +char * +pat_subst (string, pat, rep, mflags) + char *string, *pat, *rep; + int mflags; +{ + char *ret, *s, *e, *str, *rstr, *mstr; + int rsize, rptr, l, replen, mtype, rxpand, rslen, mlen; + + if (string == 0) + return (savestring ("")); + + mtype = mflags & MATCH_TYPEMASK; + +#if 0 /* bash-4.2 ? */ + rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0; +#else + rxpand = 0; +#endif + + /* Special cases: + * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING + * with REP and return the result. + * 2. A null pattern with mtype == MATCH_END means to append REP to + * STRING and return the result. + * These don't understand or process `&' in the replacement string. + */ + if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END)) + { + replen = STRLEN (rep); + l = STRLEN (string); + ret = (char *)xmalloc (replen + l + 2); + if (replen == 0) + strcpy (ret, string); + else if (mtype == MATCH_BEG) + { + strcpy (ret, rep); + strcpy (ret + replen, string); + } + else + { + strcpy (ret, string); + strcpy (ret + l, rep); + } + return (ret); + } + + ret = (char *)xmalloc (rsize = 64); + ret[0] = '\0'; + + for (replen = STRLEN (rep), rptr = 0, str = string;;) + { + if (match_pattern (str, pat, mtype, &s, &e) == 0) + break; + l = s - str; + + if (rxpand) + { + int x; + mlen = e - s; + mstr = xmalloc (mlen + 1); + for (x = 0; x < mlen; x++) + mstr[x] = s[x]; + mstr[mlen] = '\0'; + rstr = strcreplace (rep, '&', mstr, 0); + rslen = strlen (rstr); + } + else + { + rstr = rep; + rslen = replen; + } + + RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64); + + /* OK, now copy the leading unmatched portion of the string (from + str to s) to ret starting at rptr (the current offset). Then copy + the replacement string at ret + rptr + (s - str). Increment + rptr (if necessary) and str and go on. */ + if (l) + { + strncpy (ret + rptr, str, l); + rptr += l; + } + if (replen) + { + strncpy (ret + rptr, rstr, rslen); + rptr += rslen; + } + str = e; /* e == end of match */ + + if (rstr != rep) + free (rstr); + + if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY) + break; + + if (s == e) + { + /* On a zero-length match, make sure we copy one character, since + we increment one character to avoid infinite recursion. */ + RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64); + ret[rptr++] = *str++; + e++; /* avoid infinite recursion on zero-length match */ + } + } + + /* Now copy the unmatched portion of the input string */ + if (str && *str) + { + RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64); + strcpy (ret + rptr, str); + } + else + ret[rptr] = '\0'; + + return ret; +} + +/* Do pattern match and replacement on the positional parameters. */ +static char * +pos_params_pat_subst (string, pat, rep, mflags) + char *string, *pat, *rep; + int mflags; +{ + WORD_LIST *save, *params; + WORD_DESC *w; + char *ret; + int pchar, qflags; + + save = params = list_rest_of_args (); + if (save == 0) + return ((char *)NULL); + + for ( ; params; params = params->next) + { + ret = pat_subst (params->word->word, pat, rep, mflags); + w = alloc_word_desc (); + w->word = ret ? ret : savestring (""); + dispose_word (params->word); + params->word = w; + } + + pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@'; + qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0; + +#if 0 + if ((mflags & (MATCH_QUOTED|MATCH_STARSUB)) == (MATCH_QUOTED|MATCH_STARSUB)) + ret = string_list_dollar_star (quote_list (save)); + else if ((mflags & MATCH_STARSUB) == MATCH_STARSUB) + ret = string_list_dollar_star (save); + else if ((mflags & MATCH_QUOTED) == MATCH_QUOTED) + ret = string_list_dollar_at (save, qflags); + else + ret = string_list_dollar_star (save); +#else + ret = string_list_pos_params (pchar, save, qflags); +#endif + + dispose_words (save); + + return (ret); +} + +/* Perform pattern substitution on VALUE, which is the expansion of + VARNAME. PATSUB is an expression supplying the pattern to match + and the string to substitute. QUOTED is a flags word containing + the type of quoting currently in effect. */ +static char * +parameter_brace_patsub (varname, value, ind, patsub, quoted, flags) + char *varname, *value; + int ind; + char *patsub; + int quoted, flags; +{ + int vtype, mflags, starsub, delim; + char *val, *temp, *pat, *rep, *p, *lpatsub, *tt; + SHELL_VAR *v; + + if (value == 0) + return ((char *)NULL); + + this_command_name = varname; + + vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); + if (vtype == -1) + return ((char *)NULL); + + starsub = vtype & VT_STARSUB; + vtype &= ~VT_STARSUB; + + mflags = 0; + /* PATSUB is never NULL when this is called. */ + if (*patsub == '/') + { + mflags |= MATCH_GLOBREP; + patsub++; + } + + /* Malloc this because expand_string_if_necessary or one of the expansion + functions in its call chain may free it on a substitution error. */ + lpatsub = savestring (patsub); + + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + mflags |= MATCH_QUOTED; + + if (starsub) + mflags |= MATCH_STARSUB; + + /* If the pattern starts with a `/', make sure we skip over it when looking + for the replacement delimiter. */ + delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0); + if (lpatsub[delim] == '/') + { + lpatsub[delim] = 0; + rep = lpatsub + delim + 1; + } + else + rep = (char *)NULL; + + if (rep && *rep == '\0') + rep = (char *)NULL; + + /* Perform the same expansions on the pattern as performed by the + pattern removal expansions. */ + pat = getpattern (lpatsub, quoted, 1); + + if (rep) + { + if ((mflags & MATCH_QUOTED) == 0) + rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit); + else + rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit); + } + + /* ksh93 doesn't allow the match specifier to be a part of the expanded + pattern. This is an extension. Make sure we don't anchor the pattern + at the beginning or end of the string if we're doing global replacement, + though. */ + p = pat; + if (mflags & MATCH_GLOBREP) + mflags |= MATCH_ANY; + else if (pat && pat[0] == '#') + { + mflags |= MATCH_BEG; + p++; + } + else if (pat && pat[0] == '%') + { + mflags |= MATCH_END; + p++; + } + else + mflags |= MATCH_ANY; + + /* OK, we now want to substitute REP for PAT in VAL. If + flags & MATCH_GLOBREP is non-zero, the substitution is done + everywhere, otherwise only the first occurrence of PAT is + replaced. The pattern matching code doesn't understand + CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable + values passed in (VT_VARIABLE) so the pattern substitution + code works right. We need to requote special chars after + we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the + other cases if QUOTED == 0, since the posparams and arrays + indexed by * or @ do special things when QUOTED != 0. */ + + switch (vtype) + { + case VT_VARIABLE: + case VT_ARRAYMEMBER: + temp = pat_subst (val, p, rep, mflags); + if (vtype == VT_VARIABLE) + FREE (val); + if (temp) + { + tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp); + free (temp); + temp = tt; + } + break; + case VT_POSPARMS: + temp = pos_params_pat_subst (val, p, rep, mflags); + if (temp && (mflags & MATCH_QUOTED) == 0) + { + tt = quote_escapes (temp); + free (temp); + temp = tt; + } + break; +#if defined (ARRAY_VARS) + case VT_ARRAYVAR: + temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags) + : array_patsub (array_cell (v), p, rep, mflags); + /* Don't call quote_escapes anymore; array_patsub calls + array_quote_escapes as appropriate before adding the + space separators; ditto for assoc_patsub. */ + break; +#endif + } + + FREE (pat); + FREE (rep); + free (lpatsub); + + return temp; +} + +/****************************************************************/ +/* */ +/* Functions to perform case modification on variable values */ +/* */ +/****************************************************************/ + +/* Do case modification on the positional parameters. */ + +static char * +pos_params_modcase (string, pat, modop, mflags) + char *string, *pat; + int modop; + int mflags; +{ + WORD_LIST *save, *params; + WORD_DESC *w; + char *ret; + int pchar, qflags; + + save = params = list_rest_of_args (); + if (save == 0) + return ((char *)NULL); + + for ( ; params; params = params->next) + { + ret = sh_modcase (params->word->word, pat, modop); + w = alloc_word_desc (); + w->word = ret ? ret : savestring (""); + dispose_word (params->word); + params->word = w; + } + + pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@'; + qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0; + + ret = string_list_pos_params (pchar, save, qflags); + dispose_words (save); + + return (ret); +} + +/* Perform case modification on VALUE, which is the expansion of + VARNAME. MODSPEC is an expression supplying the type of modification + to perform. QUOTED is a flags word containing the type of quoting + currently in effect. */ +static char * +parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags) + char *varname, *value; + int ind, modspec; + char *patspec; + int quoted, flags; +{ + int vtype, starsub, modop, mflags, x; + char *val, *temp, *pat, *p, *lpat, *tt; + SHELL_VAR *v; + + if (value == 0) + return ((char *)NULL); + + this_command_name = varname; + + vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); + if (vtype == -1) + return ((char *)NULL); + + starsub = vtype & VT_STARSUB; + vtype &= ~VT_STARSUB; + + modop = 0; + mflags = 0; + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + mflags |= MATCH_QUOTED; + if (starsub) + mflags |= MATCH_STARSUB; + + p = patspec; + if (modspec == '^') + { + x = p && p[0] == modspec; + modop = x ? CASE_UPPER : CASE_UPFIRST; + p += x; + } + else if (modspec == ',') + { + x = p && p[0] == modspec; + modop = x ? CASE_LOWER : CASE_LOWFIRST; + p += x; + } + else if (modspec == '~') + { + x = p && p[0] == modspec; + modop = x ? CASE_TOGGLEALL : CASE_TOGGLE; + p += x; + } + + lpat = p ? savestring (p) : 0; + /* Perform the same expansions on the pattern as performed by the + pattern removal expansions. FOR LATER */ + pat = lpat ? getpattern (lpat, quoted, 1) : 0; + + /* OK, now we do the case modification. */ + switch (vtype) + { + case VT_VARIABLE: + case VT_ARRAYMEMBER: + temp = sh_modcase (val, pat, modop); + if (vtype == VT_VARIABLE) + FREE (val); + if (temp) + { + tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp); + free (temp); + temp = tt; + } + break; + + case VT_POSPARMS: + temp = pos_params_modcase (val, pat, modop, mflags); + if (temp && (mflags & MATCH_QUOTED) == 0) + { + tt = quote_escapes (temp); + free (temp); + temp = tt; + } + break; + +#if defined (ARRAY_VARS) + case VT_ARRAYVAR: + temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags) + : array_modcase (array_cell (v), pat, modop, mflags); + /* Don't call quote_escapes; array_modcase calls array_quote_escapes + as appropriate before adding the space separators; ditto for + assoc_modcase. */ + break; +#endif + } + + FREE (pat); + free (lpat); + + return temp; +} + +/* Check for unbalanced parens in S, which is the contents of $(( ... )). If + any occur, this must be a nested command substitution, so return 0. + Otherwise, return 1. A valid arithmetic expression must always have a + ( before a matching ), so any cases where there are more right parens + means that this must not be an arithmetic expression, though the parser + will not accept it without a balanced total number of parens. */ +static int +chk_arithsub (s, len) + const char *s; + int len; +{ + int i, count; + DECLARE_MBSTATE; + + i = count = 0; + while (i < len) + { + if (s[i] == LPAREN) + count++; + else if (s[i] == RPAREN) + { + count--; + if (count < 0) + return 0; + } + + switch (s[i]) + { + default: + ADVANCE_CHAR (s, len, i); + break; + + case '\\': + i++; + if (s[i]) + ADVANCE_CHAR (s, len, i); + break; + + case '\'': + i = skip_single_quoted (s, len, ++i); + break; + + case '"': + i = skip_double_quoted ((char *)s, len, ++i); + break; + } + } + + return (count == 0); +} + +/****************************************************************/ +/* */ +/* Functions to perform parameter expansion on a string */ +/* */ +/****************************************************************/ + +/* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */ +static WORD_DESC * +parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at) + char *string; + int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at, pflags; +{ + int check_nullness, var_is_set, var_is_null, var_is_special; + int want_substring, want_indir, want_patsub, want_casemod; + char *name, *value, *temp, *temp1; + WORD_DESC *tdesc, *ret; + int t_index, sindex, c, tflag, modspec; + intmax_t number; + arrayind_t ind; + + temp = temp1 = value = (char *)NULL; + var_is_set = var_is_null = var_is_special = check_nullness = 0; + want_substring = want_indir = want_patsub = want_casemod = 0; + + sindex = *indexp; + t_index = ++sindex; + /* ${#var} doesn't have any of the other parameter expansions on it. */ + if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */ + name = string_extract (string, &t_index, "}", SX_VARNAME); + else +#if defined (CASEMOD_EXPANSIONS) + /* To enable case-toggling expansions using the `~' operator character + change the 1 to 0. */ +# if defined (CASEMOD_CAPCASE) + name = string_extract (string, &t_index, "#%^,~:-=?+/}", SX_VARNAME); +# else + name = string_extract (string, &t_index, "#%^,:-=?+/}", SX_VARNAME); +# endif /* CASEMOD_CAPCASE */ +#else + name = string_extract (string, &t_index, "#%:-=?+/}", SX_VARNAME); +#endif /* CASEMOD_EXPANSIONS */ + + ret = 0; + tflag = 0; + + ind = INTMAX_MIN; + + /* If the name really consists of a special variable, then make sure + that we have the entire name. We don't allow indirect references + to special variables except `#', `?', `@' and `*'. */ + if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) || + (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index]))) + { + t_index++; + temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0); + name = (char *)xrealloc (name, 3 + (strlen (temp1))); + *name = string[sindex]; + if (string[sindex] == '!') + { + /* indirect reference of $#, $?, $@, or $* */ + name[1] = string[sindex + 1]; + strcpy (name + 2, temp1); + } + else + strcpy (name + 1, temp1); + free (temp1); + } + sindex = t_index; + + /* Find out what character ended the variable name. Then + do the appropriate thing. */ + if (c = string[sindex]) + sindex++; + + /* If c is followed by one of the valid parameter expansion + characters, move past it as normal. If not, assume that + a substring specification is being given, and do not move + past it. */ + if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex])) + { + check_nullness++; + if (c = string[sindex]) + sindex++; + } + else if (c == ':' && string[sindex] != RBRACE) + want_substring = 1; + else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */ + want_patsub = 1; +#if defined (CASEMOD_EXPANSIONS) + else if (c == '^' || c == ',' || c == '~') + { + modspec = c; + want_casemod = 1; + } +#endif + + /* Catch the valid and invalid brace expressions that made it through the + tests above. */ + /* ${#-} is a valid expansion and means to take the length of $-. + Similarly for ${#?} and ${##}... */ + if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 && + VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE) + { + name = (char *)xrealloc (name, 3); + name[1] = c; + name[2] = '\0'; + c = string[sindex++]; + } + + /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */ + if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 && + member (c, "%:=+/") && string[sindex] == RBRACE) + { + temp = (char *)NULL; + goto bad_substitution; + } + + /* Indirect expansion begins with a `!'. A valid indirect expansion is + either a variable name, one of the positional parameters or a special + variable that expands to one of the positional parameters. */ + want_indir = *name == '!' && + (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) + || VALID_INDIR_PARAM (name[1])); + + /* Determine the value of this variable. */ + + /* Check for special variables, directly referenced. */ + if (SPECIAL_VAR (name, want_indir)) + var_is_special++; + + /* Check for special expansion things, like the length of a parameter */ + if (*name == '#' && name[1]) + { + /* If we are not pointing at the character just after the + closing brace, then we haven't gotten all of the name. + Since it begins with a special character, this is a bad + substitution. Also check NAME for validity before trying + to go on. */ + if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0)) + { + temp = (char *)NULL; + goto bad_substitution; + } + + number = parameter_brace_expand_length (name); + if (number == INTMAX_MIN && unbound_vars_is_error) + { + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (name+1); + free (name); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } + free (name); + + *indexp = sindex; + if (number < 0) + return (&expand_wdesc_error); + else + { + ret = alloc_word_desc (); + ret->word = itos (number); + return ret; + } + } + + /* ${@} is identical to $@. */ + if (name[0] == '@' && name[1] == '\0') + { + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 1; + + if (contains_dollar_at) + *contains_dollar_at = 1; + } + + /* Process ${!PREFIX*} expansion. */ + if (want_indir && string[sindex - 1] == RBRACE && + (string[sindex - 2] == '*' || string[sindex - 2] == '@') && + legal_variable_starter ((unsigned char) name[1])) + { + char **x; + WORD_LIST *xlist; + + temp1 = savestring (name + 1); + number = strlen (temp1); + temp1[number - 1] = '\0'; + x = all_variables_matching_prefix (temp1); + xlist = strvec_to_word_list (x, 0, 0); + if (string[sindex - 2] == '*') + temp = string_list_dollar_star (xlist); + else + { + temp = string_list_dollar_at (xlist, quoted); + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 1; + if (contains_dollar_at) + *contains_dollar_at = 1; + } + free (x); + dispose_words (xlist); + free (temp1); + *indexp = sindex; + + free (name); + + ret = alloc_word_desc (); + ret->word = temp; + return ret; + } + +#if defined (ARRAY_VARS) + /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */ + if (want_indir && string[sindex - 1] == RBRACE && + string[sindex - 2] == ']' && valid_array_reference (name+1)) + { + char *x, *x1; + + temp1 = savestring (name + 1); + x = array_variable_name (temp1, &x1, (int *)0); /* [ */ + FREE (x); + if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']') + { + temp = array_keys (temp1, quoted); /* handles assoc vars too */ + if (x1[0] == '@') + { + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 1; + if (contains_dollar_at) + *contains_dollar_at = 1; + } + + free (temp1); + *indexp = sindex; + + ret = alloc_word_desc (); + ret->word = temp; + return ret; + } + + free (temp1); + } +#endif /* ARRAY_VARS */ + + /* Make sure that NAME is valid before trying to go on. */ + if (valid_brace_expansion_word (want_indir ? name + 1 : name, + var_is_special) == 0) + { + temp = (char *)NULL; + goto bad_substitution; + } + + if (want_indir) + tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at); + else + tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&PF_NOSPLIT2), &ind); + + if (tdesc) + { + temp = tdesc->word; + tflag = tdesc->flags; + dispose_word_desc (tdesc); + } + else + temp = (char *)0; + +#if defined (ARRAY_VARS) + if (valid_array_reference (name)) + chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at); +#endif + + var_is_set = temp != (char *)0; + var_is_null = check_nullness && (var_is_set == 0 || *temp == 0); + + /* Get the rest of the stuff inside the braces. */ + if (c && c != RBRACE) + { + /* Extract the contents of the ${ ... } expansion + according to the Posix.2 rules. */ + value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD); + if (string[sindex] == RBRACE) + sindex++; + else + goto bad_substitution; + } + else + value = (char *)NULL; + + *indexp = sindex; + + /* All the cases where an expansion can possibly generate an unbound + variable error. */ + if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE) + { + if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1])) + { + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (name); + FREE (value); + FREE (temp); + free (name); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } + } + + /* If this is a substring spec, process it and add the result. */ + if (want_substring) + { + temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); + FREE (name); + FREE (value); + FREE (temp); + + if (temp1 == &expand_param_error) + return (&expand_wdesc_error); + else if (temp1 == &expand_param_fatal) + return (&expand_wdesc_fatal); + + ret = alloc_word_desc (); + ret->word = temp1; + if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ret->flags |= W_QUOTED|W_HASQUOTEDNULL; + return ret; + } + else if (want_patsub) + { + temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); + FREE (name); + FREE (value); + FREE (temp); + + if (temp1 == &expand_param_error) + return (&expand_wdesc_error); + else if (temp1 == &expand_param_fatal) + return (&expand_wdesc_fatal); + + ret = alloc_word_desc (); + ret->word = temp1; + if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ret->flags |= W_QUOTED|W_HASQUOTEDNULL; + return ret; + } +#if defined (CASEMOD_EXPANSIONS) + else if (want_casemod) + { + temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); + FREE (name); + FREE (value); + FREE (temp); + + if (temp1 == &expand_param_error) + return (&expand_wdesc_error); + else if (temp1 == &expand_param_fatal) + return (&expand_wdesc_fatal); + + ret = alloc_word_desc (); + ret->word = temp1; + if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ret->flags |= W_QUOTED|W_HASQUOTEDNULL; + return ret; + } +#endif + + /* Do the right thing based on which character ended the variable name. */ + switch (c) + { + default: + case '\0': + bad_substitution: + last_command_exit_value = EXECUTION_FAILURE; + report_error (_("%s: bad substitution"), string ? string : "??"); + FREE (value); + FREE (temp); + free (name); + return &expand_wdesc_error; + + case RBRACE: + break; + + case '#': /* ${param#[#]pattern} */ + case '%': /* ${param%[%]pattern} */ + if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0') + { + FREE (value); + break; + } + temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); + free (temp); + free (value); + free (name); + + ret = alloc_word_desc (); + ret->word = temp1; + if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ret->flags |= W_QUOTED|W_HASQUOTEDNULL; + return ret; + + case '-': + case '=': + case '?': + case '+': + if (var_is_set && var_is_null == 0) + { + /* If the operator is `+', we don't want the value of the named + variable for anything, just the value of the right hand side. */ + if (c == '+') + { + /* XXX -- if we're double-quoted and the named variable is "$@", + we want to turn off any special handling of "$@" -- + we're not using it, so whatever is on the rhs applies. */ + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 0; + if (contains_dollar_at) + *contains_dollar_at = 0; + + FREE (temp); + if (value) + { + /* From Posix discussion on austin-group list. Issue 221 + requires that backslashes escaping `}' inside + double-quoted ${...} be removed. */ + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + quoted |= Q_DOLBRACE; + ret = parameter_brace_expand_rhs (name, value, c, + quoted, + quoted_dollar_atp, + contains_dollar_at); + /* XXX - fix up later, esp. noting presence of + W_HASQUOTEDNULL in ret->flags */ + free (value); + } + else + temp = (char *)NULL; + } + else + { + FREE (value); + } + /* Otherwise do nothing; just use the value in TEMP. */ + } + else /* VAR not set or VAR is NULL. */ + { + FREE (temp); + temp = (char *)NULL; + if (c == '=' && var_is_special) + { + last_command_exit_value = EXECUTION_FAILURE; + report_error (_("$%s: cannot assign in this way"), name); + free (name); + free (value); + return &expand_wdesc_error; + } + else if (c == '?') + { + parameter_brace_expand_error (name, value); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } + else if (c != '+') + { + /* XXX -- if we're double-quoted and the named variable is "$@", + we want to turn off any special handling of "$@" -- + we're not using it, so whatever is on the rhs applies. */ + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) + *quoted_dollar_atp = 0; + if (contains_dollar_at) + *contains_dollar_at = 0; + + /* From Posix discussion on austin-group list. Issue 221 requires + that backslashes escaping `}' inside double-quoted ${...} be + removed. */ + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + quoted |= Q_DOLBRACE; + ret = parameter_brace_expand_rhs (name, value, c, quoted, + quoted_dollar_atp, + contains_dollar_at); + /* XXX - fix up later, esp. noting presence of + W_HASQUOTEDNULL in tdesc->flags */ + } + free (value); + } + + break; + } + free (name); + + if (ret == 0) + { + ret = alloc_word_desc (); + ret->flags = tflag; + ret->word = temp; + } + return (ret); +} + +/* Expand a single ${xxx} expansion. The braces are optional. When + the braces are used, parameter_brace_expand() does the work, + possibly calling param_expand recursively. */ +static WORD_DESC * +param_expand (string, sindex, quoted, expanded_something, + contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p, + pflags) + char *string; + int *sindex, quoted, *expanded_something, *contains_dollar_at; + int *quoted_dollar_at_p, *had_quoted_null_p, pflags; +{ + char *temp, *temp1, uerror[3]; + int zindex, t_index, expok; + unsigned char c; + intmax_t number; + SHELL_VAR *var; + WORD_LIST *list; + WORD_DESC *tdesc, *ret; + int tflag; + + zindex = *sindex; + c = string[++zindex]; + + temp = (char *)NULL; + ret = tdesc = (WORD_DESC *)NULL; + tflag = 0; + + /* Do simple cases first. Switch on what follows '$'. */ + switch (c) + { + /* $0 .. $9? */ + case '0': + case '1': + case '2': + case '3': + case '4': + case '5': + case '6': + case '7': + case '8': + case '9': + temp1 = dollar_vars[TODIGIT (c)]; + if (unbound_vars_is_error && temp1 == (char *)NULL) + { + uerror[0] = '$'; + uerror[1] = c; + uerror[2] = '\0'; + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (uerror); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } + if (temp1) + temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ? quote_string (temp1) + : quote_escapes (temp1); + else + temp = (char *)NULL; + + break; + + /* $$ -- pid of the invoking shell. */ + case '$': + temp = itos (dollar_dollar_pid); + break; + + /* $# -- number of positional parameters. */ + case '#': + temp = itos (number_of_args ()); + break; + + /* $? -- return value of the last synchronous command. */ + case '?': + temp = itos (last_command_exit_value); + break; + + /* $- -- flags supplied to the shell on invocation or by `set'. */ + case '-': + temp = which_set_flags (); + break; + + /* $! -- Pid of the last asynchronous command. */ + case '!': + /* If no asynchronous pids have been created, expand to nothing. + If `set -u' has been executed, and no async processes have + been created, this is an expansion error. */ + if (last_asynchronous_pid == NO_PID) + { + if (expanded_something) + *expanded_something = 0; + temp = (char *)NULL; + if (unbound_vars_is_error) + { + uerror[0] = '$'; + uerror[1] = c; + uerror[2] = '\0'; + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (uerror); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } + } + else + temp = itos (last_asynchronous_pid); + break; + + /* The only difference between this and $@ is when the arg is quoted. */ + case '*': /* `$*' */ + list = list_rest_of_args (); + +#if 0 + /* According to austin-group posix proposal by Geoff Clare in + <20090505091501.GA10097@squonk.masqnet> of 5 May 2009: + + "The shell shall write a message to standard error and + immediately exit when it tries to expand an unset parameter + other than the '@' and '*' special parameters." + */ + + if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0) + { + uerror[0] = '$'; + uerror[1] = '*'; + uerror[2] = '\0'; + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (uerror); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } +#endif + + /* If there are no command-line arguments, this should just + disappear if there are other characters in the expansion, + even if it's quoted. */ + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0) + temp = (char *)NULL; + else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE)) + { + /* If we have "$*" we want to make a string of the positional + parameters, separated by the first character of $IFS, and + quote the whole string, including the separators. If IFS + is unset, the parameters are separated by ' '; if $IFS is + null, the parameters are concatenated. */ + temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list); + if (temp) + { + temp1 = quote_string (temp); + if (*temp == 0) + tflag |= W_HASQUOTEDNULL; + free (temp); + temp = temp1; + } + } + else + { + /* We check whether or not we're eventually going to split $* here, + for example when IFS is empty and we are processing the rhs of + an assignment statement. In that case, we don't separate the + arguments at all. Otherwise, if the $* is not quoted it is + identical to $@ */ +#if 1 +# if defined (HANDLE_MULTIBYTE) + if (expand_no_split_dollar_star && ifs_firstc[0] == 0) +# else + if (expand_no_split_dollar_star && ifs_firstc == 0) +# endif + temp = string_list_dollar_star (list); + else + temp = string_list_dollar_at (list, quoted); +#else + temp = string_list_dollar_at (list, quoted); +#endif + if (expand_no_split_dollar_star == 0 && contains_dollar_at) + *contains_dollar_at = 1; + } + + dispose_words (list); + break; + + /* When we have "$@" what we want is "$1" "$2" "$3" ... This + means that we have to turn quoting off after we split into + the individually quoted arguments so that the final split + on the first character of $IFS is still done. */ + case '@': /* `$@' */ + list = list_rest_of_args (); + +#if 0 + /* According to austin-group posix proposal by Geoff Clare in + <20090505091501.GA10097@squonk.masqnet> of 5 May 2009: + + "The shell shall write a message to standard error and + immediately exit when it tries to expand an unset parameter + other than the '@' and '*' special parameters." + */ + + if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0) + { + uerror[0] = '$'; + uerror[1] = '@'; + uerror[2] = '\0'; + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (uerror); + return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); + } +#endif + + /* We want to flag the fact that we saw this. We can't turn + off quoting entirely, because other characters in the + string might need it (consider "\"$@\""), but we need some + way to signal that the final split on the first character + of $IFS should be done, even though QUOTED is 1. */ + /* XXX - should this test include Q_PATQUOTE? */ + if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + *quoted_dollar_at_p = 1; + if (contains_dollar_at) + *contains_dollar_at = 1; + + /* We want to separate the positional parameters with the first + character of $IFS in case $IFS is something other than a space. + We also want to make sure that splitting is done no matter what -- + according to POSIX.2, this expands to a list of the positional + parameters no matter what IFS is set to. */ +#if 0 + temp = string_list_dollar_at (list, quoted); +#else + temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted); +#endif + + dispose_words (list); + break; + + case LBRACE: + tdesc = parameter_brace_expand (string, &zindex, quoted, pflags, + quoted_dollar_at_p, + contains_dollar_at); + + if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal) + return (tdesc); + temp = tdesc ? tdesc->word : (char *)0; + + /* XXX */ + /* Quoted nulls should be removed if there is anything else + in the string. */ + /* Note that we saw the quoted null so we can add one back at + the end of this function if there are no other characters + in the string, discard TEMP, and go on. The exception to + this is when we have "${@}" and $1 is '', since $@ needs + special handling. */ + if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp)) + { + if (had_quoted_null_p) + *had_quoted_null_p = 1; + if (*quoted_dollar_at_p == 0) + { + free (temp); + tdesc->word = temp = (char *)NULL; + } + + } + + ret = tdesc; + goto return0; + + /* Do command or arithmetic substitution. */ + case LPAREN: + /* We have to extract the contents of this paren substitution. */ + t_index = zindex + 1; + temp = extract_command_subst (string, &t_index, 0); + zindex = t_index; + + /* For Posix.2-style `$(( ))' arithmetic substitution, + extract the expression and pass it to the evaluator. */ + if (temp && *temp == LPAREN) + { + char *temp2; + temp1 = temp + 1; + temp2 = savestring (temp1); + t_index = strlen (temp2) - 1; + + if (temp2[t_index] != RPAREN) + { + free (temp2); + goto comsub; + } + + /* Cut off ending `)' */ + temp2[t_index] = '\0'; + + if (chk_arithsub (temp2, t_index) == 0) + { + free (temp2); +#if 0 + internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution")); +#endif + goto comsub; + } + + /* Expand variables found inside the expression. */ + temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES); + free (temp2); + +arithsub: + /* No error messages. */ + this_command_name = (char *)NULL; + number = evalexp (temp1, &expok); + free (temp); + free (temp1); + if (expok == 0) + { + if (interactive_shell == 0 && posixly_correct) + { + last_command_exit_value = EXECUTION_FAILURE; + return (&expand_wdesc_fatal); + } + else + return (&expand_wdesc_error); + } + temp = itos (number); + break; + } + +comsub: + if (pflags & PF_NOCOMSUB) + /* we need zindex+1 because string[zindex] == RPAREN */ + temp1 = substring (string, *sindex, zindex+1); + else + { + tdesc = command_substitute (temp, quoted); + temp1 = tdesc ? tdesc->word : (char *)NULL; + if (tdesc) + dispose_word_desc (tdesc); + } + FREE (temp); + temp = temp1; + break; + + /* Do POSIX.2d9-style arithmetic substitution. This will probably go + away in a future bash release. */ + case '[': + /* Extract the contents of this arithmetic substitution. */ + t_index = zindex + 1; + temp = extract_arithmetic_subst (string, &t_index); + zindex = t_index; + if (temp == 0) + { + temp = savestring (string); + if (expanded_something) + *expanded_something = 0; + goto return0; + } + + /* Do initial variable expansion. */ + temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES); + + goto arithsub; + + default: + /* Find the variable in VARIABLE_LIST. */ + temp = (char *)NULL; + + for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++) + ; + temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL; + + /* If this isn't a variable name, then just output the `$'. */ + if (temp1 == 0 || *temp1 == '\0') + { + FREE (temp1); + temp = (char *)xmalloc (2); + temp[0] = '$'; + temp[1] = '\0'; + if (expanded_something) + *expanded_something = 0; + goto return0; + } + + /* If the variable exists, return its value cell. */ + var = find_variable (temp1); + + if (var && invisible_p (var) == 0 && var_isset (var)) + { +#if defined (ARRAY_VARS) + if (assoc_p (var) || array_p (var)) + { + temp = array_p (var) ? array_reference (array_cell (var), 0) + : assoc_reference (assoc_cell (var), "0"); + if (temp) + temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ? quote_string (temp) + : quote_escapes (temp); + else if (unbound_vars_is_error) + goto unbound_variable; + } + else +#endif + { + temp = value_cell (var); + + temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) + ? quote_string (temp) + : quote_escapes (temp); + } + + free (temp1); + + goto return0; + } + + temp = (char *)NULL; + +unbound_variable: + if (unbound_vars_is_error) + { + last_command_exit_value = EXECUTION_FAILURE; + err_unboundvar (temp1); + } + else + { + free (temp1); + goto return0; + } + + free (temp1); + last_command_exit_value = EXECUTION_FAILURE; + return ((unbound_vars_is_error && interactive_shell == 0) + ? &expand_wdesc_fatal + : &expand_wdesc_error); + } + + if (string[zindex]) + zindex++; + +return0: + *sindex = zindex; + + if (ret == 0) + { + ret = alloc_word_desc (); + ret->flags = tflag; /* XXX */ + ret->word = temp; + } + return ret; +} + +/* Make a word list which is the result of parameter and variable + expansion, command substitution, arithmetic substitution, and + quote removal of WORD. Return a pointer to a WORD_LIST which is + the result of the expansion. If WORD contains a null word, the + word list returned is also null. + + QUOTED contains flag values defined in shell.h. + + ISEXP is used to tell expand_word_internal that the word should be + treated as the result of an expansion. This has implications for + how IFS characters in the word are treated. + + CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null + they point to an integer value which receives information about expansion. + CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero. + EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions, + else zero. + + This only does word splitting in the case of $@ expansion. In that + case, we split on ' '. */ + +/* Values for the local variable quoted_state. */ +#define UNQUOTED 0 +#define PARTIALLY_QUOTED 1 +#define WHOLLY_QUOTED 2 + +static WORD_LIST * +expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something) + WORD_DESC *word; + int quoted, isexp; + int *contains_dollar_at; + int *expanded_something; +{ + WORD_LIST *list; + WORD_DESC *tword; + + /* The intermediate string that we build while expanding. */ + char *istring; + + /* The current size of the above object. */ + int istring_size; + + /* Index into ISTRING. */ + int istring_index; + + /* Temporary string storage. */ + char *temp, *temp1; + + /* The text of WORD. */ + register char *string; + + /* The size of STRING. */ + size_t string_size; + + /* The index into STRING. */ + int sindex; + + /* This gets 1 if we see a $@ while quoted. */ + int quoted_dollar_at; + + /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on + whether WORD contains no quoting characters, a partially quoted + string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */ + int quoted_state; + + /* State flags */ + int had_quoted_null; + int has_dollar_at; + int tflag; + int pflags; /* flags passed to param_expand */ + + int assignoff; /* If assignment, offset of `=' */ + + register unsigned char c; /* Current character. */ + int t_index; /* For calls to string_extract_xxx. */ + + char twochars[2]; + + DECLARE_MBSTATE; + + istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE); + istring[istring_index = 0] = '\0'; + quoted_dollar_at = had_quoted_null = has_dollar_at = 0; + quoted_state = UNQUOTED; + + string = word->word; + if (string == 0) + goto finished_with_string; + /* Don't need the string length for the SADD... and COPY_ macros unless + multibyte characters are possible. */ + string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1; + + if (contains_dollar_at) + *contains_dollar_at = 0; + + assignoff = -1; + + /* Begin the expansion. */ + + for (sindex = 0; ;) + { + c = string[sindex]; + + /* Case on toplevel character. */ + switch (c) + { + case '\0': + goto finished_with_string; + + case CTLESC: + sindex++; +#if HANDLE_MULTIBYTE + if (MB_CUR_MAX > 1 && string[sindex]) + { + SADD_MBQCHAR_BODY(temp, string, sindex, string_size); + } + else +#endif + { + temp = (char *)xmalloc (3); + temp[0] = CTLESC; + temp[1] = c = string[sindex]; + temp[2] = '\0'; + } + +dollar_add_string: + if (string[sindex]) + sindex++; + +add_string: + if (temp) + { + istring = sub_append_string (temp, istring, &istring_index, &istring_size); + temp = (char *)0; + } + + break; + +#if defined (PROCESS_SUBSTITUTION) + /* Process substitution. */ + case '<': + case '>': + { + if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct) + { + sindex--; /* add_character: label increments sindex */ + goto add_character; + } + else + t_index = sindex + 1; /* skip past both '<' and LPAREN */ + + temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index); /*))*/ + sindex = t_index; + + /* If the process substitution specification is `<()', we want to + open the pipe for writing in the child and produce output; if + it is `>()', we want to open the pipe for reading in the child + and consume input. */ + temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0; + + FREE (temp1); + + goto dollar_add_string; + } +#endif /* PROCESS_SUBSTITUTION */ + + case '=': + /* Posix.2 section 3.6.1 says that tildes following `=' in words + which are not assignment statements are not expanded. If the + shell isn't in posix mode, though, we perform tilde expansion + on `likely candidate' unquoted assignment statements (flags + include W_ASSIGNMENT but not W_QUOTED). A likely candidate + contains an unquoted :~ or =~. Something to think about: we + now have a flag that says to perform tilde expansion on arguments + to `assignment builtins' like declare and export that look like + assignment statements. We now do tilde expansion on such words + even in POSIX mode. */ + if (word->flags & (W_ASSIGNRHS|W_NOTILDE)) + { + if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) + goto add_ifs_character; + else + goto add_character; + } + /* If we're not in posix mode or forcing assignment-statement tilde + expansion, note where the `=' appears in the word and prepare to + do tilde expansion following the first `='. */ + if ((word->flags & W_ASSIGNMENT) && + (posixly_correct == 0 || (word->flags & W_TILDEEXP)) && + assignoff == -1 && sindex > 0) + assignoff = sindex; + if (sindex == assignoff && string[sindex+1] == '~') /* XXX */ + word->flags |= W_ITILDE; +#if 0 + else if ((word->flags & W_ASSIGNMENT) && + (posixly_correct == 0 || (word->flags & W_TILDEEXP)) && + string[sindex+1] == '~') + word->flags |= W_ITILDE; +#endif + if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) + goto add_ifs_character; + else + goto add_character; + + case ':': + if (word->flags & W_NOTILDE) + { + if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) + goto add_ifs_character; + else + goto add_character; + } + + if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) && + string[sindex+1] == '~') + word->flags |= W_ITILDE; + + if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) + goto add_ifs_character; + else + goto add_character; + + case '~': + /* If the word isn't supposed to be tilde expanded, or we're not + at the start of a word or after an unquoted : or = in an + assignment statement, we don't do tilde expansion. */ + if ((word->flags & (W_NOTILDE|W_DQUOTE)) || + (sindex > 0 && ((word->flags & W_ITILDE) == 0)) || + (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) + { + word->flags &= ~W_ITILDE; + if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0) + goto add_ifs_character; + else + goto add_character; + } + + if (word->flags & W_ASSIGNRHS) + tflag = 2; + else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP)) + tflag = 1; + else + tflag = 0; + + temp = bash_tilde_find_word (string + sindex, tflag, &t_index); + + word->flags &= ~W_ITILDE; + + if (temp && *temp && t_index > 0) + { + temp1 = bash_tilde_expand (temp, tflag); + if (temp1 && *temp1 == '~' && STREQ (temp, temp1)) + { + FREE (temp); + FREE (temp1); + goto add_character; /* tilde expansion failed */ + } + free (temp); + temp = temp1; + sindex += t_index; + goto add_quoted_string; /* XXX was add_string */ + } + else + { + FREE (temp); + goto add_character; + } + + case '$': + if (expanded_something) + *expanded_something = 1; + + has_dollar_at = 0; + pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0; + if (word->flags & W_NOSPLIT2) + pflags |= PF_NOSPLIT2; + if (word->flags & W_ASSIGNRHS) + pflags |= PF_ASSIGNRHS; + tword = param_expand (string, &sindex, quoted, expanded_something, + &has_dollar_at, "ed_dollar_at, + &had_quoted_null, pflags); + + if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal) + { + free (string); + free (istring); + return ((tword == &expand_wdesc_error) ? &expand_word_error + : &expand_word_fatal); + } + if (contains_dollar_at && has_dollar_at) + *contains_dollar_at = 1; + + if (tword && (tword->flags & W_HASQUOTEDNULL)) + had_quoted_null = 1; + + temp = tword ? tword->word : (char *)NULL; + dispose_word_desc (tword); + + goto add_string; + break; + + case '`': /* Backquoted command substitution. */ + { + t_index = sindex++; + + temp = string_extract (string, &sindex, "`", SX_REQMATCH); + /* The test of sindex against t_index is to allow bare instances of + ` to pass through, for backwards compatibility. */ + if (temp == &extract_string_error || temp == &extract_string_fatal) + { + if (sindex - 1 == t_index) + { + sindex = t_index; + goto add_character; + } + last_command_exit_value = EXECUTION_FAILURE; + report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index); + free (string); + free (istring); + return ((temp == &extract_string_error) ? &expand_word_error + : &expand_word_fatal); + } + + if (expanded_something) + *expanded_something = 1; + + if (word->flags & W_NOCOMSUB) + /* sindex + 1 because string[sindex] == '`' */ + temp1 = substring (string, t_index, sindex + 1); + else + { + de_backslash (temp); + tword = command_substitute (temp, quoted); + temp1 = tword ? tword->word : (char *)NULL; + if (tword) + dispose_word_desc (tword); + } + FREE (temp); + temp = temp1; + goto dollar_add_string; + } + + case '\\': + if (string[sindex + 1] == '\n') + { + sindex += 2; + continue; + } + + c = string[++sindex]; + + if (quoted & Q_HERE_DOCUMENT) + tflag = CBSHDOC; + else if (quoted & Q_DOUBLE_QUOTES) + tflag = CBSDQUOTE; + else + tflag = 0; + + /* From Posix discussion on austin-group list: Backslash escaping + a } in ${...} is removed. Issue 0000221 */ + if ((quoted & Q_DOLBRACE) && c == RBRACE) + { + SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size); + } + else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0)) + { + SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size); + } + else if (c == 0) + { + c = CTLNUL; + sindex--; /* add_character: label increments sindex */ + goto add_character; + } + else + { + SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size); + } + + sindex++; +add_twochars: + /* BEFORE jumping here, we need to increment sindex if appropriate */ + RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, + DEFAULT_ARRAY_SIZE); + istring[istring_index++] = twochars[0]; + istring[istring_index++] = twochars[1]; + istring[istring_index] = '\0'; + + break; + + case '"': +#if 0 + if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE)) +#else + if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) +#endif + goto add_character; + + t_index = ++sindex; + temp = string_extract_double_quoted (string, &sindex, 0); + + /* If the quotes surrounded the entire string, then the + whole word was quoted. */ + quoted_state = (t_index == 1 && string[sindex] == '\0') + ? WHOLLY_QUOTED + : PARTIALLY_QUOTED; + + if (temp && *temp) + { + tword = alloc_word_desc (); + tword->word = temp; + + temp = (char *)NULL; + + has_dollar_at = 0; + /* Need to get W_HASQUOTEDNULL flag through this function. */ + list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &has_dollar_at, (int *)NULL); + + if (list == &expand_word_error || list == &expand_word_fatal) + { + free (istring); + free (string); + /* expand_word_internal has already freed temp_word->word + for us because of the way it prints error messages. */ + tword->word = (char *)NULL; + dispose_word (tword); + return list; + } + + dispose_word (tword); + + /* "$@" (a double-quoted dollar-at) expands into nothing, + not even a NULL word, when there are no positional + parameters. */ + if (list == 0 && has_dollar_at) + { + quoted_dollar_at++; + break; + } + + /* If we get "$@", we know we have expanded something, so we + need to remember it for the final split on $IFS. This is + a special case; it's the only case where a quoted string + can expand into more than one word. It's going to come back + from the above call to expand_word_internal as a list with + a single word, in which all characters are quoted and + separated by blanks. What we want to do is to turn it back + into a list for the next piece of code. */ + if (list) + dequote_list (list); + + if (list && list->word && (list->word->flags & W_HASQUOTEDNULL)) + had_quoted_null = 1; + + if (has_dollar_at) + { + quoted_dollar_at++; + if (contains_dollar_at) + *contains_dollar_at = 1; + if (expanded_something) + *expanded_something = 1; + } + } + else + { + /* What we have is "". This is a minor optimization. */ + FREE (temp); + list = (WORD_LIST *)NULL; + } + + /* The code above *might* return a list (consider the case of "$@", + where it returns "$1", "$2", etc.). We can't throw away the + rest of the list, and we have to make sure each word gets added + as quoted. We test on tresult->next: if it is non-NULL, we + quote the whole list, save it to a string with string_list, and + add that string. We don't need to quote the results of this + (and it would be wrong, since that would quote the separators + as well), so we go directly to add_string. */ + if (list) + { + if (list->next) + { +#if 0 + if (quoted_dollar_at && (word->flags & W_NOSPLIT2)) + temp = string_list_internal (quote_list (list), " "); + else +#endif + /* Testing quoted_dollar_at makes sure that "$@" is + split correctly when $IFS does not contain a space. */ + temp = quoted_dollar_at + ? string_list_dollar_at (list, Q_DOUBLE_QUOTES) + : string_list (quote_list (list)); + dispose_words (list); + goto add_string; + } + else + { + temp = savestring (list->word->word); + tflag = list->word->flags; + dispose_words (list); + + /* If the string is not a quoted null string, we want + to remove any embedded unquoted CTLNUL characters. + We do not want to turn quoted null strings back into + the empty string, though. We do this because we + want to remove any quoted nulls from expansions that + contain other characters. For example, if we have + x"$*"y or "x$*y" and there are no positional parameters, + the $* should expand into nothing. */ + /* We use the W_HASQUOTEDNULL flag to differentiate the + cases: a quoted null character as above and when + CTLNUL is contained in the (non-null) expansion + of some variable. We use the had_quoted_null flag to + pass the value through this function to its caller. */ + if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0) + remove_quoted_nulls (temp); /* XXX */ + } + } + else + temp = (char *)NULL; + + /* We do not want to add quoted nulls to strings that are only + partially quoted; we can throw them away. The execption to + this is when we are going to be performing word splitting, + since we have to preserve a null argument if the next character + will cause word splitting. */ + if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2))) + continue; + + add_quoted_string: + + if (temp) + { + temp1 = temp; + temp = quote_string (temp); + free (temp1); + goto add_string; + } + else + { + /* Add NULL arg. */ + c = CTLNUL; + sindex--; /* add_character: label increments sindex */ + goto add_character; + } + + /* break; */ + + case '\'': +#if 0 + if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE)) +#else + if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) +#endif + goto add_character; + + t_index = ++sindex; + temp = string_extract_single_quoted (string, &sindex); + + /* If the entire STRING was surrounded by single quotes, + then the string is wholly quoted. */ + quoted_state = (t_index == 1 && string[sindex] == '\0') + ? WHOLLY_QUOTED + : PARTIALLY_QUOTED; + + /* If all we had was '', it is a null expansion. */ + if (*temp == '\0') + { + free (temp); + temp = (char *)NULL; + } + else + remove_quoted_escapes (temp); /* ??? */ + + /* We do not want to add quoted nulls to strings that are only + partially quoted; such nulls are discarded. */ + if (temp == 0 && (quoted_state == PARTIALLY_QUOTED)) + continue; + + /* If we have a quoted null expansion, add a quoted NULL to istring. */ + if (temp == 0) + { + c = CTLNUL; + sindex--; /* add_character: label increments sindex */ + goto add_character; + } + else + goto add_quoted_string; + + /* break; */ + + default: + /* This is the fix for " $@ " */ + add_ifs_character: + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c))) + { + if (string[sindex]) /* from old goto dollar_add_string */ + sindex++; + if (c == 0) + { + c = CTLNUL; + goto add_character; + } + else + { +#if HANDLE_MULTIBYTE + if (MB_CUR_MAX > 1) + sindex--; + + if (MB_CUR_MAX > 1) + { + SADD_MBQCHAR_BODY(temp, string, sindex, string_size); + } + else +#endif + { + twochars[0] = CTLESC; + twochars[1] = c; + goto add_twochars; + } + } + } + + SADD_MBCHAR (temp, string, sindex, string_size); + + add_character: + RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size, + DEFAULT_ARRAY_SIZE); + istring[istring_index++] = c; + istring[istring_index] = '\0'; + + /* Next character. */ + sindex++; + } + } + +finished_with_string: + /* OK, we're ready to return. If we have a quoted string, and + quoted_dollar_at is not set, we do no splitting at all; otherwise + we split on ' '. The routines that call this will handle what to + do if nothing has been expanded. */ + + /* Partially and wholly quoted strings which expand to the empty + string are retained as an empty arguments. Unquoted strings + which expand to the empty string are discarded. The single + exception is the case of expanding "$@" when there are no + positional parameters. In that case, we discard the expansion. */ + + /* Because of how the code that handles "" and '' in partially + quoted strings works, we need to make ISTRING into a QUOTED_NULL + if we saw quoting characters, but the expansion was empty. + "" and '' are tossed away before we get to this point when + processing partially quoted strings. This makes "" and $xxx"" + equivalent when xxx is unset. We also look to see whether we + saw a quoted null from a ${} expansion and add one back if we + need to. */ + + /* If we expand to nothing and there were no single or double quotes + in the word, we throw it away. Otherwise, we return a NULL word. + The single exception is for $@ surrounded by double quotes when + there are no positional parameters. In that case, we also throw + the word away. */ + + if (*istring == '\0') + { + if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED)) + { + istring[0] = CTLNUL; + istring[1] = '\0'; + tword = make_bare_word (istring); + tword->flags |= W_HASQUOTEDNULL; /* XXX */ + list = make_word_list (tword, (WORD_LIST *)NULL); + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + tword->flags |= W_QUOTED; + } + /* According to sh, ksh, and Posix.2, if a word expands into nothing + and a double-quoted "$@" appears anywhere in it, then the entire + word is removed. */ + else if (quoted_state == UNQUOTED || quoted_dollar_at) + list = (WORD_LIST *)NULL; +#if 0 + else + { + tword = make_bare_word (istring); + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + tword->flags |= W_QUOTED; + list = make_word_list (tword, (WORD_LIST *)NULL); + } +#else + else + list = (WORD_LIST *)NULL; +#endif + } + else if (word->flags & W_NOSPLIT) + { + tword = make_bare_word (istring); + if (word->flags & W_ASSIGNMENT) + tword->flags |= W_ASSIGNMENT; /* XXX */ + if (word->flags & W_COMPASSIGN) + tword->flags |= W_COMPASSIGN; /* XXX */ + if (word->flags & W_NOGLOB) + tword->flags |= W_NOGLOB; /* XXX */ + if (word->flags & W_NOEXPAND) + tword->flags |= W_NOEXPAND; /* XXX */ + if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) + tword->flags |= W_QUOTED; + if (had_quoted_null) + tword->flags |= W_HASQUOTEDNULL; + list = make_word_list (tword, (WORD_LIST *)NULL); + } + else + { + char *ifs_chars; + + ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL; + + /* If we have $@, we need to split the results no matter what. If + IFS is unset or NULL, string_list_dollar_at has separated the + positional parameters with a space, so we split on space (we have + set ifs_chars to " \t\n" above if ifs is unset). If IFS is set, + string_list_dollar_at has separated the positional parameters + with the first character of $IFS, so we split on $IFS. */ + if (has_dollar_at && ifs_chars) + list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1); + else + { + tword = make_bare_word (istring); + if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED)) + tword->flags |= W_QUOTED; + if (word->flags & W_ASSIGNMENT) + tword->flags |= W_ASSIGNMENT; + if (word->flags & W_COMPASSIGN) + tword->flags |= W_COMPASSIGN; + if (word->flags & W_NOGLOB) + tword->flags |= W_NOGLOB; + if (word->flags & W_NOEXPAND) + tword->flags |= W_NOEXPAND; + if (had_quoted_null) + tword->flags |= W_HASQUOTEDNULL; /* XXX */ + list = make_word_list (tword, (WORD_LIST *)NULL); + } + } + + free (istring); + return (list); +} + +/* **************************************************************** */ +/* */ +/* Functions for Quote Removal */ +/* */ +/* **************************************************************** */ + +/* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the + backslash quoting rules for within double quotes or a here document. */ +char * +string_quote_removal (string, quoted) + char *string; + int quoted; +{ + size_t slen; + char *r, *result_string, *temp, *send; + int sindex, tindex, dquote; + unsigned char c; + DECLARE_MBSTATE; + + /* The result can be no longer than the original string. */ + slen = strlen (string); + send = string + slen; + + r = result_string = (char *)xmalloc (slen + 1); + + for (dquote = sindex = 0; c = string[sindex];) + { + switch (c) + { + case '\\': + c = string[++sindex]; + if (c == 0) + { + *r++ = '\\'; + break; + } + if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0) + *r++ = '\\'; + /* FALLTHROUGH */ + + default: + SCOPY_CHAR_M (r, string, send, sindex); + break; + + case '\'': + if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) + { + *r++ = c; + sindex++; + break; + } + tindex = sindex + 1; + temp = string_extract_single_quoted (string, &tindex); + if (temp) + { + strcpy (r, temp); + r += strlen (r); + free (temp); + } + sindex = tindex; + break; + + case '"': + dquote = 1 - dquote; + sindex++; + break; + } + } + *r = '\0'; + return (result_string); +} + +#if 0 +/* UNUSED */ +/* Perform quote removal on word WORD. This allocates and returns a new + WORD_DESC *. */ +WORD_DESC * +word_quote_removal (word, quoted) + WORD_DESC *word; + int quoted; +{ + WORD_DESC *w; + char *t; + + t = string_quote_removal (word->word, quoted); + w = alloc_word_desc (); + w->word = t ? t : savestring (""); + return (w); +} + +/* Perform quote removal on all words in LIST. If QUOTED is non-zero, + the members of the list are treated as if they are surrounded by + double quotes. Return a new list, or NULL if LIST is NULL. */ +WORD_LIST * +word_list_quote_removal (list, quoted) + WORD_LIST *list; + int quoted; +{ + WORD_LIST *result, *t, *tresult, *e; + + for (t = list, result = (WORD_LIST *)NULL; t; t = t->next) + { + tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL); +#if 0 + result = (WORD_LIST *) list_append (result, tresult); +#else + if (result == 0) + result = e = tresult; + else + { + e->next = tresult; + while (e->next) + e = e->next; + } +#endif + } + return (result); +} +#endif + +/******************************************* + * * + * Functions to perform word splitting * + * * + *******************************************/ + +void +setifs (v) + SHELL_VAR *v; +{ + char *t; + unsigned char uc; + + ifs_var = v; + ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n"; + + /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet + handle multibyte chars in IFS */ + memset (ifs_cmap, '\0', sizeof (ifs_cmap)); + for (t = ifs_value ; t && *t; t++) + { + uc = *t; + ifs_cmap[uc] = 1; + } + +#if defined (HANDLE_MULTIBYTE) + if (ifs_value == 0) + { + ifs_firstc[0] = '\0'; + ifs_firstc_len = 1; + } + else + { + size_t ifs_len; + ifs_len = strnlen (ifs_value, MB_CUR_MAX); + ifs_firstc_len = MBLEN (ifs_value, ifs_len); + if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len)) + { + ifs_firstc[0] = ifs_value[0]; + ifs_firstc[1] = '\0'; + ifs_firstc_len = 1; + } + else + memcpy (ifs_firstc, ifs_value, ifs_firstc_len); + } +#else + ifs_firstc = ifs_value ? *ifs_value : 0; +#endif +} + +char * +getifs () +{ + return ifs_value; +} + +/* This splits a single word into a WORD LIST on $IFS, but only if the word + is not quoted. list_string () performs quote removal for us, even if we + don't do any splitting. */ +WORD_LIST * +word_split (w, ifs_chars) + WORD_DESC *w; + char *ifs_chars; +{ + WORD_LIST *result; + + if (w) + { + char *xifs; + + xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars; + result = list_string (w->word, xifs, w->flags & W_QUOTED); + } + else + result = (WORD_LIST *)NULL; + + return (result); +} + +/* Perform word splitting on LIST and return the RESULT. It is possible + to return (WORD_LIST *)NULL. */ +static WORD_LIST * +word_list_split (list) + WORD_LIST *list; +{ + WORD_LIST *result, *t, *tresult, *e; + + for (t = list, result = (WORD_LIST *)NULL; t; t = t->next) + { + tresult = word_split (t->word, ifs_value); + if (result == 0) + result = e = tresult; + else + { + e->next = tresult; + while (e->next) + e = e->next; + } + } + return (result); +} + +/************************************************** + * * + * Functions to expand an entire WORD_LIST * + * * + **************************************************/ + +/* Do any word-expansion-specific cleanup and jump to top_level */ +static void +exp_jump_to_top_level (v) + int v; +{ + set_pipestatus_from_exit (last_command_exit_value); + + /* Cleanup code goes here. */ + expand_no_split_dollar_star = 0; /* XXX */ + expanding_redir = 0; + assigning_in_environment = 0; + + if (parse_and_execute_level == 0) + top_level_cleanup (); /* from sig.c */ + + jump_to_top_level (v); +} + +/* Put NLIST (which is a WORD_LIST * of only one element) at the front of + ELIST, and set ELIST to the new list. */ +#define PREPEND_LIST(nlist, elist) \ + do { nlist->next = elist; elist = nlist; } while (0) + +/* Separate out any initial variable assignments from TLIST. If set -k has + been executed, remove all assignment statements from TLIST. Initial + variable assignments and other environment assignments are placed + on SUBST_ASSIGN_VARLIST. */ +static WORD_LIST * +separate_out_assignments (tlist) + WORD_LIST *tlist; +{ + register WORD_LIST *vp, *lp; + + if (tlist == 0) + return ((WORD_LIST *)NULL); + + if (subst_assign_varlist) + dispose_words (subst_assign_varlist); /* Clean up after previous error */ + + subst_assign_varlist = (WORD_LIST *)NULL; + vp = lp = tlist; + + /* Separate out variable assignments at the start of the command. + Loop invariant: vp->next == lp + Loop postcondition: + lp = list of words left after assignment statements skipped + tlist = original list of words + */ + while (lp && (lp->word->flags & W_ASSIGNMENT)) + { + vp = lp; + lp = lp->next; + } + + /* If lp != tlist, we have some initial assignment statements. + We make SUBST_ASSIGN_VARLIST point to the list of assignment + words and TLIST point to the remaining words. */ + if (lp != tlist) + { + subst_assign_varlist = tlist; + /* ASSERT(vp->next == lp); */ + vp->next = (WORD_LIST *)NULL; /* terminate variable list */ + tlist = lp; /* remainder of word list */ + } + + /* vp == end of variable list */ + /* tlist == remainder of original word list without variable assignments */ + if (!tlist) + /* All the words in tlist were assignment statements */ + return ((WORD_LIST *)NULL); + + /* ASSERT(tlist != NULL); */ + /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */ + + /* If the -k option is in effect, we need to go through the remaining + words, separate out the assignment words, and place them on + SUBST_ASSIGN_VARLIST. */ + if (place_keywords_in_env) + { + WORD_LIST *tp; /* tp == running pointer into tlist */ + + tp = tlist; + lp = tlist->next; + + /* Loop Invariant: tp->next == lp */ + /* Loop postcondition: tlist == word list without assignment statements */ + while (lp) + { + if (lp->word->flags & W_ASSIGNMENT) + { + /* Found an assignment statement, add this word to end of + subst_assign_varlist (vp). */ + if (!subst_assign_varlist) + subst_assign_varlist = vp = lp; + else + { + vp->next = lp; + vp = lp; + } + + /* Remove the word pointed to by LP from TLIST. */ + tp->next = lp->next; + /* ASSERT(vp == lp); */ + lp->next = (WORD_LIST *)NULL; + lp = tp->next; + } + else + { + tp = lp; + lp = lp->next; + } + } + } + return (tlist); +} + +#define WEXP_VARASSIGN 0x001 +#define WEXP_BRACEEXP 0x002 +#define WEXP_TILDEEXP 0x004 +#define WEXP_PARAMEXP 0x008 +#define WEXP_PATHEXP 0x010 + +/* All of the expansions, including variable assignments at the start of + the list. */ +#define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP) + +/* All of the expansions except variable assignments at the start of + the list. */ +#define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP) + +/* All of the `shell expansions': brace expansion, tilde expansion, parameter + expansion, command substitution, arithmetic expansion, word splitting, and + quote removal. */ +#define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP) + +/* Take the list of words in LIST and do the various substitutions. Return + a new list of words which is the expanded list, and without things like + variable assignments. */ + +WORD_LIST * +expand_words (list) + WORD_LIST *list; +{ + return (expand_word_list_internal (list, WEXP_ALL)); +} + +/* Same as expand_words (), but doesn't hack variable or environment + variables. */ +WORD_LIST * +expand_words_no_vars (list) + WORD_LIST *list; +{ + return (expand_word_list_internal (list, WEXP_NOVARS)); +} + +WORD_LIST * +expand_words_shellexp (list) + WORD_LIST *list; +{ + return (expand_word_list_internal (list, WEXP_SHELLEXP)); +} + +static WORD_LIST * +glob_expand_word_list (tlist, eflags) + WORD_LIST *tlist; + int eflags; +{ + char **glob_array, *temp_string; + register int glob_index; + WORD_LIST *glob_list, *output_list, *disposables, *next; + WORD_DESC *tword; + + output_list = disposables = (WORD_LIST *)NULL; + glob_array = (char **)NULL; + while (tlist) + { + /* For each word, either globbing is attempted or the word is + added to orig_list. If globbing succeeds, the results are + added to orig_list and the word (tlist) is added to the list + of disposable words. If globbing fails and failed glob + expansions are left unchanged (the shell default), the + original word is added to orig_list. If globbing fails and + failed glob expansions are removed, the original word is + added to the list of disposable words. orig_list ends up + in reverse order and requires a call to REVERSE_LIST to + be set right. After all words are examined, the disposable + words are freed. */ + next = tlist->next; + + /* If the word isn't an assignment and contains an unquoted + pattern matching character, then glob it. */ + if ((tlist->word->flags & W_NOGLOB) == 0 && + unquoted_glob_pattern_p (tlist->word->word)) + { + glob_array = shell_glob_filename (tlist->word->word); + + /* Handle error cases. + I don't think we should report errors like "No such file + or directory". However, I would like to report errors + like "Read failed". */ + + if (glob_array == 0 || GLOB_FAILED (glob_array)) + { + glob_array = (char **)xmalloc (sizeof (char *)); + glob_array[0] = (char *)NULL; + } + + /* Dequote the current word in case we have to use it. */ + if (glob_array[0] == NULL) + { + temp_string = dequote_string (tlist->word->word); + free (tlist->word->word); + tlist->word->word = temp_string; + } + + /* Make the array into a word list. */ + glob_list = (WORD_LIST *)NULL; + for (glob_index = 0; glob_array[glob_index]; glob_index++) + { + tword = make_bare_word (glob_array[glob_index]); + tword->flags |= W_GLOBEXP; /* XXX */ + glob_list = make_word_list (tword, glob_list); + } + + if (glob_list) + { + output_list = (WORD_LIST *)list_append (glob_list, output_list); + PREPEND_LIST (tlist, disposables); + } + else if (fail_glob_expansion != 0) + { + last_command_exit_value = EXECUTION_FAILURE; + report_error (_("no match: %s"), tlist->word->word); + exp_jump_to_top_level (DISCARD); + } + else if (allow_null_glob_expansion == 0) + { + /* Failed glob expressions are left unchanged. */ + PREPEND_LIST (tlist, output_list); + } + else + { + /* Failed glob expressions are removed. */ + PREPEND_LIST (tlist, disposables); + } + } + else + { + /* Dequote the string. */ + temp_string = dequote_string (tlist->word->word); + free (tlist->word->word); + tlist->word->word = temp_string; + PREPEND_LIST (tlist, output_list); + } + + strvec_dispose (glob_array); + glob_array = (char **)NULL; + + tlist = next; + } + + if (disposables) + dispose_words (disposables); + + if (output_list) + output_list = REVERSE_LIST (output_list, WORD_LIST *); + + return (output_list); +} + +#if defined (BRACE_EXPANSION) +static WORD_LIST * +brace_expand_word_list (tlist, eflags) + WORD_LIST *tlist; + int eflags; +{ + register char **expansions; + char *temp_string; + WORD_LIST *disposables, *output_list, *next; + WORD_DESC *w; + int eindex; + + for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next) + { + next = tlist->next; + + if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG)) + { +/*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/ + PREPEND_LIST (tlist, output_list); + continue; + } + + /* Only do brace expansion if the word has a brace character. If + not, just add the word list element to BRACES and continue. In + the common case, at least when running shell scripts, this will + degenerate to a bunch of calls to `mbschr', and then what is + basically a reversal of TLIST into BRACES, which is corrected + by a call to REVERSE_LIST () on BRACES when the end of TLIST + is reached. */ + if (mbschr (tlist->word->word, LBRACE)) + { + expansions = brace_expand (tlist->word->word); + + for (eindex = 0; temp_string = expansions[eindex]; eindex++) + { + w = alloc_word_desc (); + w->word = temp_string; + + /* If brace expansion didn't change the word, preserve + the flags. We may want to preserve the flags + unconditionally someday -- XXX */ + if (STREQ (temp_string, tlist->word->word)) + w->flags = tlist->word->flags; + else + w = make_word_flags (w, temp_string); + + output_list = make_word_list (w, output_list); + } + free (expansions); + + /* Add TLIST to the list of words to be freed after brace + expansion has been performed. */ + PREPEND_LIST (tlist, disposables); + } + else + PREPEND_LIST (tlist, output_list); + } + + if (disposables) + dispose_words (disposables); + + if (output_list) + output_list = REVERSE_LIST (output_list, WORD_LIST *); + + return (output_list); +} +#endif + +#if defined (ARRAY_VARS) +/* Take WORD, a compound associative array assignment, and internally run + 'declare -A w', where W is the variable name portion of WORD. */ +static int +make_internal_declare (word, option) + char *word; + char *option; +{ + int t; + WORD_LIST *wl; + WORD_DESC *w; + + w = make_word (word); + + t = assignment (w->word, 0); + w->word[t] = '\0'; + + wl = make_word_list (w, (WORD_LIST *)NULL); + wl = make_word_list (make_word (option), wl); + + return (declare_builtin (wl)); +} +#endif + +static WORD_LIST * +shell_expand_word_list (tlist, eflags) + WORD_LIST *tlist; + int eflags; +{ + WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list; + int expanded_something, has_dollar_at; + char *temp_string; + + /* We do tilde expansion all the time. This is what 1003.2 says. */ + new_list = (WORD_LIST *)NULL; + for (orig_list = tlist; tlist; tlist = next) + { + temp_string = tlist->word->word; + + next = tlist->next; + +#if defined (ARRAY_VARS) + /* If this is a compound array assignment to a builtin that accepts + such assignments (e.g., `declare'), take the assignment and perform + it separately, handling the semantics of declarations inside shell + functions. This avoids the double-evaluation of such arguments, + because `declare' does some evaluation of compound assignments on + its own. */ + if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG)) + { + int t; + + if (tlist->word->flags & W_ASSIGNASSOC) + make_internal_declare (tlist->word->word, "-A"); + + t = do_word_assignment (tlist->word, 0); + if (t == 0) + { + last_command_exit_value = EXECUTION_FAILURE; + exp_jump_to_top_level (DISCARD); + } + + /* Now transform the word as ksh93 appears to do and go on */ + t = assignment (tlist->word->word, 0); + tlist->word->word[t] = '\0'; + tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC); + } +#endif + + expanded_something = 0; + expanded = expand_word_internal + (tlist->word, 0, 0, &has_dollar_at, &expanded_something); + + if (expanded == &expand_word_error || expanded == &expand_word_fatal) + { + /* By convention, each time this error is returned, + tlist->word->word has already been freed. */ + tlist->word->word = (char *)NULL; + + /* Dispose our copy of the original list. */ + dispose_words (orig_list); + /* Dispose the new list we're building. */ + dispose_words (new_list); + + last_command_exit_value = EXECUTION_FAILURE; + if (expanded == &expand_word_error) + exp_jump_to_top_level (DISCARD); + else + exp_jump_to_top_level (FORCE_EOF); + } + + /* Don't split words marked W_NOSPLIT. */ + if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0) + { + temp_list = word_list_split (expanded); + dispose_words (expanded); + } + else + { + /* If no parameter expansion, command substitution, process + substitution, or arithmetic substitution took place, then + do not do word splitting. We still have to remove quoted + null characters from the result. */ + word_list_remove_quoted_nulls (expanded); + temp_list = expanded; + } + + expanded = REVERSE_LIST (temp_list, WORD_LIST *); + new_list = (WORD_LIST *)list_append (expanded, new_list); + } + + if (orig_list) + dispose_words (orig_list); + + if (new_list) + new_list = REVERSE_LIST (new_list, WORD_LIST *); + + return (new_list); +} + +/* The workhorse for expand_words () and expand_words_no_vars (). + First arg is LIST, a WORD_LIST of words. + Second arg EFLAGS is a flags word controlling which expansions are + performed. + + This does all of the substitutions: brace expansion, tilde expansion, + parameter expansion, command substitution, arithmetic expansion, + process substitution, word splitting, and pathname expansion, according + to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits + set, or for which no expansion is done, do not undergo word splitting. + Words with the W_NOGLOB bit set do not undergo pathname expansion. */ +static WORD_LIST * +expand_word_list_internal (list, eflags) + WORD_LIST *list; + int eflags; +{ + WORD_LIST *new_list, *temp_list; + int tint; + + if (list == 0) + return ((WORD_LIST *)NULL); + + garglist = new_list = copy_word_list (list); + if (eflags & WEXP_VARASSIGN) + { + garglist = new_list = separate_out_assignments (new_list); + if (new_list == 0) + { + if (subst_assign_varlist) + { + /* All the words were variable assignments, so they are placed + into the shell's environment. */ + for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next) + { + this_command_name = (char *)NULL; /* no arithmetic errors */ + tint = do_word_assignment (temp_list->word, 0); + /* Variable assignment errors in non-interactive shells + running in Posix.2 mode cause the shell to exit. */ + if (tint == 0) + { + last_command_exit_value = EXECUTION_FAILURE; + if (interactive_shell == 0 && posixly_correct) + exp_jump_to_top_level (FORCE_EOF); + else + exp_jump_to_top_level (DISCARD); + } + } + dispose_words (subst_assign_varlist); + subst_assign_varlist = (WORD_LIST *)NULL; + } + return ((WORD_LIST *)NULL); + } + } + + /* Begin expanding the words that remain. The expansions take place on + things that aren't really variable assignments. */ + +#if defined (BRACE_EXPANSION) + /* Do brace expansion on this word if there are any brace characters + in the string. */ + if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list) + new_list = brace_expand_word_list (new_list, eflags); +#endif /* BRACE_EXPANSION */ + + /* Perform the `normal' shell expansions: tilde expansion, parameter and + variable substitution, command substitution, arithmetic expansion, + and word splitting. */ + new_list = shell_expand_word_list (new_list, eflags); + + /* Okay, we're almost done. Now let's just do some filename + globbing. */ + if (new_list) + { + if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0) + /* Glob expand the word list unless globbing has been disabled. */ + new_list = glob_expand_word_list (new_list, eflags); + else + /* Dequote the words, because we're not performing globbing. */ + new_list = dequote_list (new_list); + } + + if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist) + { + sh_wassign_func_t *assign_func; + int is_special_builtin, is_builtin_or_func; + + /* If the remainder of the words expand to nothing, Posix.2 requires + that the variable and environment assignments affect the shell's + environment. */ + assign_func = new_list ? assign_in_env : do_word_assignment; + tempenv_assign_error = 0; + + is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word))); + /* Posix says that special builtins exit if a variable assignment error + occurs in an assignment preceding it. */ + is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word)); + + for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next) + { + this_command_name = (char *)NULL; + assigning_in_environment = (assign_func == assign_in_env); + tint = (*assign_func) (temp_list->word, is_builtin_or_func); + assigning_in_environment = 0; + /* Variable assignment errors in non-interactive shells running + in Posix.2 mode cause the shell to exit. */ + if (tint == 0) + { + if (assign_func == do_word_assignment) + { + last_command_exit_value = EXECUTION_FAILURE; + if (interactive_shell == 0 && posixly_correct && is_special_builtin) + exp_jump_to_top_level (FORCE_EOF); + else + exp_jump_to_top_level (DISCARD); + } + else + tempenv_assign_error++; + } + } + + dispose_words (subst_assign_varlist); + subst_assign_varlist = (WORD_LIST *)NULL; + } + +#if 0 + tint = list_length (new_list) + 1; + RESIZE_MALLOCED_BUFFER (glob_argv_flags, 0, tint, glob_argv_flags_size, 16); + for (tint = 0, temp_list = new_list; temp_list; temp_list = temp_list->next) + glob_argv_flags[tint++] = (temp_list->word->flags & W_GLOBEXP) ? '1' : '0'; + glob_argv_flags[tint] = '\0'; +#endif + + return (new_list); +} diff --git a/tests/assoc.right b/tests/assoc.right index 861551949..b9fd001c5 100644 --- a/tests/assoc.right +++ b/tests/assoc.right @@ -187,3 +187,6 @@ declare -A foo='(["bar\\]bie"]="doll" )' bar${foo}bie doll declare -A foo='(["bar\${foo}bie"]="doll" )' +bar +after printf +after use: 0 diff --git a/tests/assoc.tests b/tests/assoc.tests index 8b6fd5535..c3fb18e60 100644 --- a/tests/assoc.tests +++ b/tests/assoc.tests @@ -184,3 +184,5 @@ ${THIS_SH} ./assoc4.sub ${THIS_SH} ./assoc5.sub ${THIS_SH} ./assoc6.sub + +${THIS_SH} ./assoc7.sub diff --git a/tests/assoc7.sub b/tests/assoc7.sub new file mode 100644 index 000000000..5bc5150ec --- /dev/null +++ b/tests/assoc7.sub @@ -0,0 +1,16 @@ +# problem with bash versions through bash-4.2 +foo() +{ + declare -A hash + declare hash[baz]=bar #bash crashes here + + echo ${hash[@]} +} + +foo + +declare -a ary +printf -v ary[0] "%b" "" +echo "after printf" +x="${ary[*]}" # segfaults here +echo "after use: $?" diff --git a/tests/case.right b/tests/case.right index d57925312..52fb8a149 100644 --- a/tests/case.right +++ b/tests/case.right @@ -4,3 +4,6 @@ and here retest and match no more clauses +1.0 +./case.tests: line 29: xx: readonly variable +1.1 diff --git a/tests/case.tests b/tests/case.tests index 7c501658f..37e7fb924 100644 --- a/tests/case.tests +++ b/tests/case.tests @@ -16,3 +16,15 @@ esac case a in a) echo no more clauses;& esac + +x=0 y=1 +case 1 in + $((y=0)) ) ;; + $((x=1)) ) ;& + $((x=2)) ) echo $x.$y ;; +esac + +unset x +readonly xx=1 +case 1 in $((xx++)) ) echo hi1 ;; *) echo hi2; esac +echo ${xx}.$? diff --git a/tests/errors.right b/tests/errors.right index 07c95de96..a9722156d 100644 --- a/tests/errors.right +++ b/tests/errors.right @@ -98,9 +98,13 @@ trap: usage: trap [-lp] [[arg] signal_spec ...] ./errors.tests: line 250: kill: `': not a pid or valid job spec kill: usage: kill [-s sigspec | -n signum | -sigspec] pid | jobspec ... or kill -l [sigspec] ./errors.tests: line 255: set: trackall: invalid option name +./errors.tests: line 259: xx: readonly variable +1 ./errors1.sub: line 1: .: -i: invalid option .: usage: . filename [arguments] ./errors1.sub: line 9: shift: -4: shift count out of range ./errors1.sub: line 14: break: -1: loop count out of range after f -./errors.tests: line 264: `!!': not a valid identifier +./errors2.sub: line 3: ${$NO_SUCH_VAR}: bad substitution +1 +./errors.tests: line 270: `!!': not a valid identifier diff --git a/tests/errors.tests b/tests/errors.tests index e339f60df..d23d5c4f0 100644 --- a/tests/errors.tests +++ b/tests/errors.tests @@ -254,7 +254,13 @@ kill -INT # bad shell option names set -o trackall # bash is not ksh +# problem with versions through bash-4.2 +readonly xx=5 +echo $((xx=5)) +echo $? + ${THIS_SH} ./errors1.sub +${THIS_SH} ./errors2.sub # this must be last! # in posix mode, a function name must be a valid identifier diff --git a/tests/errors2.sub b/tests/errors2.sub new file mode 100644 index 000000000..79ef33350 --- /dev/null +++ b/tests/errors2.sub @@ -0,0 +1,3 @@ +set -e +trap 'echo $?' EXIT +echo ${$NO_SUCH_VAR} # Bad substitution expected here diff --git a/variables.c b/variables.c index 904ed133d..ef09a80de 100644 --- a/variables.c +++ b/variables.c @@ -87,6 +87,7 @@ extern int line_number, line_number_base; extern int subshell_environment, indirection_level, subshell_level; extern int build_version, patch_level; extern int expanding_redir; +extern int last_command_exit_value; extern char *dist_version, *release_status; extern char *shell_name; extern char *primary_prompt, *secondary_prompt; @@ -360,7 +361,10 @@ initialize_shell_variables (env, privmode) array_needs_making = 1; } else - report_error (_("error importing function definition for `%s'"), name); + { + last_command_exit_value = 1; + report_error (_("error importing function definition for `%s'"), name); + } /* ( */ if (name[char_index - 1] == ')' && name[char_index - 2] == '\0')