]> git.ipfire.org Git - thirdparty/bash.git/blame - subst.c
fix for SIGINT in sourced script
[thirdparty/bash.git] / subst.c
CommitLineData
95732b49
JA
1/* subst.c -- The part of the shell that does parameter, command, arithmetic,
2 and globbing substitutions. */
726f6388 3
bb70624e
JA
4/* ``Have a little faith, there's magic in the night. You ain't a
5 beauty, but, hey, you're alright.'' */
6
a0c0a00f 7/* Copyright (C) 1987-2016 Free Software Foundation, Inc.
726f6388
JA
8
9 This file is part of GNU Bash, the Bourne Again SHell.
10
3185942a
JA
11 Bash is free software: you can redistribute it and/or modify
12 it under the terms of the GNU General Public License as published by
13 the Free Software Foundation, either version 3 of the License, or
14 (at your option) any later version.
726f6388 15
3185942a
JA
16 Bash is distributed in the hope that it will be useful,
17 but WITHOUT ANY WARRANTY; without even the implied warranty of
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 GNU General Public License for more details.
726f6388 20
3185942a
JA
21 You should have received a copy of the GNU General Public License
22 along with Bash. If not, see <http://www.gnu.org/licenses/>.
23*/
726f6388 24
ccc6cda3
JA
25#include "config.h"
26
726f6388
JA
27#include "bashtypes.h"
28#include <stdio.h>
f73dda09 29#include "chartypes.h"
3185942a
JA
30#if defined (HAVE_PWD_H)
31# include <pwd.h>
32#endif
726f6388
JA
33#include <signal.h>
34#include <errno.h>
ccc6cda3
JA
35
36#if defined (HAVE_UNISTD_H)
37# include <unistd.h>
38#endif
726f6388 39
a0c0a00f
CR
40#define NEED_FPURGE_DECL
41
726f6388
JA
42#include "bashansi.h"
43#include "posixstat.h"
b80f6443 44#include "bashintl.h"
726f6388
JA
45
46#include "shell.h"
495aee44 47#include "parser.h"
726f6388
JA
48#include "flags.h"
49#include "jobs.h"
50#include "execute_cmd.h"
51#include "filecntl.h"
ccc6cda3
JA
52#include "trap.h"
53#include "pathexp.h"
54#include "mailcheck.h"
55
7117c2d2 56#include "shmbutil.h"
a0c0a00f
CR
57#if defined (HAVE_MBSTR_H) && defined (HAVE_MBSCHR)
58# include <mbstr.h> /* mbschr */
59#endif
495aee44 60#include "typemax.h"
7117c2d2 61
ccc6cda3
JA
62#include "builtins/getopt.h"
63#include "builtins/common.h"
726f6388 64
3185942a
JA
65#include "builtins/builtext.h"
66
cce855bc 67#include <tilde/tilde.h>
f73dda09 68#include <glob/strmatch.h>
ccc6cda3
JA
69
70#if !defined (errno)
71extern int errno;
72#endif /* !errno */
726f6388
JA
73
74/* The size that strings change by. */
d166f048 75#define DEFAULT_INITIAL_ARRAY_SIZE 112
ccc6cda3
JA
76#define DEFAULT_ARRAY_SIZE 128
77
78/* Variable types. */
79#define VT_VARIABLE 0
80#define VT_POSPARMS 1
81#define VT_ARRAYVAR 2
d166f048 82#define VT_ARRAYMEMBER 3
3185942a 83#define VT_ASSOCVAR 4
726f6388 84
b80f6443
JA
85#define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */
86
ccc6cda3
JA
87/* Flags for quoted_strchr */
88#define ST_BACKSL 0x01
89#define ST_CTLESC 0x02
7117c2d2
JA
90#define ST_SQUOTE 0x04 /* unused yet */
91#define ST_DQUOTE 0x08 /* unused yet */
92
cce855bc
JA
93/* These defs make it easier to use the editor. */
94#define LBRACE '{'
95#define RBRACE '}'
96#define LPAREN '('
97#define RPAREN ')'
ac50fbac
CR
98#define LBRACK '['
99#define RBRACK ']'
726f6388 100
0001803f
CR
101#if defined (HANDLE_MULTIBYTE)
102#define WLPAREN L'('
103#define WRPAREN L')'
104#endif
105
a0c0a00f
CR
106#define DOLLAR_AT_STAR(c) ((c) == '@' || (c) == '*')
107#define STR_DOLLAR_AT_STAR(s) (DOLLAR_AT_STAR ((s)[0]) && (s)[1] == '\0')
108
28ef6c31
JA
109/* Evaluates to 1 if C is one of the shell's special parameters whose length
110 can be taken, but is also one of the special expansion characters. */
111#define VALID_SPECIAL_LENGTH_PARAM(c) \
a0c0a00f 112 ((c) == '-' || (c) == '?' || (c) == '#' || (c) == '@')
28ef6c31
JA
113
114/* Evaluates to 1 if C is one of the shell's special parameters for which an
115 indirect variable reference may be made. */
116#define VALID_INDIR_PARAM(c) \
495aee44 117 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
28ef6c31
JA
118
119/* Evaluates to 1 if C is one of the OP characters that follows the parameter
120 in ${parameter[:]OPword}. */
7117c2d2 121#define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
28ef6c31 122
bb70624e
JA
123/* Evaluates to 1 if this is one of the shell's special variables. */
124#define SPECIAL_VAR(name, wi) \
a0c0a00f 125 (*name && ((DIGIT (*name) && all_digits (name)) || \
f73dda09 126 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
a0c0a00f
CR
127 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1]))))
128
129/* This can be used by all of the *_extract_* functions that have a similar
130 structure. It can't just be wrapped in a do...while(0) loop because of
131 the embedded `break'. The dangling else accommodates a trailing semicolon;
132 we could also put in a do ; while (0) */
133
134
135#define CHECK_STRING_OVERRUN(oind, ind, len, ch) \
136 if (ind >= len) \
137 { \
138 oind = len; \
139 ch = 0; \
140 break; \
141 } \
142 else \
bb70624e 143
f73dda09
JA
144/* An expansion function that takes a string and a quoted flag and returns
145 a WORD_LIST *. Used as the type of the third argument to
146 expand_string_if_necessary(). */
147typedef WORD_LIST *EXPFUNC __P((char *, int));
148
726f6388
JA
149/* Process ID of the last command executed within command substitution. */
150pid_t last_command_subst_pid = NO_PID;
b72432fd 151pid_t current_command_subst_pid = NO_PID;
726f6388 152
7117c2d2
JA
153/* Variables used to keep track of the characters in IFS. */
154SHELL_VAR *ifs_var;
155char *ifs_value;
156unsigned char ifs_cmap[UCHAR_MAX + 1];
ac50fbac 157int ifs_is_set, ifs_is_null;
95732b49
JA
158
159#if defined (HANDLE_MULTIBYTE)
160unsigned char ifs_firstc[MB_LEN_MAX];
161size_t ifs_firstc_len;
162#else
7117c2d2 163unsigned char ifs_firstc;
95732b49 164#endif
7117c2d2 165
a0c0a00f
CR
166/* If non-zero, command substitution inherits the value of errexit option */
167int inherit_errexit = 0;
168
0001803f
CR
169/* Sentinel to tell when we are performing variable assignments preceding a
170 command name and putting them into the environment. Used to make sure
171 we use the temporary environment when looking up variable values. */
3185942a
JA
172int assigning_in_environment;
173
0001803f
CR
174/* Used to hold a list of variable assignments preceding a command. Global
175 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
176 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
177WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
178
a0c0a00f
CR
179/* Tell the expansion functions to not longjmp back to top_level on fatal
180 errors. Enabled when doing completion and prompt string expansion. */
181int no_longjmp_on_fatal_error = 0;
182
726f6388 183/* Extern functions and variables from different files. */
b80f6443 184extern int last_command_exit_value, last_command_exit_signal;
a0c0a00f 185extern int subshell_environment, running_in_background;
0001803f 186extern int subshell_level, parse_and_execute_level, sourcelevel;
a0c0a00f 187extern int eof_encountered, line_number;
bb70624e 188extern int return_catch_flag, return_catch_value;
f73dda09 189extern pid_t dollar_dollar_pid;
726f6388 190extern int posixly_correct;
726f6388 191extern char *this_command_name;
ccc6cda3 192extern struct fd_bitmap *current_fds_to_close;
cce855bc 193extern int wordexp_only;
b80f6443
JA
194extern int expanding_redir;
195extern int tempenv_assign_error;
ac50fbac 196extern int builtin_ignoring_errexit;
726f6388 197
a0c0a00f
CR
198#if defined (JOB_CONTROL) && defined (PROCESS_SUBSTITUTION)
199extern PROCESS *last_procsub_child;
200#endif
201
0628567a
JA
202#if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
203extern wchar_t *wcsdup __P((const wchar_t *));
204#endif
205
ccc6cda3
JA
206/* Non-zero means to allow unmatched globbed filenames to expand to
207 a null file. */
208int allow_null_glob_expansion;
209
b80f6443
JA
210/* Non-zero means to throw an error when globbing fails to match anything. */
211int fail_glob_expansion;
212
f73dda09 213#if 0
ccc6cda3
JA
214/* Variables to keep track of which words in an expanded word list (the
215 output of expand_word_list_internal) are the result of globbing
f73dda09
JA
216 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
217 (CURRENTLY UNUSED). */
ccc6cda3
JA
218char *glob_argv_flags;
219static int glob_argv_flags_size;
f73dda09 220#endif
726f6388 221
a0c0a00f
CR
222static WORD_LIST *cached_quoted_dollar_at = 0;
223
726f6388 224static WORD_LIST expand_word_error, expand_word_fatal;
95732b49 225static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
726f6388 226static char expand_param_error, expand_param_fatal;
95732b49 227static char extract_string_error, extract_string_fatal;
726f6388 228
28ef6c31
JA
229/* Set by expand_word_unsplit; used to inhibit splitting and re-joining
230 $* on $IFS, primarily when doing assignment statements. */
231static int expand_no_split_dollar_star = 0;
bb70624e 232
bb70624e
JA
233/* A WORD_LIST of words to be expanded by expand_word_list_internal,
234 without any leading variable assignments. */
235static WORD_LIST *garglist = (WORD_LIST *)NULL;
b72432fd 236
f73dda09 237static char *quoted_substring __P((char *, int, int));
7117c2d2
JA
238static int quoted_strlen __P((char *));
239static char *quoted_strchr __P((char *, int, int));
f73dda09
JA
240
241static char *expand_string_if_necessary __P((char *, int, EXPFUNC *));
242static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *));
243static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
244static WORD_LIST *expand_string_internal __P((char *, int));
245static WORD_LIST *expand_string_leave_quoted __P((char *, int));
246static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *));
247
f73dda09 248static WORD_LIST *list_quote_escapes __P((WORD_LIST *));
ac50fbac
CR
249static WORD_LIST *list_dequote_escapes __P((WORD_LIST *));
250
f73dda09
JA
251static char *make_quoted_char __P((int));
252static WORD_LIST *quote_list __P((WORD_LIST *));
f73dda09
JA
253
254static int unquoted_substring __P((char *, char *));
255static int unquoted_member __P((int, char *));
256
95732b49
JA
257#if defined (ARRAY_VARS)
258static SHELL_VAR *do_compound_assignment __P((char *, char *, int));
259#endif
260static int do_assignment_internal __P((const WORD_DESC *, int));
f73dda09 261
3185942a 262static char *string_extract_verbatim __P((char *, size_t, int *, char *, int));
f73dda09
JA
263static char *string_extract __P((char *, int *, char *, int));
264static char *string_extract_double_quoted __P((char *, int *, int));
7117c2d2 265static inline char *string_extract_single_quoted __P((char *, int *));
a0c0a00f
CR
266static inline int skip_single_quoted __P((const char *, size_t, int, int));
267static int skip_double_quoted __P((char *, size_t, int, int));
7117c2d2
JA
268static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int));
269static char *extract_dollar_brace_string __P((char *, int *, int, int));
89a92869 270static int skip_matched_pair __P((const char *, int, int, int, int));
f73dda09 271
f73dda09
JA
272static char *pos_params __P((char *, int, int, int));
273
b80f6443
JA
274static unsigned char *mb_getcharlens __P((char *, int));
275
276static char *remove_upattern __P((char *, char *, int));
0628567a 277#if defined (HANDLE_MULTIBYTE)
b80f6443
JA
278static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int));
279#endif
f73dda09 280static char *remove_pattern __P((char *, char *, int));
b80f6443 281
b80f6443
JA
282static int match_upattern __P((char *, char *, int, char **, char **));
283#if defined (HANDLE_MULTIBYTE)
b80f6443
JA
284static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
285#endif
f73dda09
JA
286static int match_pattern __P((char *, char *, int, char **, char **));
287static int getpatspec __P((int, char *));
288static char *getpattern __P((char *, int, int));
7117c2d2 289static char *variable_remove_pattern __P((char *, char *, int, int));
f73dda09 290static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int));
7117c2d2 291static char *parameter_list_remove_pattern __P((int, char *, int, int));
f73dda09 292#ifdef ARRAY_VARS
3185942a 293static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int));
f73dda09 294#endif
495aee44 295static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int));
f73dda09 296
a0c0a00f
CR
297static char *string_var_assignment __P((SHELL_VAR *, char *));
298#if defined (ARRAY_VARS)
299static char *array_var_assignment __P((SHELL_VAR *, int, int));
300#endif
301static char *pos_params_assignment __P((WORD_LIST *, int, int));
302static char *string_transform __P((int, SHELL_VAR *, char *));
303static char *list_transform __P((int, SHELL_VAR *, WORD_LIST *, int, int));
304static char *parameter_list_transform __P((int, int, int));
305#if defined ARRAY_VARS
306static char *array_transform __P((int, SHELL_VAR *, char *, int));
307#endif
308static char *parameter_brace_transform __P((char *, char *, int, char *, int, int, int));
309
f73dda09
JA
310static char *process_substitute __P((char *, int));
311
3185942a 312static char *read_comsub __P((int, int, int *));
f73dda09
JA
313
314#ifdef ARRAY_VARS
315static arrayind_t array_length_reference __P((char *));
316#endif
317
318static int valid_brace_expansion_word __P((char *, int));
b80f6443 319static int chk_atstar __P((char *, int, int *, int *));
0628567a 320static int chk_arithsub __P((const char *, int));
b80f6443 321
495aee44 322static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *));
ac50fbac 323static char *parameter_brace_find_indir __P((char *, int, int, int));
95732b49 324static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *));
a0c0a00f 325static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int, int *, int *));
f73dda09
JA
326static void parameter_brace_expand_error __P((char *, char *));
327
328static int valid_length_expression __P((char *));
7117c2d2 329static intmax_t parameter_brace_expand_length __P((char *));
f73dda09
JA
330
331static char *skiparith __P((char *, int));
3185942a 332static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
495aee44 333static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
b80f6443 334static char *mb_substring __P((char *, int, int));
495aee44
CR
335static char *parameter_brace_substring __P((char *, char *, int, char *, int, int));
336
337static int shouldexp_replacement __P((char *));
f73dda09
JA
338
339static char *pos_params_pat_subst __P((char *, char *, char *, int));
340
a0c0a00f 341static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int, int));
f73dda09 342
3185942a 343static char *pos_params_casemod __P((char *, char *, int, int));
495aee44 344static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int));
3185942a 345
0001803f 346static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *));
95732b49 347static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int));
f73dda09
JA
348
349static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
350
f73dda09
JA
351static WORD_LIST *word_list_split __P((WORD_LIST *));
352
b80f6443
JA
353static void exp_jump_to_top_level __P((int));
354
f73dda09
JA
355static WORD_LIST *separate_out_assignments __P((WORD_LIST *));
356static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int));
357#ifdef BRACE_EXPANSION
358static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int));
359#endif
3185942a 360#if defined (ARRAY_VARS)
a0c0a00f 361static int make_internal_declare __P((char *, char *, char *));
3185942a 362#endif
f73dda09
JA
363static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int));
364static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int));
726f6388
JA
365
366/* **************************************************************** */
367/* */
368/* Utility Functions */
369/* */
370/* **************************************************************** */
371
0001803f
CR
372#if defined (DEBUG)
373void
374dump_word_flags (flags)
375 int flags;
376{
377 int f;
378
379 f = flags;
380 fprintf (stderr, "%d -> ", f);
a0c0a00f
CR
381 if (f & W_ARRAYIND)
382 {
383 f &= ~W_ARRAYIND;
384 fprintf (stderr, "W_ARRAYIND%s", f ? "|" : "");
385 }
0001803f
CR
386 if (f & W_ASSIGNASSOC)
387 {
388 f &= ~W_ASSIGNASSOC;
389 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
390 }
ac50fbac
CR
391 if (f & W_ASSIGNARRAY)
392 {
393 f &= ~W_ASSIGNARRAY;
394 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
395 }
0001803f
CR
396 if (f & W_HASCTLESC)
397 {
398 f &= ~W_HASCTLESC;
399 fprintf (stderr, "W_HASCTLESC%s", f ? "|" : "");
400 }
401 if (f & W_NOPROCSUB)
402 {
403 f &= ~W_NOPROCSUB;
404 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
405 }
406 if (f & W_DQUOTE)
407 {
408 f &= ~W_DQUOTE;
409 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
410 }
411 if (f & W_HASQUOTEDNULL)
412 {
413 f &= ~W_HASQUOTEDNULL;
414 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
415 }
416 if (f & W_ASSIGNARG)
417 {
418 f &= ~W_ASSIGNARG;
419 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
420 }
421 if (f & W_ASSNBLTIN)
422 {
423 f &= ~W_ASSNBLTIN;
424 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
425 }
6d41b715
CR
426 if (f & W_ASSNGLOBAL)
427 {
428 f &= ~W_ASSNGLOBAL;
429 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
430 }
0001803f
CR
431 if (f & W_COMPASSIGN)
432 {
433 f &= ~W_COMPASSIGN;
434 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
435 }
436 if (f & W_NOEXPAND)
437 {
438 f &= ~W_NOEXPAND;
439 fprintf (stderr, "W_NOEXPAND%s", f ? "|" : "");
440 }
441 if (f & W_ITILDE)
442 {
443 f &= ~W_ITILDE;
444 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
445 }
446 if (f & W_NOTILDE)
447 {
448 f &= ~W_NOTILDE;
449 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
450 }
451 if (f & W_ASSIGNRHS)
452 {
453 f &= ~W_ASSIGNRHS;
454 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
455 }
456 if (f & W_NOCOMSUB)
457 {
458 f &= ~W_NOCOMSUB;
459 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
460 }
461 if (f & W_DOLLARSTAR)
462 {
463 f &= ~W_DOLLARSTAR;
464 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
465 }
466 if (f & W_DOLLARAT)
467 {
468 f &= ~W_DOLLARAT;
469 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
470 }
471 if (f & W_TILDEEXP)
472 {
473 f &= ~W_TILDEEXP;
474 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
475 }
476 if (f & W_NOSPLIT2)
477 {
478 f &= ~W_NOSPLIT2;
479 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
480 }
0001803f
CR
481 if (f & W_NOSPLIT)
482 {
483 f &= ~W_NOSPLIT;
484 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
485 }
ac50fbac 486 if (f & W_NOBRACE)
0001803f 487 {
ac50fbac
CR
488 f &= ~W_NOBRACE;
489 fprintf (stderr, "W_NOBRACE%s", f ? "|" : "");
490 }
491 if (f & W_NOGLOB)
492 {
493 f &= ~W_NOGLOB;
494 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
495 }
496 if (f & W_SPLITSPACE)
497 {
498 f &= ~W_SPLITSPACE;
499 fprintf (stderr, "W_SPLITSPACE%s", f ? "|" : "");
0001803f
CR
500 }
501 if (f & W_ASSIGNMENT)
502 {
503 f &= ~W_ASSIGNMENT;
504 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
505 }
506 if (f & W_QUOTED)
507 {
508 f &= ~W_QUOTED;
509 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
510 }
511 if (f & W_HASDOLLAR)
512 {
513 f &= ~W_HASDOLLAR;
514 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
515 }
a0c0a00f
CR
516 if (f & W_COMPLETE)
517 {
518 f &= ~W_COMPLETE;
519 fprintf (stderr, "W_COMPLETE%s", f ? "|" : "");
520 }
521
0001803f
CR
522 fprintf (stderr, "\n");
523 fflush (stderr);
524}
525#endif
526
7117c2d2 527#ifdef INCLUDE_UNUSED
ccc6cda3
JA
528static char *
529quoted_substring (string, start, end)
530 char *string;
531 int start, end;
532{
533 register int len, l;
534 register char *result, *s, *r;
535
536 len = end - start;
537
538 /* Move to string[start], skipping quoted characters. */
539 for (s = string, l = 0; *s && l < start; )
540 {
541 if (*s == CTLESC)
542 {
28ef6c31
JA
543 s++;
544 continue;
ccc6cda3
JA
545 }
546 l++;
547 if (*s == 0)
28ef6c31 548 break;
ccc6cda3
JA
549 }
550
f73dda09 551 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
ccc6cda3
JA
552
553 /* Copy LEN characters, including quote characters. */
554 s = string + l;
555 for (l = 0; l < len; s++)
556 {
557 if (*s == CTLESC)
28ef6c31 558 *r++ = *s++;
ccc6cda3
JA
559 *r++ = *s;
560 l++;
561 if (*s == 0)
28ef6c31 562 break;
ccc6cda3
JA
563 }
564 *r = '\0';
565 return result;
566}
7117c2d2
JA
567#endif
568
569#ifdef INCLUDE_UNUSED
570/* Return the length of S, skipping over quoted characters */
571static int
572quoted_strlen (s)
573 char *s;
574{
575 register char *p;
576 int i;
577
578 i = 0;
579 for (p = s; *p; p++)
580 {
581 if (*p == CTLESC)
582 {
583 p++;
584 if (*p == 0)
585 return (i + 1);
586 }
587 i++;
588 }
589
590 return i;
591}
592#endif
ccc6cda3
JA
593
594/* Find the first occurrence of character C in string S, obeying shell
595 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
596 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
597 escaped with CTLESC are skipped. */
7117c2d2 598static char *
ccc6cda3
JA
599quoted_strchr (s, c, flags)
600 char *s;
601 int c, flags;
602{
603 register char *p;
604
605 for (p = s; *p; p++)
606 {
607 if (((flags & ST_BACKSL) && *p == '\\')
608 || ((flags & ST_CTLESC) && *p == CTLESC))
609 {
610 p++;
611 if (*p == '\0')
612 return ((char *)NULL);
613 continue;
614 }
615 else if (*p == c)
616 return p;
617 }
618 return ((char *)NULL);
619}
620
a0c0a00f 621#if defined (INCLUDE_UNUSED)
cce855bc 622/* Return 1 if CHARACTER appears in an unquoted portion of
7117c2d2 623 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
cce855bc
JA
624static int
625unquoted_member (character, string)
626 int character;
726f6388
JA
627 char *string;
628{
7117c2d2 629 size_t slen;
cce855bc 630 int sindex, c;
7117c2d2 631 DECLARE_MBSTATE;
726f6388 632
7117c2d2
JA
633 slen = strlen (string);
634 sindex = 0;
635 while (c = string[sindex])
726f6388 636 {
cce855bc
JA
637 if (c == character)
638 return (1);
639
640 switch (c)
ccc6cda3 641 {
cce855bc 642 default:
7117c2d2 643 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
644 break;
645
646 case '\\':
647 sindex++;
648 if (string[sindex])
7117c2d2 649 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
650 break;
651
652 case '\'':
a0c0a00f 653 sindex = skip_single_quoted (string, slen, ++sindex, 0);
cce855bc
JA
654 break;
655
656 case '"':
a0c0a00f 657 sindex = skip_double_quoted (string, slen, ++sindex, 0);
cce855bc 658 break;
ccc6cda3 659 }
726f6388 660 }
cce855bc 661 return (0);
726f6388
JA
662}
663
cce855bc
JA
664/* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
665static int
666unquoted_substring (substr, string)
667 char *substr, *string;
726f6388 668{
7117c2d2 669 size_t slen;
cce855bc 670 int sindex, c, sublen;
7117c2d2 671 DECLARE_MBSTATE;
726f6388 672
cce855bc
JA
673 if (substr == 0 || *substr == '\0')
674 return (0);
675
7117c2d2 676 slen = strlen (string);
cce855bc
JA
677 sublen = strlen (substr);
678 for (sindex = 0; c = string[sindex]; )
726f6388 679 {
cce855bc
JA
680 if (STREQN (string + sindex, substr, sublen))
681 return (1);
682
683 switch (c)
684 {
685 case '\\':
686 sindex++;
cce855bc 687 if (string[sindex])
7117c2d2 688 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
689 break;
690
691 case '\'':
a0c0a00f 692 sindex = skip_single_quoted (string, slen, ++sindex, 0);
cce855bc
JA
693 break;
694
695 case '"':
a0c0a00f 696 sindex = skip_double_quoted (string, slen, ++sindex, 0);
cce855bc
JA
697 break;
698
699 default:
7117c2d2 700 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
701 break;
702 }
726f6388 703 }
cce855bc 704 return (0);
ccc6cda3 705}
a0c0a00f 706#endif
726f6388 707
cce855bc
JA
708/* Most of the substitutions must be done in parallel. In order
709 to avoid using tons of unclear goto's, I have some functions
710 for manipulating malloc'ed strings. They all take INDX, a
711 pointer to an integer which is the offset into the string
712 where manipulation is taking place. They also take SIZE, a
713 pointer to an integer which is the current length of the
714 character array for this string. */
726f6388 715
cce855bc
JA
716/* Append SOURCE to TARGET at INDEX. SIZE is the current amount
717 of space allocated to TARGET. SOURCE can be NULL, in which
718 case nothing happens. Gets rid of SOURCE by freeing it.
719 Returns TARGET in case the location has changed. */
7117c2d2 720INLINE char *
cce855bc
JA
721sub_append_string (source, target, indx, size)
722 char *source, *target;
a0c0a00f
CR
723 int *indx;
724 size_t *size;
cce855bc
JA
725{
726 if (source)
726f6388 727 {
a0c0a00f
CR
728 int n;
729 size_t srclen;
cce855bc
JA
730
731 srclen = STRLEN (source);
732 if (srclen >= (int)(*size - *indx))
726f6388 733 {
cce855bc
JA
734 n = srclen + *indx;
735 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
f73dda09 736 target = (char *)xrealloc (target, (*size = n));
726f6388 737 }
cce855bc
JA
738
739 FASTCOPY (source, target + *indx, srclen);
740 *indx += srclen;
741 target[*indx] = '\0';
742
743 free (source);
726f6388 744 }
cce855bc
JA
745 return (target);
746}
747
748#if 0
749/* UNUSED */
750/* Append the textual representation of NUMBER to TARGET.
751 INDX and SIZE are as in SUB_APPEND_STRING. */
752char *
753sub_append_number (number, target, indx, size)
7117c2d2 754 intmax_t number;
cce855bc 755 char *target;
a0c0a00f
CR
756 int *indx;
757 size_t *size;
cce855bc
JA
758{
759 char *temp;
760
761 temp = itos (number);
762 return (sub_append_string (temp, target, indx, size));
726f6388 763}
d166f048 764#endif
726f6388
JA
765
766/* Extract a substring from STRING, starting at SINDEX and ending with
767 one of the characters in CHARLIST. Don't make the ending character
768 part of the string. Leave SINDEX pointing at the ending character.
3185942a 769 Understand about backslashes in the string. If (flags & SX_VARNAME)
7117c2d2
JA
770 is non-zero, and array variables have been compiled into the shell,
771 everything between a `[' and a corresponding `]' is skipped over.
3185942a
JA
772 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
773 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
95732b49 774 contain a closing character from CHARLIST. */
726f6388 775static char *
7117c2d2 776string_extract (string, sindex, charlist, flags)
f73dda09
JA
777 char *string;
778 int *sindex;
779 char *charlist;
7117c2d2 780 int flags;
726f6388 781{
ccc6cda3 782 register int c, i;
95732b49 783 int found;
7117c2d2 784 size_t slen;
726f6388 785 char *temp;
7117c2d2 786 DECLARE_MBSTATE;
726f6388 787
95732b49 788 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
7117c2d2 789 i = *sindex;
95732b49 790 found = 0;
7117c2d2 791 while (c = string[i])
726f6388
JA
792 {
793 if (c == '\\')
7117c2d2
JA
794 {
795 if (string[i + 1])
796 i++;
797 else
798 break;
799 }
ccc6cda3 800#if defined (ARRAY_VARS)
3185942a 801 else if ((flags & SX_VARNAME) && c == '[')
ccc6cda3
JA
802 {
803 int ni;
804 /* If this is an array subscript, skip over it and continue. */
0001803f 805 ni = skipsubscript (string, i, 0);
ccc6cda3
JA
806 if (string[ni] == ']')
807 i = ni;
808 }
809#endif
810 else if (MEMBER (c, charlist))
95732b49
JA
811 {
812 found = 1;
726f6388 813 break;
95732b49 814 }
7117c2d2
JA
815
816 ADVANCE_CHAR (string, slen, i);
726f6388 817 }
bb70624e 818
95732b49
JA
819 /* If we had to have a matching delimiter and didn't find one, return an
820 error and let the caller deal with it. */
3185942a 821 if ((flags & SX_REQMATCH) && found == 0)
95732b49
JA
822 {
823 *sindex = i;
824 return (&extract_string_error);
825 }
826
3185942a 827 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
726f6388 828 *sindex = i;
95732b49 829
726f6388
JA
830 return (temp);
831}
832
ccc6cda3
JA
833/* Extract the contents of STRING as if it is enclosed in double quotes.
834 SINDEX, when passed in, is the offset of the character immediately
835 following the opening double quote; on exit, SINDEX is left pointing after
836 the closing double quote. If STRIPDQ is non-zero, unquoted double
837 quotes are stripped and the string is terminated by a null byte.
838 Backslashes between the embedded double quotes are processed. If STRIPDQ
839 is zero, an unquoted `"' terminates the string. */
7117c2d2 840static char *
a0c0a00f 841string_extract_double_quoted (string, sindex, flags)
726f6388 842 char *string;
a0c0a00f 843 int *sindex, flags;
726f6388 844{
7117c2d2
JA
845 size_t slen;
846 char *send;
f73dda09
JA
847 int j, i, t;
848 unsigned char c;
ccc6cda3
JA
849 char *temp, *ret; /* The new string we return. */
850 int pass_next, backquote, si; /* State variables for the machine. */
851 int dquote;
a0c0a00f 852 int stripdq;
7117c2d2
JA
853 DECLARE_MBSTATE;
854
855 slen = strlen (string + *sindex) + *sindex;
856 send = string + slen;
726f6388 857
a0c0a00f
CR
858 stripdq = (flags & SX_STRIPDQ);
859
ccc6cda3 860 pass_next = backquote = dquote = 0;
7117c2d2 861 temp = (char *)xmalloc (1 + slen - *sindex);
726f6388 862
7117c2d2
JA
863 j = 0;
864 i = *sindex;
865 while (c = string[i])
726f6388 866 {
ccc6cda3
JA
867 /* Process a character that was quoted by a backslash. */
868 if (pass_next)
726f6388 869 {
495aee44 870 /* XXX - take another look at this in light of Interp 221 */
ccc6cda3 871 /* Posix.2 sez:
726f6388 872
ccc6cda3
JA
873 ``The backslash shall retain its special meaning as an escape
874 character only when followed by one of the characters:
7117c2d2 875 $ ` " \ <newline>''.
726f6388 876
ccc6cda3
JA
877 If STRIPDQ is zero, we handle the double quotes here and let
878 expand_word_internal handle the rest. If STRIPDQ is non-zero,
879 we have already been through one round of backslash stripping,
880 and want to strip these backslashes only if DQUOTE is non-zero,
881 indicating that we are inside an embedded double-quoted string. */
882
883 /* If we are in an embedded quoted string, then don't strip
884 backslashes before characters for which the backslash
885 retains its special meaning, but remove backslashes in
886 front of other characters. If we are not in an
887 embedded quoted string, don't strip backslashes at all.
888 This mess is necessary because the string was already
889 surrounded by double quotes (and sh has some really weird
890 quoting rules).
891 The returned string will be run through expansion as if
892 it were double-quoted. */
893 if ((stripdq == 0 && c != '"') ||
28ef6c31 894 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
ccc6cda3 895 temp[j++] = '\\';
ccc6cda3 896 pass_next = 0;
7117c2d2
JA
897
898add_one_character:
899 COPY_CHAR_I (temp, j, string, send, i);
ccc6cda3
JA
900 continue;
901 }
726f6388 902
ccc6cda3
JA
903 /* A backslash protects the next character. The code just above
904 handles preserving the backslash in front of any character but
905 a double quote. */
906 if (c == '\\')
726f6388 907 {
ccc6cda3 908 pass_next++;
7117c2d2 909 i++;
726f6388
JA
910 continue;
911 }
912
ccc6cda3
JA
913 /* Inside backquotes, ``the portion of the quoted string from the
914 initial backquote and the characters up to the next backquote
915 that is not preceded by a backslash, having escape characters
916 removed, defines that command''. */
917 if (backquote)
726f6388 918 {
ccc6cda3
JA
919 if (c == '`')
920 backquote = 0;
921 temp[j++] = c;
7117c2d2 922 i++;
726f6388
JA
923 continue;
924 }
925
ccc6cda3 926 if (c == '`')
726f6388 927 {
ccc6cda3
JA
928 temp[j++] = c;
929 backquote++;
7117c2d2 930 i++;
ccc6cda3 931 continue;
726f6388
JA
932 }
933
ccc6cda3
JA
934 /* Pass everything between `$(' and the matching `)' or a quoted
935 ${ ... } pair through according to the Posix.2 specification. */
cce855bc 936 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
726f6388 937 {
b80f6443
JA
938 int free_ret = 1;
939
ccc6cda3 940 si = i + 2;
cce855bc 941 if (string[i + 1] == LPAREN)
a0c0a00f 942 ret = extract_command_subst (string, &si, (flags & SX_COMPLETE));
ccc6cda3 943 else
495aee44 944 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
726f6388 945
ccc6cda3
JA
946 temp[j++] = '$';
947 temp[j++] = string[i + 1];
726f6388 948
b80f6443
JA
949 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
950 is set. */
951 if (ret == 0 && no_longjmp_on_fatal_error)
952 {
953 free_ret = 0;
954 ret = string + i + 2;
955 }
956
ccc6cda3
JA
957 for (t = 0; ret[t]; t++, j++)
958 temp[j] = ret[t];
b80f6443 959 temp[j] = string[si];
726f6388 960
b80f6443
JA
961 if (string[si])
962 {
963 j++;
964 i = si + 1;
965 }
966 else
967 i = si;
968
969 if (free_ret)
970 free (ret);
ccc6cda3 971 continue;
726f6388
JA
972 }
973
ccc6cda3 974 /* Add any character but a double quote to the quoted string we're
28ef6c31 975 accumulating. */
ccc6cda3 976 if (c != '"')
7117c2d2 977 goto add_one_character;
ccc6cda3
JA
978
979 /* c == '"' */
980 if (stripdq)
726f6388 981 {
ccc6cda3 982 dquote ^= 1;
7117c2d2 983 i++;
ccc6cda3 984 continue;
726f6388 985 }
ccc6cda3
JA
986
987 break;
726f6388 988 }
ccc6cda3 989 temp[j] = '\0';
726f6388 990
ccc6cda3
JA
991 /* Point to after the closing quote. */
992 if (c)
993 i++;
726f6388
JA
994 *sindex = i;
995
ccc6cda3
JA
996 return (temp);
997}
998
999/* This should really be another option to string_extract_double_quoted. */
f73dda09 1000static int
a0c0a00f 1001skip_double_quoted (string, slen, sind, flags)
ccc6cda3 1002 char *string;
7117c2d2 1003 size_t slen;
ccc6cda3 1004 int sind;
a0c0a00f 1005 int flags;
ccc6cda3 1006{
f73dda09 1007 int c, i;
ccc6cda3
JA
1008 char *ret;
1009 int pass_next, backquote, si;
7117c2d2 1010 DECLARE_MBSTATE;
ccc6cda3
JA
1011
1012 pass_next = backquote = 0;
7117c2d2
JA
1013 i = sind;
1014 while (c = string[i])
726f6388 1015 {
ccc6cda3
JA
1016 if (pass_next)
1017 {
1018 pass_next = 0;
7117c2d2 1019 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1020 continue;
1021 }
1022 else if (c == '\\')
1023 {
1024 pass_next++;
7117c2d2 1025 i++;
ccc6cda3
JA
1026 continue;
1027 }
1028 else if (backquote)
1029 {
1030 if (c == '`')
1031 backquote = 0;
7117c2d2 1032 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1033 continue;
1034 }
1035 else if (c == '`')
1036 {
1037 backquote++;
7117c2d2 1038 i++;
ccc6cda3
JA
1039 continue;
1040 }
cce855bc 1041 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
ccc6cda3
JA
1042 {
1043 si = i + 2;
cce855bc 1044 if (string[i + 1] == LPAREN)
a0c0a00f 1045 ret = extract_command_subst (string, &si, SX_NOALLOC|(flags&SX_COMPLETE));
ccc6cda3 1046 else
495aee44 1047 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
ccc6cda3 1048
a0c0a00f
CR
1049 /* These can consume the entire string if they are unterminated */
1050 CHECK_STRING_OVERRUN (i, si, slen, c);
1051
7117c2d2 1052 i = si + 1;
ccc6cda3
JA
1053 continue;
1054 }
1055 else if (c != '"')
7117c2d2
JA
1056 {
1057 ADVANCE_CHAR (string, slen, i);
1058 continue;
1059 }
ccc6cda3
JA
1060 else
1061 break;
726f6388 1062 }
ccc6cda3
JA
1063
1064 if (c)
1065 i++;
1066
1067 return (i);
726f6388
JA
1068}
1069
ccc6cda3
JA
1070/* Extract the contents of STRING as if it is enclosed in single quotes.
1071 SINDEX, when passed in, is the offset of the character immediately
1072 following the opening single quote; on exit, SINDEX is left pointing after
1073 the closing single quote. */
1074static inline char *
1075string_extract_single_quoted (string, sindex)
1076 char *string;
1077 int *sindex;
1078{
f73dda09 1079 register int i;
7117c2d2 1080 size_t slen;
ccc6cda3 1081 char *t;
7117c2d2 1082 DECLARE_MBSTATE;
ccc6cda3 1083
95732b49
JA
1084 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1085 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
7117c2d2
JA
1086 i = *sindex;
1087 while (string[i] && string[i] != '\'')
1088 ADVANCE_CHAR (string, slen, i);
ccc6cda3 1089
bb70624e 1090 t = substring (string, *sindex, i);
ccc6cda3
JA
1091
1092 if (string[i])
1093 i++;
1094 *sindex = i;
1095
1096 return (t);
1097}
1098
a0c0a00f
CR
1099/* Skip over a single-quoted string. We overload the SX_COMPLETE flag to mean
1100 that we are splitting out words for completion and have encountered a $'...'
1101 string, which allows backslash-escaped single quotes. */
ccc6cda3 1102static inline int
a0c0a00f 1103skip_single_quoted (string, slen, sind, flags)
0628567a 1104 const char *string;
7117c2d2 1105 size_t slen;
ccc6cda3 1106 int sind;
a0c0a00f 1107 int flags;
ccc6cda3 1108{
28ef6c31 1109 register int c;
7117c2d2
JA
1110 DECLARE_MBSTATE;
1111
1112 c = sind;
1113 while (string[c] && string[c] != '\'')
a0c0a00f
CR
1114 {
1115 if ((flags & SX_COMPLETE) && string[c] == '\\' && string[c+1] == '\'' && string[c+2])
1116 ADVANCE_CHAR (string, slen, c);
1117 ADVANCE_CHAR (string, slen, c);
1118 }
ccc6cda3 1119
28ef6c31
JA
1120 if (string[c])
1121 c++;
1122 return c;
ccc6cda3
JA
1123}
1124
1125/* Just like string_extract, but doesn't hack backslashes or any of
bb70624e 1126 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
726f6388 1127static char *
3185942a 1128string_extract_verbatim (string, slen, sindex, charlist, flags)
f73dda09 1129 char *string;
95732b49 1130 size_t slen;
ccc6cda3 1131 int *sindex;
f73dda09 1132 char *charlist;
3185942a 1133 int flags;
ccc6cda3 1134{
0001803f 1135 register int i;
95732b49
JA
1136#if defined (HANDLE_MULTIBYTE)
1137 size_t clen;
1138 wchar_t *wcharlist;
1139#endif
ccc6cda3
JA
1140 int c;
1141 char *temp;
95732b49 1142 DECLARE_MBSTATE;
ccc6cda3 1143
a0c0a00f 1144 if ((flags & SX_NOCTLESC) && charlist[0] == '\'' && charlist[1] == '\0')
ccc6cda3
JA
1145 {
1146 temp = string_extract_single_quoted (string, sindex);
1147 --*sindex; /* leave *sindex at separator character */
1148 return temp;
1149 }
1150
95732b49
JA
1151 i = *sindex;
1152#if 0
1153 /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need
1154 this only if MB_CUR_MAX > 1. */
1155 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1;
1156#endif
1157#if defined (HANDLE_MULTIBYTE)
1158 clen = strlen (charlist);
1159 wcharlist = 0;
1160#endif
1161 while (c = string[i])
ccc6cda3 1162 {
95732b49
JA
1163#if defined (HANDLE_MULTIBYTE)
1164 size_t mblength;
1165#endif
3185942a
JA
1166 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1167 {
1168 i += 2;
1169 continue;
1170 }
1171 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1172 through, to protect the CTLNULs from later calls to
1173 remove_quoted_nulls. */
1174 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
ccc6cda3 1175 {
95732b49 1176 i += 2;
ccc6cda3
JA
1177 continue;
1178 }
1179
95732b49
JA
1180#if defined (HANDLE_MULTIBYTE)
1181 mblength = MBLEN (string + i, slen - i);
1182 if (mblength > 1)
1183 {
1184 wchar_t wc;
1185 mblength = mbtowc (&wc, string + i, slen - i);
1186 if (MB_INVALIDCH (mblength))
1187 {
1188 if (MEMBER (c, charlist))
1189 break;
1190 }
1191 else
1192 {
1193 if (wcharlist == 0)
1194 {
1195 size_t len;
1196 len = mbstowcs (wcharlist, charlist, 0);
1197 if (len == -1)
1198 len = 0;
0628567a
JA
1199 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1200 mbstowcs (wcharlist, charlist, len + 1);
95732b49
JA
1201 }
1202
1203 if (wcschr (wcharlist, wc))
1204 break;
1205 }
1206 }
1207 else
1208#endif
ccc6cda3
JA
1209 if (MEMBER (c, charlist))
1210 break;
95732b49
JA
1211
1212 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1213 }
1214
95732b49
JA
1215#if defined (HANDLE_MULTIBYTE)
1216 FREE (wcharlist);
1217#endif
1218
bb70624e 1219 temp = substring (string, *sindex, i);
ccc6cda3
JA
1220 *sindex = i;
1221
1222 return (temp);
1223}
1224
1225/* Extract the $( construct in STRING, and return a new string.
1226 Start extracting at (SINDEX) as if we had just seen "$(".
3185942a 1227 Make (SINDEX) get the position of the matching ")". )
0001803f 1228 XFLAGS is additional flags to pass to other extraction functions. */
ccc6cda3 1229char *
3185942a 1230extract_command_subst (string, sindex, xflags)
726f6388
JA
1231 char *string;
1232 int *sindex;
3185942a 1233 int xflags;
726f6388 1234{
a0c0a00f
CR
1235 char *ret;
1236
1237 if (string[*sindex] == LPAREN || (xflags & SX_COMPLETE))
3185942a
JA
1238 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1239 else
1240 {
1241 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
a0c0a00f
CR
1242 ret = xparse_dolparen (string, string+*sindex, sindex, xflags);
1243 return ret;
3185942a 1244 }
ccc6cda3
JA
1245}
1246
28ef6c31 1247/* Extract the $[ construct in STRING, and return a new string. (])
ccc6cda3
JA
1248 Start extracting at (SINDEX) as if we had just seen "$[".
1249 Make (SINDEX) get the position of the matching "]". */
1250char *
1251extract_arithmetic_subst (string, sindex)
1252 char *string;
1253 int *sindex;
1254{
7117c2d2 1255 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
ccc6cda3
JA
1256}
1257
1258#if defined (PROCESS_SUBSTITUTION)
1259/* Extract the <( or >( construct in STRING, and return a new string.
1260 Start extracting at (SINDEX) as if we had just seen "<(".
cce855bc 1261 Make (SINDEX) get the position of the matching ")". */ /*))*/
ccc6cda3 1262char *
85b94814 1263extract_process_subst (string, starter, sindex, xflags)
ccc6cda3
JA
1264 char *string;
1265 char *starter;
1266 int *sindex;
85b94814 1267 int xflags;
ccc6cda3 1268{
85b94814 1269#if 0
ac50fbac 1270 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
85b94814
CR
1271#else
1272 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1273 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1274#endif
ccc6cda3
JA
1275}
1276#endif /* PROCESS_SUBSTITUTION */
1277
1278#if defined (ARRAY_VARS)
95732b49
JA
1279/* This can be fooled by unquoted right parens in the passed string. If
1280 each caller verifies that the last character in STRING is a right paren,
1281 we don't even need to call extract_delimited_string. */
ccc6cda3
JA
1282char *
1283extract_array_assignment_list (string, sindex)
1284 char *string;
1285 int *sindex;
1286{
95732b49
JA
1287 int slen;
1288 char *ret;
1289
1290 slen = strlen (string); /* ( */
1291 if (string[slen - 1] == ')')
1292 {
1293 ret = substring (string, *sindex, slen - 1);
1294 *sindex = slen - 1;
1295 return ret;
1296 }
1297 return 0;
ccc6cda3
JA
1298}
1299#endif
1300
1301/* Extract and create a new string from the contents of STRING, a
1302 character string delimited with OPENER and CLOSER. SINDEX is
1303 the address of an int describing the current offset in STRING;
1304 it should point to just after the first OPENER found. On exit,
1305 SINDEX gets the position of the last character of the matching CLOSER.
1306 If OPENER is more than a single character, ALT_OPENER, if non-null,
1307 contains a character string that can also match CLOSER and thus
1308 needs to be skipped. */
1309static char *
7117c2d2 1310extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
ccc6cda3
JA
1311 char *string;
1312 int *sindex;
1313 char *opener, *alt_opener, *closer;
7117c2d2 1314 int flags;
ccc6cda3
JA
1315{
1316 int i, c, si;
7117c2d2 1317 size_t slen;
ccc6cda3 1318 char *t, *result;
0628567a 1319 int pass_character, nesting_level, in_comment;
ccc6cda3 1320 int len_closer, len_opener, len_alt_opener;
7117c2d2 1321 DECLARE_MBSTATE;
ccc6cda3 1322
7117c2d2 1323 slen = strlen (string + *sindex) + *sindex;
ccc6cda3
JA
1324 len_opener = STRLEN (opener);
1325 len_alt_opener = STRLEN (alt_opener);
1326 len_closer = STRLEN (closer);
726f6388 1327
0628567a 1328 pass_character = in_comment = 0;
726f6388
JA
1329
1330 nesting_level = 1;
ccc6cda3 1331 i = *sindex;
726f6388 1332
ccc6cda3 1333 while (nesting_level)
726f6388 1334 {
ccc6cda3
JA
1335 c = string[i];
1336
a0c0a00f
CR
1337 /* If a recursive call or a call to ADVANCE_CHAR leaves the index beyond
1338 the end of the string, catch it and cut the loop. */
1339 if (i > slen)
1340 {
1341 i = slen;
1342 c = string[i = slen];
1343 break;
1344 }
1345
ccc6cda3 1346 if (c == 0)
28ef6c31 1347 break;
ccc6cda3 1348
0628567a
JA
1349 if (in_comment)
1350 {
1351 if (c == '\n')
1352 in_comment = 0;
1353 ADVANCE_CHAR (string, slen, i);
1354 continue;
1355 }
1356
ccc6cda3 1357 if (pass_character) /* previous char was backslash */
726f6388
JA
1358 {
1359 pass_character = 0;
7117c2d2 1360 ADVANCE_CHAR (string, slen, i);
726f6388
JA
1361 continue;
1362 }
1363
0628567a 1364 /* Not exactly right yet; should handle shell metacharacters and
0001803f 1365 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
3185942a 1366 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
0628567a
JA
1367 {
1368 in_comment = 1;
1369 ADVANCE_CHAR (string, slen, i);
1370 continue;
1371 }
1372
7117c2d2 1373 if (c == CTLESC || c == '\\')
726f6388 1374 {
ccc6cda3
JA
1375 pass_character++;
1376 i++;
1377 continue;
726f6388
JA
1378 }
1379
495aee44
CR
1380 /* Process a nested command substitution, but only if we're parsing an
1381 arithmetic substitution. */
0001803f
CR
1382 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1383 {
1384 si = i + 2;
495aee44 1385 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
0001803f
CR
1386 i = si + 1;
1387 continue;
1388 }
0001803f 1389
ccc6cda3
JA
1390 /* Process a nested OPENER. */
1391 if (STREQN (string + i, opener, len_opener))
726f6388 1392 {
ccc6cda3 1393 si = i + len_opener;
3185942a 1394 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
ccc6cda3 1395 i = si + 1;
ccc6cda3 1396 continue;
726f6388
JA
1397 }
1398
ccc6cda3
JA
1399 /* Process a nested ALT_OPENER */
1400 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
726f6388 1401 {
ccc6cda3 1402 si = i + len_alt_opener;
3185942a 1403 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
ccc6cda3 1404 i = si + 1;
726f6388
JA
1405 continue;
1406 }
ccc6cda3
JA
1407
1408 /* If the current substring terminates the delimited string, decrement
1409 the nesting level. */
1410 if (STREQN (string + i, closer, len_closer))
726f6388 1411 {
7117c2d2 1412 i += len_closer - 1; /* move to last byte of the closer */
ccc6cda3
JA
1413 nesting_level--;
1414 if (nesting_level == 0)
1415 break;
726f6388 1416 }
ccc6cda3
JA
1417
1418 /* Pass old-style command substitution through verbatim. */
1419 if (c == '`')
28ef6c31
JA
1420 {
1421 si = i + 1;
3185942a 1422 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
28ef6c31 1423 i = si + 1;
28ef6c31
JA
1424 continue;
1425 }
ccc6cda3 1426
7117c2d2
JA
1427 /* Pass single-quoted and double-quoted strings through verbatim. */
1428 if (c == '\'' || c == '"')
28ef6c31
JA
1429 {
1430 si = i + 1;
a0c0a00f
CR
1431 i = (c == '\'') ? skip_single_quoted (string, slen, si, 0)
1432 : skip_double_quoted (string, slen, si, 0);
28ef6c31
JA
1433 continue;
1434 }
ccc6cda3 1435
7117c2d2
JA
1436 /* move past this character, which was not special. */
1437 ADVANCE_CHAR (string, slen, i);
726f6388
JA
1438 }
1439
b80f6443 1440 if (c == 0 && nesting_level)
726f6388 1441 {
b80f6443
JA
1442 if (no_longjmp_on_fatal_error == 0)
1443 {
b80f6443 1444 last_command_exit_value = EXECUTION_FAILURE;
ac50fbac 1445 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
b80f6443
JA
1446 exp_jump_to_top_level (DISCARD);
1447 }
1448 else
1449 {
1450 *sindex = i;
1451 return (char *)NULL;
1452 }
726f6388 1453 }
ccc6cda3 1454
cce855bc 1455 si = i - *sindex - len_closer + 1;
3185942a 1456 if (flags & SX_NOALLOC)
7117c2d2
JA
1457 result = (char *)NULL;
1458 else
1459 {
1460 result = (char *)xmalloc (1 + si);
1461 strncpy (result, string + *sindex, si);
1462 result[si] = '\0';
1463 }
cce855bc
JA
1464 *sindex = i;
1465
726f6388
JA
1466 return (result);
1467}
1468
ccc6cda3
JA
1469/* Extract a parameter expansion expression within ${ and } from STRING.
1470 Obey the Posix.2 rules for finding the ending `}': count braces while
1471 skipping over enclosed quoted strings and command substitutions.
1472 SINDEX is the address of an int describing the current offset in STRING;
1473 it should point to just after the first `{' found. On exit, SINDEX
1474 gets the position of the matching `}'. QUOTED is non-zero if this
1475 occurs inside double quotes. */
1476/* XXX -- this is very similar to extract_delimited_string -- XXX */
726f6388 1477static char *
7117c2d2 1478extract_dollar_brace_string (string, sindex, quoted, flags)
726f6388 1479 char *string;
7117c2d2 1480 int *sindex, quoted, flags;
726f6388 1481{
f73dda09 1482 register int i, c;
7117c2d2 1483 size_t slen;
495aee44 1484 int pass_character, nesting_level, si, dolbrace_state;
ccc6cda3 1485 char *result, *t;
7117c2d2 1486 DECLARE_MBSTATE;
726f6388 1487
ccc6cda3 1488 pass_character = 0;
ccc6cda3 1489 nesting_level = 1;
7117c2d2 1490 slen = strlen (string + *sindex) + *sindex;
ccc6cda3 1491
495aee44 1492 /* The handling of dolbrace_state needs to agree with the code in parse.y:
49ed961b
CR
1493 parse_matched_pair(). The different initial value is to handle the
1494 case where this function is called to parse the word in
1495 ${param op word} (SX_WORD). */
1496 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1497 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1498 dolbrace_state = DOLBRACE_QUOTE;
495aee44 1499
7117c2d2
JA
1500 i = *sindex;
1501 while (c = string[i])
726f6388 1502 {
ccc6cda3 1503 if (pass_character)
726f6388 1504 {
ccc6cda3 1505 pass_character = 0;
7117c2d2 1506 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1507 continue;
1508 }
726f6388 1509
cce855bc
JA
1510 /* CTLESCs and backslashes quote the next character. */
1511 if (c == CTLESC || c == '\\')
726f6388 1512 {
ccc6cda3 1513 pass_character++;
7117c2d2 1514 i++;
726f6388
JA
1515 continue;
1516 }
1517
cce855bc 1518 if (string[i] == '$' && string[i+1] == LBRACE)
726f6388 1519 {
ccc6cda3 1520 nesting_level++;
7117c2d2 1521 i += 2;
726f6388
JA
1522 continue;
1523 }
1524
cce855bc 1525 if (c == RBRACE)
726f6388 1526 {
ccc6cda3
JA
1527 nesting_level--;
1528 if (nesting_level == 0)
1529 break;
7117c2d2 1530 i++;
726f6388
JA
1531 continue;
1532 }
1533
ccc6cda3
JA
1534 /* Pass the contents of old-style command substitutions through
1535 verbatim. */
1536 if (c == '`')
726f6388 1537 {
ccc6cda3 1538 si = i + 1;
3185942a 1539 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
a0c0a00f
CR
1540
1541 CHECK_STRING_OVERRUN (i, si, slen, c);
1542
7117c2d2 1543 i = si + 1;
ccc6cda3
JA
1544 continue;
1545 }
726f6388 1546
cce855bc
JA
1547 /* Pass the contents of new-style command substitutions and
1548 arithmetic substitutions through verbatim. */
1549 if (string[i] == '$' && string[i+1] == LPAREN)
ccc6cda3 1550 {
726f6388 1551 si = i + 2;
3185942a 1552 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
7117c2d2 1553 i = si + 1;
726f6388
JA
1554 continue;
1555 }
1556
495aee44
CR
1557 /* Pass the contents of double-quoted strings through verbatim. */
1558 if (c == '"')
1559 {
1560 si = i + 1;
a0c0a00f 1561 i = skip_double_quoted (string, slen, si, 0);
495aee44
CR
1562 /* skip_XXX_quoted leaves index one past close quote */
1563 continue;
1564 }
1565
1566 if (c == '\'')
1567 {
1568/*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
ac50fbac 1569 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
495aee44
CR
1570 ADVANCE_CHAR (string, slen, i);
1571 else
1572 {
1573 si = i + 1;
a0c0a00f 1574 i = skip_single_quoted (string, slen, si, 0);
495aee44
CR
1575 }
1576
1577 continue;
1578 }
7117c2d2
JA
1579
1580 /* move past this character, which was not special. */
1581 ADVANCE_CHAR (string, slen, i);
495aee44
CR
1582
1583 /* This logic must agree with parse.y:parse_matched_pair, since they
1584 share the same defines. */
1585 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1586 dolbrace_state = DOLBRACE_QUOTE;
1587 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1588 dolbrace_state = DOLBRACE_QUOTE;
1589 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
ac50fbac 1590 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
495aee44
CR
1591 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1592 dolbrace_state = DOLBRACE_QUOTE;
1593 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1594 dolbrace_state = DOLBRACE_QUOTE;
a0c0a00f
CR
1595 /* This is intended to handle all of the [:]op expansions and the substring/
1596 length/pattern removal/pattern substitution expansions. */
495aee44
CR
1597 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1598 dolbrace_state = DOLBRACE_OP;
1599 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1600 dolbrace_state = DOLBRACE_WORD;
cce855bc 1601 }
726f6388 1602
b80f6443 1603 if (c == 0 && nesting_level)
cce855bc 1604 {
b80f6443
JA
1605 if (no_longjmp_on_fatal_error == 0)
1606 { /* { */
b80f6443 1607 last_command_exit_value = EXECUTION_FAILURE;
ac50fbac 1608 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
b80f6443
JA
1609 exp_jump_to_top_level (DISCARD);
1610 }
1611 else
1612 {
1613 *sindex = i;
1614 return ((char *)NULL);
1615 }
726f6388 1616 }
726f6388 1617
3185942a 1618 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
726f6388
JA
1619 *sindex = i;
1620
ccc6cda3 1621 return (result);
726f6388
JA
1622}
1623
ccc6cda3
JA
1624/* Remove backslashes which are quoting backquotes from STRING. Modifies
1625 STRING, and returns a pointer to it. */
1626char *
1627de_backslash (string)
726f6388 1628 char *string;
ccc6cda3 1629{
7117c2d2
JA
1630 register size_t slen;
1631 register int i, j, prev_i;
1632 DECLARE_MBSTATE;
726f6388 1633
7117c2d2
JA
1634 slen = strlen (string);
1635 i = j = 0;
1636
1637 /* Loop copying string[i] to string[j], i >= j. */
1638 while (i < slen)
1639 {
1640 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
ccc6cda3 1641 string[i + 1] == '$'))
7117c2d2
JA
1642 i++;
1643 prev_i = i;
1644 ADVANCE_CHAR (string, slen, i);
1645 if (j < prev_i)
b80f6443 1646 do string[j++] = string[prev_i++]; while (prev_i < i);
7117c2d2 1647 else
b80f6443 1648 j = i;
7117c2d2
JA
1649 }
1650 string[j] = '\0';
1651
ccc6cda3
JA
1652 return (string);
1653}
726f6388 1654
ccc6cda3 1655#if 0
cce855bc 1656/*UNUSED*/
ccc6cda3
JA
1657/* Replace instances of \! in a string with !. */
1658void
1659unquote_bang (string)
1660 char *string;
1661{
1662 register int i, j;
1663 register char *temp;
726f6388 1664
f73dda09 1665 temp = (char *)xmalloc (1 + strlen (string));
726f6388 1666
ccc6cda3
JA
1667 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1668 {
1669 if (string[i] == '\\' && string[i + 1] == '!')
1670 {
1671 temp[j] = '!';
1672 i++;
1673 }
1674 }
1675 strcpy (string, temp);
1676 free (temp);
726f6388 1677}
ccc6cda3 1678#endif
726f6388 1679
a0c0a00f 1680#define CQ_RETURN(x) do { no_longjmp_on_fatal_error = oldjmp; return (x); } while (0)
3185942a 1681
89a92869 1682/* This function assumes s[i] == open; returns with s[ret] == close; used to
0001803f
CR
1683 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1684 matched pairs of quotes or backquotes, or skip word expansions; it is
1685 intended to be used after expansion has been performed and during final
1686 assignment parsing (see arrayfunc.c:assign_compound_array_list()). */
89a92869
CR
1687static int
1688skip_matched_pair (string, start, open, close, flags)
1689 const char *string;
1690 int start, open, close, flags;
1691{
a0c0a00f 1692 int i, pass_next, backq, si, c, count, oldjmp;
89a92869
CR
1693 size_t slen;
1694 char *temp, *ss;
1695 DECLARE_MBSTATE;
1696
1697 slen = strlen (string + start) + start;
a0c0a00f 1698 oldjmp = no_longjmp_on_fatal_error;
89a92869
CR
1699 no_longjmp_on_fatal_error = 1;
1700
1701 i = start + 1; /* skip over leading bracket */
1702 count = 1;
1703 pass_next = backq = 0;
1704 ss = (char *)string;
1705 while (c = string[i])
1706 {
1707 if (pass_next)
1708 {
1709 pass_next = 0;
1710 if (c == 0)
1711 CQ_RETURN(i);
1712 ADVANCE_CHAR (string, slen, i);
1713 continue;
1714 }
1715 else if (c == '\\')
1716 {
1717 pass_next = 1;
1718 i++;
1719 continue;
1720 }
1721 else if (backq)
1722 {
1723 if (c == '`')
1724 backq = 0;
1725 ADVANCE_CHAR (string, slen, i);
1726 continue;
1727 }
0001803f 1728 else if ((flags & 1) == 0 && c == '`')
89a92869
CR
1729 {
1730 backq = 1;
1731 i++;
1732 continue;
1733 }
0001803f 1734 else if ((flags & 1) == 0 && c == open)
89a92869
CR
1735 {
1736 count++;
1737 i++;
1738 continue;
1739 }
1740 else if (c == close)
1741 {
1742 count--;
1743 if (count == 0)
1744 break;
1745 i++;
1746 continue;
1747 }
0001803f 1748 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
89a92869 1749 {
a0c0a00f
CR
1750 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i, 0)
1751 : skip_double_quoted (ss, slen, ++i, 0);
89a92869
CR
1752 /* no increment, the skip functions increment past the closing quote. */
1753 }
0001803f 1754 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
89a92869
CR
1755 {
1756 si = i + 2;
1757 if (string[si] == '\0')
1758 CQ_RETURN(si);
1759
1760 if (string[i+1] == LPAREN)
1761 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1762 else
1763 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
a0c0a00f
CR
1764
1765 CHECK_STRING_OVERRUN (i, si, slen, c);
1766
89a92869
CR
1767 i = si;
1768 if (string[i] == '\0') /* don't increment i past EOS in loop */
1769 break;
1770 i++;
1771 continue;
1772 }
1773 else
1774 ADVANCE_CHAR (string, slen, i);
1775 }
1776
1777 CQ_RETURN(i);
1778}
1779
1780#if defined (ARRAY_VARS)
1781int
0001803f 1782skipsubscript (string, start, flags)
89a92869 1783 const char *string;
0001803f 1784 int start, flags;
89a92869 1785{
0001803f 1786 return (skip_matched_pair (string, start, '[', ']', flags));
89a92869
CR
1787}
1788#endif
1789
3185942a
JA
1790/* Skip characters in STRING until we find a character in DELIMS, and return
1791 the index of that character. START is the index into string at which we
1792 begin. This is similar in spirit to strpbrk, but it returns an index into
1793 STRING and takes a starting index. This little piece of code knows quite
1794 a lot of shell syntax. It's very similar to skip_double_quoted and other
1795 functions of that ilk. */
1796int
1797skip_to_delim (string, start, delims, flags)
1798 char *string;
1799 int start;
1800 char *delims;
1801 int flags;
1802{
a0c0a00f
CR
1803 int i, pass_next, backq, dquote, si, c, oldjmp;
1804 int invert, skipquote, skipcmd, noprocsub, completeflag;
1805 int arithexp, skipcol;
3185942a 1806 size_t slen;
495aee44 1807 char *temp, open[3];
3185942a
JA
1808 DECLARE_MBSTATE;
1809
1810 slen = strlen (string + start) + start;
a0c0a00f 1811 oldjmp = no_longjmp_on_fatal_error;
3185942a
JA
1812 if (flags & SD_NOJMP)
1813 no_longjmp_on_fatal_error = 1;
1814 invert = (flags & SD_INVERT);
0001803f 1815 skipcmd = (flags & SD_NOSKIPCMD) == 0;
a0c0a00f
CR
1816 noprocsub = (flags & SD_NOPROCSUB);
1817 completeflag = (flags & SD_COMPLETE) ? SX_COMPLETE : 0;
1818
1819 arithexp = (flags & SD_ARITHEXP);
1820 skipcol = 0;
3185942a
JA
1821
1822 i = start;
a0c0a00f 1823 pass_next = backq = dquote = 0;
3185942a
JA
1824 while (c = string[i])
1825 {
0001803f
CR
1826 /* If this is non-zero, we should not let quote characters be delimiters
1827 and the current character is a single or double quote. We should not
1828 test whether or not it's a delimiter until after we skip single- or
1829 double-quoted strings. */
1830 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
3185942a
JA
1831 if (pass_next)
1832 {
1833 pass_next = 0;
1834 if (c == 0)
1835 CQ_RETURN(i);
1836 ADVANCE_CHAR (string, slen, i);
1837 continue;
1838 }
1839 else if (c == '\\')
1840 {
1841 pass_next = 1;
1842 i++;
1843 continue;
1844 }
1845 else if (backq)
1846 {
1847 if (c == '`')
1848 backq = 0;
1849 ADVANCE_CHAR (string, slen, i);
1850 continue;
1851 }
1852 else if (c == '`')
1853 {
1854 backq = 1;
1855 i++;
1856 continue;
1857 }
a0c0a00f 1858 else if (arithexp && skipcol && c == ':')
3185942a 1859 {
a0c0a00f
CR
1860 skipcol--;
1861 i++;
1862 continue;
3185942a 1863 }
a0c0a00f
CR
1864 else if (arithexp && c == '?')
1865 {
1866 skipcol++;
1867 i++;
1868 continue;
1869 }
1870 else if (skipquote == 0 && invert == 0 && member (c, delims))
1871 break;
1872 /* the usual case is to use skip_xxx_quoted, but we don't skip over double
1873 quoted strings when looking for the history expansion character as a
1874 delimiter. */
1875 /* special case for programmable completion which takes place before
1876 parser converts backslash-escaped single quotes between $'...' to
1877 `regular' single-quoted strings. */
1878 else if (completeflag && i > 0 && string[i-1] == '$' && c == '\'')
1879 i = skip_single_quoted (string, slen, ++i, SX_COMPLETE);
1880 else if (c == '\'')
1881 i = skip_single_quoted (string, slen, ++i, 0);
1882 else if (c == '"')
1883 i = skip_double_quoted (string, slen, ++i, completeflag);
1884 else if (c == LPAREN && arithexp)
1885 {
1886 si = i + 1;
1887 if (string[si] == '\0')
1888 CQ_RETURN(si);
1889
1890 temp = extract_delimited_string (string, &si, "(", "(", ")", SX_NOALLOC); /* ) */
1891 i = si;
1892 if (string[i] == '\0') /* don't increment i past EOS in loop */
1893 break;
1894 i++;
1895 continue;
1896 }
0001803f 1897 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
3185942a
JA
1898 {
1899 si = i + 2;
1900 if (string[si] == '\0')
1901 CQ_RETURN(si);
1902
1903 if (string[i+1] == LPAREN)
1904 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1905 else
1906 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1907 i = si;
1908 if (string[i] == '\0') /* don't increment i past EOS in loop */
1909 break;
1910 i++;
1911 continue;
1912 }
0001803f 1913#if defined (PROCESS_SUBSTITUTION)
a0c0a00f 1914 else if (skipcmd && noprocsub == 0 && (c == '<' || c == '>') && string[i+1] == LPAREN)
0001803f
CR
1915 {
1916 si = i + 2;
1917 if (string[si] == '\0')
1918 CQ_RETURN(si);
a0c0a00f
CR
1919#if 1
1920 temp = extract_delimited_string (string, &si, (c == '<') ? "<(" : ">(", "(", ")", SX_COMMAND|SX_NOALLOC); /* )) */
1921#else
85b94814 1922 temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si, 0);
a0c0a00f
CR
1923 free (temp); /* XXX - not using SX_NOALLOC here yet */
1924#endif
0001803f
CR
1925 i = si;
1926 if (string[i] == '\0')
1927 break;
1928 i++;
1929 continue;
1930 }
1931#endif /* PROCESS_SUBSTITUTION */
495aee44
CR
1932#if defined (EXTENDED_GLOB)
1933 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1934 {
1935 si = i + 2;
1936 if (string[si] == '\0')
1937 CQ_RETURN(si);
1938
1939 open[0] = c;
1940 open[1] = LPAREN;
1941 open[2] = '\0';
1942 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
1943
1944 i = si;
1945 if (string[i] == '\0') /* don't increment i past EOS in loop */
1946 break;
1947 i++;
1948 continue;
1949 }
1950#endif
ac50fbac
CR
1951 else if ((flags & SD_GLOB) && c == LBRACK)
1952 {
1953 si = i + 1;
1954 if (string[si] == '\0')
1955 CQ_RETURN(si);
1956
1957 temp = extract_delimited_string (string, &si, "[", "[", "]", SX_NOALLOC); /* ] */
1958
1959 i = si;
1960 if (string[i] == '\0') /* don't increment i past EOS in loop */
1961 break;
1962 i++;
1963 continue;
1964 }
0001803f 1965 else if ((skipquote || invert) && (member (c, delims) == 0))
3185942a
JA
1966 break;
1967 else
1968 ADVANCE_CHAR (string, slen, i);
1969 }
1970
1971 CQ_RETURN(i);
1972}
1973
a0c0a00f
CR
1974#if defined (BANG_HISTORY)
1975/* Skip to the history expansion character (delims[0]), paying attention to
1976 quoted strings and command and process substitution. This is a stripped-
1977 down version of skip_to_delims. The essential difference is that this
1978 resets the quoting state when starting a command substitution */
1979int
1980skip_to_histexp (string, start, delims, flags)
1981 char *string;
1982 int start;
1983 char *delims;
1984 int flags;
1985{
1986 int i, pass_next, backq, dquote, si, c, oldjmp;
1987 int histexp_comsub, histexp_backq, old_dquote;
1988 size_t slen;
1989 char *temp, open[3];
1990 DECLARE_MBSTATE;
1991
1992 slen = strlen (string + start) + start;
1993 oldjmp = no_longjmp_on_fatal_error;
1994 if (flags & SD_NOJMP)
1995 no_longjmp_on_fatal_error = 1;
1996
1997 histexp_comsub = histexp_backq = old_dquote = 0;
1998
1999 i = start;
2000 pass_next = backq = dquote = 0;
2001 while (c = string[i])
2002 {
2003 if (pass_next)
2004 {
2005 pass_next = 0;
2006 if (c == 0)
2007 CQ_RETURN(i);
2008 ADVANCE_CHAR (string, slen, i);
2009 continue;
2010 }
2011 else if (c == '\\')
2012 {
2013 pass_next = 1;
2014 i++;
2015 continue;
2016 }
2017 else if (backq && c == '`')
2018 {
2019 backq = 0;
2020 histexp_backq--;
2021 dquote = old_dquote;
2022 i++;
2023 continue;
2024 }
2025 else if (c == '`')
2026 {
2027 backq = 1;
2028 histexp_backq++;
2029 old_dquote = dquote; /* simple - one level for now */
2030 dquote = 0;
2031 i++;
2032 continue;
2033 }
2034 /* When in double quotes, act as if the double quote is a member of
2035 history_no_expand_chars, like the history library does */
2036 else if (dquote && c == delims[0] && string[i+1] == '"')
2037 {
2038 i++;
2039 continue;
2040 }
2041 else if (c == delims[0])
2042 break;
2043 /* the usual case is to use skip_xxx_quoted, but we don't skip over double
2044 quoted strings when looking for the history expansion character as a
2045 delimiter. */
2046 else if (dquote && c == '\'')
2047 {
2048 i++;
2049 continue;
2050 }
2051 else if (c == '\'')
2052 i = skip_single_quoted (string, slen, ++i, 0);
2053 /* The posixly_correct test makes posix-mode shells allow double quotes
2054 to quote the history expansion character */
2055 else if (posixly_correct == 0 && c == '"')
2056 {
2057 dquote = 1 - dquote;
2058 i++;
2059 continue;
2060 }
2061 else if (c == '"')
2062 i = skip_double_quoted (string, slen, ++i, 0);
2063#if defined (PROCESS_SUBSTITUTION)
2064 else if ((c == '$' || c == '<' || c == '>') && string[i+1] == LPAREN && string[i+2] != LPAREN)
2065#else
2066 else if (c == '$' && string[i+1] == LPAREN && string[i+2] != LPAREN)
2067#endif
2068 {
2069 if (string[i+2] == '\0')
2070 CQ_RETURN(i+2);
2071 i += 2;
2072 histexp_comsub++;
2073 old_dquote = dquote;
2074 dquote = 0;
2075 }
2076 else if (histexp_comsub && c == RPAREN)
2077 {
2078 histexp_comsub--;
2079 dquote = old_dquote;
2080 i++;
2081 continue;
2082 }
2083 else if (backq) /* placeholder */
2084 {
2085 ADVANCE_CHAR (string, slen, i);
2086 continue;
2087 }
2088 else
2089 ADVANCE_CHAR (string, slen, i);
2090 }
2091
2092 CQ_RETURN(i);
2093}
2094#endif /* BANG_HISTORY */
2095
ccc6cda3 2096#if defined (READLINE)
726f6388
JA
2097/* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
2098 an unclosed quoted string), or if the character at EINDEX is quoted
28ef6c31 2099 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
b72432fd 2100 single and double-quoted string parsing functions should not return an
7117c2d2
JA
2101 error if there are unclosed quotes or braces. The characters that this
2102 recognizes need to be the same as the contents of
2103 rl_completer_quote_characters. */
b72432fd 2104
726f6388
JA
2105int
2106char_is_quoted (string, eindex)
2107 char *string;
2108 int eindex;
2109{
a0c0a00f 2110 int i, pass_next, c, oldjmp;
7117c2d2
JA
2111 size_t slen;
2112 DECLARE_MBSTATE;
726f6388 2113
7117c2d2 2114 slen = strlen (string);
a0c0a00f 2115 oldjmp = no_longjmp_on_fatal_error;
28ef6c31 2116 no_longjmp_on_fatal_error = 1;
7117c2d2
JA
2117 i = pass_next = 0;
2118 while (i <= eindex)
726f6388 2119 {
7117c2d2
JA
2120 c = string[i];
2121
726f6388
JA
2122 if (pass_next)
2123 {
2124 pass_next = 0;
2125 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
b72432fd 2126 CQ_RETURN(1);
7117c2d2 2127 ADVANCE_CHAR (string, slen, i);
726f6388
JA
2128 continue;
2129 }
7117c2d2 2130 else if (c == '\\')
ccc6cda3
JA
2131 {
2132 pass_next = 1;
7117c2d2 2133 i++;
ccc6cda3
JA
2134 continue;
2135 }
a0c0a00f
CR
2136 else if (c == '$' && string[i+1] == '\'' && string[i+2])
2137 {
2138 i += 2;
2139 i = skip_single_quoted (string, slen, i, SX_COMPLETE);
2140 if (i > eindex)
2141 CQ_RETURN (i);
2142 }
7117c2d2
JA
2143 else if (c == '\'' || c == '"')
2144 {
a0c0a00f
CR
2145 i = (c == '\'') ? skip_single_quoted (string, slen, ++i, 0)
2146 : skip_double_quoted (string, slen, ++i, SX_COMPLETE);
7117c2d2
JA
2147 if (i > eindex)
2148 CQ_RETURN(1);
2149 /* no increment, the skip_xxx functions go one past end */
2150 }
2151 else
2152 ADVANCE_CHAR (string, slen, i);
726f6388 2153 }
7117c2d2 2154
b72432fd 2155 CQ_RETURN(0);
726f6388
JA
2156}
2157
726f6388
JA
2158int
2159unclosed_pair (string, eindex, openstr)
2160 char *string;
2161 int eindex;
2162 char *openstr;
2163{
ccc6cda3 2164 int i, pass_next, openc, olen;
7117c2d2
JA
2165 size_t slen;
2166 DECLARE_MBSTATE;
726f6388 2167
7117c2d2 2168 slen = strlen (string);
726f6388 2169 olen = strlen (openstr);
7117c2d2
JA
2170 i = pass_next = openc = 0;
2171 while (i <= eindex)
726f6388
JA
2172 {
2173 if (pass_next)
2174 {
2175 pass_next = 0;
2176 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
2177 return 0;
7117c2d2
JA
2178 ADVANCE_CHAR (string, slen, i);
2179 continue;
2180 }
2181 else if (string[i] == '\\')
2182 {
2183 pass_next = 1;
2184 i++;
726f6388
JA
2185 continue;
2186 }
2187 else if (STREQN (string + i, openstr, olen))
2188 {
2189 openc = 1 - openc;
7117c2d2 2190 i += olen;
726f6388 2191 }
a0c0a00f 2192 /* XXX - may want to handle $'...' specially here */
ccc6cda3 2193 else if (string[i] == '\'' || string[i] == '"')
726f6388 2194 {
a0c0a00f
CR
2195 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i, 0)
2196 : skip_double_quoted (string, slen, i, SX_COMPLETE);
726f6388
JA
2197 if (i > eindex)
2198 return 0;
2199 }
7117c2d2
JA
2200 else
2201 ADVANCE_CHAR (string, slen, i);
726f6388
JA
2202 }
2203 return (openc);
2204}
bb70624e 2205
bb70624e
JA
2206/* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
2207 individual words. If DELIMS is NULL, the current value of $IFS is used
b80f6443
JA
2208 to split the string, and the function follows the shell field splitting
2209 rules. SENTINEL is an index to look for. NWP, if non-NULL,
bb70624e
JA
2210 gets the number of words in the returned list. CWP, if non-NULL, gets
2211 the index of the word containing SENTINEL. Non-whitespace chars in
2212 DELIMS delimit separate fields. */
2213WORD_LIST *
0001803f 2214split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
bb70624e
JA
2215 char *string;
2216 int slen;
2217 char *delims;
0001803f 2218 int sentinel, flags;
bb70624e
JA
2219 int *nwp, *cwp;
2220{
0001803f 2221 int ts, te, i, nw, cw, ifs_split, dflags;
f73dda09 2222 char *token, *d, *d2;
bb70624e
JA
2223 WORD_LIST *ret, *tl;
2224
2225 if (string == 0 || *string == '\0')
2226 {
2227 if (nwp)
2228 *nwp = 0;
2229 if (cwp)
2230 *cwp = 0;
2231 return ((WORD_LIST *)NULL);
2232 }
2233
7117c2d2 2234 d = (delims == 0) ? ifs_value : delims;
b80f6443 2235 ifs_split = delims == 0;
bb70624e
JA
2236
2237 /* Make d2 the non-whitespace characters in delims */
2238 d2 = 0;
2239 if (delims)
2240 {
95732b49
JA
2241 size_t slength;
2242#if defined (HANDLE_MULTIBYTE)
2243 size_t mblength = 1;
2244#endif
2245 DECLARE_MBSTATE;
2246
2247 slength = strlen (delims);
2248 d2 = (char *)xmalloc (slength + 1);
2249 i = ts = 0;
2250 while (delims[i])
bb70624e 2251 {
95732b49 2252#if defined (HANDLE_MULTIBYTE)
0628567a
JA
2253 mbstate_t state_bak;
2254 state_bak = state;
95732b49
JA
2255 mblength = MBRLEN (delims + i, slength, &state);
2256 if (MB_INVALIDCH (mblength))
2257 state = state_bak;
2258 else if (mblength > 1)
2259 {
2260 memcpy (d2 + ts, delims + i, mblength);
2261 ts += mblength;
2262 i += mblength;
2263 slength -= mblength;
2264 continue;
2265 }
2266#endif
2267 if (whitespace (delims[i]) == 0)
bb70624e 2268 d2[ts++] = delims[i];
95732b49
JA
2269
2270 i++;
2271 slength--;
bb70624e
JA
2272 }
2273 d2[ts] = '\0';
2274 }
2275
2276 ret = (WORD_LIST *)NULL;
2277
0001803f 2278 /* Remove sequences of whitespace characters at the start of the string, as
b80f6443
JA
2279 long as those characters are delimiters. */
2280 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
bb70624e
JA
2281 ;
2282 if (string[i] == '\0')
a0c0a00f
CR
2283 {
2284 FREE (d2);
2285 return (ret);
2286 }
bb70624e
JA
2287
2288 ts = i;
2289 nw = 0;
2290 cw = -1;
0001803f 2291 dflags = flags|SD_NOJMP;
bb70624e
JA
2292 while (1)
2293 {
0001803f 2294 te = skip_to_delim (string, ts, d, dflags);
bb70624e
JA
2295
2296 /* If we have a non-whitespace delimiter character, use it to make a
2297 separate field. This is just about what $IFS splitting does and
2298 is closer to the behavior of the shell parser. */
28ef6c31 2299 if (ts == te && d2 && member (string[ts], d2))
bb70624e
JA
2300 {
2301 te = ts + 1;
b80f6443
JA
2302 /* If we're using IFS splitting, the non-whitespace delimiter char
2303 and any additional IFS whitespace delimits a field. */
2304 if (ifs_split)
2305 while (member (string[te], d) && spctabnl (string[te]))
2306 te++;
2307 else
2308 while (member (string[te], d2))
2309 te++;
bb70624e
JA
2310 }
2311
2312 token = substring (string, ts, te);
2313
2314 ret = add_string_to_list (token, ret);
2315 free (token);
2316 nw++;
2317
2318 if (sentinel >= ts && sentinel <= te)
2319 cw = nw;
2320
2321 /* If the cursor is at whitespace just before word start, set the
28ef6c31 2322 sentinel word to the current word. */
bb70624e
JA
2323 if (cwp && cw == -1 && sentinel == ts-1)
2324 cw = nw;
2325
2326 /* If the cursor is at whitespace between two words, make a new, empty
28ef6c31
JA
2327 word, add it before (well, after, since the list is in reverse order)
2328 the word we just added, and set the current word to that one. */
bb70624e 2329 if (cwp && cw == -1 && sentinel < ts)
28ef6c31 2330 {
7117c2d2 2331 tl = make_word_list (make_word (""), ret->next);
28ef6c31
JA
2332 ret->next = tl;
2333 cw = nw;
2334 nw++;
2335 }
bb70624e
JA
2336
2337 if (string[te] == 0)
2338 break;
2339
b80f6443
JA
2340 i = te;
2341 while (member (string[i], d) && (ifs_split || spctabnl(string[i])))
bb70624e
JA
2342 i++;
2343
2344 if (string[i])
2345 ts = i;
2346 else
2347 break;
2348 }
2349
2350 /* Special case for SENTINEL at the end of STRING. If we haven't found
2351 the word containing SENTINEL yet, and the index we're looking for is at
0001803f
CR
2352 the end of STRING (or past the end of the previously-found token,
2353 possible if the end of the line is composed solely of IFS whitespace)
2354 add an additional null argument and set the current word pointer to that. */
2355 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
bb70624e
JA
2356 {
2357 if (whitespace (string[sentinel - 1]))
28ef6c31
JA
2358 {
2359 token = "";
2360 ret = add_string_to_list (token, ret);
2361 nw++;
2362 }
bb70624e
JA
2363 cw = nw;
2364 }
2365
2366 if (nwp)
2367 *nwp = nw;
2368 if (cwp)
2369 *cwp = cw;
2370
ac50fbac
CR
2371 FREE (d2);
2372
bb70624e
JA
2373 return (REVERSE_LIST (ret, WORD_LIST *));
2374}
726f6388
JA
2375#endif /* READLINE */
2376
ccc6cda3
JA
2377#if 0
2378/* UNUSED */
726f6388
JA
2379/* Extract the name of the variable to bind to from the assignment string. */
2380char *
2381assignment_name (string)
2382 char *string;
2383{
ccc6cda3 2384 int offset;
726f6388
JA
2385 char *temp;
2386
b80f6443 2387 offset = assignment (string, 0);
ccc6cda3 2388 if (offset == 0)
726f6388 2389 return (char *)NULL;
bb70624e 2390 temp = substring (string, 0, offset);
726f6388
JA
2391 return (temp);
2392}
ccc6cda3 2393#endif
726f6388 2394
cce855bc
JA
2395/* **************************************************************** */
2396/* */
2397/* Functions to convert strings to WORD_LISTs and vice versa */
2398/* */
2399/* **************************************************************** */
2400
726f6388
JA
2401/* Return a single string of all the words in LIST. SEP is the separator
2402 to put between individual elements of LIST in the output string. */
7117c2d2 2403char *
726f6388
JA
2404string_list_internal (list, sep)
2405 WORD_LIST *list;
2406 char *sep;
2407{
2408 register WORD_LIST *t;
2409 char *result, *r;
a0c0a00f 2410 size_t word_len, sep_len, result_size;
726f6388 2411
ccc6cda3 2412 if (list == 0)
726f6388
JA
2413 return ((char *)NULL);
2414
b80f6443
JA
2415 /* Short-circuit quickly if we don't need to separate anything. */
2416 if (list->next == 0)
2417 return (savestring (list->word->word));
2418
726f6388
JA
2419 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2420 sep_len = STRLEN (sep);
2421 result_size = 0;
2422
2423 for (t = list; t; t = t->next)
2424 {
2425 if (t != list)
2426 result_size += sep_len;
2427 result_size += strlen (t->word->word);
2428 }
2429
f73dda09 2430 r = result = (char *)xmalloc (result_size + 1);
726f6388
JA
2431
2432 for (t = list; t; t = t->next)
2433 {
2434 if (t != list && sep_len)
2435 {
ccc6cda3
JA
2436 if (sep_len > 1)
2437 {
2438 FASTCOPY (sep, r, sep_len);
2439 r += sep_len;
2440 }
2441 else
2442 *r++ = sep[0];
726f6388
JA
2443 }
2444
2445 word_len = strlen (t->word->word);
2446 FASTCOPY (t->word->word, r, word_len);
2447 r += word_len;
2448 }
2449
ccc6cda3 2450 *r = '\0';
726f6388
JA
2451 return (result);
2452}
2453
2454/* Return a single string of all the words present in LIST, separating
2455 each word with a space. */
2456char *
2457string_list (list)
2458 WORD_LIST *list;
2459{
2460 return (string_list_internal (list, " "));
2461}
2462
3185942a
JA
2463/* An external interface that can be used by the rest of the shell to
2464 obtain a string containing the first character in $IFS. Handles all
2465 the multibyte complications. If LENP is non-null, it is set to the
2466 length of the returned string. */
2467char *
2468ifs_firstchar (lenp)
2469 int *lenp;
2470{
2471 char *ret;
2472 int len;
2473
2474 ret = xmalloc (MB_LEN_MAX + 1);
2475#if defined (HANDLE_MULTIBYTE)
2476 if (ifs_firstc_len == 1)
2477 {
2478 ret[0] = ifs_firstc[0];
2479 ret[1] = '\0';
2480 len = ret[0] ? 1 : 0;
2481 }
2482 else
2483 {
2484 memcpy (ret, ifs_firstc, ifs_firstc_len);
2485 ret[len = ifs_firstc_len] = '\0';
2486 }
2487#else
2488 ret[0] = ifs_firstc;
2489 ret[1] = '\0';
2490 len = ret[0] ? 0 : 1;
2491#endif
2492
2493 if (lenp)
2494 *lenp = len;
2495
2496 return ret;
2497}
2498
726f6388
JA
2499/* Return a single string of all the words present in LIST, obeying the
2500 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2501 expansion [of $*] appears within a double quoted string, it expands
2502 to a single field with the value of each parameter separated by the
2503 first character of the IFS variable, or by a <space> if IFS is unset." */
f73dda09 2504char *
726f6388
JA
2505string_list_dollar_star (list)
2506 WORD_LIST *list;
2507{
0628567a 2508 char *ret;
95732b49 2509#if defined (HANDLE_MULTIBYTE)
0628567a 2510# if defined (__GNUC__)
95732b49 2511 char sep[MB_CUR_MAX + 1];
0628567a
JA
2512# else
2513 char *sep = 0;
2514# endif
95732b49 2515#else
7117c2d2 2516 char sep[2];
95732b49 2517#endif
726f6388 2518
95732b49 2519#if defined (HANDLE_MULTIBYTE)
0628567a
JA
2520# if !defined (__GNUC__)
2521 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2522# endif /* !__GNUC__ */
95732b49
JA
2523 if (ifs_firstc_len == 1)
2524 {
2525 sep[0] = ifs_firstc[0];
2526 sep[1] = '\0';
2527 }
2528 else
2529 {
2530 memcpy (sep, ifs_firstc, ifs_firstc_len);
2531 sep[ifs_firstc_len] = '\0';
2532 }
2533#else
7117c2d2 2534 sep[0] = ifs_firstc;
726f6388 2535 sep[1] = '\0';
95732b49 2536#endif
726f6388 2537
0628567a
JA
2538 ret = string_list_internal (list, sep);
2539#if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2540 free (sep);
2541#endif
2542 return ret;
726f6388
JA
2543}
2544
cce855bc
JA
2545/* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2546 is non-zero, the $@ appears within double quotes, and we should quote
2547 the list before converting it into a string. If IFS is unset, and the
2548 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2549 in the words in the list, because the default value of $IFS is
2550 <space><tab><newline>, IFS characters in the words in the list should
2551 also be split. If IFS is null, and the word is not quoted, we need
2552 to quote the words in the list to preserve the positional parameters
a0c0a00f
CR
2553 exactly.
2554 Valid values for the FLAGS argument are the PF_ flags in command.h,
2555 the only one we care about is PF_ASSIGNRHS. $@ is supposed to expand
2556 to the positional parameters separated by spaces no matter what IFS is
2557 set to if in a context where word splitting is not performed. The only
2558 one that we didn't handle before is assignment statement arguments to
2559 declaration builtins like `declare'. */
f73dda09 2560char *
a0c0a00f 2561string_list_dollar_at (list, quoted, flags)
cce855bc
JA
2562 WORD_LIST *list;
2563 int quoted;
a0c0a00f 2564 int flags;
cce855bc 2565{
95732b49
JA
2566 char *ifs, *ret;
2567#if defined (HANDLE_MULTIBYTE)
0628567a 2568# if defined (__GNUC__)
95732b49 2569 char sep[MB_CUR_MAX + 1];
0628567a
JA
2570# else
2571 char *sep = 0;
2572# endif /* !__GNUC__ */
95732b49
JA
2573#else
2574 char sep[2];
2575#endif
cce855bc
JA
2576 WORD_LIST *tlist;
2577
7117c2d2
JA
2578 /* XXX this could just be ifs = ifs_value; */
2579 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
cce855bc 2580
95732b49 2581#if defined (HANDLE_MULTIBYTE)
0628567a
JA
2582# if !defined (__GNUC__)
2583 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2584# endif /* !__GNUC__ */
a0c0a00f
CR
2585 /* XXX - bash-4.4/bash-5.0 testing PF_ASSIGNRHS */
2586 if (flags & PF_ASSIGNRHS)
2587 {
2588 sep[0] = ' ';
2589 sep[1] = '\0';
2590 }
2591 else if (ifs && *ifs)
95732b49
JA
2592 {
2593 if (ifs_firstc_len == 1)
2594 {
2595 sep[0] = ifs_firstc[0];
2596 sep[1] = '\0';
2597 }
2598 else
2599 {
2600 memcpy (sep, ifs_firstc, ifs_firstc_len);
2601 sep[ifs_firstc_len] = '\0';
2602 }
2603 }
2604 else
2605 {
2606 sep[0] = ' ';
2607 sep[1] = '\0';
2608 }
2609#else
a0c0a00f
CR
2610 /* XXX - bash-4.4/bash-5.0 test PF_ASSIGNRHS */
2611 sep[0] = ((flags & PF_ASSIGNRHS) || ifs == 0 || *ifs == 0) ? ' ' : *ifs;
cce855bc 2612 sep[1] = '\0';
95732b49 2613#endif
cce855bc 2614
f1be666c
JA
2615 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2616 it now that quote_escapes quotes spaces */
0001803f 2617 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
cce855bc
JA
2618 ? quote_list (list)
2619 : list_quote_escapes (list);
0628567a
JA
2620
2621 ret = string_list_internal (tlist, sep);
2622#if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2623 free (sep);
2624#endif
2625 return ret;
cce855bc
JA
2626}
2627
ac50fbac 2628/* Turn the positional parameters into a string, understanding quoting and
3185942a
JA
2629 the various subtleties of using the first character of $IFS as the
2630 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2631 string_list as appropriate. */
2632char *
2633string_list_pos_params (pchar, list, quoted)
2634 int pchar;
2635 WORD_LIST *list;
2636 int quoted;
2637{
2638 char *ret;
2639 WORD_LIST *tlist;
2640
2641 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2642 {
2643 tlist = quote_list (list);
2644 word_list_remove_quoted_nulls (tlist);
2645 ret = string_list_dollar_star (tlist);
2646 }
2647 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2648 {
2649 tlist = quote_list (list);
2650 word_list_remove_quoted_nulls (tlist);
2651 ret = string_list (tlist);
2652 }
2653 else if (pchar == '*')
2654 {
2655 /* Even when unquoted, string_list_dollar_star does the right thing
2656 making sure that the first character of $IFS is used as the
2657 separator. */
2658 ret = string_list_dollar_star (list);
2659 }
2660 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2661 /* We use string_list_dollar_at, but only if the string is quoted, since
2662 that quotes the escapes if it's not, which we don't want. We could
2663 use string_list (the old code did), but that doesn't do the right
2664 thing if the first character of $IFS is not a space. We use
2665 string_list_dollar_star if the string is unquoted so we make sure that
2666 the elements of $@ are separated by the first character of $IFS for
2667 later splitting. */
a0c0a00f 2668 ret = string_list_dollar_at (list, quoted, 0);
3185942a
JA
2669 else if (pchar == '@')
2670 ret = string_list_dollar_star (list);
2671 else
2672 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2673
2674 return ret;
2675}
2676
726f6388
JA
2677/* Return the list of words present in STRING. Separate the string into
2678 words at any of the characters found in SEPARATORS. If QUOTED is
2679 non-zero then word in the list will have its quoted flag set, otherwise
2680 the quoted flag is left as make_word () deemed fit.
2681
2682 This obeys the P1003.2 word splitting semantics. If `separators' is
2683 exactly <space><tab><newline>, then the splitting algorithm is that of
2684 the Bourne shell, which treats any sequence of characters from `separators'
2685 as a delimiter. If IFS is unset, which results in `separators' being set
2686 to "", no splitting occurs. If separators has some other value, the
2687 following rules are applied (`IFS white space' means zero or more
2688 occurrences of <space>, <tab>, or <newline>, as long as those characters
2689 are in `separators'):
2690
2691 1) IFS white space is ignored at the start and the end of the
2692 string.
2693 2) Each occurrence of a character in `separators' that is not
2694 IFS white space, along with any adjacent occurrences of
2695 IFS white space delimits a field.
2696 3) Any nonzero-length sequence of IFS white space delimits a field.
2697 */
2698
2699/* BEWARE! list_string strips null arguments. Don't call it twice and
2700 expect to have "" preserved! */
2701
726f6388
JA
2702/* This performs word splitting and quoted null character removal on
2703 STRING. */
b80f6443
JA
2704#define issep(c) \
2705 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2706 : (c) == (separators)[0]) \
2707 : 0)
726f6388
JA
2708
2709WORD_LIST *
2710list_string (string, separators, quoted)
2711 register char *string, *separators;
2712 int quoted;
2713{
ccc6cda3
JA
2714 WORD_LIST *result;
2715 WORD_DESC *t;
2716 char *current_word, *s;
3185942a 2717 int sindex, sh_style_split, whitesep, xflags;
95732b49 2718 size_t slen;
726f6388
JA
2719
2720 if (!string || !*string)
2721 return ((WORD_LIST *)NULL);
2722
7117c2d2
JA
2723 sh_style_split = separators && separators[0] == ' ' &&
2724 separators[1] == '\t' &&
2725 separators[2] == '\n' &&
2726 separators[3] == '\0';
3185942a
JA
2727 for (xflags = 0, s = ifs_value; s && *s; s++)
2728 {
2729 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2730 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2731 }
726f6388 2732
95732b49 2733 slen = 0;
726f6388
JA
2734 /* Remove sequences of whitespace at the beginning of STRING, as
2735 long as those characters appear in IFS. Do not do this if
2736 STRING is quoted or if there are no separator characters. */
2737 if (!quoted || !separators || !*separators)
2738 {
2739 for (s = string; *s && spctabnl (*s) && issep (*s); s++);
2740
2741 if (!*s)
2742 return ((WORD_LIST *)NULL);
2743
2744 string = s;
2745 }
2746
2747 /* OK, now STRING points to a word that does not begin with white space.
2748 The splitting algorithm is:
7117c2d2
JA
2749 extract a word, stopping at a separator
2750 skip sequences of spc, tab, or nl as long as they are separators
726f6388 2751 This obeys the field splitting rules in Posix.2. */
a0c0a00f 2752 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 1;
ccc6cda3 2753 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
726f6388 2754 {
95732b49
JA
2755 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2756 unless multibyte chars are possible. */
3185942a 2757 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
ccc6cda3 2758 if (current_word == 0)
726f6388
JA
2759 break;
2760
2761 /* If we have a quoted empty string, add a quoted null argument. We
2762 want to preserve the quoted null character iff this is a quoted
2763 empty string; otherwise the quoted null characters are removed
2764 below. */
2765 if (QUOTED_NULL (current_word))
2766 {
95732b49 2767 t = alloc_word_desc ();
726f6388 2768 t->word = make_quoted_char ('\0');
95732b49 2769 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
726f6388
JA
2770 result = make_word_list (t, result);
2771 }
ccc6cda3 2772 else if (current_word[0] != '\0')
726f6388
JA
2773 {
2774 /* If we have something, then add it regardless. However,
2775 perform quoted null character removal on the current word. */
2776 remove_quoted_nulls (current_word);
cce855bc 2777 result = add_string_to_list (current_word, result);
95732b49 2778 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
ccc6cda3
JA
2779 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2780 result->word->flags |= W_QUOTED;
726f6388
JA
2781 }
2782
2783 /* If we're not doing sequences of separators in the traditional
2784 Bourne shell style, then add a quoted null argument. */
726f6388
JA
2785 else if (!sh_style_split && !spctabnl (string[sindex]))
2786 {
95732b49 2787 t = alloc_word_desc ();
ccc6cda3 2788 t->word = make_quoted_char ('\0');
95732b49 2789 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
ccc6cda3 2790 result = make_word_list (t, result);
726f6388
JA
2791 }
2792
2793 free (current_word);
2794
28ef6c31
JA
2795 /* Note whether or not the separator is IFS whitespace, used later. */
2796 whitesep = string[sindex] && spctabnl (string[sindex]);
2797
726f6388
JA
2798 /* Move past the current separator character. */
2799 if (string[sindex])
95732b49
JA
2800 {
2801 DECLARE_MBSTATE;
2802 ADVANCE_CHAR (string, slen, sindex);
2803 }
726f6388
JA
2804
2805 /* Now skip sequences of space, tab, or newline characters if they are
2806 in the list of separators. */
2807 while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex]))
2808 sindex++;
28ef6c31 2809
7117c2d2
JA
2810 /* If the first separator was IFS whitespace and the current character
2811 is a non-whitespace IFS character, it should be part of the current
2812 field delimiter, not a separate delimiter that would result in an
2813 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
28ef6c31 2814 if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex]))
95732b49
JA
2815 {
2816 sindex++;
2817 /* An IFS character that is not IFS white space, along with any
2818 adjacent IFS white space, shall delimit a field. (SUSv3) */
0628567a 2819 while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex]))
95732b49
JA
2820 sindex++;
2821 }
726f6388
JA
2822 }
2823 return (REVERSE_LIST (result, WORD_LIST *));
2824}
2825
2826/* Parse a single word from STRING, using SEPARATORS to separate fields.
2827 ENDPTR is set to the first character after the word. This is used by
bc007799
CR
2828 the `read' builtin.
2829
2830 This is never called with SEPARATORS != $IFS, and takes advantage of that.
7117c2d2 2831
726f6388
JA
2832 XXX - this function is very similar to list_string; they should be
2833 combined - XXX */
bc007799
CR
2834
2835#define islocalsep(c) (local_cmap[(unsigned char)(c)] != 0)
2836
726f6388
JA
2837char *
2838get_word_from_string (stringp, separators, endptr)
2839 char **stringp, *separators, **endptr;
2840{
2841 register char *s;
2842 char *current_word;
3185942a 2843 int sindex, sh_style_split, whitesep, xflags;
bc007799 2844 unsigned char local_cmap[UCHAR_MAX+1]; /* really only need single-byte chars here */
95732b49 2845 size_t slen;
726f6388
JA
2846
2847 if (!stringp || !*stringp || !**stringp)
2848 return ((char *)NULL);
ccc6cda3 2849
7117c2d2
JA
2850 sh_style_split = separators && separators[0] == ' ' &&
2851 separators[1] == '\t' &&
2852 separators[2] == '\n' &&
2853 separators[3] == '\0';
bc007799
CR
2854 memset (local_cmap, '\0', sizeof (local_cmap));
2855 for (xflags = 0, s = separators; s && *s; s++)
3185942a
JA
2856 {
2857 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2858 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
bc007799 2859 local_cmap[(unsigned char)*s] = 1; /* local charmap of separators */
3185942a 2860 }
726f6388 2861
3185942a 2862 s = *stringp;
95732b49
JA
2863 slen = 0;
2864
726f6388 2865 /* Remove sequences of whitespace at the beginning of STRING, as
bc007799
CR
2866 long as those characters appear in SEPARATORS. This happens if
2867 SEPARATORS == $' \t\n' or if IFS is unset. */
2868 if (sh_style_split || separators == 0)
726f6388 2869 {
bc007799 2870 for (; *s && spctabnl (*s) && islocalsep (*s); s++);
726f6388
JA
2871
2872 /* If the string is nothing but whitespace, update it and return. */
2873 if (!*s)
2874 {
2875 *stringp = s;
2876 if (endptr)
2877 *endptr = s;
2878 return ((char *)NULL);
2879 }
2880 }
2881
2882 /* OK, S points to a word that does not begin with white space.
2883 Now extract a word, stopping at a separator, save a pointer to
2884 the first character after the word, then skip sequences of spc,
2885 tab, or nl as long as they are separators.
ccc6cda3 2886
726f6388
JA
2887 This obeys the field splitting rules in Posix.2. */
2888 sindex = 0;
bc007799
CR
2889 /* Don't need string length in ADVANCE_CHAR unless multibyte chars are
2890 possible, but need it in string_extract_verbatim for bounds checking */
2891 slen = STRLEN (s);
3185942a 2892 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
726f6388
JA
2893
2894 /* Set ENDPTR to the first character after the end of the word. */
2895 if (endptr)
2896 *endptr = s + sindex;
2897
28ef6c31
JA
2898 /* Note whether or not the separator is IFS whitespace, used later. */
2899 whitesep = s[sindex] && spctabnl (s[sindex]);
2900
726f6388
JA
2901 /* Move past the current separator character. */
2902 if (s[sindex])
95732b49
JA
2903 {
2904 DECLARE_MBSTATE;
2905 ADVANCE_CHAR (s, slen, sindex);
2906 }
726f6388
JA
2907
2908 /* Now skip sequences of space, tab, or newline characters if they are
2909 in the list of separators. */
bc007799 2910 while (s[sindex] && spctabnl (s[sindex]) && islocalsep (s[sindex]))
726f6388
JA
2911 sindex++;
2912
28ef6c31
JA
2913 /* If the first separator was IFS whitespace and the current character is
2914 a non-whitespace IFS character, it should be part of the current field
2915 delimiter, not a separate delimiter that would result in an empty field.
2916 Look at POSIX.2, 3.6.5, (3)(b). */
bc007799 2917 if (s[sindex] && whitesep && islocalsep (s[sindex]) && !spctabnl (s[sindex]))
95732b49
JA
2918 {
2919 sindex++;
2920 /* An IFS character that is not IFS white space, along with any adjacent
2921 IFS white space, shall delimit a field. */
bc007799 2922 while (s[sindex] && spctabnl (s[sindex]) && islocalsep(s[sindex]))
95732b49
JA
2923 sindex++;
2924 }
28ef6c31 2925
726f6388
JA
2926 /* Update STRING to point to the next field. */
2927 *stringp = s + sindex;
2928 return (current_word);
2929}
2930
2931/* Remove IFS white space at the end of STRING. Start at the end
2932 of the string and walk backwards until the beginning of the string
2933 or we find a character that's not IFS white space and not CTLESC.
2934 Only let CTLESC escape a white space character if SAW_ESCAPE is
2935 non-zero. */
2936char *
2937strip_trailing_ifs_whitespace (string, separators, saw_escape)
2938 char *string, *separators;
2939 int saw_escape;
2940{
2941 char *s;
ccc6cda3 2942
726f6388 2943 s = string + STRLEN (string) - 1;
7117c2d2 2944 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
726f6388
JA
2945 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2946 s--;
2947 *++s = '\0';
2948 return string;
2949}
2950
bb70624e
JA
2951#if 0
2952/* UNUSED */
2953/* Split STRING into words at whitespace. Obeys shell-style quoting with
2954 backslashes, single and double quotes. */
ccc6cda3
JA
2955WORD_LIST *
2956list_string_with_quotes (string)
2957 char *string;
2958{
2959 WORD_LIST *list;
2960 char *token, *s;
7117c2d2 2961 size_t s_len;
ccc6cda3
JA
2962 int c, i, tokstart, len;
2963
2964 for (s = string; s && *s && spctabnl (*s); s++)
2965 ;
2966 if (s == 0 || *s == 0)
2967 return ((WORD_LIST *)NULL);
2968
7117c2d2 2969 s_len = strlen (s);
ccc6cda3
JA
2970 tokstart = i = 0;
2971 list = (WORD_LIST *)NULL;
2972 while (1)
2973 {
2974 c = s[i];
2975 if (c == '\\')
2976 {
2977 i++;
2978 if (s[i])
2979 i++;
2980 }
2981 else if (c == '\'')
a0c0a00f 2982 i = skip_single_quoted (s, s_len, ++i, 0);
ccc6cda3 2983 else if (c == '"')
a0c0a00f 2984 i = skip_double_quoted (s, s_len, ++i, 0);
ccc6cda3
JA
2985 else if (c == 0 || spctabnl (c))
2986 {
2987 /* We have found the end of a token. Make a word out of it and
2988 add it to the word list. */
bb70624e 2989 token = substring (s, tokstart, i);
cce855bc 2990 list = add_string_to_list (token, list);
ccc6cda3
JA
2991 free (token);
2992 while (spctabnl (s[i]))
2993 i++;
2994 if (s[i])
2995 tokstart = i;
2996 else
2997 break;
2998 }
2999 else
3000 i++; /* normal character */
3001 }
3002 return (REVERSE_LIST (list, WORD_LIST *));
3003}
bb70624e 3004#endif
d166f048 3005
cce855bc
JA
3006/********************************************************/
3007/* */
3008/* Functions to perform assignment statements */
3009/* */
3010/********************************************************/
d166f048 3011
95732b49
JA
3012#if defined (ARRAY_VARS)
3013static SHELL_VAR *
3014do_compound_assignment (name, value, flags)
3015 char *name, *value;
3016 int flags;
3017{
3018 SHELL_VAR *v;
ac50fbac 3019 int mklocal, mkassoc, mkglobal;
0628567a 3020 WORD_LIST *list;
95732b49
JA
3021
3022 mklocal = flags & ASS_MKLOCAL;
3185942a 3023 mkassoc = flags & ASS_MKASSOC;
ac50fbac 3024 mkglobal = flags & ASS_MKGLOBAL;
95732b49
JA
3025
3026 if (mklocal && variable_context)
3027 {
3028 v = find_variable (name);
a0c0a00f
CR
3029 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3030 {
3031 if (v && readonly_p (v))
3032 err_readonly (name);
3033 return (v); /* XXX */
3034 }
3185942a
JA
3035 list = expand_compound_array_assignment (v, value, flags);
3036 if (mkassoc)
3037 v = make_local_assoc_variable (name);
3038 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
ac50fbac
CR
3039 v = make_local_array_variable (name, 0);
3040 if (v)
3041 assign_compound_array_list (v, list, flags);
a0c0a00f
CR
3042 if (list)
3043 dispose_words (list);
ac50fbac
CR
3044 }
3045 /* In a function but forcing assignment in global context */
3046 else if (mkglobal && variable_context)
3047 {
3048 v = find_global_variable (name);
a0c0a00f
CR
3049 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3050 {
3051 if (v && readonly_p (v))
3052 err_readonly (name);
3053 return (v); /* XXX */
3054 }
ac50fbac
CR
3055 list = expand_compound_array_assignment (v, value, flags);
3056 if (v == 0 && mkassoc)
3057 v = make_new_assoc_variable (name);
3058 else if (v && mkassoc && assoc_p (v) == 0)
3059 v = convert_var_to_assoc (v);
3060 else if (v == 0)
3061 v = make_new_array_variable (name);
3062 else if (v && mkassoc == 0 && array_p (v) == 0)
3063 v = convert_var_to_array (v);
3064 if (v)
3065 assign_compound_array_list (v, list, flags);
a0c0a00f
CR
3066 if (list)
3067 dispose_words (list);
95732b49
JA
3068 }
3069 else
a0c0a00f
CR
3070 {
3071 v = assign_array_from_string (name, value, flags);
3072 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3073 {
3074 if (v && readonly_p (v))
3075 err_readonly (name);
3076 return (v); /* XXX */
3077 }
3078 }
95732b49
JA
3079
3080 return (v);
3081}
3082#endif
3083
726f6388
JA
3084/* Given STRING, an assignment string, get the value of the right side
3085 of the `=', and bind it to the left side. If EXPAND is true, then
3086 perform parameter expansion, command substitution, and arithmetic
3087 expansion on the right-hand side. Perform tilde expansion in any
3088 case. Do not perform word splitting on the result of expansion. */
3089static int
95732b49
JA
3090do_assignment_internal (word, expand)
3091 const WORD_DESC *word;
726f6388
JA
3092 int expand;
3093{
495aee44
CR
3094 int offset, appendop, assign_list, aflags, retval;
3095 char *name, *value, *temp;
ccc6cda3
JA
3096 SHELL_VAR *entry;
3097#if defined (ARRAY_VARS)
3098 char *t;
b80f6443 3099 int ni;
ccc6cda3 3100#endif
95732b49 3101 const char *string;
ccc6cda3 3102
95732b49
JA
3103 if (word == 0 || word->word == 0)
3104 return 0;
3105
3106 appendop = assign_list = aflags = 0;
3107 string = word->word;
b80f6443 3108 offset = assignment (string, 0);
ccc6cda3
JA
3109 name = savestring (string);
3110 value = (char *)NULL;
726f6388
JA
3111
3112 if (name[offset] == '=')
3113 {
95732b49
JA
3114 if (name[offset - 1] == '+')
3115 {
3116 appendop = 1;
3117 name[offset - 1] = '\0';
3118 }
3119
3120 name[offset] = 0; /* might need this set later */
726f6388
JA
3121 temp = name + offset + 1;
3122
ccc6cda3 3123#if defined (ARRAY_VARS)
95732b49 3124 if (expand && (word->flags & W_COMPASSIGN))
726f6388 3125 {
ccc6cda3 3126 assign_list = ni = 1;
95732b49 3127 value = extract_array_assignment_list (temp, &ni);
ccc6cda3
JA
3128 }
3129 else
3130#endif
ccc6cda3 3131 if (expand && temp[0])
95732b49 3132 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
726f6388
JA
3133 else
3134 value = savestring (temp);
3135 }
3136
3137 if (value == 0)
d166f048 3138 {
f73dda09 3139 value = (char *)xmalloc (1);
d166f048
JA
3140 value[0] = '\0';
3141 }
726f6388 3142
726f6388 3143 if (echo_command_at_execute)
95732b49
JA
3144 {
3145 if (appendop)
3146 name[offset - 1] = '+';
3147 xtrace_print_assignment (name, value, assign_list, 1);
3148 if (appendop)
3149 name[offset - 1] = '\0';
3150 }
726f6388 3151
d166f048 3152#define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
ccc6cda3 3153
95732b49
JA
3154 if (appendop)
3155 aflags |= ASS_APPEND;
3156
ccc6cda3 3157#if defined (ARRAY_VARS)
0001803f 3158 if (t = mbschr (name, '[')) /*]*/
ccc6cda3
JA
3159 {
3160 if (assign_list)
3161 {
b80f6443 3162 report_error (_("%s: cannot assign list to array member"), name);
ccc6cda3
JA
3163 ASSIGN_RETURN (0);
3164 }
95732b49 3165 entry = assign_array_element (name, value, aflags);
ccc6cda3 3166 if (entry == 0)
28ef6c31 3167 ASSIGN_RETURN (0);
ccc6cda3
JA
3168 }
3169 else if (assign_list)
95732b49 3170 {
6d41b715 3171 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
95732b49 3172 aflags |= ASS_MKLOCAL;
ac50fbac
CR
3173 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
3174 aflags |= ASS_MKGLOBAL;
3185942a
JA
3175 if (word->flags & W_ASSIGNASSOC)
3176 aflags |= ASS_MKASSOC;
95732b49
JA
3177 entry = do_compound_assignment (name, value, aflags);
3178 }
ccc6cda3
JA
3179 else
3180#endif /* ARRAY_VARS */
95732b49 3181 entry = bind_variable (name, value, aflags);
ccc6cda3 3182
726f6388
JA
3183 stupidly_hack_special_variables (name);
3184
3185942a
JA
3185 /* Return 1 if the assignment seems to have been performed correctly. */
3186 if (entry == 0 || readonly_p (entry))
3187 retval = 0; /* assignment failure */
3188 else if (noassign_p (entry))
3189 {
3190 last_command_exit_value = EXECUTION_FAILURE;
3191 retval = 1; /* error status, but not assignment failure */
3192 }
3193 else
3194 retval = 1;
3195
3196 if (entry && retval != 0 && noassign_p (entry) == 0)
3197 VUNSETATTR (entry, att_invisible);
3198
3199 ASSIGN_RETURN (retval);
726f6388
JA
3200}
3201
3202/* Perform the assignment statement in STRING, and expand the
95732b49 3203 right side by doing tilde, command and parameter expansion. */
ccc6cda3 3204int
726f6388 3205do_assignment (string)
95732b49 3206 char *string;
726f6388 3207{
95732b49
JA
3208 WORD_DESC td;
3209
3210 td.flags = W_ASSIGNMENT;
3211 td.word = string;
3212
3213 return do_assignment_internal (&td, 1);
3214}
3215
3216int
495aee44 3217do_word_assignment (word, flags)
95732b49 3218 WORD_DESC *word;
495aee44 3219 int flags;
95732b49
JA
3220{
3221 return do_assignment_internal (word, 1);
726f6388
JA
3222}
3223
3224/* Given STRING, an assignment string, get the value of the right side
95732b49
JA
3225 of the `=', and bind it to the left side. Do not perform any word
3226 expansions on the right hand side. */
ccc6cda3 3227int
726f6388 3228do_assignment_no_expand (string)
95732b49 3229 char *string;
726f6388 3230{
95732b49
JA
3231 WORD_DESC td;
3232
3233 td.flags = W_ASSIGNMENT;
3234 td.word = string;
3235
3236 return (do_assignment_internal (&td, 0));
726f6388
JA
3237}
3238
cce855bc
JA
3239/***************************************************
3240 * *
3241 * Functions to manage the positional parameters *
3242 * *
3243 ***************************************************/
726f6388
JA
3244
3245/* Return the word list that corresponds to `$*'. */
3246WORD_LIST *
3247list_rest_of_args ()
3248{
ccc6cda3 3249 register WORD_LIST *list, *args;
726f6388
JA
3250 int i;
3251
3252 /* Break out of the loop as soon as one of the dollar variables is null. */
ccc6cda3
JA
3253 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
3254 list = make_word_list (make_bare_word (dollar_vars[i]), list);
3255
3256 for (args = rest_of_args; args; args = args->next)
3257 list = make_word_list (make_bare_word (args->word->word), list);
726f6388 3258
726f6388
JA
3259 return (REVERSE_LIST (list, WORD_LIST *));
3260}
3261
ccc6cda3
JA
3262int
3263number_of_args ()
3264{
3265 register WORD_LIST *list;
3266 int n;
3267
3268 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
3269 ;
3270 for (list = rest_of_args; list; list = list->next)
3271 n++;
3272 return n;
3273}
3274
cce855bc
JA
3275/* Return the value of a positional parameter. This handles values > 10. */
3276char *
3277get_dollar_var_value (ind)
7117c2d2 3278 intmax_t ind;
cce855bc
JA
3279{
3280 char *temp;
3281 WORD_LIST *p;
3282
3283 if (ind < 10)
3284 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
3285 else /* We want something like ${11} */
3286 {
3287 ind -= 10;
3288 for (p = rest_of_args; p && ind--; p = p->next)
28ef6c31 3289 ;
cce855bc
JA
3290 temp = p ? savestring (p->word->word) : (char *)NULL;
3291 }
3292 return (temp);
3293}
3294
726f6388
JA
3295/* Make a single large string out of the dollar digit variables,
3296 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
3297 case of "$*" with respect to IFS. */
3298char *
3299string_rest_of_args (dollar_star)
3300 int dollar_star;
3301{
ccc6cda3 3302 register WORD_LIST *list;
726f6388
JA
3303 char *string;
3304
ccc6cda3 3305 list = list_rest_of_args ();
726f6388
JA
3306 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
3307 dispose_words (list);
3308 return (string);
3309}
3310
cce855bc
JA
3311/* Return a string containing the positional parameters from START to
3312 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
7117c2d2
JA
3313 which only makes a difference if QUOTED is non-zero. If QUOTED includes
3314 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3315 no quoting chars are added. */
cce855bc
JA
3316static char *
3317pos_params (string, start, end, quoted)
3318 char *string;
3319 int start, end, quoted;
726f6388 3320{
cce855bc
JA
3321 WORD_LIST *save, *params, *h, *t;
3322 char *ret;
3323 int i;
726f6388 3324
bb70624e
JA
3325 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3326 if (start == end)
3327 return ((char *)NULL);
3328
cce855bc 3329 save = params = list_rest_of_args ();
a0c0a00f 3330 if (save == 0 && start > 0)
cce855bc
JA
3331 return ((char *)NULL);
3332
3185942a
JA
3333 if (start == 0) /* handle ${@:0[:x]} specially */
3334 {
3335 t = make_word_list (make_word (dollar_vars[0]), params);
3336 save = params = t;
3337 }
3338
0001803f 3339 for (i = start ? 1 : 0; params && i < start; i++)
cce855bc
JA
3340 params = params->next;
3341 if (params == 0)
a0c0a00f
CR
3342 {
3343 dispose_words (save);
3344 return ((char *)NULL);
3345 }
cce855bc 3346 for (h = t = params; params && i < end; i++)
d166f048 3347 {
cce855bc
JA
3348 t = params;
3349 params = params->next;
d166f048 3350 }
726f6388 3351
cce855bc 3352 t->next = (WORD_LIST *)NULL;
3185942a
JA
3353
3354 ret = string_list_pos_params (string[0], h, quoted);
3355
bb70624e
JA
3356 if (t != params)
3357 t->next = params;
726f6388 3358
cce855bc
JA
3359 dispose_words (save);
3360 return (ret);
3361}
3362
3363/******************************************************************/
3364/* */
3365/* Functions to expand strings to strings or WORD_LISTs */
3366/* */
3367/******************************************************************/
3368
3369#if defined (PROCESS_SUBSTITUTION)
95732b49 3370#define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
cce855bc 3371#else
95732b49 3372#define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
cce855bc
JA
3373#endif
3374
3375/* If there are any characters in STRING that require full expansion,
3376 then call FUNC to expand STRING; otherwise just perform quote
3377 removal if necessary. This returns a new string. */
3378static char *
f73dda09 3379expand_string_if_necessary (string, quoted, func)
cce855bc
JA
3380 char *string;
3381 int quoted;
f73dda09 3382 EXPFUNC *func;
cce855bc
JA
3383{
3384 WORD_LIST *list;
7117c2d2 3385 size_t slen;
cce855bc
JA
3386 int i, saw_quote;
3387 char *ret;
7117c2d2 3388 DECLARE_MBSTATE;
cce855bc 3389
95732b49
JA
3390 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3391 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
7117c2d2
JA
3392 i = saw_quote = 0;
3393 while (string[i])
cce855bc
JA
3394 {
3395 if (EXP_CHAR (string[i]))
3396 break;
3397 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3398 saw_quote = 1;
7117c2d2 3399 ADVANCE_CHAR (string, slen, i);
cce855bc
JA
3400 }
3401
3402 if (string[i])
3403 {
3404 list = (*func) (string, quoted);
3405 if (list)
3406 {
3407 ret = string_list (list);
3408 dispose_words (list);
3409 }
3410 else
3411 ret = (char *)NULL;
3412 }
3413 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3414 ret = string_quote_removal (string, quoted);
3415 else
3416 ret = savestring (string);
7117c2d2 3417
cce855bc
JA
3418 return ret;
3419}
3420
3421static inline char *
f73dda09 3422expand_string_to_string_internal (string, quoted, func)
cce855bc
JA
3423 char *string;
3424 int quoted;
f73dda09 3425 EXPFUNC *func;
cce855bc
JA
3426{
3427 WORD_LIST *list;
3428 char *ret;
3429
3430 if (string == 0 || *string == '\0')
3431 return ((char *)NULL);
3432
3433 list = (*func) (string, quoted);
3434 if (list)
3435 {
3436 ret = string_list (list);
3437 dispose_words (list);
3438 }
3439 else
3440 ret = (char *)NULL;
3441
3442 return (ret);
3443}
3444
f73dda09
JA
3445char *
3446expand_string_to_string (string, quoted)
3447 char *string;
3448 int quoted;
3449{
3450 return (expand_string_to_string_internal (string, quoted, expand_string));
3451}
3452
3453char *
3454expand_string_unsplit_to_string (string, quoted)
3455 char *string;
3456 int quoted;
3457{
3458 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3459}
3460
95732b49
JA
3461char *
3462expand_assignment_string_to_string (string, quoted)
3463 char *string;
3464 int quoted;
3465{
3466 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3467}
3468
0628567a
JA
3469char *
3470expand_arith_string (string, quoted)
3471 char *string;
3185942a 3472 int quoted;
0628567a 3473{
ac50fbac
CR
3474 WORD_DESC td;
3475 WORD_LIST *list, *tlist;
3476 size_t slen;
3477 int i, saw_quote;
3478 char *ret;
3479 DECLARE_MBSTATE;
3480
3481 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3482 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3483 i = saw_quote = 0;
3484 while (string[i])
3485 {
3486 if (EXP_CHAR (string[i]))
3487 break;
3488 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3489 saw_quote = 1;
3490 ADVANCE_CHAR (string, slen, i);
3491 }
3492
3493 if (string[i])
3494 {
3495 /* This is expanded version of expand_string_internal as it's called by
3496 expand_string_leave_quoted */
3497 td.flags = W_NOPROCSUB; /* don't want process substitution */
3498 td.word = savestring (string);
3499 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3500 /* This takes care of the calls from expand_string_leave_quoted and
3501 expand_string */
3502 if (list)
3503 {
3504 tlist = word_list_split (list);
3505 dispose_words (list);
3506 list = tlist;
3507 if (list)
3508 dequote_list (list);
3509 }
3510 /* This comes from expand_string_if_necessary */
3511 if (list)
3512 {
3513 ret = string_list (list);
3514 dispose_words (list);
3515 }
3516 else
3517 ret = (char *)NULL;
3518 FREE (td.word);
3519 }
a0c0a00f
CR
3520 else if (saw_quote && (quoted & Q_ARITH))
3521 ret = string_quote_removal (string, quoted);
ac50fbac
CR
3522 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3523 ret = string_quote_removal (string, quoted);
3524 else
3525 ret = savestring (string);
3526
3527 return ret;
0628567a
JA
3528}
3529
cce855bc
JA
3530#if defined (COND_COMMAND)
3531/* Just remove backslashes in STRING. Returns a new string. */
3532char *
3533remove_backslashes (string)
3534 char *string;
3535{
3536 char *r, *ret, *s;
3537
f73dda09 3538 r = ret = (char *)xmalloc (strlen (string) + 1);
cce855bc
JA
3539 for (s = string; s && *s; )
3540 {
3541 if (*s == '\\')
28ef6c31 3542 s++;
cce855bc 3543 if (*s == 0)
28ef6c31 3544 break;
cce855bc
JA
3545 *r++ = *s++;
3546 }
3547 *r = '\0';
3548 return ret;
3549}
3550
3551/* This needs better error handling. */
3552/* Expand W for use as an argument to a unary or binary operator in a
f1be666c 3553 [[...]] expression. If SPECIAL is 1, this is the rhs argument
cce855bc 3554 to the != or == operator, and should be treated as a pattern. In
f1be666c
JA
3555 this case, we quote the string specially for the globbing code. If
3556 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3557 be quoted appropriately for regcomp/regexec. The caller is responsible
3558 for removing the backslashes if the unquoted word is needed later. */
cce855bc
JA
3559char *
3560cond_expand_word (w, special)
3561 WORD_DESC *w;
3562 int special;
3563{
3564 char *r, *p;
3565 WORD_LIST *l;
f1be666c 3566 int qflags;
cce855bc
JA
3567
3568 if (w->word == 0 || w->word[0] == '\0')
3569 return ((char *)NULL);
3570
3b34f6e6 3571 expand_no_split_dollar_star = 1;
0001803f 3572 w->flags |= W_NOSPLIT2;
b72432fd 3573 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3b34f6e6 3574 expand_no_split_dollar_star = 0;
cce855bc
JA
3575 if (l)
3576 {
ac50fbac 3577 if (special == 0) /* LHS */
cce855bc
JA
3578 {
3579 dequote_list (l);
3580 r = string_list (l);
3581 }
3582 else
28ef6c31 3583 {
ac50fbac
CR
3584 /* Need to figure out whether or not we should call dequote_escapes
3585 or a new dequote_ctlnul function here, and under what
3586 circumstances. */
f1be666c
JA
3587 qflags = QGLOB_CVTNULL;
3588 if (special == 2)
3589 qflags |= QGLOB_REGEXP;
a0c0a00f 3590 word_list_remove_quoted_nulls (l);
28ef6c31 3591 p = string_list (l);
f1be666c 3592 r = quote_string_for_globbing (p, qflags);
28ef6c31
JA
3593 free (p);
3594 }
cce855bc
JA
3595 dispose_words (l);
3596 }
3597 else
3598 r = (char *)NULL;
3599
3600 return r;
3601}
3602#endif
3603
3604/* Call expand_word_internal to expand W and handle error returns.
3605 A convenience function for functions that don't want to handle
3606 any errors or free any memory before aborting. */
3607static WORD_LIST *
b72432fd 3608call_expand_word_internal (w, q, i, c, e)
cce855bc 3609 WORD_DESC *w;
b72432fd 3610 int q, i, *c, *e;
cce855bc
JA
3611{
3612 WORD_LIST *result;
3613
b72432fd 3614 result = expand_word_internal (w, q, i, c, e);
bb70624e 3615 if (result == &expand_word_error || result == &expand_word_fatal)
cce855bc
JA
3616 {
3617 /* By convention, each time this error is returned, w->word has
bb70624e
JA
3618 already been freed (it sometimes may not be in the fatal case,
3619 but that doesn't result in a memory leak because we're going
3620 to exit in most cases). */
cce855bc 3621 w->word = (char *)NULL;
28ef6c31 3622 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 3623 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
cce855bc 3624 /* NOTREACHED */
ac50fbac 3625 return (NULL);
cce855bc 3626 }
cce855bc
JA
3627 else
3628 return (result);
3629}
3630
3631/* Perform parameter expansion, command substitution, and arithmetic
ac50fbac
CR
3632 expansion on STRING, as if it were a word. Leave the result quoted.
3633 Since this does not perform word splitting, it leaves quoted nulls
3634 in the result. */
cce855bc
JA
3635static WORD_LIST *
3636expand_string_internal (string, quoted)
3637 char *string;
3638 int quoted;
3639{
3640 WORD_DESC td;
3641 WORD_LIST *tresult;
3642
3643 if (string == 0 || *string == 0)
3644 return ((WORD_LIST *)NULL);
3645
28ef6c31
JA
3646 td.flags = 0;
3647 td.word = savestring (string);
3648
b72432fd 3649 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31
JA
3650
3651 FREE (td.word);
cce855bc 3652 return (tresult);
726f6388
JA
3653}
3654
3655/* Expand STRING by performing parameter expansion, command substitution,
3656 and arithmetic expansion. Dequote the resulting WORD_LIST before
3657 returning it, but do not perform word splitting. The call to
3658 remove_quoted_nulls () is in here because word splitting normally
3659 takes care of quote removal. */
3660WORD_LIST *
3661expand_string_unsplit (string, quoted)
3662 char *string;
3663 int quoted;
3664{
3665 WORD_LIST *value;
3666
28ef6c31 3667 if (string == 0 || *string == '\0')
726f6388
JA
3668 return ((WORD_LIST *)NULL);
3669
28ef6c31 3670 expand_no_split_dollar_star = 1;
726f6388 3671 value = expand_string_internal (string, quoted);
28ef6c31
JA
3672 expand_no_split_dollar_star = 0;
3673
726f6388
JA
3674 if (value)
3675 {
3676 if (value->word)
95732b49
JA
3677 {
3678 remove_quoted_nulls (value->word->word);
3679 value->word->flags &= ~W_HASQUOTEDNULL;
3680 }
3681 dequote_list (value);
3682 }
3683 return (value);
3684}
3685
3686/* Expand the rhs of an assignment statement */
3687WORD_LIST *
3688expand_string_assignment (string, quoted)
3689 char *string;
3690 int quoted;
3691{
3692 WORD_DESC td;
3693 WORD_LIST *value;
3694
3695 if (string == 0 || *string == '\0')
3696 return ((WORD_LIST *)NULL);
3697
3698 expand_no_split_dollar_star = 1;
3699
3700 td.flags = W_ASSIGNRHS;
3701 td.word = savestring (string);
3702 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3703 FREE (td.word);
3704
3705 expand_no_split_dollar_star = 0;
3706
3707 if (value)
3708 {
3709 if (value->word)
3710 {
3711 remove_quoted_nulls (value->word->word);
3712 value->word->flags &= ~W_HASQUOTEDNULL;
3713 }
726f6388
JA
3714 dequote_list (value);
3715 }
3716 return (value);
3717}
3718
bb70624e
JA
3719
3720/* Expand one of the PS? prompt strings. This is a sort of combination of
3721 expand_string_unsplit and expand_string_internal, but returns the
3722 passed string when an error occurs. Might want to trap other calls
3723 to jump_to_top_level here so we don't endlessly loop. */
3724WORD_LIST *
f1be666c 3725expand_prompt_string (string, quoted, wflags)
bb70624e
JA
3726 char *string;
3727 int quoted;
f1be666c 3728 int wflags;
bb70624e
JA
3729{
3730 WORD_LIST *value;
3731 WORD_DESC td;
3732
3733 if (string == 0 || *string == 0)
3734 return ((WORD_LIST *)NULL);
3735
f1be666c 3736 td.flags = wflags;
bb70624e 3737 td.word = savestring (string);
28ef6c31
JA
3738
3739 no_longjmp_on_fatal_error = 1;
bb70624e 3740 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31
JA
3741 no_longjmp_on_fatal_error = 0;
3742
bb70624e
JA
3743 if (value == &expand_word_error || value == &expand_word_fatal)
3744 {
3745 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3746 return value;
3747 }
3748 FREE (td.word);
3749 if (value)
3750 {
3751 if (value->word)
95732b49
JA
3752 {
3753 remove_quoted_nulls (value->word->word);
3754 value->word->flags &= ~W_HASQUOTEDNULL;
3755 }
bb70624e
JA
3756 dequote_list (value);
3757 }
3758 return (value);
3759}
3760
726f6388
JA
3761/* Expand STRING just as if you were expanding a word, but do not dequote
3762 the resultant WORD_LIST. This is called only from within this file,
3763 and is used to correctly preserve quoted characters when expanding
3764 things like ${1+"$@"}. This does parameter expansion, command
b72432fd 3765 substitution, arithmetic expansion, and word splitting. */
726f6388
JA
3766static WORD_LIST *
3767expand_string_leave_quoted (string, quoted)
3768 char *string;
3769 int quoted;
3770{
3771 WORD_LIST *tlist;
3772 WORD_LIST *tresult;
3773
ccc6cda3 3774 if (string == 0 || *string == '\0')
726f6388
JA
3775 return ((WORD_LIST *)NULL);
3776
3777 tlist = expand_string_internal (string, quoted);
3778
3779 if (tlist)
3780 {
3781 tresult = word_list_split (tlist);
3782 dispose_words (tlist);
3783 return (tresult);
3784 }
3785 return ((WORD_LIST *)NULL);
3786}
3787
ccc6cda3
JA
3788/* This does not perform word splitting or dequote the WORD_LIST
3789 it returns. */
3790static WORD_LIST *
a0c0a00f 3791expand_string_for_rhs (string, quoted, dollar_at_p, expanded_p)
ccc6cda3 3792 char *string;
a0c0a00f 3793 int quoted, *dollar_at_p, *expanded_p;
ccc6cda3
JA
3794{
3795 WORD_DESC td;
3796 WORD_LIST *tresult;
3797
3798 if (string == 0 || *string == '\0')
3799 return (WORD_LIST *)NULL;
3800
a0c0a00f 3801 expand_no_split_dollar_star = 1;
aeb26a67 3802 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
ccc6cda3 3803 td.word = string;
a0c0a00f
CR
3804 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, expanded_p);
3805 expand_no_split_dollar_star = 0;
3806
ccc6cda3
JA
3807 return (tresult);
3808}
3809
726f6388
JA
3810/* Expand STRING just as if you were expanding a word. This also returns
3811 a list of words. Note that filename globbing is *NOT* done for word
3812 or string expansion, just when the shell is expanding a command. This
3813 does parameter expansion, command substitution, arithmetic expansion,
3814 and word splitting. Dequote the resultant WORD_LIST before returning. */
3815WORD_LIST *
3816expand_string (string, quoted)
3817 char *string;
3818 int quoted;
3819{
3820 WORD_LIST *result;
3821
28ef6c31 3822 if (string == 0 || *string == '\0')
726f6388
JA
3823 return ((WORD_LIST *)NULL);
3824
3825 result = expand_string_leave_quoted (string, quoted);
ccc6cda3 3826 return (result ? dequote_list (result) : result);
726f6388
JA
3827}
3828
3829/***************************************************
3830 * *
3831 * Functions to handle quoting chars *
3832 * *
3833 ***************************************************/
3834
cce855bc
JA
3835/* Conventions:
3836
3837 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3838 The parser passes CTLNUL as CTLESC CTLNUL. */
3839
cce855bc
JA
3840/* Quote escape characters in string s, but no other characters. This is
3841 used to protect CTLESC and CTLNUL in variable values from the rest of
3185942a
JA
3842 the word expansion process after the variable is expanded (word splitting
3843 and filename generation). If IFS is null, we quote spaces as well, just
3844 in case we split on spaces later (in the case of unquoted $@, we will
3845 eventually attempt to split the entire word on spaces). Corresponding
3846 code exists in dequote_escapes. Even if we don't end up splitting on
3847 spaces, quoting spaces is not a problem. This should never be called on
3848 a string that is quoted with single or double quotes or part of a here
3849 document (effectively double-quoted). */
f73dda09 3850char *
cce855bc
JA
3851quote_escapes (string)
3852 char *string;
3853{
3854 register char *s, *t;
7117c2d2
JA
3855 size_t slen;
3856 char *result, *send;
3185942a 3857 int quote_spaces, skip_ctlesc, skip_ctlnul;
7117c2d2 3858 DECLARE_MBSTATE;
cce855bc 3859
7117c2d2
JA
3860 slen = strlen (string);
3861 send = string + slen;
3862
f1be666c 3863 quote_spaces = (ifs_value && *ifs_value == 0);
3185942a
JA
3864
3865 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3866 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3867
7117c2d2
JA
3868 t = result = (char *)xmalloc ((slen * 2) + 1);
3869 s = string;
3870
3871 while (*s)
cce855bc 3872 {
3185942a 3873 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
cce855bc 3874 *t++ = CTLESC;
7117c2d2 3875 COPY_CHAR_P (t, s, send);
cce855bc
JA
3876 }
3877 *t = '\0';
ac50fbac 3878
cce855bc
JA
3879 return (result);
3880}
3881
3882static WORD_LIST *
3883list_quote_escapes (list)
3884 WORD_LIST *list;
3885{
3886 register WORD_LIST *w;
3887 char *t;
3888
3889 for (w = list; w; w = w->next)
3890 {
3891 t = w->word->word;
3892 w->word->word = quote_escapes (t);
3893 free (t);
3894 }
3895 return list;
3896}
3897
7117c2d2
JA
3898/* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3899
3900 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3901 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3902 data stream pass through properly.
3903
3904 We need to remove doubled CTLESC characters inside quoted strings before
3905 quoting the entire string, so we do not double the number of CTLESC
3906 characters.
3907
3908 Also used by parts of the pattern substitution code. */
3185942a 3909char *
cce855bc
JA
3910dequote_escapes (string)
3911 char *string;
3912{
3185942a 3913 register char *s, *t, *s1;
7117c2d2
JA
3914 size_t slen;
3915 char *result, *send;
f1be666c 3916 int quote_spaces;
7117c2d2 3917 DECLARE_MBSTATE;
cce855bc 3918
7117c2d2
JA
3919 if (string == 0)
3920 return string;
3921
3922 slen = strlen (string);
3923 send = string + slen;
3924
3925 t = result = (char *)xmalloc (slen + 1);
7117c2d2
JA
3926
3927 if (strchr (string, CTLESC) == 0)
3185942a 3928 return (strcpy (result, string));
7117c2d2 3929
f1be666c 3930 quote_spaces = (ifs_value && *ifs_value == 0);
3185942a
JA
3931
3932 s = string;
7117c2d2 3933 while (*s)
cce855bc 3934 {
f1be666c 3935 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
cce855bc
JA
3936 {
3937 s++;
3938 if (*s == '\0')
3939 break;
3940 }
7117c2d2 3941 COPY_CHAR_P (t, s, send);
cce855bc
JA
3942 }
3943 *t = '\0';
ac50fbac 3944
cce855bc
JA
3945 return result;
3946}
726f6388 3947
ac50fbac
CR
3948static WORD_LIST *
3949list_dequote_escapes (list)
3950 WORD_LIST *list;
3951{
3952 register WORD_LIST *w;
3953 char *t;
3954
3955 for (w = list; w; w = w->next)
3956 {
3957 t = w->word->word;
3958 w->word->word = dequote_escapes (t);
3959 free (t);
3960 }
3961 return list;
3962}
3963
0628567a
JA
3964/* Return a new string with the quoted representation of character C.
3965 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3966 set in any resultant WORD_DESC where this value is the word. */
726f6388
JA
3967static char *
3968make_quoted_char (c)
3969 int c;
3970{
3971 char *temp;
3972
f73dda09 3973 temp = (char *)xmalloc (3);
726f6388
JA
3974 if (c == 0)
3975 {
3976 temp[0] = CTLNUL;
3977 temp[1] = '\0';
3978 }
3979 else
3980 {
3981 temp[0] = CTLESC;
3982 temp[1] = c;
3983 temp[2] = '\0';
3984 }
3985 return (temp);
3986}
3987
0628567a
JA
3988/* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3989 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3990 this value is the word. */
ccc6cda3 3991char *
726f6388
JA
3992quote_string (string)
3993 char *string;
3994{
ccc6cda3 3995 register char *t;
7117c2d2
JA
3996 size_t slen;
3997 char *result, *send;
726f6388 3998
ccc6cda3 3999 if (*string == 0)
726f6388 4000 {
f73dda09 4001 result = (char *)xmalloc (2);
726f6388
JA
4002 result[0] = CTLNUL;
4003 result[1] = '\0';
4004 }
4005 else
4006 {
7117c2d2 4007 DECLARE_MBSTATE;
726f6388 4008
7117c2d2
JA
4009 slen = strlen (string);
4010 send = string + slen;
4011
4012 result = (char *)xmalloc ((slen * 2) + 1);
4013
4014 for (t = result; string < send; )
726f6388
JA
4015 {
4016 *t++ = CTLESC;
7117c2d2 4017 COPY_CHAR_P (t, string, send);
726f6388
JA
4018 }
4019 *t = '\0';
4020 }
4021 return (result);
4022}
4023
0628567a 4024/* De-quote quoted characters in STRING. */
726f6388
JA
4025char *
4026dequote_string (string)
4027 char *string;
4028{
7117c2d2
JA
4029 register char *s, *t;
4030 size_t slen;
4031 char *result, *send;
4032 DECLARE_MBSTATE;
726f6388 4033
a0c0a00f
CR
4034#if defined (DEBUG)
4035 if (string[0] == CTLESC && string[1] == 0)
4036 internal_inform ("dequote_string: string with bare CTLESC");
4037#endif
4038
7117c2d2
JA
4039 slen = strlen (string);
4040
4041 t = result = (char *)xmalloc (slen + 1);
726f6388
JA
4042
4043 if (QUOTED_NULL (string))
4044 {
4045 result[0] = '\0';
4046 return (result);
4047 }
4048
a0c0a00f
CR
4049 /* A string consisting of only a single CTLESC should pass through unchanged */
4050 if (string[0] == CTLESC && string[1] == 0)
4051 {
4052 result[0] = CTLESC;
4053 result[1] = '\0';
4054 return (result);
4055 }
4056
726f6388
JA
4057 /* If no character in the string can be quoted, don't bother examining
4058 each character. Just return a copy of the string passed to us. */
7117c2d2
JA
4059 if (strchr (string, CTLESC) == NULL)
4060 return (strcpy (result, string));
726f6388 4061
7117c2d2
JA
4062 send = string + slen;
4063 s = string;
4064 while (*s)
726f6388 4065 {
7117c2d2 4066 if (*s == CTLESC)
726f6388 4067 {
7117c2d2
JA
4068 s++;
4069 if (*s == '\0')
726f6388
JA
4070 break;
4071 }
7117c2d2 4072 COPY_CHAR_P (t, s, send);
726f6388
JA
4073 }
4074
4075 *t = '\0';
4076 return (result);
4077}
4078
4079/* Quote the entire WORD_LIST list. */
ccc6cda3 4080static WORD_LIST *
726f6388
JA
4081quote_list (list)
4082 WORD_LIST *list;
4083{
4084 register WORD_LIST *w;
ccc6cda3 4085 char *t;
726f6388
JA
4086
4087 for (w = list; w; w = w->next)
4088 {
ccc6cda3 4089 t = w->word->word;
726f6388 4090 w->word->word = quote_string (t);
3185942a
JA
4091 if (*t == 0)
4092 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
ccc6cda3 4093 w->word->flags |= W_QUOTED;
3185942a 4094 free (t);
726f6388 4095 }
ccc6cda3 4096 return list;
726f6388
JA
4097}
4098
0628567a
JA
4099/* De-quote quoted characters in each word in LIST. */
4100WORD_LIST *
7117c2d2
JA
4101dequote_list (list)
4102 WORD_LIST *list;
4103{
4104 register char *s;
4105 register WORD_LIST *tlist;
4106
4107 for (tlist = list; tlist; tlist = tlist->next)
4108 {
4109 s = dequote_string (tlist->word->word);
3185942a
JA
4110 if (QUOTED_NULL (tlist->word->word))
4111 tlist->word->flags &= ~W_HASQUOTEDNULL;
7117c2d2
JA
4112 free (tlist->word->word);
4113 tlist->word->word = s;
4114 }
4115 return list;
4116}
4117
4118/* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
4119 string. */
3185942a 4120char *
7117c2d2
JA
4121remove_quoted_escapes (string)
4122 char *string;
4123{
4124 char *t;
4125
4126 if (string)
4127 {
4128 t = dequote_escapes (string);
4129 strcpy (string, t);
4130 free (t);
4131 }
4132
4133 return (string);
4134}
4135
cce855bc
JA
4136/* Perform quoted null character removal on STRING. We don't allow any
4137 quoted null characters in the middle or at the ends of strings because
4138 of how expand_word_internal works. remove_quoted_nulls () turns
4139 STRING into an empty string iff it only consists of a quoted null,
4140 and removes all unquoted CTLNUL characters. */
3185942a 4141char *
cce855bc
JA
4142remove_quoted_nulls (string)
4143 char *string;
4144{
7117c2d2
JA
4145 register size_t slen;
4146 register int i, j, prev_i;
4147 DECLARE_MBSTATE;
4148
4149 if (strchr (string, CTLNUL) == 0) /* XXX */
4150 return string; /* XXX */
4151
4152 slen = strlen (string);
4153 i = j = 0;
4154
4155 while (i < slen)
4156 {
4157 if (string[i] == CTLESC)
b80f6443
JA
4158 {
4159 /* Old code had j++, but we cannot assume that i == j at this
4160 point -- what if a CTLNUL has already been removed from the
4161 string? We don't want to drop the CTLESC or recopy characters
4162 that we've already copied down. */
4163 i++; string[j++] = CTLESC;
4164 if (i == slen)
4165 break;
4166 }
7117c2d2 4167 else if (string[i] == CTLNUL)
a48a8ac3
CR
4168 {
4169 i++;
4170 continue;
4171 }
7117c2d2
JA
4172
4173 prev_i = i;
4174 ADVANCE_CHAR (string, slen, i);
4175 if (j < prev_i)
cce855bc 4176 {
7117c2d2 4177 do string[j++] = string[prev_i++]; while (prev_i < i);
cce855bc 4178 }
7117c2d2
JA
4179 else
4180 j = i;
cce855bc 4181 }
7117c2d2
JA
4182 string[j] = '\0';
4183
4184 return (string);
cce855bc
JA
4185}
4186
4187/* Perform quoted null character removal on each element of LIST.
4188 This modifies LIST. */
4189void
4190word_list_remove_quoted_nulls (list)
4191 WORD_LIST *list;
4192{
4193 register WORD_LIST *t;
4194
4195 for (t = list; t; t = t->next)
95732b49
JA
4196 {
4197 remove_quoted_nulls (t->word->word);
4198 t->word->flags &= ~W_HASQUOTEDNULL;
4199 }
cce855bc
JA
4200}
4201
4202/* **************************************************************** */
4203/* */
4204/* Functions for Matching and Removing Patterns */
4205/* */
4206/* **************************************************************** */
4207
b80f6443
JA
4208#if defined (HANDLE_MULTIBYTE)
4209#if 0 /* Currently unused */
4210static unsigned char *
4211mb_getcharlens (string, len)
4212 char *string;
4213 int len;
4214{
4215 int i, offset, last;
4216 unsigned char *ret;
4217 char *p;
4218 DECLARE_MBSTATE;
4219
4220 i = offset = 0;
4221 last = 0;
4222 ret = (unsigned char *)xmalloc (len);
4223 memset (ret, 0, len);
4224 while (string[last])
4225 {
4226 ADVANCE_CHAR (string, len, offset);
4227 ret[last] = offset - last;
4228 last = offset;
4229 }
4230 return ret;
4231}
4232#endif
4233#endif
4234
cce855bc
JA
4235/* Remove the portion of PARAM matched by PATTERN according to OP, where OP
4236 can have one of 4 values:
4237 RP_LONG_LEFT remove longest matching portion at start of PARAM
726f6388
JA
4238 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
4239 RP_LONG_RIGHT remove longest matching portion at end of PARAM
4240 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
4241*/
4242
4243#define RP_LONG_LEFT 1
4244#define RP_SHORT_LEFT 2
4245#define RP_LONG_RIGHT 3
4246#define RP_SHORT_RIGHT 4
4247
495aee44 4248/* Returns its first argument if nothing matched; new memory otherwise */
726f6388 4249static char *
b80f6443 4250remove_upattern (param, pattern, op)
726f6388
JA
4251 char *param, *pattern;
4252 int op;
4253{
a0c0a00f 4254 register size_t len;
ccc6cda3 4255 register char *end;
726f6388
JA
4256 register char *p, *ret, c;
4257
ccc6cda3
JA
4258 len = STRLEN (param);
4259 end = param + len;
726f6388
JA
4260
4261 switch (op)
4262 {
4263 case RP_LONG_LEFT: /* remove longest match at start */
4264 for (p = end; p >= param; p--)
4265 {
4266 c = *p; *p = '\0';
f73dda09 4267 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
726f6388
JA
4268 {
4269 *p = c;
4270 return (savestring (p));
4271 }
4272 *p = c;
b80f6443 4273
726f6388
JA
4274 }
4275 break;
4276
4277 case RP_SHORT_LEFT: /* remove shortest match at start */
4278 for (p = param; p <= end; p++)
4279 {
4280 c = *p; *p = '\0';
f73dda09 4281 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
726f6388
JA
4282 {
4283 *p = c;
4284 return (savestring (p));
4285 }
4286 *p = c;
4287 }
4288 break;
4289
ccc6cda3
JA
4290 case RP_LONG_RIGHT: /* remove longest match at end */
4291 for (p = param; p <= end; p++)
4292 {
f73dda09 4293 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
ccc6cda3
JA
4294 {
4295 c = *p; *p = '\0';
4296 ret = savestring (param);
4297 *p = c;
4298 return (ret);
4299 }
4300 }
4301 break;
4302
4303 case RP_SHORT_RIGHT: /* remove shortest match at end */
4304 for (p = end; p >= param; p--)
4305 {
f73dda09 4306 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
ccc6cda3
JA
4307 {
4308 c = *p; *p = '\0';
4309 ret = savestring (param);
4310 *p = c;
4311 return (ret);
4312 }
4313 }
4314 break;
4315 }
b80f6443 4316
495aee44 4317 return (param); /* no match, return original string */
ccc6cda3
JA
4318}
4319
b80f6443 4320#if defined (HANDLE_MULTIBYTE)
495aee44 4321/* Returns its first argument if nothing matched; new memory otherwise */
b80f6443
JA
4322static wchar_t *
4323remove_wpattern (wparam, wstrlen, wpattern, op)
4324 wchar_t *wparam;
4325 size_t wstrlen;
4326 wchar_t *wpattern;
4327 int op;
4328{
0628567a
JA
4329 wchar_t wc, *ret;
4330 int n;
b80f6443
JA
4331
4332 switch (op)
4333 {
4334 case RP_LONG_LEFT: /* remove longest match at start */
4335 for (n = wstrlen; n >= 0; n--)
4336 {
4337 wc = wparam[n]; wparam[n] = L'\0';
4338 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4339 {
4340 wparam[n] = wc;
4341 return (wcsdup (wparam + n));
4342 }
4343 wparam[n] = wc;
4344 }
4345 break;
4346
4347 case RP_SHORT_LEFT: /* remove shortest match at start */
4348 for (n = 0; n <= wstrlen; n++)
4349 {
4350 wc = wparam[n]; wparam[n] = L'\0';
4351 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4352 {
4353 wparam[n] = wc;
4354 return (wcsdup (wparam + n));
4355 }
4356 wparam[n] = wc;
4357 }
4358 break;
4359
4360 case RP_LONG_RIGHT: /* remove longest match at end */
4361 for (n = 0; n <= wstrlen; n++)
4362 {
4363 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4364 {
4365 wc = wparam[n]; wparam[n] = L'\0';
4366 ret = wcsdup (wparam);
4367 wparam[n] = wc;
4368 return (ret);
4369 }
4370 }
4371 break;
4372
4373 case RP_SHORT_RIGHT: /* remove shortest match at end */
4374 for (n = wstrlen; n >= 0; n--)
4375 {
4376 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4377 {
4378 wc = wparam[n]; wparam[n] = L'\0';
4379 ret = wcsdup (wparam);
4380 wparam[n] = wc;
4381 return (ret);
4382 }
4383 }
4384 break;
4385 }
4386
495aee44 4387 return (wparam); /* no match, return original string */
b80f6443
JA
4388}
4389#endif /* HANDLE_MULTIBYTE */
4390
4391static char *
4392remove_pattern (param, pattern, op)
4393 char *param, *pattern;
4394 int op;
4395{
495aee44
CR
4396 char *xret;
4397
b80f6443
JA
4398 if (param == NULL)
4399 return (param);
4400 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4401 return (savestring (param));
4402
4403#if defined (HANDLE_MULTIBYTE)
4404 if (MB_CUR_MAX > 1)
4405 {
4406 wchar_t *ret, *oret;
4407 size_t n;
4408 wchar_t *wparam, *wpattern;
4409 mbstate_t ps;
b80f6443
JA
4410
4411 n = xdupmbstowcs (&wpattern, NULL, pattern);
4412 if (n == (size_t)-1)
495aee44
CR
4413 {
4414 xret = remove_upattern (param, pattern, op);
4415 return ((xret == param) ? savestring (param) : xret);
4416 }
b80f6443 4417 n = xdupmbstowcs (&wparam, NULL, param);
ac50fbac 4418
b80f6443
JA
4419 if (n == (size_t)-1)
4420 {
4421 free (wpattern);
495aee44
CR
4422 xret = remove_upattern (param, pattern, op);
4423 return ((xret == param) ? savestring (param) : xret);
b80f6443
JA
4424 }
4425 oret = ret = remove_wpattern (wparam, n, wpattern, op);
495aee44
CR
4426 /* Don't bother to convert wparam back to multibyte string if nothing
4427 matched; just return copy of original string */
4428 if (ret == wparam)
4429 {
4430 free (wparam);
4431 free (wpattern);
4432 return (savestring (param));
4433 }
b80f6443
JA
4434
4435 free (wparam);
4436 free (wpattern);
4437
4438 n = strlen (param);
0628567a 4439 xret = (char *)xmalloc (n + 1);
b80f6443
JA
4440 memset (&ps, '\0', sizeof (mbstate_t));
4441 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4442 xret[n] = '\0'; /* just to make sure */
4443 free (oret);
4444 return xret;
4445 }
4446 else
4447#endif
ccc6cda3 4448 {
495aee44
CR
4449 xret = remove_upattern (param, pattern, op);
4450 return ((xret == param) ? savestring (param) : xret);
ccc6cda3
JA
4451 }
4452}
4453
4454/* Match PAT anywhere in STRING and return the match boundaries.
4455 This returns 1 in case of a successful match, 0 otherwise. SP
4456 and EP are pointers into the string where the match begins and
4457 ends, respectively. MTYPE controls what kind of match is attempted.
4458 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4459 of the string, respectively. The longest match is returned. */
4460static int
b80f6443 4461match_upattern (string, pat, mtype, sp, ep)
ccc6cda3
JA
4462 char *string, *pat;
4463 int mtype;
4464 char **sp, **ep;
4465{
a0c0a00f
CR
4466 int c, mlen;
4467 size_t len;
95732b49 4468 register char *p, *p1, *npat;
ccc6cda3 4469 char *end;
495aee44 4470 int n1;
ccc6cda3 4471
95732b49
JA
4472 /* If the pattern doesn't match anywhere in the string, go ahead and
4473 short-circuit right away. A minor optimization, saves a bunch of
4474 unnecessary calls to strmatch (up to N calls for a string of N
4475 characters) if the match is unsuccessful. To preserve the semantics
4476 of the substring matches below, we make sure that the pattern has
4477 `*' as first and last character, making a new pattern if necessary. */
4478 /* XXX - check this later if I ever implement `**' with special meaning,
4479 since this will potentially result in `**' at the beginning or end */
4480 len = STRLEN (pat);
0001803f 4481 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
95732b49 4482 {
a0c0a00f
CR
4483 int unescaped_backslash;
4484 char *pp;
4485
0628567a 4486 p = npat = (char *)xmalloc (len + 3);
95732b49 4487 p1 = pat;
0001803f 4488 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
95732b49
JA
4489 *p++ = '*';
4490 while (*p1)
4491 *p++ = *p1++;
a0c0a00f
CR
4492#if 1
4493 /* Need to also handle a pattern that ends with an unescaped backslash.
4494 For right now, we ignore it because the pattern matching code will
4495 fail the match anyway */
4496 /* If the pattern ends with a `*' we leave it alone if it's preceded by
4497 an even number of backslashes, but if it's escaped by a backslash
4498 we need to add another `*'. */
4499 if (p1[-1] == '*' && (unescaped_backslash = p1[-2] == '\\'))
4500 {
4501 pp = p1 - 3;
4502 while (pp >= pat && *pp-- == '\\')
4503 unescaped_backslash = 1 - unescaped_backslash;
4504 if (unescaped_backslash)
4505 *p++ = '*';
4506 }
4507 else if (p1[-1] != '*')
95732b49 4508 *p++ = '*';
a0c0a00f
CR
4509#else
4510 if (p1[-1] != '*' || p1[-2] == '\\')
4511 *p++ = '*';
4512#endif
95732b49
JA
4513 *p = '\0';
4514 }
4515 else
4516 npat = pat;
a0c0a00f 4517 c = strmatch (npat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
95732b49
JA
4518 if (npat != pat)
4519 free (npat);
4520 if (c == FNM_NOMATCH)
4521 return (0);
4522
b80f6443
JA
4523 len = STRLEN (string);
4524 end = string + len;
ccc6cda3 4525
495aee44
CR
4526 mlen = umatchlen (pat, len);
4527
ccc6cda3
JA
4528 switch (mtype)
4529 {
4530 case MATCH_ANY:
4531 for (p = string; p <= end; p++)
4532 {
a0c0a00f 4533 if (match_pattern_char (pat, p, FNMATCH_IGNCASE))
ccc6cda3 4534 {
495aee44
CR
4535 p1 = (mlen == -1) ? end : p + mlen;
4536 /* p1 - p = length of portion of string to be considered
4537 p = current position in string
4538 mlen = number of characters consumed by match (-1 for entire string)
4539 end = end of string
4540 we want to break immediately if the potential match len
4541 is greater than the number of characters remaining in the
4542 string
4543 */
4544 if (p1 > end)
4545 break;
4546 for ( ; p1 >= p; p1--)
ccc6cda3
JA
4547 {
4548 c = *p1; *p1 = '\0';
a0c0a00f 4549 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
ccc6cda3
JA
4550 {
4551 *p1 = c;
4552 *sp = p;
4553 *ep = p1;
4554 return 1;
4555 }
4556 *p1 = c;
495aee44
CR
4557#if 1
4558 /* If MLEN != -1, we have a fixed length pattern. */
4559 if (mlen != -1)
4560 break;
4561#endif
ccc6cda3
JA
4562 }
4563 }
4564 }
b80f6443 4565
ccc6cda3
JA
4566 return (0);
4567
4568 case MATCH_BEG:
a0c0a00f 4569 if (match_pattern_char (pat, string, FNMATCH_IGNCASE) == 0)
28ef6c31 4570 return (0);
b80f6443 4571
495aee44 4572 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
ccc6cda3
JA
4573 {
4574 c = *p; *p = '\0';
a0c0a00f 4575 if (strmatch (pat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
ccc6cda3
JA
4576 {
4577 *p = c;
4578 *sp = string;
4579 *ep = p;
4580 return 1;
4581 }
4582 *p = c;
495aee44
CR
4583 /* If MLEN != -1, we have a fixed length pattern. */
4584 if (mlen != -1)
4585 break;
ccc6cda3 4586 }
b80f6443 4587
ccc6cda3 4588 return (0);
726f6388 4589
ccc6cda3 4590 case MATCH_END:
495aee44 4591 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
b80f6443 4592 {
a0c0a00f 4593 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4594 {
4595 *sp = p;
4596 *ep = end;
4597 return 1;
4598 }
495aee44
CR
4599 /* If MLEN != -1, we have a fixed length pattern. */
4600 if (mlen != -1)
4601 break;
b80f6443
JA
4602 }
4603
4604 return (0);
4605 }
4606
4607 return (0);
4608}
4609
4610#if defined (HANDLE_MULTIBYTE)
a0c0a00f
CR
4611
4612#define WFOLD(c) (match_ignore_case && iswupper (c) ? towlower (c) : (c))
4613
b80f6443
JA
4614/* Match WPAT anywhere in WSTRING and return the match boundaries.
4615 This returns 1 in case of a successful match, 0 otherwise. Wide
4616 character version. */
4617static int
4618match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4619 wchar_t *wstring;
4620 char **indices;
4621 size_t wstrlen;
4622 wchar_t *wpat;
4623 int mtype;
4624 char **sp, **ep;
4625{
95732b49 4626 wchar_t wc, *wp, *nwpat, *wp1;
495aee44
CR
4627 size_t len;
4628 int mlen;
4629 int n, n1, n2, simple;
4630
4631 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4632#if defined (EXTENDED_GLOB)
4633 if (extended_glob)
91717ba3 4634 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
b80f6443
JA
4635#endif
4636
95732b49
JA
4637 /* If the pattern doesn't match anywhere in the string, go ahead and
4638 short-circuit right away. A minor optimization, saves a bunch of
4639 unnecessary calls to strmatch (up to N calls for a string of N
4640 characters) if the match is unsuccessful. To preserve the semantics
4641 of the substring matches below, we make sure that the pattern has
4642 `*' as first and last character, making a new pattern if necessary. */
95732b49 4643 len = wcslen (wpat);
0001803f 4644 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
95732b49 4645 {
a0c0a00f
CR
4646 int unescaped_backslash;
4647 wchar_t *wpp;
4648
0628567a 4649 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
95732b49 4650 wp1 = wpat;
0001803f 4651 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
95732b49
JA
4652 *wp++ = L'*';
4653 while (*wp1 != L'\0')
4654 *wp++ = *wp1++;
a0c0a00f
CR
4655#if 1
4656 /* See comments above in match_upattern. */
4657 if (wp1[-1] == L'*' && (unescaped_backslash = wp1[-2] == L'\\'))
4658 {
4659 wpp = wp1 - 3;
4660 while (wpp >= wpat && *wpp-- == L'\\')
4661 unescaped_backslash = 1 - unescaped_backslash;
4662 if (unescaped_backslash)
4663 *wp++ = L'*';
4664 }
4665 else if (wp1[-1] != L'*')
4666 *wp++ = L'*';
4667#else
95732b49
JA
4668 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4669 *wp++ = L'*';
a0c0a00f 4670#endif
95732b49
JA
4671 *wp = '\0';
4672 }
4673 else
4674 nwpat = wpat;
a0c0a00f 4675 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
95732b49
JA
4676 if (nwpat != wpat)
4677 free (nwpat);
4678 if (len == FNM_NOMATCH)
4679 return (0);
4680
495aee44
CR
4681 mlen = wmatchlen (wpat, wstrlen);
4682
4683/* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
b80f6443
JA
4684 switch (mtype)
4685 {
4686 case MATCH_ANY:
4687 for (n = 0; n <= wstrlen; n++)
4688 {
a0c0a00f 4689 n2 = simple ? (WFOLD(*wpat) == WFOLD(wstring[n])) : match_pattern_wchar (wpat, wstring + n, FNMATCH_IGNCASE);
495aee44 4690 if (n2)
b80f6443 4691 {
495aee44
CR
4692 n1 = (mlen == -1) ? wstrlen : n + mlen;
4693 if (n1 > wstrlen)
4694 break;
4695
4696 for ( ; n1 >= n; n1--)
b80f6443
JA
4697 {
4698 wc = wstring[n1]; wstring[n1] = L'\0';
a0c0a00f 4699 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4700 {
4701 wstring[n1] = wc;
4702 *sp = indices[n];
4703 *ep = indices[n1];
4704 return 1;
4705 }
4706 wstring[n1] = wc;
495aee44
CR
4707 /* If MLEN != -1, we have a fixed length pattern. */
4708 if (mlen != -1)
4709 break;
b80f6443
JA
4710 }
4711 }
4712 }
4713
4714 return (0);
4715
4716 case MATCH_BEG:
a0c0a00f 4717 if (match_pattern_wchar (wpat, wstring, FNMATCH_IGNCASE) == 0)
b80f6443
JA
4718 return (0);
4719
495aee44 4720 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
b80f6443
JA
4721 {
4722 wc = wstring[n]; wstring[n] = L'\0';
a0c0a00f 4723 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4724 {
4725 wstring[n] = wc;
4726 *sp = indices[0];
4727 *ep = indices[n];
4728 return 1;
4729 }
4730 wstring[n] = wc;
495aee44
CR
4731 /* If MLEN != -1, we have a fixed length pattern. */
4732 if (mlen != -1)
4733 break;
b80f6443
JA
4734 }
4735
4736 return (0);
4737
4738 case MATCH_END:
495aee44 4739 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
b80f6443 4740 {
a0c0a00f 4741 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4742 {
4743 *sp = indices[n];
4744 *ep = indices[wstrlen];
4745 return 1;
4746 }
495aee44
CR
4747 /* If MLEN != -1, we have a fixed length pattern. */
4748 if (mlen != -1)
4749 break;
b80f6443
JA
4750 }
4751
ccc6cda3 4752 return (0);
726f6388 4753 }
ccc6cda3
JA
4754
4755 return (0);
726f6388 4756}
a0c0a00f 4757#undef WFOLD
b80f6443
JA
4758#endif /* HANDLE_MULTIBYTE */
4759
4760static int
4761match_pattern (string, pat, mtype, sp, ep)
4762 char *string, *pat;
4763 int mtype;
4764 char **sp, **ep;
4765{
4766#if defined (HANDLE_MULTIBYTE)
4767 int ret;
4768 size_t n;
4769 wchar_t *wstring, *wpat;
4770 char **indices;
495aee44 4771 size_t slen, plen, mslen, mplen;
b80f6443
JA
4772#endif
4773
a0c0a00f 4774 if (string == 0 || pat == 0 || *pat == 0)
b80f6443
JA
4775 return (0);
4776
4777#if defined (HANDLE_MULTIBYTE)
4778 if (MB_CUR_MAX > 1)
4779 {
495aee44 4780 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
495aee44
CR
4781 return (match_upattern (string, pat, mtype, sp, ep));
4782
b80f6443
JA
4783 n = xdupmbstowcs (&wpat, NULL, pat);
4784 if (n == (size_t)-1)
4785 return (match_upattern (string, pat, mtype, sp, ep));
4786 n = xdupmbstowcs (&wstring, &indices, string);
4787 if (n == (size_t)-1)
4788 {
4789 free (wpat);
4790 return (match_upattern (string, pat, mtype, sp, ep));
4791 }
4792 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4793
4794 free (wpat);
4795 free (wstring);
4796 free (indices);
4797
4798 return (ret);
4799 }
4800 else
4801#endif
4802 return (match_upattern (string, pat, mtype, sp, ep));
4803}
726f6388 4804
cce855bc
JA
4805static int
4806getpatspec (c, value)
4807 int c;
4808 char *value;
4809{
4810 if (c == '#')
4811 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4812 else /* c == '%' */
4813 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4814}
4815
4816/* Posix.2 says that the WORD should be run through tilde expansion,
4817 parameter expansion, command substitution and arithmetic expansion.
4818 This leaves the result quoted, so quote_string_for_globbing () has
f73dda09 4819 to be called to fix it up for strmatch (). If QUOTED is non-zero,
cce855bc
JA
4820 it means that the entire expression was enclosed in double quotes.
4821 This means that quoting characters in the pattern do not make any
4822 special pattern characters quoted. For example, the `*' in the
4823 following retains its special meaning: "${foo#'*'}". */
4824static char *
4825getpattern (value, quoted, expandpat)
4826 char *value;
4827 int quoted, expandpat;
4828{
4829 char *pat, *tword;
4830 WORD_LIST *l;
0628567a 4831#if 0
cce855bc 4832 int i;
0628567a 4833#endif
7117c2d2
JA
4834 /* There is a problem here: how to handle single or double quotes in the
4835 pattern string when the whole expression is between double quotes?
4836 POSIX.2 says that enclosing double quotes do not cause the pattern to
4837 be quoted, but does that leave us a problem with @ and array[@] and their
4838 expansions inside a pattern? */
4839#if 0
cce855bc
JA
4840 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4841 {
4842 i = 0;
a0c0a00f 4843 pat = string_extract_double_quoted (tword, &i, SX_STRIPDQ);
cce855bc
JA
4844 free (tword);
4845 tword = pat;
4846 }
7117c2d2 4847#endif
cce855bc 4848
7117c2d2
JA
4849 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4850 word splitting. */
95732b49 4851 l = *value ? expand_string_for_rhs (value,
7117c2d2 4852 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
cce855bc 4853 (int *)NULL, (int *)NULL)
cce855bc 4854 : (WORD_LIST *)0;
cce855bc
JA
4855 pat = string_list (l);
4856 dispose_words (l);
4857 if (pat)
4858 {
4859 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4860 free (pat);
4861 pat = tword;
4862 }
4863 return (pat);
4864}
4865
7117c2d2 4866#if 0
cce855bc
JA
4867/* Handle removing a pattern from a string as a result of ${name%[%]value}
4868 or ${name#[#]value}. */
4869static char *
7117c2d2
JA
4870variable_remove_pattern (value, pattern, patspec, quoted)
4871 char *value, *pattern;
4872 int patspec, quoted;
cce855bc 4873{
7117c2d2 4874 char *tword;
cce855bc 4875
7117c2d2 4876 tword = remove_pattern (value, pattern, patspec);
cce855bc 4877
cce855bc
JA
4878 return (tword);
4879}
a0c0a00f
CR
4880#endif
4881
4882static char *
4883list_remove_pattern (list, pattern, patspec, itype, quoted)
4884 WORD_LIST *list;
4885 char *pattern;
4886 int patspec, itype, quoted;
4887{
4888 WORD_LIST *new, *l;
4889 WORD_DESC *w;
4890 char *tword;
4891
4892 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4893 {
4894 tword = remove_pattern (l->word->word, pattern, patspec);
4895 w = alloc_word_desc ();
4896 w->word = tword ? tword : savestring ("");
4897 new = make_word_list (w, new);
4898 }
4899
4900 l = REVERSE_LIST (new, WORD_LIST *);
4901 tword = string_list_pos_params (itype, l, quoted);
4902 dispose_words (l);
4903
4904 return (tword);
4905}
4906
4907static char *
4908parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4909 int itype;
4910 char *pattern;
4911 int patspec, quoted;
4912{
4913 char *ret;
4914 WORD_LIST *list;
4915
4916 list = list_rest_of_args ();
4917 if (list == 0)
4918 return ((char *)NULL);
4919 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4920 dispose_words (list);
4921 return (ret);
4922}
4923
4924#if defined (ARRAY_VARS)
4925static char *
4926array_remove_pattern (var, pattern, patspec, varname, quoted)
4927 SHELL_VAR *var;
4928 char *pattern;
4929 int patspec;
4930 char *varname; /* so we can figure out how it's indexed */
4931 int quoted;
4932{
4933 ARRAY *a;
4934 HASH_TABLE *h;
4935 int itype;
4936 char *ret;
4937 WORD_LIST *list;
4938 SHELL_VAR *v;
4939
4940 /* compute itype from varname here */
4941 v = array_variable_part (varname, &ret, 0);
4942
4943 /* XXX */
4944 if (v && invisible_p (v))
4945 return ((char *)NULL);
4946
4947 itype = ret[0];
4948
4949 a = (v && array_p (v)) ? array_cell (v) : 0;
4950 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4951
4952 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4953 if (list == 0)
4954 return ((char *)NULL);
4955 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4956 dispose_words (list);
4957
4958 return ret;
4959}
4960#endif /* ARRAY_VARS */
4961
4962static char *
4963parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4964 char *varname, *value;
4965 int ind;
4966 char *patstr;
4967 int rtype, quoted, flags;
4968{
4969 int vtype, patspec, starsub;
4970 char *temp1, *val, *pattern;
4971 SHELL_VAR *v;
4972
4973 if (value == 0)
4974 return ((char *)NULL);
4975
4976 this_command_name = varname;
4977
4978 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4979 if (vtype == -1)
4980 return ((char *)NULL);
4981
4982 starsub = vtype & VT_STARSUB;
4983 vtype &= ~VT_STARSUB;
4984
4985 patspec = getpatspec (rtype, patstr);
4986 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4987 patstr++;
4988
4989 /* Need to pass getpattern newly-allocated memory in case of expansion --
4990 the expansion code will free the passed string on an error. */
4991 temp1 = savestring (patstr);
4992 pattern = getpattern (temp1, quoted, 1);
4993 free (temp1);
4994
4995 temp1 = (char *)NULL; /* shut up gcc */
4996 switch (vtype)
4997 {
4998 case VT_VARIABLE:
4999 case VT_ARRAYMEMBER:
5000 temp1 = remove_pattern (val, pattern, patspec);
5001 if (vtype == VT_VARIABLE)
5002 FREE (val);
5003 if (temp1)
5004 {
5005 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5006 ? quote_string (temp1)
5007 : quote_escapes (temp1);
5008 free (temp1);
5009 temp1 = val;
5010 }
5011 break;
5012#if defined (ARRAY_VARS)
5013 case VT_ARRAYVAR:
5014 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
5015 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5016 {
5017 val = quote_escapes (temp1);
5018 free (temp1);
5019 temp1 = val;
5020 }
5021 break;
5022#endif
5023 case VT_POSPARMS:
5024 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
5025 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5026 {
5027 val = quote_escapes (temp1);
5028 free (temp1);
5029 temp1 = val;
5030 }
5031 break;
5032 }
5033
5034 FREE (pattern);
5035 return temp1;
5036}
5037
5038static char *
5039string_var_assignment (v, s)
5040 SHELL_VAR *v;
5041 char *s;
5042{
5043 char flags[MAX_ATTRIBUTES], *ret, *val;
5044 int i;
5045
5046 val = sh_quote_reusable (s, 0);
5047 i = var_attribute_string (v, 0, flags);
5048 ret = (char *)xmalloc (i + strlen (val) + strlen (v->name) + 16 + MAX_ATTRIBUTES);
5049 if (i > 0)
5050 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
5051 else
5052 sprintf (ret, "%s=%s", v->name, val);
5053 free (val);
5054 return ret;
5055}
5056
5057#if defined (ARRAY_VARS)
5058static char *
5059array_var_assignment (v, itype, quoted)
5060 SHELL_VAR *v;
5061 int itype, quoted;
5062{
5063 char *ret, *val, flags[MAX_ATTRIBUTES];
5064 int i;
5065
5066 if (v == 0)
5067 return (char *)NULL;
5068 val = array_p (v) ? array_to_assign (array_cell (v), 0)
5069 : assoc_to_assign (assoc_cell (v), 0);
5070 if (val == 0)
5071 {
5072 val = (char *)xmalloc (3);
5073 val[0] = '(';
5074 val[1] = ')';
5075 val[2] = 0;
5076 }
5077 else
5078 {
5079 ret = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (val) : quote_escapes (val);
5080 free (val);
5081 val = ret;
5082 }
5083 i = var_attribute_string (v, 0, flags);
5084 ret = (char *)xmalloc (i + strlen (val) + strlen (v->name) + 16);
5085 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
5086 free (val);
5087 return ret;
5088}
5089#endif
5090
5091static char *
5092pos_params_assignment (list, itype, quoted)
5093 WORD_LIST *list;
5094 int itype;
5095 int quoted;
5096{
5097 char *temp, *ret;
5098
5099 /* first, we transform the list to quote each word. */
5100 temp = list_transform ('Q', (SHELL_VAR *)0, list, itype, quoted);
5101 ret = (char *)xmalloc (strlen (temp) + 8);
5102 strcpy (ret, "set -- ");
5103 strcpy (ret + 7, temp);
5104 free (temp);
5105 return ret;
5106}
5107
5108static char *
5109string_transform (xc, v, s)
5110 int xc;
5111 SHELL_VAR *v;
5112 char *s;
5113{
5114 char *ret, flags[MAX_ATTRIBUTES];
5115 int i;
5116
5117 if (((xc == 'A' || xc == 'a') && v == 0) || (xc != 'a' && s == 0))
5118 return (char *)NULL;
5119
5120 switch (xc)
5121 {
5122 /* Transformations that interrogate the variable */
5123 case 'a':
5124 i = var_attribute_string (v, 0, flags);
5125 ret = (i > 0) ? savestring (flags) : (char *)NULL;
5126 break;
5127 case 'A':
5128 ret = string_var_assignment (v, s);
5129 break;
5130 /* Transformations that modify the variable's value */
5131 case 'E':
5132 ret = ansiexpand (s, 0, strlen (s), (int *)0);
5133 break;
5134 case 'P':
5135 ret = decode_prompt_string (s);
5136 break;
5137 case 'Q':
5138 ret = sh_quote_reusable (s, 0);
5139 break;
5140 default:
5141 ret = (char *)NULL;
5142 break;
5143 }
5144 return ret;
5145}
cce855bc
JA
5146
5147static char *
a0c0a00f
CR
5148list_transform (xc, v, list, itype, quoted)
5149 int xc;
5150 SHELL_VAR *v;
cce855bc 5151 WORD_LIST *list;
a0c0a00f 5152 int itype, quoted;
cce855bc
JA
5153{
5154 WORD_LIST *new, *l;
5155 WORD_DESC *w;
5156 char *tword;
5157
5158 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
5159 {
a0c0a00f 5160 tword = string_transform (xc, v, l->word->word);
95732b49 5161 w = alloc_word_desc ();
a0c0a00f 5162 w->word = tword ? tword : savestring (""); /* XXX */
cce855bc
JA
5163 new = make_word_list (w, new);
5164 }
5165
5166 l = REVERSE_LIST (new, WORD_LIST *);
3185942a 5167 tword = string_list_pos_params (itype, l, quoted);
cce855bc 5168 dispose_words (l);
3185942a 5169
cce855bc
JA
5170 return (tword);
5171}
5172
5173static char *
a0c0a00f
CR
5174parameter_list_transform (xc, itype, quoted)
5175 int xc;
7117c2d2 5176 int itype;
a0c0a00f 5177 int quoted;
cce855bc 5178{
7117c2d2 5179 char *ret;
cce855bc
JA
5180 WORD_LIST *list;
5181
cce855bc 5182 list = list_rest_of_args ();
7117c2d2
JA
5183 if (list == 0)
5184 return ((char *)NULL);
a0c0a00f
CR
5185 if (xc == 'A')
5186 return (pos_params_assignment (list, itype, quoted));
5187 ret = list_transform (xc, (SHELL_VAR *)0, list, itype, quoted);
cce855bc 5188 dispose_words (list);
cce855bc
JA
5189 return (ret);
5190}
5191
5192#if defined (ARRAY_VARS)
5193static char *
a0c0a00f
CR
5194array_transform (xc, var, varname, quoted)
5195 int xc;
3185942a 5196 SHELL_VAR *var;
7117c2d2
JA
5197 char *varname; /* so we can figure out how it's indexed */
5198 int quoted;
cce855bc 5199{
3185942a
JA
5200 ARRAY *a;
5201 HASH_TABLE *h;
7117c2d2
JA
5202 int itype;
5203 char *ret;
5204 WORD_LIST *list;
5205 SHELL_VAR *v;
cce855bc 5206
7117c2d2
JA
5207 /* compute itype from varname here */
5208 v = array_variable_part (varname, &ret, 0);
ac50fbac
CR
5209
5210 /* XXX */
a0c0a00f 5211 if (v && invisible_p (v))
ac50fbac
CR
5212 return ((char *)NULL);
5213
7117c2d2
JA
5214 itype = ret[0];
5215
a0c0a00f
CR
5216 if (xc == 'A')
5217 return (array_var_assignment (v, itype, quoted));
5218
3185942a
JA
5219 a = (v && array_p (v)) ? array_cell (v) : 0;
5220 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
5221
5222 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
7117c2d2
JA
5223 if (list == 0)
5224 return ((char *)NULL);
a0c0a00f 5225 ret = list_transform (xc, v, list, itype, quoted);
7117c2d2
JA
5226 dispose_words (list);
5227
5228 return ret;
5229}
5230#endif /* ARRAY_VARS */
5231
5232static char *
a0c0a00f 5233parameter_brace_transform (varname, value, ind, xform, rtype, quoted, flags)
495aee44
CR
5234 char *varname, *value;
5235 int ind;
a0c0a00f 5236 char *xform;
495aee44 5237 int rtype, quoted, flags;
7117c2d2 5238{
a0c0a00f
CR
5239 int vtype, xc;
5240 char *temp1, *val;
7117c2d2
JA
5241 SHELL_VAR *v;
5242
a0c0a00f
CR
5243 xc = xform[0];
5244 if (value == 0 && xc != 'A' && xc != 'a')
7117c2d2
JA
5245 return ((char *)NULL);
5246
5247 this_command_name = varname;
5248
495aee44 5249 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7117c2d2 5250 if (vtype == -1)
cce855bc
JA
5251 return ((char *)NULL);
5252
a0c0a00f
CR
5253 /* check for valid values of xc */
5254 switch (xc)
5255 {
5256 case 'a': /* expand to a string with just attributes */
5257 case 'A': /* expand as an assignment statement with attributes */
5258 case 'E': /* expand like $'...' */
5259 case 'P': /* expand like prompt string */
5260 case 'Q': /* quote reusably */
5261 break;
5262 default:
5263 return &expand_param_error;
5264 }
cce855bc 5265
7117c2d2 5266 temp1 = (char *)NULL; /* shut up gcc */
a0c0a00f 5267 switch (vtype & ~VT_STARSUB)
cce855bc 5268 {
7117c2d2
JA
5269 case VT_VARIABLE:
5270 case VT_ARRAYMEMBER:
a0c0a00f 5271 temp1 = string_transform (xc, v, val);
7117c2d2
JA
5272 if (vtype == VT_VARIABLE)
5273 FREE (val);
5274 if (temp1)
28ef6c31 5275 {
3185942a
JA
5276 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5277 ? quote_string (temp1)
5278 : quote_escapes (temp1);
7117c2d2
JA
5279 free (temp1);
5280 temp1 = val;
28ef6c31 5281 }
7117c2d2
JA
5282 break;
5283#if defined (ARRAY_VARS)
5284 case VT_ARRAYVAR:
a0c0a00f 5285 temp1 = array_transform (xc, v, varname, quoted);
7117c2d2 5286 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
cce855bc 5287 {
7117c2d2
JA
5288 val = quote_escapes (temp1);
5289 free (temp1);
5290 temp1 = val;
cce855bc 5291 }
7117c2d2
JA
5292 break;
5293#endif
5294 case VT_POSPARMS:
a0c0a00f 5295 temp1 = parameter_list_transform (xc, varname[0], quoted);
7117c2d2
JA
5296 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5297 {
5298 val = quote_escapes (temp1);
5299 free (temp1);
5300 temp1 = val;
5301 }
5302 break;
cce855bc
JA
5303 }
5304
7117c2d2 5305 return temp1;
a0c0a00f 5306}
cce855bc 5307
726f6388
JA
5308/*******************************************
5309 * *
5310 * Functions to expand WORD_DESCs *
5311 * *
5312 *******************************************/
5313
5314/* Expand WORD, performing word splitting on the result. This does
5315 parameter expansion, command substitution, arithmetic expansion,
5316 word splitting, and quote removal. */
5317
5318WORD_LIST *
5319expand_word (word, quoted)
5320 WORD_DESC *word;
5321 int quoted;
5322{
5323 WORD_LIST *result, *tresult;
5324
b72432fd 5325 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
726f6388
JA
5326 result = word_list_split (tresult);
5327 dispose_words (tresult);
ccc6cda3 5328 return (result ? dequote_list (result) : result);
726f6388
JA
5329}
5330
5331/* Expand WORD, but do not perform word splitting on the result. This
5332 does parameter expansion, command substitution, arithmetic expansion,
5333 and quote removal. */
5334WORD_LIST *
28ef6c31 5335expand_word_unsplit (word, quoted)
726f6388
JA
5336 WORD_DESC *word;
5337 int quoted;
5338{
5339 WORD_LIST *result;
5340
28ef6c31 5341 expand_no_split_dollar_star = 1;
0001803f
CR
5342#if defined (HANDLE_MULTIBYTE)
5343 if (ifs_firstc[0] == 0)
5344#else
5345 if (ifs_firstc == 0)
5346#endif
5347 word->flags |= W_NOSPLIT;
1cc06898 5348 word->flags |= W_NOSPLIT2;
b72432fd 5349 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31 5350 expand_no_split_dollar_star = 0;
7117c2d2 5351
ccc6cda3 5352 return (result ? dequote_list (result) : result);
726f6388
JA
5353}
5354
5355/* Perform shell expansions on WORD, but do not perform word splitting or
0001803f
CR
5356 quote removal on the result. Virtually identical to expand_word_unsplit;
5357 could be combined if implementations don't diverge. */
726f6388
JA
5358WORD_LIST *
5359expand_word_leave_quoted (word, quoted)
5360 WORD_DESC *word;
5361 int quoted;
5362{
0001803f
CR
5363 WORD_LIST *result;
5364
5365 expand_no_split_dollar_star = 1;
5366#if defined (HANDLE_MULTIBYTE)
5367 if (ifs_firstc[0] == 0)
5368#else
5369 if (ifs_firstc == 0)
5370#endif
5371 word->flags |= W_NOSPLIT;
5372 word->flags |= W_NOSPLIT2;
5373 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
5374 expand_no_split_dollar_star = 0;
5375
5376 return result;
726f6388
JA
5377}
5378
726f6388
JA
5379#if defined (PROCESS_SUBSTITUTION)
5380
cce855bc
JA
5381/*****************************************************************/
5382/* */
5383/* Hacking Process Substitution */
5384/* */
5385/*****************************************************************/
726f6388 5386
726f6388
JA
5387#if !defined (HAVE_DEV_FD)
5388/* Named pipes must be removed explicitly with `unlink'. This keeps a list
5389 of FIFOs the shell has open. unlink_fifo_list will walk the list and
5390 unlink all of them. add_fifo_list adds the name of an open FIFO to the
5391 list. NFIFO is a count of the number of FIFOs in the list. */
5392#define FIFO_INCR 20
5393
f73dda09
JA
5394struct temp_fifo {
5395 char *file;
5396 pid_t proc;
5397};
5398
5399static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
ccc6cda3
JA
5400static int nfifo;
5401static int fifo_list_size;
726f6388 5402
a0c0a00f
CR
5403void
5404clear_fifo_list ()
5405{
5406}
5407
495aee44
CR
5408char *
5409copy_fifo_list (sizep)
5410 int *sizep;
5411{
5412 if (sizep)
5413 *sizep = 0;
5414 return (char *)NULL;
5415}
5416
726f6388
JA
5417static void
5418add_fifo_list (pathname)
5419 char *pathname;
5420{
5421 if (nfifo >= fifo_list_size - 1)
5422 {
5423 fifo_list_size += FIFO_INCR;
f73dda09
JA
5424 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
5425 fifo_list_size * sizeof (struct temp_fifo));
726f6388
JA
5426 }
5427
f73dda09
JA
5428 fifo_list[nfifo].file = savestring (pathname);
5429 nfifo++;
726f6388
JA
5430}
5431
495aee44
CR
5432void
5433unlink_fifo (i)
5434 int i;
5435{
5436 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
5437 {
5438 unlink (fifo_list[i].file);
5439 free (fifo_list[i].file);
5440 fifo_list[i].file = (char *)NULL;
5441 fifo_list[i].proc = -1;
5442 }
5443}
5444
726f6388
JA
5445void
5446unlink_fifo_list ()
5447{
f73dda09
JA
5448 int saved, i, j;
5449
ccc6cda3 5450 if (nfifo == 0)
726f6388
JA
5451 return;
5452
f73dda09 5453 for (i = saved = 0; i < nfifo; i++)
726f6388 5454 {
f73dda09
JA
5455 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
5456 {
7117c2d2
JA
5457 unlink (fifo_list[i].file);
5458 free (fifo_list[i].file);
5459 fifo_list[i].file = (char *)NULL;
5460 fifo_list[i].proc = -1;
f73dda09
JA
5461 }
5462 else
7117c2d2 5463 saved++;
f73dda09
JA
5464 }
5465
5466 /* If we didn't remove some of the FIFOs, compact the list. */
5467 if (saved)
5468 {
5469 for (i = j = 0; i < nfifo; i++)
5470 if (fifo_list[i].file)
5471 {
5472 fifo_list[j].file = fifo_list[i].file;
5473 fifo_list[j].proc = fifo_list[i].proc;
5474 j++;
5475 }
5476 nfifo = j;
726f6388 5477 }
f73dda09
JA
5478 else
5479 nfifo = 0;
726f6388
JA
5480}
5481
495aee44
CR
5482/* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
5483 from some point in the past, and close all open FIFOs in fifo_list
5484 that are not marked as active in LIST. If LIST is NULL, close
5485 everything in fifo_list. LSIZE is the number of elements in LIST, in
5486 case it's larger than fifo_list_size (size of fifo_list). */
5487void
5488close_new_fifos (list, lsize)
5489 char *list;
5490 int lsize;
5491{
5492 int i;
5493
5494 if (list == 0)
5495 {
5496 unlink_fifo_list ();
5497 return;
5498 }
5499
5500 for (i = 0; i < lsize; i++)
5501 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
5502 unlink_fifo (i);
5503
5504 for (i = lsize; i < fifo_list_size; i++)
5505 unlink_fifo (i);
5506}
5507
f1be666c
JA
5508int
5509fifos_pending ()
5510{
5511 return nfifo;
5512}
5513
495aee44
CR
5514int
5515num_fifos ()
5516{
5517 return nfifo;
5518}
5519
726f6388
JA
5520static char *
5521make_named_pipe ()
5522{
5523 char *tname;
5524
3185942a 5525 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
726f6388
JA
5526 if (mkfifo (tname, 0600) < 0)
5527 {
5528 free (tname);
5529 return ((char *)NULL);
5530 }
5531
5532 add_fifo_list (tname);
5533 return (tname);
5534}
5535
726f6388
JA
5536#else /* HAVE_DEV_FD */
5537
5538/* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
5539 has open to children. NFDS is a count of the number of bits currently
5540 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
5541 of open files. */
5542static char *dev_fd_list = (char *)NULL;
ccc6cda3 5543static int nfds;
726f6388
JA
5544static int totfds; /* The highest possible number of open files. */
5545
a0c0a00f
CR
5546void
5547clear_fifo (i)
5548 int i;
5549{
5550 if (dev_fd_list[i])
5551 {
5552 dev_fd_list[i] = 0;
5553 nfds--;
5554 }
5555}
5556
5557void
5558clear_fifo_list ()
5559{
5560 register int i;
5561
5562 if (nfds == 0)
5563 return;
5564
5565 for (i = 0; nfds && i < totfds; i++)
5566 clear_fifo (i);
5567
5568 nfds = 0;
5569}
5570
495aee44
CR
5571char *
5572copy_fifo_list (sizep)
5573 int *sizep;
5574{
5575 char *ret;
5576
5577 if (nfds == 0 || totfds == 0)
5578 {
5579 if (sizep)
5580 *sizep = 0;
5581 return (char *)NULL;
5582 }
5583
5584 if (sizep)
5585 *sizep = totfds;
5586 ret = (char *)xmalloc (totfds);
5587 return (memcpy (ret, dev_fd_list, totfds));
5588}
5589
726f6388
JA
5590static void
5591add_fifo_list (fd)
5592 int fd;
5593{
495aee44 5594 if (dev_fd_list == 0 || fd >= totfds)
726f6388
JA
5595 {
5596 int ofds;
5597
5598 ofds = totfds;
5599 totfds = getdtablesize ();
5600 if (totfds < 0 || totfds > 256)
5601 totfds = 256;
3185942a 5602 if (fd >= totfds)
726f6388
JA
5603 totfds = fd + 2;
5604
f73dda09 5605 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
7117c2d2 5606 memset (dev_fd_list + ofds, '\0', totfds - ofds);
726f6388
JA
5607 }
5608
5609 dev_fd_list[fd] = 1;
5610 nfds++;
5611}
5612
f1be666c
JA
5613int
5614fifos_pending ()
5615{
5616 return 0; /* used for cleanup; not needed with /dev/fd */
5617}
5618
495aee44
CR
5619int
5620num_fifos ()
5621{
5622 return nfds;
5623}
5624
5625void
5626unlink_fifo (fd)
5627 int fd;
5628{
5629 if (dev_fd_list[fd])
5630 {
5631 close (fd);
5632 dev_fd_list[fd] = 0;
5633 nfds--;
5634 }
5635}
5636
726f6388
JA
5637void
5638unlink_fifo_list ()
5639{
5640 register int i;
5641
ccc6cda3 5642 if (nfds == 0)
726f6388
JA
5643 return;
5644
5645 for (i = 0; nfds && i < totfds; i++)
495aee44 5646 unlink_fifo (i);
726f6388
JA
5647
5648 nfds = 0;
5649}
5650
495aee44
CR
5651/* Take LIST, which is a snapshot copy of dev_fd_list from some point in
5652 the past, and close all open fds in dev_fd_list that are not marked
5653 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
5654 LSIZE is the number of elements in LIST, in case it's larger than
5655 totfds (size of dev_fd_list). */
5656void
5657close_new_fifos (list, lsize)
5658 char *list;
5659 int lsize;
5660{
5661 int i;
5662
5663 if (list == 0)
5664 {
5665 unlink_fifo_list ();
5666 return;
5667 }
5668
5669 for (i = 0; i < lsize; i++)
5670 if (list[i] == 0 && i < totfds && dev_fd_list[i])
5671 unlink_fifo (i);
5672
5673 for (i = lsize; i < totfds; i++)
5674 unlink_fifo (i);
5675}
5676
726f6388
JA
5677#if defined (NOTDEF)
5678print_dev_fd_list ()
5679{
5680 register int i;
5681
f73dda09 5682 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
726f6388
JA
5683 fflush (stderr);
5684
5685 for (i = 0; i < totfds; i++)
5686 {
5687 if (dev_fd_list[i])
5688 fprintf (stderr, " %d", i);
5689 }
5690 fprintf (stderr, "\n");
5691}
5692#endif /* NOTDEF */
5693
5694static char *
5695make_dev_fd_filename (fd)
5696 int fd;
5697{
f73dda09 5698 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
726f6388 5699
17345e5a 5700 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
bb70624e
JA
5701
5702 strcpy (ret, DEV_FD_PREFIX);
5703 p = inttostr (fd, intbuf, sizeof (intbuf));
5704 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5705
726f6388
JA
5706 add_fifo_list (fd);
5707 return (ret);
5708}
5709
5710#endif /* HAVE_DEV_FD */
5711
5712/* Return a filename that will open a connection to the process defined by
5713 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5714 a filename in /dev/fd corresponding to a descriptor that is one of the
5715 ends of the pipe. If not defined, we use named pipes on systems that have
5716 them. Systems without /dev/fd and named pipes are out of luck.
5717
5718 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5719 use the read end of the pipe and dup that file descriptor to fd 0 in
5720 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5721 writing or use the write end of the pipe in the child, and dup that
5722 file descriptor to fd 1 in the child. The parent does the opposite. */
5723
5724static char *
5725process_substitute (string, open_for_read_in_child)
5726 char *string;
5727 int open_for_read_in_child;
5728{
5729 char *pathname;
5730 int fd, result;
5731 pid_t old_pid, pid;
5732#if defined (HAVE_DEV_FD)
5733 int parent_pipe_fd, child_pipe_fd;
5734 int fildes[2];
5735#endif /* HAVE_DEV_FD */
5736#if defined (JOB_CONTROL)
5737 pid_t old_pipeline_pgrp;
ccc6cda3 5738#endif
726f6388 5739
cce855bc 5740 if (!string || !*string || wordexp_only)
726f6388
JA
5741 return ((char *)NULL);
5742
5743#if !defined (HAVE_DEV_FD)
5744 pathname = make_named_pipe ();
5745#else /* HAVE_DEV_FD */
5746 if (pipe (fildes) < 0)
5747 {
a0c0a00f 5748 sys_error ("%s", _("cannot make pipe for process substitution"));
726f6388
JA
5749 return ((char *)NULL);
5750 }
5751 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5752 the pipe in the parent, otherwise the read end. */
5753 parent_pipe_fd = fildes[open_for_read_in_child];
5754 child_pipe_fd = fildes[1 - open_for_read_in_child];
d166f048
JA
5755 /* Move the parent end of the pipe to some high file descriptor, to
5756 avoid clashes with FDs used by the script. */
5757 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5758
726f6388
JA
5759 pathname = make_dev_fd_filename (parent_pipe_fd);
5760#endif /* HAVE_DEV_FD */
5761
3185942a 5762 if (pathname == 0)
726f6388 5763 {
a0c0a00f 5764 sys_error ("%s", _("cannot make pipe for process substitution"));
726f6388
JA
5765 return ((char *)NULL);
5766 }
5767
5768 old_pid = last_made_pid;
5769
5770#if defined (JOB_CONTROL)
5771 old_pipeline_pgrp = pipeline_pgrp;
a0c0a00f
CR
5772 if (pipeline_pgrp == 0 || (subshell_environment & (SUBSHELL_PIPE|SUBSHELL_FORK|SUBSHELL_ASYNC)) == 0)
5773 pipeline_pgrp = shell_pgrp;
ccc6cda3 5774 save_pipeline (1);
ccc6cda3
JA
5775#endif /* JOB_CONTROL */
5776
726f6388
JA
5777 pid = make_child ((char *)NULL, 1);
5778 if (pid == 0)
5779 {
ccc6cda3 5780 reset_terminating_signals (); /* XXX */
b80f6443 5781 free_pushed_string_input ();
726f6388 5782 /* Cancel traps, in trap.c. */
495aee44 5783 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
a0c0a00f 5784 QUIT; /* catch any interrupts we got post-fork */
726f6388 5785 setup_async_signals ();
3185942a 5786 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
a0c0a00f
CR
5787
5788 /* if we're expanding a redirection, we shouldn't have access to the
5789 temporary environment, but commands in the subshell should have
5790 access to their own temporary environment. */
5791 if (expanding_redir)
5792 flush_temporary_env ();
726f6388 5793 }
ccc6cda3
JA
5794
5795#if defined (JOB_CONTROL)
726f6388
JA
5796 set_sigchld_handler ();
5797 stop_making_children ();
3185942a 5798 /* XXX - should we only do this in the parent? (as in command subst) */
726f6388 5799 pipeline_pgrp = old_pipeline_pgrp;
a0c0a00f
CR
5800#else
5801 stop_making_children ();
ccc6cda3 5802#endif /* JOB_CONTROL */
726f6388
JA
5803
5804 if (pid < 0)
5805 {
a0c0a00f 5806 sys_error ("%s", _("cannot make child for process substitution"));
726f6388
JA
5807 free (pathname);
5808#if defined (HAVE_DEV_FD)
5809 close (parent_pipe_fd);
5810 close (child_pipe_fd);
5811#endif /* HAVE_DEV_FD */
5812 return ((char *)NULL);
5813 }
5814
5815 if (pid > 0)
5816 {
ccc6cda3 5817#if defined (JOB_CONTROL)
a0c0a00f 5818 if (last_procsub_child)
2965eca9 5819 discard_last_procsub_child ();
a0c0a00f 5820 last_procsub_child = restore_pipeline (0);
ccc6cda3
JA
5821#endif
5822
f73dda09
JA
5823#if !defined (HAVE_DEV_FD)
5824 fifo_list[nfifo-1].proc = pid;
5825#endif
5826
726f6388
JA
5827 last_made_pid = old_pid;
5828
5829#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5830 close_pgrp_pipe ();
5831#endif /* JOB_CONTROL && PGRP_PIPE */
5832
5833#if defined (HAVE_DEV_FD)
5834 close (child_pipe_fd);
5835#endif /* HAVE_DEV_FD */
5836
5837 return (pathname);
5838 }
5839
5840 set_sigint_handler ();
5841
5842#if defined (JOB_CONTROL)
5843 set_job_control (0);
5844#endif /* JOB_CONTROL */
5845
5846#if !defined (HAVE_DEV_FD)
5847 /* Open the named pipe in the child. */
ac50fbac 5848 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
726f6388
JA
5849 if (fd < 0)
5850 {
b80f6443
JA
5851 /* Two separate strings for ease of translation. */
5852 if (open_for_read_in_child)
5853 sys_error (_("cannot open named pipe %s for reading"), pathname);
5854 else
5855 sys_error (_("cannot open named pipe %s for writing"), pathname);
5856
726f6388
JA
5857 exit (127);
5858 }
bb70624e
JA
5859 if (open_for_read_in_child)
5860 {
28ef6c31 5861 if (sh_unset_nodelay_mode (fd) < 0)
bb70624e 5862 {
3185942a 5863 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
bb70624e
JA
5864 exit (127);
5865 }
5866 }
726f6388
JA
5867#else /* HAVE_DEV_FD */
5868 fd = child_pipe_fd;
5869#endif /* HAVE_DEV_FD */
5870
a0c0a00f
CR
5871 /* Discard buffered stdio output before replacing the underlying file
5872 descriptor. */
5873 if (open_for_read_in_child == 0)
5874 fpurge (stdout);
5875
726f6388
JA
5876 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5877 {
b80f6443 5878 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
ccc6cda3 5879 open_for_read_in_child ? 0 : 1);
726f6388
JA
5880 exit (127);
5881 }
5882
f73dda09
JA
5883 if (fd != (open_for_read_in_child ? 0 : 1))
5884 close (fd);
726f6388
JA
5885
5886 /* Need to close any files that this process has open to pipes inherited
5887 from its parent. */
5888 if (current_fds_to_close)
5889 {
5890 close_fd_bitmap (current_fds_to_close);
5891 current_fds_to_close = (struct fd_bitmap *)NULL;
5892 }
5893
5894#if defined (HAVE_DEV_FD)
5895 /* Make sure we close the parent's end of the pipe and clear the slot
5896 in the fd list so it is not closed later, if reallocated by, for
5897 instance, pipe(2). */
5898 close (parent_pipe_fd);
5899 dev_fd_list[parent_pipe_fd] = 0;
5900#endif /* HAVE_DEV_FD */
5901
8dea6e87 5902 /* subshells shouldn't have this flag, which controls using the temporary
a0c0a00f
CR
5903 environment for variable lookups. We have already flushed the temporary
5904 environment above in the case we're expanding a redirection, so processes
5905 executed by this command need to be able to set it independently of their
5906 parent. */
8dea6e87
CR
5907 expanding_redir = 0;
5908
9cce630e
CR
5909 remove_quoted_escapes (string);
5910
a0c0a00f 5911 subshell_level++;
d166f048 5912 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
a0c0a00f 5913 subshell_level--;
726f6388
JA
5914
5915#if !defined (HAVE_DEV_FD)
5916 /* Make sure we close the named pipe in the child before we exit. */
5917 close (open_for_read_in_child ? 0 : 1);
5918#endif /* !HAVE_DEV_FD */
5919
ac50fbac
CR
5920 last_command_exit_value = result;
5921 result = run_exit_trap ();
726f6388
JA
5922 exit (result);
5923 /*NOTREACHED*/
5924}
5925#endif /* PROCESS_SUBSTITUTION */
5926
cce855bc
JA
5927/***********************************/
5928/* */
5929/* Command Substitution */
5930/* */
5931/***********************************/
5932
d166f048 5933static char *
3185942a 5934read_comsub (fd, quoted, rflag)
d166f048 5935 int fd, quoted;
3185942a 5936 int *rflag;
d166f048 5937{
3185942a
JA
5938 char *istring, buf[128], *bufp, *s;
5939 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
f73dda09 5940 ssize_t bufn;
280bd77d 5941 int nullbyte;
d166f048
JA
5942
5943 istring = (char *)NULL;
3185942a
JA
5944 istring_index = istring_size = bufn = tflag = 0;
5945
5946 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5947 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
d166f048 5948
280bd77d
CR
5949 nullbyte = 0;
5950
3185942a
JA
5951 /* Read the output of the command through the pipe. This may need to be
5952 changed to understand multibyte characters in the future. */
d166f048
JA
5953 while (1)
5954 {
5955 if (fd < 0)
28ef6c31 5956 break;
d166f048
JA
5957 if (--bufn <= 0)
5958 {
bb70624e 5959 bufn = zread (fd, buf, sizeof (buf));
d166f048
JA
5960 if (bufn <= 0)
5961 break;
5962 bufp = buf;
5963 }
5964 c = *bufp++;
5965
28ef6c31
JA
5966 if (c == 0)
5967 {
a0c0a00f 5968#if 1
280bd77d
CR
5969 if (nullbyte == 0)
5970 {
5971 internal_warning ("%s", _("command substitution: ignored null byte in input"));
5972 nullbyte = 1;
5973 }
28ef6c31
JA
5974#endif
5975 continue;
5976 }
5977
d166f048
JA
5978 /* Add the character to ISTRING, possibly after resizing it. */
5979 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5980
f1be666c
JA
5981 /* This is essentially quote_string inline */
5982 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5983 istring[istring_index++] = CTLESC;
5984 /* Escape CTLESC and CTLNUL in the output to protect those characters
5985 from the rest of the word expansions (word splitting and globbing.)
5986 This is essentially quote_escapes inline. */
3185942a
JA
5987 else if (skip_ctlesc == 0 && c == CTLESC)
5988 {
5989 tflag |= W_HASCTLESC;
5990 istring[istring_index++] = CTLESC;
5991 }
5992 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
d166f048
JA
5993 istring[istring_index++] = CTLESC;
5994
5995 istring[istring_index++] = c;
28ef6c31
JA
5996
5997#if 0
5998#if defined (__CYGWIN__)
5999 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
6000 {
6001 istring_index--;
6002 istring[istring_index - 1] = '\n';
6003 }
6004#endif
6005#endif
d166f048
JA
6006 }
6007
6008 if (istring)
6009 istring[istring_index] = '\0';
6010
6011 /* If we read no output, just return now and save ourselves some
6012 trouble. */
6013 if (istring_index == 0)
6014 {
6015 FREE (istring);
3185942a
JA
6016 if (rflag)
6017 *rflag = tflag;
d166f048
JA
6018 return (char *)NULL;
6019 }
6020
6021 /* Strip trailing newlines from the output of the command. */
6022 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6023 {
6024 while (istring_index > 0)
6025 {
6026 if (istring[istring_index - 1] == '\n')
6027 {
6028 --istring_index;
6029
6030 /* If the newline was quoted, remove the quoting char. */
6031 if (istring[istring_index - 1] == CTLESC)
6032 --istring_index;
6033 }
6034 else
6035 break;
6036 }
6037 istring[istring_index] = '\0';
6038 }
6039 else
6040 strip_trailing (istring, istring_index - 1, 1);
6041
3185942a
JA
6042 if (rflag)
6043 *rflag = tflag;
d166f048
JA
6044 return istring;
6045}
6046
3185942a
JA
6047/* Perform command substitution on STRING. This returns a WORD_DESC * with the
6048 contained string possibly quoted. */
6049WORD_DESC *
726f6388
JA
6050command_substitute (string, quoted)
6051 char *string;
6052 int quoted;
6053{
95732b49 6054 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
a0c0a00f 6055 char *istring, *s;
3185942a
JA
6056 int result, fildes[2], function_value, pflags, rc, tflag;
6057 WORD_DESC *ret;
726f6388 6058
ccc6cda3 6059 istring = (char *)NULL;
726f6388
JA
6060
6061 /* Don't fork () if there is no need to. In the case of no command to
6062 run, just return NULL. */
a0c0a00f
CR
6063#if 1
6064 for (s = string; s && *s && (shellblank (*s) || *s == '\n'); s++)
6065 ;
6066 if (s == 0 || *s == 0)
6067 return ((WORD_DESC *)NULL);
6068#else
726f6388 6069 if (!string || !*string || (string[0] == '\n' && !string[1]))
3185942a 6070 return ((WORD_DESC *)NULL);
a0c0a00f 6071#endif
726f6388 6072
cce855bc
JA
6073 if (wordexp_only && read_but_dont_execute)
6074 {
0001803f 6075 last_command_exit_value = EX_WEXPCOMSUB;
cce855bc
JA
6076 jump_to_top_level (EXITPROG);
6077 }
6078
bb70624e
JA
6079 /* We're making the assumption here that the command substitution will
6080 eventually run a command from the file system. Since we'll run
6081 maybe_make_export_env in this subshell before executing that command,
6082 the parent shell and any other shells it starts will have to remake
6083 the environment. If we make it before we fork, other shells won't
6084 have to. Don't bother if we have any temporary variable assignments,
6085 though, because the export environment will be remade after this
6086 command completes anyway, but do it if all the words to be expanded
6087 are variable assignments. */
6088 if (subst_assign_varlist == 0 || garglist == 0)
6089 maybe_make_export_env (); /* XXX */
6090
b80f6443 6091 /* Flags to pass to parse_and_execute() */
0001803f 6092 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
b80f6443 6093
726f6388
JA
6094 /* Pipe the output of executing STRING into the current shell. */
6095 if (pipe (fildes) < 0)
6096 {
a0c0a00f 6097 sys_error ("%s", _("cannot make pipe for command substitution"));
726f6388
JA
6098 goto error_exit;
6099 }
6100
6101 old_pid = last_made_pid;
6102#if defined (JOB_CONTROL)
ccc6cda3 6103 old_pipeline_pgrp = pipeline_pgrp;
28ef6c31
JA
6104 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
6105 if ((subshell_environment & SUBSHELL_PIPE) == 0)
6106 pipeline_pgrp = shell_pgrp;
ccc6cda3 6107 cleanup_the_pipeline ();
95732b49 6108#endif /* JOB_CONTROL */
726f6388 6109
95732b49 6110 old_async_pid = last_asynchronous_pid;
95732b49 6111 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
95732b49
JA
6112 last_asynchronous_pid = old_async_pid;
6113
726f6388 6114 if (pid == 0)
495aee44
CR
6115 {
6116 /* Reset the signal handlers in the child, but don't free the
6117 trap strings. Set a flag noting that we have to free the
6118 trap strings if we run trap to change a signal disposition. */
6119 reset_signal_handlers ();
a0c0a00f
CR
6120 if (ISINTERRUPT)
6121 {
6122 kill (getpid (), SIGINT);
6123 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6124 }
6125 QUIT; /* catch any interrupts we got post-fork */
495aee44
CR
6126 subshell_environment |= SUBSHELL_RESETTRAP;
6127 }
ccc6cda3
JA
6128
6129#if defined (JOB_CONTROL)
3185942a 6130 /* XXX DO THIS ONLY IN PARENT ? XXX */
ccc6cda3
JA
6131 set_sigchld_handler ();
6132 stop_making_children ();
f1be666c
JA
6133 if (pid != 0)
6134 pipeline_pgrp = old_pipeline_pgrp;
f73dda09
JA
6135#else
6136 stop_making_children ();
ccc6cda3 6137#endif /* JOB_CONTROL */
726f6388
JA
6138
6139 if (pid < 0)
6140 {
b80f6443 6141 sys_error (_("cannot make child for command substitution"));
726f6388
JA
6142 error_exit:
6143
ac50fbac
CR
6144 last_made_pid = old_pid;
6145
726f6388
JA
6146 FREE (istring);
6147 close (fildes[0]);
6148 close (fildes[1]);
3185942a 6149 return ((WORD_DESC *)NULL);
726f6388
JA
6150 }
6151
6152 if (pid == 0)
6153 {
a0c0a00f
CR
6154 /* The currently executing shell is not interactive. */
6155 interactive = 0;
6156
726f6388 6157 set_sigint_handler (); /* XXX */
28ef6c31 6158
b80f6443
JA
6159 free_pushed_string_input ();
6160
a0c0a00f
CR
6161 /* Discard buffered stdio output before replacing the underlying file
6162 descriptor. */
6163 fpurge (stdout);
6164
726f6388
JA
6165 if (dup2 (fildes[1], 1) < 0)
6166 {
a0c0a00f 6167 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
726f6388
JA
6168 exit (EXECUTION_FAILURE);
6169 }
6170
6171 /* If standard output is closed in the parent shell
6172 (such as after `exec >&-'), file descriptor 1 will be
6173 the lowest available file descriptor, and end up in
6174 fildes[0]. This can happen for stdin and stderr as well,
6175 but stdout is more important -- it will cause no output
6176 to be generated from this command. */
6177 if ((fildes[1] != fileno (stdin)) &&
6178 (fildes[1] != fileno (stdout)) &&
6179 (fildes[1] != fileno (stderr)))
6180 close (fildes[1]);
6181
6182 if ((fildes[0] != fileno (stdin)) &&
6183 (fildes[0] != fileno (stdout)) &&
6184 (fildes[0] != fileno (stderr)))
6185 close (fildes[0]);
6186
495aee44
CR
6187#ifdef __CYGWIN__
6188 /* Let stdio know the fd may have changed from text to binary mode, and
6189 make sure to preserve stdout line buffering. */
6190 freopen (NULL, "w", stdout);
6191 sh_setlinebuf (stdout);
6192#endif /* __CYGWIN__ */
6193
ccc6cda3 6194 /* This is a subshell environment. */
28ef6c31 6195 subshell_environment |= SUBSHELL_COMSUB;
ccc6cda3 6196
a0c0a00f
CR
6197 /* Many shells do not appear to inherit the -v option for command
6198 substitutions. */
6199 change_flag ('v', FLAG_OFF);
6200
6201 /* When inherit_errexit option is not enabled, command substitution does
6202 not inherit the -e flag. It is enabled when Posix mode is enabled */
6203 if (inherit_errexit == 0)
ac50fbac
CR
6204 {
6205 builtin_ignoring_errexit = 0;
6206 change_flag ('e', FLAG_OFF);
ac50fbac 6207 }
a0c0a00f
CR
6208 set_shellopts ();
6209
6210 /* If we are expanding a redirection, we can dispose of any temporary
6211 environment we received, since redirections are not supposed to have
6212 access to the temporary environment. We will have to see whether this
6213 affects temporary environments supplied to `eval', but the temporary
6214 environment gets copied to builtin_env at some point. */
6215 if (expanding_redir)
6216 {
6217 flush_temporary_env ();
6218 expanding_redir = 0;
6219 }
726f6388
JA
6220
6221 remove_quoted_escapes (string);
6222
ccc6cda3 6223 startup_state = 2; /* see if we can avoid a fork */
726f6388
JA
6224 /* Give command substitution a place to jump back to on failure,
6225 so we don't go back up to main (). */
ac50fbac 6226 result = setjmp_nosigs (top_level);
726f6388 6227
bb70624e
JA
6228 /* If we're running a command substitution inside a shell function,
6229 trap `return' so we don't return from the function in the subshell
6230 and go off to never-never land. */
6231 if (result == 0 && return_catch_flag)
ac50fbac 6232 function_value = setjmp_nosigs (return_catch);
bb70624e
JA
6233 else
6234 function_value = 0;
6235
b80f6443
JA
6236 if (result == ERREXIT)
6237 rc = last_command_exit_value;
6238 else if (result == EXITPROG)
6239 rc = last_command_exit_value;
726f6388 6240 else if (result)
b80f6443 6241 rc = EXECUTION_FAILURE;
bb70624e 6242 else if (function_value)
b80f6443 6243 rc = return_catch_value;
726f6388 6244 else
b80f6443
JA
6245 {
6246 subshell_level++;
6247 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
6248 subshell_level--;
6249 }
6250
6251 last_command_exit_value = rc;
6252 rc = run_exit_trap ();
f1be666c
JA
6253#if defined (PROCESS_SUBSTITUTION)
6254 unlink_fifo_list ();
6255#endif
b80f6443 6256 exit (rc);
726f6388
JA
6257 }
6258 else
6259 {
726f6388
JA
6260#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6261 close_pgrp_pipe ();
6262#endif /* JOB_CONTROL && PGRP_PIPE */
6263
6264 close (fildes[1]);
6265
3185942a
JA
6266 tflag = 0;
6267 istring = read_comsub (fildes[0], quoted, &tflag);
ccc6cda3 6268
726f6388
JA
6269 close (fildes[0]);
6270
b72432fd 6271 current_command_subst_pid = pid;
726f6388
JA
6272 last_command_exit_value = wait_for (pid);
6273 last_command_subst_pid = pid;
6274 last_made_pid = old_pid;
6275
6276#if defined (JOB_CONTROL)
6277 /* If last_command_exit_value > 128, then the substituted command
6278 was terminated by a signal. If that signal was SIGINT, then send
6279 SIGINT to ourselves. This will break out of loops, for instance. */
b80f6443 6280 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
726f6388
JA
6281 kill (getpid (), SIGINT);
6282
6283 /* wait_for gives the terminal back to shell_pgrp. If some other
cce855bc
JA
6284 process group should have it, give it away to that group here.
6285 pipeline_pgrp is non-zero only while we are constructing a
ac50fbac 6286 pipeline, so what we are concerned about is whether or not that
cce855bc 6287 pipeline was started in the background. A pipeline started in
a0c0a00f
CR
6288 the background should never get the tty back here. We duplicate
6289 the conditions that wait_for tests to make sure we only give
6290 the terminal back to pipeline_pgrp under the conditions that wait_for
6291 gave it to shell_pgrp. If wait_for doesn't mess with the terminal
6292 pgrp, we should not either. */
6293 if (interactive && pipeline_pgrp != (pid_t)0 && running_in_background == 0 &&
6294 (subshell_environment & (SUBSHELL_ASYNC|SUBSHELL_PIPE)) == 0)
28ef6c31 6295 give_terminal_to (pipeline_pgrp, 0);
726f6388
JA
6296#endif /* JOB_CONTROL */
6297
3185942a
JA
6298 ret = alloc_word_desc ();
6299 ret->word = istring;
6300 ret->flags = tflag;
6301
6302 return ret;
726f6388
JA
6303 }
6304}
6305
6306/********************************************************
6307 * *
6308 * Utility functions for parameter expansion *
6309 * *
6310 ********************************************************/
6311
ccc6cda3 6312#if defined (ARRAY_VARS)
ccc6cda3 6313
f73dda09 6314static arrayind_t
ccc6cda3
JA
6315array_length_reference (s)
6316 char *s;
6317{
f73dda09
JA
6318 int len;
6319 arrayind_t ind;
3185942a 6320 char *akey;
f73dda09 6321 char *t, c;
ccc6cda3 6322 ARRAY *array;
495aee44 6323 HASH_TABLE *h;
ccc6cda3
JA
6324 SHELL_VAR *var;
6325
6326 var = array_variable_part (s, &t, &len);
726f6388 6327
ccc6cda3
JA
6328 /* If unbound variables should generate an error, report one and return
6329 failure. */
ac50fbac 6330 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
726f6388 6331 {
f73dda09 6332 c = *--t;
ccc6cda3 6333 *t = '\0';
0001803f 6334 last_command_exit_value = EXECUTION_FAILURE;
7117c2d2 6335 err_unboundvar (s);
f73dda09 6336 *t = c;
ccc6cda3 6337 return (-1);
726f6388 6338 }
ac50fbac 6339 else if (var == 0 || invisible_p (var))
ccc6cda3 6340 return 0;
726f6388 6341
28ef6c31
JA
6342 /* We support a couple of expansions for variables that are not arrays.
6343 We'll return the length of the value for v[0], and 1 for v[@] or
6344 v[*]. Return 0 for everything else. */
6345
6346 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
495aee44 6347 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
726f6388 6348
ccc6cda3 6349 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
ccc6cda3 6350 {
3185942a 6351 if (assoc_p (var))
495aee44 6352 return (h ? assoc_num_elements (h) : 0);
3185942a 6353 else if (array_p (var))
495aee44 6354 return (array ? array_num_elements (array) : 0);
3185942a 6355 else
495aee44 6356 return (var_isset (var) ? 1 : 0);
ccc6cda3 6357 }
ccc6cda3 6358
3185942a
JA
6359 if (assoc_p (var))
6360 {
6361 t[len - 1] = '\0';
6362 akey = expand_assignment_string_to_string (t, 0); /* [ */
6363 t[len - 1] = ']';
6364 if (akey == 0 || *akey == 0)
6365 {
6366 err_badarraysub (t);
ac50fbac 6367 FREE (akey);
3185942a
JA
6368 return (-1);
6369 }
6370 t = assoc_reference (assoc_cell (var), akey);
ac50fbac 6371 free (akey);
3185942a 6372 }
28ef6c31 6373 else
3185942a 6374 {
ac50fbac
CR
6375 ind = array_expand_index (var, t, len);
6376 /* negative subscripts to indexed arrays count back from end */
6377 if (var && array_p (var) && ind < 0)
6378 ind = array_max_index (array_cell (var)) + 1 + ind;
3185942a
JA
6379 if (ind < 0)
6380 {
6381 err_badarraysub (t);
6382 return (-1);
6383 }
6384 if (array_p (var))
6385 t = array_reference (array, ind);
6386 else
6387 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6388 }
28ef6c31 6389
f1be666c 6390 len = MB_STRLEN (t);
ccc6cda3 6391 return (len);
726f6388 6392}
ccc6cda3 6393#endif /* ARRAY_VARS */
726f6388
JA
6394
6395static int
6396valid_brace_expansion_word (name, var_is_special)
6397 char *name;
6398 int var_is_special;
6399{
f73dda09 6400 if (DIGIT (*name) && all_digits (name))
726f6388
JA
6401 return 1;
6402 else if (var_is_special)
6403 return 1;
ccc6cda3 6404#if defined (ARRAY_VARS)
a0c0a00f 6405 else if (valid_array_reference (name, 0))
ccc6cda3
JA
6406 return 1;
6407#endif /* ARRAY_VARS */
726f6388
JA
6408 else if (legal_identifier (name))
6409 return 1;
6410 else
6411 return 0;
6412}
ccc6cda3 6413
b80f6443
JA
6414static int
6415chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
6416 char *name;
6417 int quoted;
6418 int *quoted_dollar_atp, *contains_dollar_at;
6419{
6420 char *temp1;
6421
6422 if (name == 0)
6423 {
6424 if (quoted_dollar_atp)
6425 *quoted_dollar_atp = 0;
6426 if (contains_dollar_at)
6427 *contains_dollar_at = 0;
6428 return 0;
6429 }
6430
6431 /* check for $@ and $* */
6432 if (name[0] == '@' && name[1] == 0)
6433 {
6434 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6435 *quoted_dollar_atp = 1;
6436 if (contains_dollar_at)
6437 *contains_dollar_at = 1;
6438 return 1;
6439 }
6440 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6441 {
6442 if (contains_dollar_at)
6443 *contains_dollar_at = 1;
6444 return 1;
6445 }
6446
6447 /* Now check for ${array[@]} and ${array[*]} */
6448#if defined (ARRAY_VARS)
a0c0a00f 6449 else if (valid_array_reference (name, 0))
b80f6443 6450 {
0001803f 6451 temp1 = mbschr (name, '[');
b80f6443
JA
6452 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
6453 {
6454 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6455 *quoted_dollar_atp = 1;
6456 if (contains_dollar_at)
6457 *contains_dollar_at = 1;
6458 return 1;
6459 } /* [ */
6460 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6461 which should result in separate words even when IFS is unset. */
6462 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
6463 {
6464 if (contains_dollar_at)
6465 *contains_dollar_at = 1;
6466 return 1;
6467 }
6468 }
6469#endif
6470 return 0;
6471}
6472
726f6388
JA
6473/* Parameter expand NAME, and return a new string which is the expansion,
6474 or NULL if there was no expansion.
6475 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6476 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6477 NAME was found inside of a double-quoted expression. */
95732b49 6478static WORD_DESC *
495aee44 6479parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
726f6388 6480 char *name;
89a92869 6481 int var_is_special, quoted, pflags;
495aee44 6482 arrayind_t *indp;
726f6388 6483{
95732b49 6484 WORD_DESC *ret;
ccc6cda3 6485 char *temp, *tt;
7117c2d2 6486 intmax_t arg_index;
ccc6cda3 6487 SHELL_VAR *var;
f1be666c 6488 int atype, rflags;
495aee44 6489 arrayind_t ind;
726f6388 6490
95732b49
JA
6491 ret = 0;
6492 temp = 0;
f1be666c 6493 rflags = 0;
95732b49 6494
495aee44
CR
6495 if (indp)
6496 *indp = INTMAX_MIN;
6497
95732b49 6498 /* Handle multiple digit arguments, as in ${11}. */
f73dda09 6499 if (legal_number (name, &arg_index))
7117c2d2
JA
6500 {
6501 tt = get_dollar_var_value (arg_index);
b80f6443
JA
6502 if (tt)
6503 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6504 ? quote_string (tt)
6505 : quote_escapes (tt);
6506 else
6507 temp = (char *)NULL;
7117c2d2
JA
6508 FREE (tt);
6509 }
726f6388
JA
6510 else if (var_is_special) /* ${@} */
6511 {
cce855bc 6512 int sindex;
f73dda09 6513 tt = (char *)xmalloc (2 + strlen (name));
cce855bc 6514 tt[sindex = 0] = '$';
726f6388 6515 strcpy (tt + 1, name);
7117c2d2 6516
95732b49 6517 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
89a92869 6518 (int *)NULL, (int *)NULL, pflags);
cce855bc 6519 free (tt);
726f6388 6520 }
ccc6cda3 6521#if defined (ARRAY_VARS)
a0c0a00f 6522 else if (valid_array_reference (name, 0))
ccc6cda3 6523 {
ac50fbac 6524expand_arrayref:
ac50fbac 6525 if (pflags & PF_ASSIGNRHS)
a0c0a00f
CR
6526 {
6527 var = array_variable_part (name, &tt, (int *)0);
6528 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
6529 {
6530 /* Only treat as double quoted if array variable */
6531 if (var && (array_p (var) || assoc_p (var)))
6532 /* XXX - bash-4.4/bash-5.0 pass AV_ASSIGNRHS */
6533 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6534 else
6535 temp = array_value (name, quoted, 0, &atype, &ind);
6536 }
ac50fbac
CR
6537 else
6538 temp = array_value (name, quoted, 0, &atype, &ind);
a0c0a00f 6539 }
ac50fbac
CR
6540 else
6541 temp = array_value (name, quoted, 0, &atype, &ind);
7117c2d2 6542 if (atype == 0 && temp)
495aee44
CR
6543 {
6544 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6545 ? quote_string (temp)
6546 : quote_escapes (temp);
6547 rflags |= W_ARRAYIND;
6548 if (indp)
6549 *indp = ind;
6550 }
f1be666c
JA
6551 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6552 rflags |= W_HASQUOTEDNULL;
ccc6cda3
JA
6553 }
6554#endif
6555 else if (var = find_variable (name))
6556 {
7117c2d2 6557 if (var_isset (var) && invisible_p (var) == 0)
28ef6c31 6558 {
ccc6cda3 6559#if defined (ARRAY_VARS)
3185942a
JA
6560 if (assoc_p (var))
6561 temp = assoc_reference (assoc_cell (var), "0");
6562 else if (array_p (var))
6563 temp = array_reference (array_cell (var), 0);
6564 else
6565 temp = value_cell (var);
ccc6cda3
JA
6566#else
6567 temp = value_cell (var);
6568#endif
6569
6570 if (temp)
b80f6443
JA
6571 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6572 ? quote_string (temp)
6573 : quote_escapes (temp);
28ef6c31 6574 }
ccc6cda3
JA
6575 else
6576 temp = (char *)NULL;
6577 }
a0c0a00f 6578 else if (var = find_variable_last_nameref (name, 0))
ac50fbac
CR
6579 {
6580 temp = nameref_cell (var);
6581#if defined (ARRAY_VARS)
6582 /* Handle expanding nameref whose value is x[n] */
a0c0a00f 6583 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
6584 {
6585 name = temp;
6586 goto expand_arrayref;
6587 }
6588 else
6589#endif
6590 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6591 if (temp && *temp && legal_identifier (temp) == 0)
6592 {
6593 last_command_exit_value = EXECUTION_FAILURE;
6594 report_error (_("%s: invalid variable name for name reference"), temp);
6595 temp = &expand_param_error;
6596 }
6597 else
6598 temp = (char *)NULL;
6599 }
726f6388 6600 else
ccc6cda3 6601 temp = (char *)NULL;
726f6388 6602
95732b49
JA
6603 if (ret == 0)
6604 {
6605 ret = alloc_word_desc ();
6606 ret->word = temp;
f1be666c 6607 ret->flags |= rflags;
95732b49
JA
6608 }
6609 return ret;
726f6388
JA
6610}
6611
ac50fbac
CR
6612static char *
6613parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
ccc6cda3 6614 char *name;
ac50fbac 6615 int var_is_special, quoted, find_nameref;
ccc6cda3
JA
6616{
6617 char *temp, *t;
95732b49 6618 WORD_DESC *w;
ac50fbac
CR
6619 SHELL_VAR *v;
6620
a0c0a00f 6621 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
ac50fbac
CR
6622 nameref_p (v) && (t = nameref_cell (v)) && *t)
6623 return (savestring (t));
ccc6cda3 6624
ac50fbac
CR
6625 /* If var_is_special == 0, and name is not an array reference, this does
6626 more expansion than necessary. It should really look up the variable's
6627 value and not try to expand it. */
495aee44 6628 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
95732b49 6629 t = w->word;
b80f6443
JA
6630 /* Have to dequote here if necessary */
6631 if (t)
6632 {
6633 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6634 ? dequote_string (t)
6635 : dequote_escapes (t);
6636 free (t);
6637 t = temp;
6638 }
95732b49
JA
6639 dispose_word_desc (w);
6640
ac50fbac
CR
6641 return t;
6642}
6643
6644/* Expand an indirect reference to a variable: ${!NAME} expands to the
6645 value of the variable whose name is the value of NAME. */
6646static WORD_DESC *
6647parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
6648 char *name;
6649 int var_is_special, quoted;
6650 int *quoted_dollar_atp, *contains_dollar_at;
6651{
6652 char *temp, *t;
6653 WORD_DESC *w;
6654 SHELL_VAR *v;
6655
6656 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6657 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6658 bash performs an indirect lookup on foo[0] and expands the result;
6659 ksh93 expands bar[0]. We could do that here -- there are enough usable
6660 primitives to do that -- but do not at this point. */
a0c0a00f 6661 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
ac50fbac
CR
6662 {
6663 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6664 {
6665 w = alloc_word_desc ();
6666 w->word = savestring (t);
6667 w->flags = 0;
6668 return w;
6669 }
6670 }
6671
6672 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6673
b80f6443 6674 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
ccc6cda3 6675 if (t == 0)
95732b49
JA
6676 return (WORD_DESC *)NULL;
6677
a0c0a00f
CR
6678 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
6679 {
6680 report_error (_("%s: bad substitution"), t);
6681 free (t);
6682 w = alloc_word_desc ();
6683 w->word = &expand_param_error;
6684 w->flags = 0;
6685 return (w);
6686 }
6687
495aee44 6688 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
ccc6cda3 6689 free (t);
95732b49
JA
6690
6691 return w;
ccc6cda3
JA
6692}
6693
726f6388
JA
6694/* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
6695 depending on the value of C, the separating character. C can be one of
ccc6cda3
JA
6696 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
6697 between double quotes. */
95732b49 6698static WORD_DESC *
a0c0a00f 6699parameter_brace_expand_rhs (name, value, c, quoted, pflags, qdollaratp, hasdollarat)
726f6388 6700 char *name, *value;
a0c0a00f 6701 int c, quoted, pflags, *qdollaratp, *hasdollarat;
726f6388 6702{
95732b49 6703 WORD_DESC *w;
726f6388 6704 WORD_LIST *l;
a0c0a00f
CR
6705 char *t, *t1, *temp, *vname;
6706 int l_hasdollat, sindex;
726f6388 6707
a0c0a00f 6708/*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
ccc6cda3
JA
6709 /* If the entire expression is between double quotes, we want to treat
6710 the value as a double-quoted string, with the exception that we strip
3185942a 6711 embedded unescaped double quotes (for sh backwards compatibility). */
95732b49 6712 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
726f6388 6713 {
a0c0a00f
CR
6714 sindex = 0;
6715 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
726f6388 6716 }
95732b49
JA
6717 else
6718 temp = value;
ccc6cda3 6719
95732b49 6720 w = alloc_word_desc ();
a0c0a00f 6721 l_hasdollat = 0;
ccc6cda3 6722 /* XXX was 0 not quoted */
a0c0a00f 6723 l = *temp ? expand_string_for_rhs (temp, quoted, &l_hasdollat, (int *)NULL)
ccc6cda3
JA
6724 : (WORD_LIST *)0;
6725 if (hasdollarat)
a0c0a00f 6726 *hasdollarat = l_hasdollat || (l && l->next);
95732b49
JA
6727 if (temp != value)
6728 free (temp);
726f6388
JA
6729 if (l)
6730 {
a0c0a00f
CR
6731 /* If l->next is not null, we know that TEMP contained "$@", since that
6732 is the only expansion that creates more than one word. */
6733 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
6734 {
6735/*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
6736 *qdollaratp = 1;
6737 }
6738
ccc6cda3 6739 /* The expansion of TEMP returned something. We need to treat things
a0c0a00f
CR
6740 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
6741 the individual words have already been quoted. We need to turn them
b80f6443
JA
6742 into a string with the words separated by the first character of
6743 $IFS without any additional quoting, so string_list_dollar_at won't
a0c0a00f
CR
6744 do the right thing. If IFS is null, we want "$@" to split into
6745 separate arguments, not be concatenated, so we use string_list_internal
6746 and mark the word to be split on spaces later. We use
6747 string_list_dollar_star for "$@" otherwise. */
6748 if (l->next && ifs_is_null)
6749 {
6750 temp = string_list_internal (l, " ");
6751 w->flags |= W_SPLITSPACE;
6752 }
6753 else
6754 temp = (l_hasdollat || l->next) ? string_list_dollar_star (l) : string_list (l);
b80f6443 6755
cd110fdf
CR
6756 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
6757 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
6758 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
6759 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6760 (which is more paranoia than anything else), we need to return the
6761 quoted null string and set the flags to indicate it. */
ac50fbac 6762 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
cd110fdf
CR
6763 {
6764 w->flags |= W_HASQUOTEDNULL;
a0c0a00f
CR
6765/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
6766 /* If we return a quoted null with L_HASDOLLARAT, we either have a
6767 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
6768 parameters or a quoted expansion of "$@" with $1 == ''. In either
6769 case, we don't want to enable special handling of $@. */
6770 if (qdollaratp && l_hasdollat)
6771 *qdollaratp = 0;
cd110fdf 6772 }
726f6388
JA
6773 dispose_words (l);
6774 }
a0c0a00f 6775 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
726f6388 6776 {
a0c0a00f
CR
6777 /* Posix interp 221 changed the rules on this. The idea is that
6778 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
6779 when foo and xxx are unset. The problem is that it's not in any
6780 way backwards compatible and few other shells do it. We're eventually
6781 going to try and split the difference (heh) a little bit here. */
6782 /* l_hasdollat == 1 means we saw a quoted dollar at. */
6783
ccc6cda3
JA
6784 /* The brace expansion occurred between double quotes and there was
6785 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
7117c2d2 6786 it does not expand to anything. In this case, we want to return
a0c0a00f 6787 a quoted empty string. Posix interp 888 */
0628567a 6788 temp = make_quoted_char ('\0');
95732b49 6789 w->flags |= W_HASQUOTEDNULL;
a0c0a00f 6790/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
726f6388
JA
6791 }
6792 else
6793 temp = (char *)NULL;
6794
6795 if (c == '-' || c == '+')
95732b49
JA
6796 {
6797 w->word = temp;
6798 return w;
6799 }
726f6388
JA
6800
6801 /* c == '=' */
ccc6cda3 6802 t = temp ? savestring (temp) : savestring ("");
726f6388
JA
6803 t1 = dequote_string (t);
6804 free (t);
a0c0a00f
CR
6805
6806 /* bash-4.4/5.0 */
6807 vname = name;
6808 if (*name == '!' &&
6809 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
6810 {
6811 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
6812 if (vname == 0 || *vname == 0)
6813 {
6814 report_error (_("%s: invalid indirect expansion"), name);
6815 free (vname);
6816 dispose_word (w);
6817 return &expand_wdesc_error;
6818 }
6819 if (legal_identifier (vname) == 0)
6820 {
6821 report_error (_("%s: invalid variable name"), vname);
6822 free (vname);
6823 dispose_word (w);
6824 return &expand_wdesc_error;
6825 }
6826 }
6827
b80f6443 6828#if defined (ARRAY_VARS)
a0c0a00f
CR
6829 if (valid_array_reference (vname, 0))
6830 assign_array_element (vname, t1, 0);
b80f6443
JA
6831 else
6832#endif /* ARRAY_VARS */
a0c0a00f
CR
6833 bind_variable (vname, t1, 0);
6834
6835 stupidly_hack_special_variables (vname);
6836
6837 if (vname != name)
6838 free (vname);
95732b49 6839
495aee44
CR
6840 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6841 free (temp);
6842
6843 w->word = t1;
95732b49 6844 return w;
726f6388
JA
6845}
6846
6847/* Deal with the right hand side of a ${name:?value} expansion in the case
6848 that NAME is null or not set. If VALUE is non-null it is expanded and
6849 used as the error message to print, otherwise a standard message is
6850 printed. */
6851static void
6852parameter_brace_expand_error (name, value)
6853 char *name, *value;
6854{
ccc6cda3
JA
6855 WORD_LIST *l;
6856 char *temp;
6857
ac50fbac 6858 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
726f6388
JA
6859 if (value && *value)
6860 {
95732b49 6861 l = expand_string (value, 0);
ccc6cda3
JA
6862 temp = string_list (l);
6863 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6864 FREE (temp);
726f6388
JA
6865 dispose_words (l);
6866 }
6867 else
b80f6443 6868 report_error (_("%s: parameter null or not set"), name);
726f6388
JA
6869
6870 /* Free the data we have allocated during this expansion, since we
6871 are about to longjmp out. */
6872 free (name);
6873 FREE (value);
6874}
6875
6876/* Return 1 if NAME is something for which parameter_brace_expand_length is
6877 OK to do. */
6878static int
6879valid_length_expression (name)
6880 char *name;
6881{
28ef6c31 6882 return (name[1] == '\0' || /* ${#} */
f73dda09
JA
6883 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6884 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
ccc6cda3 6885#if defined (ARRAY_VARS)
a0c0a00f 6886 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
ccc6cda3 6887#endif
726f6388
JA
6888 legal_identifier (name + 1)); /* ${#PS1} */
6889}
6890
6891/* Handle the parameter brace expansion that requires us to return the
6892 length of a parameter. */
7117c2d2 6893static intmax_t
726f6388
JA
6894parameter_brace_expand_length (name)
6895 char *name;
6896{
ccc6cda3 6897 char *t, *newname;
7117c2d2 6898 intmax_t number, arg_index;
ccc6cda3
JA
6899 WORD_LIST *list;
6900#if defined (ARRAY_VARS)
6901 SHELL_VAR *var;
6902#endif
6903
6904 if (name[1] == '\0') /* ${#} */
6905 number = number_of_args ();
a0c0a00f 6906 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
cce855bc 6907 number = number_of_args ();
f73dda09 6908 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
cce855bc
JA
6909 {
6910 /* Take the lengths of some of the shell's special parameters. */
6911 switch (name[1])
6912 {
6913 case '-':
6914 t = which_set_flags ();
6915 break;
6916 case '?':
6917 t = itos (last_command_exit_value);
6918 break;
6919 case '$':
6920 t = itos (dollar_dollar_pid);
6921 break;
6922 case '!':
6923 if (last_asynchronous_pid == NO_PID)
495aee44 6924 t = (char *)NULL; /* XXX - error if set -u set? */
cce855bc 6925 else
f73dda09 6926 t = itos (last_asynchronous_pid);
cce855bc
JA
6927 break;
6928 case '#':
6929 t = itos (number_of_args ());
6930 break;
6931 }
6932 number = STRLEN (t);
6933 FREE (t);
6934 }
ccc6cda3 6935#if defined (ARRAY_VARS)
a0c0a00f 6936 else if (valid_array_reference (name + 1, 0))
ccc6cda3
JA
6937 number = array_length_reference (name + 1);
6938#endif /* ARRAY_VARS */
cce855bc 6939 else
ccc6cda3
JA
6940 {
6941 number = 0;
6942
f73dda09 6943 if (legal_number (name + 1, &arg_index)) /* ${#1} */
ccc6cda3 6944 {
f73dda09 6945 t = get_dollar_var_value (arg_index);
495aee44
CR
6946 if (t == 0 && unbound_vars_is_error)
6947 return INTMAX_MIN;
eb873671 6948 number = MB_STRLEN (t);
ccc6cda3
JA
6949 FREE (t);
6950 }
6951#if defined (ARRAY_VARS)
3185942a 6952 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
ccc6cda3 6953 {
3185942a
JA
6954 if (assoc_p (var))
6955 t = assoc_reference (assoc_cell (var), "0");
6956 else
6957 t = array_reference (array_cell (var), 0);
495aee44
CR
6958 if (t == 0 && unbound_vars_is_error)
6959 return INTMAX_MIN;
eb873671 6960 number = MB_STRLEN (t);
ccc6cda3
JA
6961 }
6962#endif
6963 else /* ${#PS1} */
6964 {
6965 newname = savestring (name);
6966 newname[0] = '$';
6967 list = expand_string (newname, Q_DOUBLE_QUOTES);
6968 t = list ? string_list (list) : (char *)NULL;
6969 free (newname);
6970 if (list)
6971 dispose_words (list);
6972
495aee44 6973 number = t ? MB_STRLEN (t) : 0;
ccc6cda3
JA
6974 FREE (t);
6975 }
6976 }
ccc6cda3
JA
6977
6978 return (number);
6979}
6980
28ef6c31
JA
6981/* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6982 so we do some ad-hoc parsing of an arithmetic expression to find
6983 the first DELIM, instead of using strchr(3). Two rules:
6984 1. If the substring contains a `(', read until closing `)'.
6985 2. If the substring contains a `?', read past one `:' for each `?'.
a0c0a00f 6986 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
28ef6c31
JA
6987*/
6988
6989static char *
6990skiparith (substr, delim)
6991 char *substr;
6992 int delim;
6993{
a0c0a00f
CR
6994 int i;
6995 char delims[2];
28ef6c31 6996
a0c0a00f
CR
6997 delims[0] = delim;
6998 delims[1] = '\0';
7117c2d2 6999
a0c0a00f 7000 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
7117c2d2 7001 return (substr + i);
28ef6c31
JA
7002}
7003
ccc6cda3
JA
7004/* Verify and limit the start and end of the desired substring. If
7005 VTYPE == 0, a regular shell variable is being used; if it is 1,
cce855bc 7006 then the positional parameters are being used; if it is 2, then
e8ce775d
JA
7007 VALUE is really a pointer to an array variable that should be used.
7008 Return value is 1 if both values were OK, 0 if there was a problem
7009 with an invalid expression, or -1 if the values were out of range. */
ccc6cda3 7010static int
3185942a
JA
7011verify_substring_values (v, value, substr, vtype, e1p, e2p)
7012 SHELL_VAR *v;
ccc6cda3 7013 char *value, *substr;
f73dda09 7014 int vtype;
7117c2d2 7015 intmax_t *e1p, *e2p;
ccc6cda3 7016{
bb70624e 7017 char *t, *temp1, *temp2;
f73dda09
JA
7018 arrayind_t len;
7019 int expok;
ccc6cda3
JA
7020#if defined (ARRAY_VARS)
7021 ARRAY *a;
3185942a 7022 HASH_TABLE *h;
ccc6cda3
JA
7023#endif
7024
28ef6c31
JA
7025 /* duplicate behavior of strchr(3) */
7026 t = skiparith (substr, ':');
7027 if (*t && *t == ':')
7117c2d2 7028 *t = '\0';
28ef6c31
JA
7029 else
7030 t = (char *)0;
f73dda09 7031
0628567a 7032 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
d166f048 7033 *e1p = evalexp (temp1, &expok);
ccc6cda3 7034 free (temp1);
d166f048
JA
7035 if (expok == 0)
7036 return (0);
ccc6cda3 7037
f73dda09 7038 len = -1; /* paranoia */
ccc6cda3
JA
7039 switch (vtype)
7040 {
7041 case VT_VARIABLE:
d166f048 7042 case VT_ARRAYMEMBER:
eb873671 7043 len = MB_STRLEN (value);
ccc6cda3
JA
7044 break;
7045 case VT_POSPARMS:
7046 len = number_of_args () + 1;
3185942a
JA
7047 if (*e1p == 0)
7048 len++; /* add one arg if counting from $0 */
ccc6cda3
JA
7049 break;
7050#if defined (ARRAY_VARS)
7051 case VT_ARRAYVAR:
eb873671 7052 /* For arrays, the first value deals with array indices. Negative
3185942a
JA
7053 offsets count from one past the array's maximum index. Associative
7054 arrays treat the number of elements as the maximum index. */
7055 if (assoc_p (v))
7056 {
7057 h = assoc_cell (v);
7058 len = assoc_num_elements (h) + (*e1p < 0);
7059 }
7060 else
7061 {
7062 a = (ARRAY *)value;
7063 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7064 }
ccc6cda3
JA
7065 break;
7066#endif
7067 }
7068
f73dda09
JA
7069 if (len == -1) /* paranoia */
7070 return -1;
7071
ccc6cda3
JA
7072 if (*e1p < 0) /* negative offsets count from end */
7073 *e1p += len;
7074
eb873671 7075 if (*e1p > len || *e1p < 0)
e8ce775d 7076 return (-1);
d166f048 7077
b80f6443
JA
7078#if defined (ARRAY_VARS)
7079 /* For arrays, the second offset deals with the number of elements. */
7080 if (vtype == VT_ARRAYVAR)
3185942a 7081 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
b80f6443
JA
7082#endif
7083
ccc6cda3
JA
7084 if (t)
7085 {
7086 t++;
bb70624e 7087 temp2 = savestring (t);
0628567a 7088 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
bb70624e 7089 free (temp2);
ccc6cda3 7090 t[-1] = ':';
d166f048 7091 *e2p = evalexp (temp1, &expok);
ccc6cda3 7092 free (temp1);
d166f048 7093 if (expok == 0)
28ef6c31 7094 return (0);
ac50fbac 7095#if 1
495aee44 7096 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
ac50fbac
CR
7097#else
7098 /* bash-4.3: allow positional parameter length < 0 to count backwards
7099 from end of positional parameters */
7100 if (vtype == VT_ARRAYVAR && *e2p < 0)
7101#endif
28ef6c31 7102 {
b80f6443 7103 internal_error (_("%s: substring expression < 0"), t);
ccc6cda3 7104 return (0);
28ef6c31 7105 }
b80f6443
JA
7106#if defined (ARRAY_VARS)
7107 /* In order to deal with sparse arrays, push the intelligence about how
7108 to deal with the number of elements desired down to the array-
7109 specific functions. */
7110 if (vtype != VT_ARRAYVAR)
7111#endif
7112 {
495aee44
CR
7113 if (*e2p < 0)
7114 {
7115 *e2p += len;
7116 if (*e2p < 0 || *e2p < *e1p)
7117 {
7118 internal_error (_("%s: substring expression < 0"), t);
7119 return (0);
7120 }
7121 }
7122 else
7123 *e2p += *e1p; /* want E2 chars starting at E1 */
b80f6443
JA
7124 if (*e2p > len)
7125 *e2p = len;
7126 }
ccc6cda3
JA
7127 }
7128 else
7129 *e2p = len;
7130
7131 return (1);
7132}
7133
ccc6cda3 7134/* Return the type of variable specified by VARNAME (simple variable,
cce855bc 7135 positional param, or array variable). Also return the value specified
7117c2d2 7136 by VARNAME (value of a variable or a reference to an array element).
495aee44
CR
7137 QUOTED is the standard description of quoting state, using Q_* defines.
7138 FLAGS is currently a set of flags to pass to array_value. If IND is
7139 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7140 passed to array_value so the array index is not computed again.
7117c2d2
JA
7141 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7142 characters in the value are quoted with CTLESC and takes appropriate
7143 steps. For convenience, *VALP is set to the dequoted VALUE. */
ccc6cda3 7144static int
495aee44 7145get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
ccc6cda3 7146 char *varname, *value;
495aee44
CR
7147 arrayind_t ind;
7148 int quoted, flags;
ccc6cda3
JA
7149 SHELL_VAR **varp;
7150 char **valp;
7151{
ac50fbac
CR
7152 int vtype, want_indir;
7153 char *temp, *vname;
7154 WORD_DESC *wd;
ccc6cda3 7155 SHELL_VAR *v;
495aee44 7156 arrayind_t lind;
ccc6cda3 7157
ac50fbac
CR
7158 want_indir = *varname == '!' &&
7159 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7160 || VALID_INDIR_PARAM (varname[1]));
7161 if (want_indir)
7162 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
a0c0a00f 7163 /* XXX - what if vname == 0 || *vname == 0 ? */
ac50fbac
CR
7164 else
7165 vname = varname;
a0c0a00f
CR
7166
7167 if (vname == 0)
7168 {
7169 vtype = VT_VARIABLE;
7170 *varp = (SHELL_VAR *)NULL;
7171 *valp = (char *)NULL;
7172 return (vtype);
7173 }
7174
7117c2d2 7175 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
a0c0a00f 7176 vtype = STR_DOLLAR_AT_STAR (vname);
ac50fbac 7177 if (vtype == VT_POSPARMS && vname[0] == '*')
b80f6443 7178 vtype |= VT_STARSUB;
ccc6cda3
JA
7179 *varp = (SHELL_VAR *)NULL;
7180
7181#if defined (ARRAY_VARS)
a0c0a00f 7182 if (valid_array_reference (vname, 0))
ccc6cda3 7183 {
ac50fbac 7184 v = array_variable_part (vname, &temp, (int *)0);
495aee44
CR
7185 /* If we want to signal array_value to use an already-computed index,
7186 set LIND to that index */
7187 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
ac50fbac
CR
7188 if (v && invisible_p (v))
7189 {
7190 vtype = VT_ARRAYMEMBER;
7191 *varp = (SHELL_VAR *)NULL;
7192 *valp = (char *)NULL;
7193 }
3185942a 7194 if (v && (array_p (v) || assoc_p (v)))
f73dda09
JA
7195 { /* [ */
7196 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
ccc6cda3 7197 {
ac50fbac 7198 /* Callers have to differentiate between indexed and associative */
ccc6cda3 7199 vtype = VT_ARRAYVAR;
b80f6443
JA
7200 if (temp[0] == '*')
7201 vtype |= VT_STARSUB;
3185942a 7202 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
ccc6cda3
JA
7203 }
7204 else
7205 {
d166f048 7206 vtype = VT_ARRAYMEMBER;
ac50fbac 7207 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
ccc6cda3
JA
7208 }
7209 *varp = v;
7210 }
95732b49
JA
7211 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
7212 {
7213 vtype = VT_VARIABLE;
7214 *varp = v;
7215 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7216 *valp = dequote_string (value);
7217 else
7218 *valp = dequote_escapes (value);
7219 }
ccc6cda3 7220 else
3185942a
JA
7221 {
7222 vtype = VT_ARRAYMEMBER;
7223 *varp = v;
ac50fbac 7224 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
3185942a 7225 }
ccc6cda3 7226 }
ac50fbac 7227 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
ccc6cda3 7228 {
7117c2d2 7229 vtype = VT_ARRAYMEMBER;
ccc6cda3 7230 *varp = v;
3185942a 7231 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
ccc6cda3
JA
7232 }
7233 else
7234#endif
b80f6443
JA
7235 {
7236 if (value && vtype == VT_VARIABLE)
7237 {
a0c0a00f 7238 *varp = find_variable (vname);
b80f6443
JA
7239 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7240 *valp = dequote_string (value);
7241 else
7242 *valp = dequote_escapes (value);
7243 }
7244 else
7245 *valp = value;
7246 }
ccc6cda3 7247
ac50fbac
CR
7248 if (want_indir)
7249 free (vname);
7250
ccc6cda3
JA
7251 return vtype;
7252}
7253
cce855bc
JA
7254/******************************************************/
7255/* */
7256/* Functions to extract substrings of variable values */
7257/* */
7258/******************************************************/
7259
b80f6443
JA
7260#if defined (HANDLE_MULTIBYTE)
7261/* Character-oriented rather than strictly byte-oriented substrings. S and
7262 E, rather being strict indices into STRING, indicate character (possibly
7263 multibyte character) positions that require calculation.
7264 Used by the ${param:offset[:length]} expansion. */
7265static char *
7266mb_substring (string, s, e)
7267 char *string;
7268 int s, e;
7269{
7270 char *tt;
a0c0a00f
CR
7271 int start, stop, i;
7272 size_t slen;
b80f6443
JA
7273 DECLARE_MBSTATE;
7274
7275 start = 0;
95732b49
JA
7276 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
7277 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
b80f6443
JA
7278
7279 i = s;
7280 while (string[start] && i--)
7281 ADVANCE_CHAR (string, slen, start);
7282 stop = start;
7283 i = e - s;
7284 while (string[stop] && i--)
7285 ADVANCE_CHAR (string, slen, stop);
7286 tt = substring (string, start, stop);
7287 return tt;
7288}
7289#endif
7290
ccc6cda3
JA
7291/* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
7292 is `@', use the positional parameters; otherwise, use the value of
7293 VARNAME. If VARNAME is an array variable, use the array elements. */
7294
7295static char *
495aee44
CR
7296parameter_brace_substring (varname, value, ind, substr, quoted, flags)
7297 char *varname, *value;
7298 int ind;
7299 char *substr;
7300 int quoted, flags;
ccc6cda3 7301{
7117c2d2 7302 intmax_t e1, e2;
b80f6443 7303 int vtype, r, starsub;
0628567a 7304 char *temp, *val, *tt, *oname;
ccc6cda3
JA
7305 SHELL_VAR *v;
7306
a0c0a00f 7307 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
ccc6cda3
JA
7308 return ((char *)NULL);
7309
0628567a 7310 oname = this_command_name;
ccc6cda3
JA
7311 this_command_name = varname;
7312
495aee44 7313 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3 7314 if (vtype == -1)
0628567a
JA
7315 {
7316 this_command_name = oname;
7317 return ((char *)NULL);
7318 }
ccc6cda3 7319
b80f6443
JA
7320 starsub = vtype & VT_STARSUB;
7321 vtype &= ~VT_STARSUB;
7322
3185942a 7323 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
0628567a 7324 this_command_name = oname;
e8ce775d 7325 if (r <= 0)
ac50fbac
CR
7326 {
7327 if (vtype == VT_VARIABLE)
7328 FREE (val);
7329 return ((r == 0) ? &expand_param_error : (char *)NULL);
7330 }
ccc6cda3
JA
7331
7332 switch (vtype)
7333 {
7334 case VT_VARIABLE:
d166f048 7335 case VT_ARRAYMEMBER:
b80f6443
JA
7336#if defined (HANDLE_MULTIBYTE)
7337 if (MB_CUR_MAX > 1)
7338 tt = mb_substring (val, e1, e2);
7339 else
7340#endif
7117c2d2 7341 tt = substring (val, e1, e2);
b80f6443 7342
7117c2d2
JA
7343 if (vtype == VT_VARIABLE)
7344 FREE (val);
7345 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7346 temp = quote_string (tt);
7347 else
7348 temp = tt ? quote_escapes (tt) : (char *)NULL;
7349 FREE (tt);
ccc6cda3
JA
7350 break;
7351 case VT_POSPARMS:
7117c2d2
JA
7352 tt = pos_params (varname, e1, e2, quoted);
7353 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
7354 {
7355 temp = tt ? quote_escapes (tt) : (char *)NULL;
7356 FREE (tt);
7357 }
7358 else
7359 temp = tt;
ccc6cda3
JA
7360 break;
7361#if defined (ARRAY_VARS)
7362 case VT_ARRAYVAR:
3185942a
JA
7363 if (assoc_p (v))
7364 /* we convert to list and take first e2 elements starting at e1th
7365 element -- officially undefined for now */
7366 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
7367 else
b80f6443
JA
7368 /* We want E2 to be the number of elements desired (arrays can be sparse,
7369 so verify_substring_values just returns the numbers specified and we
7370 rely on array_subrange to understand how to deal with them). */
3185942a 7371 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
f1be666c
JA
7372 /* array_subrange now calls array_quote_escapes as appropriate, so the
7373 caller no longer needs to. */
ccc6cda3
JA
7374 break;
7375#endif
f73dda09
JA
7376 default:
7377 temp = (char *)NULL;
ccc6cda3
JA
7378 }
7379
7380 return temp;
7381}
7382
cce855bc
JA
7383/****************************************************************/
7384/* */
7385/* Functions to perform pattern substitution on variable values */
7386/* */
7387/****************************************************************/
7388
a0c0a00f 7389#if 0 /* Unused */
495aee44
CR
7390static int
7391shouldexp_replacement (s)
7392 char *s;
7393{
7394 register char *p;
7395
7396 for (p = s; p && *p; p++)
7397 {
7398 if (*p == '\\')
7399 p++;
7400 else if (*p == '&')
7401 return 1;
7402 }
7403 return 0;
7404}
a0c0a00f 7405#endif
495aee44 7406
ccc6cda3
JA
7407char *
7408pat_subst (string, pat, rep, mflags)
7409 char *string, *pat, *rep;
7410 int mflags;
7411{
495aee44 7412 char *ret, *s, *e, *str, *rstr, *mstr;
a0c0a00f
CR
7413 int rptr, mtype, rxpand, mlen;
7414 size_t rsize, l, replen, rslen;
495aee44
CR
7415
7416 if (string == 0)
7417 return (savestring (""));
ccc6cda3 7418
b72432fd
JA
7419 mtype = mflags & MATCH_TYPEMASK;
7420
495aee44
CR
7421#if 0 /* bash-4.2 ? */
7422 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
7423#else
7424 rxpand = 0;
7425#endif
7426
b72432fd
JA
7427 /* Special cases:
7428 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
7429 * with REP and return the result.
7430 * 2. A null pattern with mtype == MATCH_END means to append REP to
7431 * STRING and return the result.
a0c0a00f
CR
7432 * 3. A null STRING with a matching pattern means to append REP to
7433 * STRING and return the result.
495aee44 7434 * These don't understand or process `&' in the replacement string.
b72432fd
JA
7435 */
7436 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
7437 {
7438 replen = STRLEN (rep);
495aee44 7439 l = STRLEN (string);
f73dda09 7440 ret = (char *)xmalloc (replen + l + 2);
bb70624e
JA
7441 if (replen == 0)
7442 strcpy (ret, string);
7443 else if (mtype == MATCH_BEG)
b72432fd
JA
7444 {
7445 strcpy (ret, rep);
7446 strcpy (ret + replen, string);
7447 }
7448 else
7449 {
7450 strcpy (ret, string);
7451 strcpy (ret + l, rep);
7452 }
7453 return (ret);
7454 }
a0c0a00f
CR
7455 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
7456 {
7457 replen = STRLEN (rep);
7458 ret = (char *)xmalloc (replen + 1);
7459 if (replen == 0)
7460 ret[0] = '\0';
7461 else
7462 strcpy (ret, rep);
7463 return (ret);
7464 }
b72432fd 7465
f73dda09 7466 ret = (char *)xmalloc (rsize = 64);
ccc6cda3
JA
7467 ret[0] = '\0';
7468
a0c0a00f 7469 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
ccc6cda3
JA
7470 {
7471 if (match_pattern (str, pat, mtype, &s, &e) == 0)
7472 break;
7473 l = s - str;
495aee44 7474
a0c0a00f 7475 if (rep && rxpand)
495aee44 7476 {
a0c0a00f
CR
7477 int x;
7478 mlen = e - s;
7479 mstr = xmalloc (mlen + 1);
495aee44
CR
7480 for (x = 0; x < mlen; x++)
7481 mstr[x] = s[x];
a0c0a00f
CR
7482 mstr[mlen] = '\0';
7483 rstr = strcreplace (rep, '&', mstr, 0);
7484 free (mstr);
7485 rslen = strlen (rstr);
495aee44
CR
7486 }
7487 else
a0c0a00f
CR
7488 {
7489 rstr = rep;
7490 rslen = replen;
7491 }
495aee44
CR
7492
7493 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
ccc6cda3
JA
7494
7495 /* OK, now copy the leading unmatched portion of the string (from
7496 str to s) to ret starting at rptr (the current offset). Then copy
28ef6c31
JA
7497 the replacement string at ret + rptr + (s - str). Increment
7498 rptr (if necessary) and str and go on. */
ccc6cda3
JA
7499 if (l)
7500 {
7501 strncpy (ret + rptr, str, l);
7502 rptr += l;
7503 }
7504 if (replen)
7505 {
495aee44
CR
7506 strncpy (ret + rptr, rstr, rslen);
7507 rptr += rslen;
ccc6cda3
JA
7508 }
7509 str = e; /* e == end of match */
b80f6443 7510
495aee44
CR
7511 if (rstr != rep)
7512 free (rstr);
7513
ccc6cda3 7514 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
28ef6c31 7515 break;
b80f6443
JA
7516
7517 if (s == e)
0001803f
CR
7518 {
7519 /* On a zero-length match, make sure we copy one character, since
7520 we increment one character to avoid infinite recursion. */
7521 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
7522 ret[rptr++] = *str++;
7523 e++; /* avoid infinite recursion on zero-length match */
7524 }
ccc6cda3
JA
7525 }
7526
7527 /* Now copy the unmatched portion of the input string */
495aee44 7528 if (str && *str)
d166f048
JA
7529 {
7530 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
7531 strcpy (ret + rptr, str);
7532 }
ccc6cda3
JA
7533 else
7534 ret[rptr] = '\0';
7535
7536 return ret;
7537}
7538
7539/* Do pattern match and replacement on the positional parameters. */
7540static char *
7541pos_params_pat_subst (string, pat, rep, mflags)
7542 char *string, *pat, *rep;
7543 int mflags;
7544{
7545 WORD_LIST *save, *params;
7546 WORD_DESC *w;
0628567a 7547 char *ret;
3185942a 7548 int pchar, qflags;
ccc6cda3
JA
7549
7550 save = params = list_rest_of_args ();
7551 if (save == 0)
7552 return ((char *)NULL);
7553
7554 for ( ; params; params = params->next)
7555 {
7556 ret = pat_subst (params->word->word, pat, rep, mflags);
95732b49
JA
7557 w = alloc_word_desc ();
7558 w->word = ret ? ret : savestring ("");
ccc6cda3
JA
7559 dispose_word (params->word);
7560 params->word = w;
ccc6cda3
JA
7561 }
7562
3185942a
JA
7563 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7564 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7565
3185942a 7566 ret = string_list_pos_params (pchar, save, qflags);
3185942a 7567
ccc6cda3
JA
7568 dispose_words (save);
7569
7570 return (ret);
7571}
7572
cce855bc
JA
7573/* Perform pattern substitution on VALUE, which is the expansion of
7574 VARNAME. PATSUB is an expression supplying the pattern to match
7575 and the string to substitute. QUOTED is a flags word containing
7576 the type of quoting currently in effect. */
ccc6cda3 7577static char *
a0c0a00f 7578parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
495aee44
CR
7579 char *varname, *value;
7580 int ind;
7581 char *patsub;
a0c0a00f 7582 int quoted, pflags, flags;
ccc6cda3 7583{
3185942a 7584 int vtype, mflags, starsub, delim;
7117c2d2 7585 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
ccc6cda3
JA
7586 SHELL_VAR *v;
7587
7588 if (value == 0)
7589 return ((char *)NULL);
7590
7591 this_command_name = varname;
7592
495aee44 7593 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3
JA
7594 if (vtype == -1)
7595 return ((char *)NULL);
7596
b80f6443
JA
7597 starsub = vtype & VT_STARSUB;
7598 vtype &= ~VT_STARSUB;
7599
ccc6cda3 7600 mflags = 0;
ac50fbac
CR
7601 /* PATSUB is never NULL when this is called. */
7602 if (*patsub == '/')
f1be666c
JA
7603 {
7604 mflags |= MATCH_GLOBREP;
7605 patsub++;
7606 }
7117c2d2
JA
7607
7608 /* Malloc this because expand_string_if_necessary or one of the expansion
7609 functions in its call chain may free it on a substitution error. */
bb70624e 7610 lpatsub = savestring (patsub);
ccc6cda3
JA
7611
7612 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7613 mflags |= MATCH_QUOTED;
7614
b80f6443
JA
7615 if (starsub)
7616 mflags |= MATCH_STARSUB;
7617
a0c0a00f
CR
7618 if (pflags & PF_ASSIGNRHS)
7619 mflags |= MATCH_ASSIGNRHS;
7620
0628567a
JA
7621 /* If the pattern starts with a `/', make sure we skip over it when looking
7622 for the replacement delimiter. */
3185942a
JA
7623 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
7624 if (lpatsub[delim] == '/')
7625 {
7626 lpatsub[delim] = 0;
7627 rep = lpatsub + delim + 1;
7628 }
7629 else
7630 rep = (char *)NULL;
ccc6cda3
JA
7631
7632 if (rep && *rep == '\0')
7633 rep = (char *)NULL;
7634
b80f6443
JA
7635 /* Perform the same expansions on the pattern as performed by the
7636 pattern removal expansions. */
7637 pat = getpattern (lpatsub, quoted, 1);
bb70624e 7638
ccc6cda3 7639 if (rep)
d166f048 7640 {
ac50fbac
CR
7641 /* We want to perform quote removal on the expanded replacement even if
7642 the entire expansion is double-quoted because the parser and string
7643 extraction functions treated quotes in the replacement string as
7644 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
7645 if (shell_compatibility_level > 42)
7646 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
7647 /* This is the bash-4.2 code. */
7648 else if ((mflags & MATCH_QUOTED) == 0)
f73dda09 7649 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
d166f048 7650 else
f73dda09 7651 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
d166f048 7652 }
ccc6cda3 7653
0628567a 7654 /* ksh93 doesn't allow the match specifier to be a part of the expanded
f1be666c
JA
7655 pattern. This is an extension. Make sure we don't anchor the pattern
7656 at the beginning or end of the string if we're doing global replacement,
7657 though. */
ccc6cda3 7658 p = pat;
f1be666c
JA
7659 if (mflags & MATCH_GLOBREP)
7660 mflags |= MATCH_ANY;
0628567a 7661 else if (pat && pat[0] == '#')
ccc6cda3
JA
7662 {
7663 mflags |= MATCH_BEG;
7664 p++;
7665 }
d166f048 7666 else if (pat && pat[0] == '%')
ccc6cda3
JA
7667 {
7668 mflags |= MATCH_END;
7669 p++;
7670 }
7671 else
7672 mflags |= MATCH_ANY;
7673
cce855bc
JA
7674 /* OK, we now want to substitute REP for PAT in VAL. If
7675 flags & MATCH_GLOBREP is non-zero, the substitution is done
7676 everywhere, otherwise only the first occurrence of PAT is
7117c2d2
JA
7677 replaced. The pattern matching code doesn't understand
7678 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
7679 values passed in (VT_VARIABLE) so the pattern substitution
7680 code works right. We need to requote special chars after
7681 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
7682 other cases if QUOTED == 0, since the posparams and arrays
7683 indexed by * or @ do special things when QUOTED != 0. */
7684
ccc6cda3
JA
7685 switch (vtype)
7686 {
7687 case VT_VARIABLE:
d166f048 7688 case VT_ARRAYMEMBER:
ccc6cda3 7689 temp = pat_subst (val, p, rep, mflags);
7117c2d2
JA
7690 if (vtype == VT_VARIABLE)
7691 FREE (val);
7692 if (temp)
7693 {
3185942a 7694 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7117c2d2
JA
7695 free (temp);
7696 temp = tt;
7697 }
ccc6cda3
JA
7698 break;
7699 case VT_POSPARMS:
7700 temp = pos_params_pat_subst (val, p, rep, mflags);
7117c2d2
JA
7701 if (temp && (mflags & MATCH_QUOTED) == 0)
7702 {
7703 tt = quote_escapes (temp);
7704 free (temp);
7705 temp = tt;
7706 }
ccc6cda3
JA
7707 break;
7708#if defined (ARRAY_VARS)
7709 case VT_ARRAYVAR:
3185942a
JA
7710 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
7711 : array_patsub (array_cell (v), p, rep, mflags);
7712 /* Don't call quote_escapes anymore; array_patsub calls
7713 array_quote_escapes as appropriate before adding the
7714 space separators; ditto for assoc_patsub. */
7715 break;
7716#endif
7717 }
7718
7719 FREE (pat);
7720 FREE (rep);
7721 free (lpatsub);
7722
7723 return temp;
7724}
7725
7726/****************************************************************/
7727/* */
7728/* Functions to perform case modification on variable values */
7729/* */
7730/****************************************************************/
7731
7732/* Do case modification on the positional parameters. */
7733
7734static char *
7735pos_params_modcase (string, pat, modop, mflags)
7736 char *string, *pat;
7737 int modop;
7738 int mflags;
7739{
7740 WORD_LIST *save, *params;
7741 WORD_DESC *w;
7742 char *ret;
7743 int pchar, qflags;
7744
7745 save = params = list_rest_of_args ();
7746 if (save == 0)
7747 return ((char *)NULL);
7748
7749 for ( ; params; params = params->next)
7750 {
7751 ret = sh_modcase (params->word->word, pat, modop);
7752 w = alloc_word_desc ();
7753 w->word = ret ? ret : savestring ("");
7754 dispose_word (params->word);
7755 params->word = w;
7756 }
7757
7758 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7759 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7760
7761 ret = string_list_pos_params (pchar, save, qflags);
7762 dispose_words (save);
7763
7764 return (ret);
7765}
7766
7767/* Perform case modification on VALUE, which is the expansion of
7768 VARNAME. MODSPEC is an expression supplying the type of modification
7769 to perform. QUOTED is a flags word containing the type of quoting
7770 currently in effect. */
7771static char *
495aee44 7772parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
3185942a 7773 char *varname, *value;
495aee44 7774 int ind, modspec;
3185942a 7775 char *patspec;
495aee44 7776 int quoted, flags;
3185942a
JA
7777{
7778 int vtype, starsub, modop, mflags, x;
7779 char *val, *temp, *pat, *p, *lpat, *tt;
7780 SHELL_VAR *v;
7781
7782 if (value == 0)
7783 return ((char *)NULL);
7784
7785 this_command_name = varname;
7786
495aee44 7787 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
3185942a
JA
7788 if (vtype == -1)
7789 return ((char *)NULL);
7790
7791 starsub = vtype & VT_STARSUB;
7792 vtype &= ~VT_STARSUB;
7793
7794 modop = 0;
7795 mflags = 0;
7796 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7797 mflags |= MATCH_QUOTED;
7798 if (starsub)
7799 mflags |= MATCH_STARSUB;
7800
7801 p = patspec;
7802 if (modspec == '^')
7803 {
7804 x = p && p[0] == modspec;
17345e5a 7805 modop = x ? CASE_UPPER : CASE_UPFIRST;
3185942a
JA
7806 p += x;
7807 }
7808 else if (modspec == ',')
7809 {
7810 x = p && p[0] == modspec;
17345e5a 7811 modop = x ? CASE_LOWER : CASE_LOWFIRST;
3185942a
JA
7812 p += x;
7813 }
7814 else if (modspec == '~')
7815 {
7816 x = p && p[0] == modspec;
7817 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7818 p += x;
7819 }
7820
7821 lpat = p ? savestring (p) : 0;
7822 /* Perform the same expansions on the pattern as performed by the
7823 pattern removal expansions. FOR LATER */
7824 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7825
7826 /* OK, now we do the case modification. */
7827 switch (vtype)
7828 {
7829 case VT_VARIABLE:
7830 case VT_ARRAYMEMBER:
7831 temp = sh_modcase (val, pat, modop);
7832 if (vtype == VT_VARIABLE)
7833 FREE (val);
7834 if (temp)
7835 {
7836 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7837 free (temp);
7838 temp = tt;
7839 }
7840 break;
7841
7842 case VT_POSPARMS:
7843 temp = pos_params_modcase (val, pat, modop, mflags);
7844 if (temp && (mflags & MATCH_QUOTED) == 0)
7117c2d2
JA
7845 {
7846 tt = quote_escapes (temp);
7847 free (temp);
7848 temp = tt;
7849 }
3185942a
JA
7850 break;
7851
7852#if defined (ARRAY_VARS)
7853 case VT_ARRAYVAR:
7854 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7855 : array_modcase (array_cell (v), pat, modop, mflags);
7856 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7857 as appropriate before adding the space separators; ditto for
7858 assoc_modcase. */
ccc6cda3
JA
7859 break;
7860#endif
7861 }
7862
7863 FREE (pat);
3185942a 7864 free (lpat);
ccc6cda3
JA
7865
7866 return temp;
7867}
7868
0628567a
JA
7869/* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7870 any occur, this must be a nested command substitution, so return 0.
7871 Otherwise, return 1. A valid arithmetic expression must always have a
7872 ( before a matching ), so any cases where there are more right parens
7873 means that this must not be an arithmetic expression, though the parser
7874 will not accept it without a balanced total number of parens. */
7875static int
7876chk_arithsub (s, len)
7877 const char *s;
7878 int len;
7879{
7880 int i, count;
7881 DECLARE_MBSTATE;
7882
7883 i = count = 0;
7884 while (i < len)
7885 {
0001803f 7886 if (s[i] == LPAREN)
0628567a 7887 count++;
0001803f 7888 else if (s[i] == RPAREN)
0628567a
JA
7889 {
7890 count--;
7891 if (count < 0)
7892 return 0;
7893 }
7894
7895 switch (s[i])
7896 {
7897 default:
7898 ADVANCE_CHAR (s, len, i);
7899 break;
7900
7901 case '\\':
7902 i++;
7903 if (s[i])
7904 ADVANCE_CHAR (s, len, i);
7905 break;
7906
7907 case '\'':
a0c0a00f 7908 i = skip_single_quoted (s, len, ++i, 0);
0628567a
JA
7909 break;
7910
7911 case '"':
a0c0a00f 7912 i = skip_double_quoted ((char *)s, len, ++i, 0);
0628567a
JA
7913 break;
7914 }
7915 }
7916
7917 return (count == 0);
7918}
7919
cce855bc
JA
7920/****************************************************************/
7921/* */
7922/* Functions to perform parameter expansion on a string */
7923/* */
7924/****************************************************************/
7925
3185942a 7926/* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
95732b49 7927static WORD_DESC *
0001803f 7928parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
ccc6cda3 7929 char *string;
a0c0a00f 7930 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
ccc6cda3
JA
7931{
7932 int check_nullness, var_is_set, var_is_null, var_is_special;
3185942a 7933 int want_substring, want_indir, want_patsub, want_casemod;
ccc6cda3 7934 char *name, *value, *temp, *temp1;
95732b49 7935 WORD_DESC *tdesc, *ret;
a0c0a00f 7936 int t_index, sindex, c, tflag, modspec, all_element_arrayref;
7117c2d2 7937 intmax_t number;
495aee44 7938 arrayind_t ind;
ccc6cda3 7939
3185942a 7940 temp = temp1 = value = (char *)NULL;
ccc6cda3 7941 var_is_set = var_is_null = var_is_special = check_nullness = 0;
3185942a 7942 want_substring = want_indir = want_patsub = want_casemod = 0;
ccc6cda3 7943
a0c0a00f
CR
7944 all_element_arrayref = 0;
7945
cce855bc
JA
7946 sindex = *indexp;
7947 t_index = ++sindex;
0628567a
JA
7948 /* ${#var} doesn't have any of the other parameter expansions on it. */
7949 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
3185942a 7950 name = string_extract (string, &t_index, "}", SX_VARNAME);
0628567a 7951 else
3185942a
JA
7952#if defined (CASEMOD_EXPANSIONS)
7953 /* To enable case-toggling expansions using the `~' operator character
7954 change the 1 to 0. */
7955# if defined (CASEMOD_CAPCASE)
a0c0a00f 7956 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
3185942a 7957# else
a0c0a00f 7958 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
3185942a
JA
7959# endif /* CASEMOD_CAPCASE */
7960#else
a0c0a00f 7961 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
3185942a 7962#endif /* CASEMOD_EXPANSIONS */
cce855bc 7963
a0c0a00f
CR
7964 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
7965 the cleanest code ever. */
7966 if (*name == 0 && sindex == t_index && string[sindex] == '@')
7967 {
7968 name = (char *)xrealloc (name, 2);
7969 name[0] = '@';
7970 name[1] = '\0';
7971 t_index++;
7972 }
7973 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == '}')
7974 {
7975 name = (char *)xrealloc (name, t_index - sindex + 2);
7976 name[t_index - sindex] = '@';
7977 name[t_index - sindex + 1] = '\0';
7978 t_index++;
7979 }
7980
95732b49
JA
7981 ret = 0;
7982 tflag = 0;
7983
495aee44
CR
7984 ind = INTMAX_MIN;
7985
cce855bc
JA
7986 /* If the name really consists of a special variable, then make sure
7987 that we have the entire name. We don't allow indirect references
a0c0a00f
CR
7988 to special variables except `#', `?', `@' and `*'. This clause is
7989 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
7990 general. */
495aee44 7991 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
a0c0a00f 7992 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
495aee44 7993 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
ccc6cda3
JA
7994 {
7995 t_index++;
a0c0a00f 7996 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
ac50fbac 7997 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
ccc6cda3
JA
7998 *name = string[sindex];
7999 if (string[sindex] == '!')
8000 {
28ef6c31
JA
8001 /* indirect reference of $#, $?, $@, or $* */
8002 name[1] = string[sindex + 1];
8003 strcpy (name + 2, temp1);
ccc6cda3 8004 }
cce855bc 8005 else
ccc6cda3
JA
8006 strcpy (name + 1, temp1);
8007 free (temp1);
8008 }
8009 sindex = t_index;
8010
8011 /* Find out what character ended the variable name. Then
8012 do the appropriate thing. */
8013 if (c = string[sindex])
8014 sindex++;
8015
8016 /* If c is followed by one of the valid parameter expansion
8017 characters, move past it as normal. If not, assume that
8018 a substring specification is being given, and do not move
8019 past it. */
28ef6c31 8020 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
ccc6cda3
JA
8021 {
8022 check_nullness++;
8023 if (c = string[sindex])
8024 sindex++;
8025 }
cce855bc 8026 else if (c == ':' && string[sindex] != RBRACE)
ccc6cda3 8027 want_substring = 1;
ac50fbac 8028 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
ccc6cda3 8029 want_patsub = 1;
3185942a
JA
8030#if defined (CASEMOD_EXPANSIONS)
8031 else if (c == '^' || c == ',' || c == '~')
8032 {
8033 modspec = c;
8034 want_casemod = 1;
8035 }
8036#endif
ccc6cda3 8037
cce855bc
JA
8038 /* Catch the valid and invalid brace expressions that made it through the
8039 tests above. */
8040 /* ${#-} is a valid expansion and means to take the length of $-.
8041 Similarly for ${#?} and ${##}... */
8042 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
28ef6c31 8043 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
cce855bc 8044 {
f73dda09 8045 name = (char *)xrealloc (name, 3);
cce855bc
JA
8046 name[1] = c;
8047 name[2] = '\0';
8048 c = string[sindex++];
8049 }
8050
8051 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8052 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8053 member (c, "%:=+/") && string[sindex] == RBRACE)
8054 {
8055 temp = (char *)NULL;
a0c0a00f 8056 goto bad_substitution; /* XXX - substitution error */
cce855bc
JA
8057 }
8058
8059 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8060 either a variable name, one of the positional parameters or a special
8061 variable that expands to one of the positional parameters. */
8062 want_indir = *name == '!' &&
f73dda09 8063 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7117c2d2 8064 || VALID_INDIR_PARAM (name[1]));
ccc6cda3
JA
8065
8066 /* Determine the value of this variable. */
8067
cce855bc 8068 /* Check for special variables, directly referenced. */
bb70624e 8069 if (SPECIAL_VAR (name, want_indir))
ccc6cda3
JA
8070 var_is_special++;
8071
cce855bc
JA
8072 /* Check for special expansion things, like the length of a parameter */
8073 if (*name == '#' && name[1])
ccc6cda3 8074 {
cce855bc 8075 /* If we are not pointing at the character just after the
28ef6c31
JA
8076 closing brace, then we haven't gotten all of the name.
8077 Since it begins with a special character, this is a bad
8078 substitution. Also check NAME for validity before trying
8079 to go on. */
cce855bc 8080 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
ccc6cda3
JA
8081 {
8082 temp = (char *)NULL;
a0c0a00f 8083 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8084 }
8085
8086 number = parameter_brace_expand_length (name);
495aee44
CR
8087 if (number == INTMAX_MIN && unbound_vars_is_error)
8088 {
8089 last_command_exit_value = EXECUTION_FAILURE;
8090 err_unboundvar (name+1);
8091 free (name);
8092 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8093 }
ccc6cda3
JA
8094 free (name);
8095
8096 *indexp = sindex;
95732b49
JA
8097 if (number < 0)
8098 return (&expand_wdesc_error);
8099 else
8100 {
8101 ret = alloc_word_desc ();
8102 ret->word = itos (number);
8103 return ret;
8104 }
ccc6cda3
JA
8105 }
8106
8107 /* ${@} is identical to $@. */
8108 if (name[0] == '@' && name[1] == '\0')
8109 {
8110 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8111 *quoted_dollar_atp = 1;
8112
8113 if (contains_dollar_at)
8114 *contains_dollar_at = 1;
ac50fbac
CR
8115
8116 tflag |= W_DOLLARAT;
ccc6cda3
JA
8117 }
8118
b80f6443 8119 /* Process ${!PREFIX*} expansion. */
bb70624e
JA
8120 if (want_indir && string[sindex - 1] == RBRACE &&
8121 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
f73dda09 8122 legal_variable_starter ((unsigned char) name[1]))
bb70624e
JA
8123 {
8124 char **x;
8125 WORD_LIST *xlist;
8126
8127 temp1 = savestring (name + 1);
8128 number = strlen (temp1);
8129 temp1[number - 1] = '\0';
8130 x = all_variables_matching_prefix (temp1);
7117c2d2 8131 xlist = strvec_to_word_list (x, 0, 0);
28ef6c31
JA
8132 if (string[sindex - 2] == '*')
8133 temp = string_list_dollar_star (xlist);
8134 else
8135 {
a0c0a00f 8136 temp = string_list_dollar_at (xlist, quoted, 0);
28ef6c31
JA
8137 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8138 *quoted_dollar_atp = 1;
8139 if (contains_dollar_at)
8140 *contains_dollar_at = 1;
ac50fbac
CR
8141
8142 tflag |= W_DOLLARAT;
28ef6c31 8143 }
bb70624e 8144 free (x);
89a92869 8145 dispose_words (xlist);
bb70624e
JA
8146 free (temp1);
8147 *indexp = sindex;
95732b49 8148
ac50fbac
CR
8149 free (name);
8150
95732b49
JA
8151 ret = alloc_word_desc ();
8152 ret->word = temp;
ac50fbac 8153 ret->flags = tflag; /* XXX */
95732b49 8154 return ret;
bb70624e 8155 }
b80f6443
JA
8156
8157#if defined (ARRAY_VARS)
8158 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
8159 if (want_indir && string[sindex - 1] == RBRACE &&
a0c0a00f 8160 string[sindex - 2] == ']' && valid_array_reference (name+1, 0))
b80f6443
JA
8161 {
8162 char *x, *x1;
8163
8164 temp1 = savestring (name + 1);
8165 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
8166 FREE (x);
8167 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
8168 {
3185942a 8169 temp = array_keys (temp1, quoted); /* handles assoc vars too */
b80f6443
JA
8170 if (x1[0] == '@')
8171 {
8172 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8173 *quoted_dollar_atp = 1;
8174 if (contains_dollar_at)
8175 *contains_dollar_at = 1;
ac50fbac
CR
8176
8177 tflag |= W_DOLLARAT;
b80f6443
JA
8178 }
8179
a0c0a00f 8180 free (name);
b80f6443
JA
8181 free (temp1);
8182 *indexp = sindex;
95732b49
JA
8183
8184 ret = alloc_word_desc ();
8185 ret->word = temp;
ac50fbac 8186 ret->flags = tflag; /* XXX */
95732b49 8187 return ret;
b80f6443
JA
8188 }
8189
8190 free (temp1);
8191 }
8192#endif /* ARRAY_VARS */
bb70624e 8193
ccc6cda3
JA
8194 /* Make sure that NAME is valid before trying to go on. */
8195 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
8196 var_is_special) == 0)
8197 {
8198 temp = (char *)NULL;
a0c0a00f 8199 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8200 }
8201
8202 if (want_indir)
1a1f8b54
CR
8203 {
8204 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
a0c0a00f
CR
8205 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8206 {
8207 temp = (char *)NULL;
8208 goto bad_substitution;
8209 }
1a1f8b54
CR
8210 /* Turn off the W_ARRAYIND flag because there is no way for this function
8211 to return the index we're supposed to be using. */
8212 if (tdesc && tdesc->flags)
8213 tdesc->flags &= ~W_ARRAYIND;
8214 }
95732b49 8215 else
ac50fbac 8216 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
95732b49
JA
8217
8218 if (tdesc)
8219 {
8220 temp = tdesc->word;
8221 tflag = tdesc->flags;
8222 dispose_word_desc (tdesc);
8223 }
ccc6cda3 8224 else
95732b49 8225 temp = (char *)0;
ccc6cda3 8226
ac50fbac
CR
8227 if (temp == &expand_param_error || temp == &expand_param_fatal)
8228 {
8229 FREE (name);
8230 FREE (value);
8231 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8232 }
8233
ccc6cda3 8234#if defined (ARRAY_VARS)
a0c0a00f
CR
8235 if (valid_array_reference (name, 0))
8236 {
8237 int qflags;
8238 char *t;
8239
8240 qflags = quoted;
8241 /* If in a context where word splitting will not take place, treat as
8242 if double-quoted. Has effects with $* and ${array[*]} */
8243 if (pflags & PF_ASSIGNRHS)
8244 qflags |= Q_DOUBLE_QUOTES;
8245 chk_atstar (name, qflags, quoted_dollar_atp, contains_dollar_at);
8246 /* We duplicate a little code here */
8247 t = mbschr (name, '[');
8248 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == ']')
8249 all_element_arrayref = 1;
8250 }
ccc6cda3
JA
8251#endif
8252
8253 var_is_set = temp != (char *)0;
8254 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
ac50fbac
CR
8255 /* XXX - this may not need to be restricted to special variables */
8256 if (check_nullness)
8257 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
ccc6cda3
JA
8258
8259 /* Get the rest of the stuff inside the braces. */
cce855bc 8260 if (c && c != RBRACE)
ccc6cda3
JA
8261 {
8262 /* Extract the contents of the ${ ... } expansion
28ef6c31 8263 according to the Posix.2 rules. */
49ed961b 8264 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
cce855bc 8265 if (string[sindex] == RBRACE)
28ef6c31 8266 sindex++;
ccc6cda3 8267 else
a0c0a00f 8268 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8269 }
8270 else
8271 value = (char *)NULL;
726f6388 8272
ccc6cda3
JA
8273 *indexp = sindex;
8274
495aee44
CR
8275 /* All the cases where an expansion can possibly generate an unbound
8276 variable error. */
8277 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
8278 {
a0c0a00f 8279 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
495aee44
CR
8280 {
8281 last_command_exit_value = EXECUTION_FAILURE;
8282 err_unboundvar (name);
8283 FREE (value);
8284 FREE (temp);
8285 free (name);
8286 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8287 }
8288 }
8289
ccc6cda3
JA
8290 /* If this is a substring spec, process it and add the result. */
8291 if (want_substring)
726f6388 8292 {
495aee44 8293 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8294 FREE (name);
8295 FREE (value);
8296 FREE (temp);
95732b49
JA
8297
8298 if (temp1 == &expand_param_error)
8299 return (&expand_wdesc_error);
8300 else if (temp1 == &expand_param_fatal)
8301 return (&expand_wdesc_fatal);
8302
8303 ret = alloc_word_desc ();
8304 ret->word = temp1;
ac50fbac
CR
8305 /* We test quoted_dollar_atp because we want variants with double-quoted
8306 "$@" to take a different code path. In fact, we make sure at the end
8307 of expand_word_internal that we're only looking at these flags if
8308 quoted_dollar_at == 0. */
8309 if (temp1 &&
8310 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8311 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
0628567a 8312 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
95732b49 8313 return ret;
726f6388 8314 }
ccc6cda3 8315 else if (want_patsub)
726f6388 8316 {
a0c0a00f 8317 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8318 FREE (name);
8319 FREE (value);
8320 FREE (temp);
95732b49
JA
8321
8322 if (temp1 == &expand_param_error)
8323 return (&expand_wdesc_error);
8324 else if (temp1 == &expand_param_fatal)
8325 return (&expand_wdesc_fatal);
8326
8327 ret = alloc_word_desc ();
8328 ret->word = temp1;
ac50fbac
CR
8329 if (temp1 &&
8330 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8331 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a 8332 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
95732b49 8333 return ret;
ccc6cda3 8334 }
3185942a
JA
8335#if defined (CASEMOD_EXPANSIONS)
8336 else if (want_casemod)
8337 {
495aee44 8338 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
3185942a
JA
8339 FREE (name);
8340 FREE (value);
8341 FREE (temp);
8342
8343 if (temp1 == &expand_param_error)
8344 return (&expand_wdesc_error);
8345 else if (temp1 == &expand_param_fatal)
8346 return (&expand_wdesc_fatal);
8347
8348 ret = alloc_word_desc ();
8349 ret->word = temp1;
ac50fbac
CR
8350 if (temp1 &&
8351 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8352 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a
JA
8353 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8354 return ret;
8355 }
8356#endif
726f6388 8357
ccc6cda3
JA
8358 /* Do the right thing based on which character ended the variable name. */
8359 switch (c)
8360 {
8361 default:
8362 case '\0':
a0c0a00f 8363bad_substitution:
98043138 8364 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 8365 report_error (_("%s: bad substitution"), string ? string : "??");
ccc6cda3
JA
8366 FREE (value);
8367 FREE (temp);
8368 free (name);
a0c0a00f
CR
8369 if (shell_compatibility_level <= 43)
8370 return &expand_wdesc_error;
8371 else
8372 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
ccc6cda3 8373
cce855bc 8374 case RBRACE:
ccc6cda3 8375 break;
726f6388 8376
a0c0a00f
CR
8377 case '@':
8378 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8379 free (temp);
8380 free (value);
8381 free (name);
8382 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
8383 {
8384 last_command_exit_value = EXECUTION_FAILURE;
8385 report_error (_("%s: bad substitution"), string ? string : "??");
8386 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8387 }
8388
8389 ret = alloc_word_desc ();
8390 ret->word = temp1;
8391 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8392 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8393 return ret;
8394
ccc6cda3
JA
8395 case '#': /* ${param#[#]pattern} */
8396 case '%': /* ${param%[%]pattern} */
8397 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
28ef6c31
JA
8398 {
8399 FREE (value);
ccc6cda3 8400 break;
28ef6c31 8401 }
495aee44 8402 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8403 free (temp);
8404 free (value);
495aee44 8405 free (name);
3185942a
JA
8406
8407 ret = alloc_word_desc ();
8408 ret->word = temp1;
8409 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8410 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8411 return ret;
ccc6cda3
JA
8412
8413 case '-':
8414 case '=':
8415 case '?':
8416 case '+':
8417 if (var_is_set && var_is_null == 0)
28ef6c31
JA
8418 {
8419 /* If the operator is `+', we don't want the value of the named
8420 variable for anything, just the value of the right hand side. */
ccc6cda3
JA
8421 if (c == '+')
8422 {
28ef6c31
JA
8423 /* XXX -- if we're double-quoted and the named variable is "$@",
8424 we want to turn off any special handling of "$@" --
8425 we're not using it, so whatever is on the rhs applies. */
8426 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8427 *quoted_dollar_atp = 0;
8428 if (contains_dollar_at)
8429 *contains_dollar_at = 0;
8430
ccc6cda3
JA
8431 FREE (temp);
8432 if (value)
28ef6c31 8433 {
495aee44
CR
8434 /* From Posix discussion on austin-group list. Issue 221
8435 requires that backslashes escaping `}' inside
8436 double-quoted ${...} be removed. */
8437 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8438 quoted |= Q_DOLBRACE;
95732b49
JA
8439 ret = parameter_brace_expand_rhs (name, value, c,
8440 quoted,
a0c0a00f 8441 pflags,
95732b49
JA
8442 quoted_dollar_atp,
8443 contains_dollar_at);
8444 /* XXX - fix up later, esp. noting presence of
8445 W_HASQUOTEDNULL in ret->flags */
ccc6cda3
JA
8446 free (value);
8447 }
8448 else
28ef6c31 8449 temp = (char *)NULL;
ccc6cda3
JA
8450 }
8451 else
8452 {
8453 FREE (value);
8454 }
8455 /* Otherwise do nothing; just use the value in TEMP. */
726f6388 8456 }
ccc6cda3 8457 else /* VAR not set or VAR is NULL. */
28ef6c31 8458 {
ccc6cda3
JA
8459 FREE (temp);
8460 temp = (char *)NULL;
8461 if (c == '=' && var_is_special)
8462 {
ac50fbac 8463 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 8464 report_error (_("$%s: cannot assign in this way"), name);
ccc6cda3
JA
8465 free (name);
8466 free (value);
95732b49 8467 return &expand_wdesc_error;
ccc6cda3
JA
8468 }
8469 else if (c == '?')
8470 {
8471 parameter_brace_expand_error (name, value);
95732b49 8472 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
ccc6cda3
JA
8473 }
8474 else if (c != '+')
28ef6c31
JA
8475 {
8476 /* XXX -- if we're double-quoted and the named variable is "$@",
8477 we want to turn off any special handling of "$@" --
8478 we're not using it, so whatever is on the rhs applies. */
8479 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8480 *quoted_dollar_atp = 0;
8481 if (contains_dollar_at)
8482 *contains_dollar_at = 0;
8483
495aee44
CR
8484 /* From Posix discussion on austin-group list. Issue 221 requires
8485 that backslashes escaping `}' inside double-quoted ${...} be
8486 removed. */
8487 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8488 quoted |= Q_DOLBRACE;
a0c0a00f 8489 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
95732b49
JA
8490 quoted_dollar_atp,
8491 contains_dollar_at);
8492 /* XXX - fix up later, esp. noting presence of
8493 W_HASQUOTEDNULL in tdesc->flags */
28ef6c31 8494 }
ccc6cda3 8495 free (value);
726f6388 8496 }
28ef6c31 8497
ccc6cda3 8498 break;
726f6388 8499 }
ccc6cda3 8500 free (name);
95732b49
JA
8501
8502 if (ret == 0)
8503 {
8504 ret = alloc_word_desc ();
8505 ret->flags = tflag;
8506 ret->word = temp;
8507 }
8508 return (ret);
726f6388
JA
8509}
8510
cce855bc
JA
8511/* Expand a single ${xxx} expansion. The braces are optional. When
8512 the braces are used, parameter_brace_expand() does the work,
8513 possibly calling param_expand recursively. */
95732b49 8514static WORD_DESC *
cce855bc
JA
8515param_expand (string, sindex, quoted, expanded_something,
8516 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
8517 pflags)
8518 char *string;
8519 int *sindex, quoted, *expanded_something, *contains_dollar_at;
8520 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
8521{
a0c0a00f 8522 char *temp, *temp1, uerror[3], *savecmd;
f73dda09
JA
8523 int zindex, t_index, expok;
8524 unsigned char c;
7117c2d2 8525 intmax_t number;
cce855bc 8526 SHELL_VAR *var;
f73dda09 8527 WORD_LIST *list;
95732b49
JA
8528 WORD_DESC *tdesc, *ret;
8529 int tflag;
cce855bc 8530
a0c0a00f 8531/*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
cce855bc
JA
8532 zindex = *sindex;
8533 c = string[++zindex];
8534
8535 temp = (char *)NULL;
95732b49
JA
8536 ret = tdesc = (WORD_DESC *)NULL;
8537 tflag = 0;
cce855bc
JA
8538
8539 /* Do simple cases first. Switch on what follows '$'. */
8540 switch (c)
8541 {
8542 /* $0 .. $9? */
8543 case '0':
8544 case '1':
8545 case '2':
8546 case '3':
8547 case '4':
8548 case '5':
8549 case '6':
8550 case '7':
8551 case '8':
8552 case '9':
f73dda09 8553 temp1 = dollar_vars[TODIGIT (c)];
cce855bc
JA
8554 if (unbound_vars_is_error && temp1 == (char *)NULL)
8555 {
7117c2d2
JA
8556 uerror[0] = '$';
8557 uerror[1] = c;
8558 uerror[2] = '\0';
cce855bc 8559 last_command_exit_value = EXECUTION_FAILURE;
0001803f 8560 err_unboundvar (uerror);
95732b49 8561 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc 8562 }
b80f6443
JA
8563 if (temp1)
8564 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8565 ? quote_string (temp1)
8566 : quote_escapes (temp1);
8567 else
8568 temp = (char *)NULL;
95732b49 8569
cce855bc
JA
8570 break;
8571
8572 /* $$ -- pid of the invoking shell. */
8573 case '$':
8574 temp = itos (dollar_dollar_pid);
8575 break;
8576
8577 /* $# -- number of positional parameters. */
8578 case '#':
8579 temp = itos (number_of_args ());
8580 break;
8581
8582 /* $? -- return value of the last synchronous command. */
8583 case '?':
8584 temp = itos (last_command_exit_value);
8585 break;
8586
8587 /* $- -- flags supplied to the shell on invocation or by `set'. */
8588 case '-':
8589 temp = which_set_flags ();
8590 break;
8591
8592 /* $! -- Pid of the last asynchronous command. */
8593 case '!':
8594 /* If no asynchronous pids have been created, expand to nothing.
8595 If `set -u' has been executed, and no async processes have
8596 been created, this is an expansion error. */
8597 if (last_asynchronous_pid == NO_PID)
8598 {
8599 if (expanded_something)
8600 *expanded_something = 0;
8601 temp = (char *)NULL;
8602 if (unbound_vars_is_error)
8603 {
7117c2d2
JA
8604 uerror[0] = '$';
8605 uerror[1] = c;
8606 uerror[2] = '\0';
cce855bc 8607 last_command_exit_value = EXECUTION_FAILURE;
0001803f 8608 err_unboundvar (uerror);
95732b49 8609 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc
JA
8610 }
8611 }
8612 else
f73dda09 8613 temp = itos (last_asynchronous_pid);
cce855bc
JA
8614 break;
8615
8616 /* The only difference between this and $@ is when the arg is quoted. */
8617 case '*': /* `$*' */
8618 list = list_rest_of_args ();
8619
89a92869
CR
8620#if 0
8621 /* According to austin-group posix proposal by Geoff Clare in
8622 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8623
8624 "The shell shall write a message to standard error and
8625 immediately exit when it tries to expand an unset parameter
8626 other than the '@' and '*' special parameters."
8627 */
8628
8629 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
8630 {
8631 uerror[0] = '$';
8632 uerror[1] = '*';
8633 uerror[2] = '\0';
3185942a 8634 last_command_exit_value = EXECUTION_FAILURE;
89a92869 8635 err_unboundvar (uerror);
3185942a
JA
8636 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8637 }
89a92869 8638#endif
3185942a 8639
cce855bc
JA
8640 /* If there are no command-line arguments, this should just
8641 disappear if there are other characters in the expansion,
8642 even if it's quoted. */
8643 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
8644 temp = (char *)NULL;
0001803f 8645 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
cce855bc
JA
8646 {
8647 /* If we have "$*" we want to make a string of the positional
8648 parameters, separated by the first character of $IFS, and
8649 quote the whole string, including the separators. If IFS
8650 is unset, the parameters are separated by ' '; if $IFS is
8651 null, the parameters are concatenated. */
0001803f 8652 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
495aee44
CR
8653 if (temp)
8654 {
a0c0a00f 8655 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
495aee44
CR
8656 if (*temp == 0)
8657 tflag |= W_HASQUOTEDNULL;
a0c0a00f
CR
8658 if (temp != temp1)
8659 free (temp);
495aee44
CR
8660 temp = temp1;
8661 }
cce855bc
JA
8662 }
8663 else
28ef6c31 8664 {
95732b49
JA
8665 /* We check whether or not we're eventually going to split $* here,
8666 for example when IFS is empty and we are processing the rhs of
8667 an assignment statement. In that case, we don't separate the
8668 arguments at all. Otherwise, if the $* is not quoted it is
8669 identical to $@ */
95732b49
JA
8670# if defined (HANDLE_MULTIBYTE)
8671 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
8672# else
8673 if (expand_no_split_dollar_star && ifs_firstc == 0)
8674# endif
8675 temp = string_list_dollar_star (list);
8676 else
ac50fbac 8677 {
a0c0a00f 8678 temp = string_list_dollar_at (list, quoted, 0);
ac50fbac
CR
8679 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
8680 tflag |= W_SPLITSPACE;
a0c0a00f
CR
8681 /* If we're not quoted but we still don't want word splitting, make
8682 we quote the IFS characters to protect them from splitting (e.g.,
8683 when $@ is in the string as well). */
8684 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
8685 {
8686 temp1 = quote_string (temp);
8687 free (temp);
8688 temp = temp1;
8689 }
ac50fbac
CR
8690 }
8691
28ef6c31
JA
8692 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
8693 *contains_dollar_at = 1;
8694 }
cce855bc
JA
8695
8696 dispose_words (list);
8697 break;
8698
8699 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
8700 means that we have to turn quoting off after we split into
8701 the individually quoted arguments so that the final split
8702 on the first character of $IFS is still done. */
8703 case '@': /* `$@' */
8704 list = list_rest_of_args ();
8705
89a92869
CR
8706#if 0
8707 /* According to austin-group posix proposal by Geoff Clare in
8708 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8709
8710 "The shell shall write a message to standard error and
8711 immediately exit when it tries to expand an unset parameter
8712 other than the '@' and '*' special parameters."
8713 */
8714
8715 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
8716 {
8717 uerror[0] = '$';
8718 uerror[1] = '@';
8719 uerror[2] = '\0';
3185942a 8720 last_command_exit_value = EXECUTION_FAILURE;
89a92869 8721 err_unboundvar (uerror);
3185942a
JA
8722 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8723 }
89a92869 8724#endif
3185942a 8725
cce855bc
JA
8726 /* We want to flag the fact that we saw this. We can't turn
8727 off quoting entirely, because other characters in the
8728 string might need it (consider "\"$@\""), but we need some
8729 way to signal that the final split on the first character
8730 of $IFS should be done, even though QUOTED is 1. */
0001803f 8731 /* XXX - should this test include Q_PATQUOTE? */
cce855bc
JA
8732 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8733 *quoted_dollar_at_p = 1;
8734 if (contains_dollar_at)
8735 *contains_dollar_at = 1;
8736
8737 /* We want to separate the positional parameters with the first
8738 character of $IFS in case $IFS is something other than a space.
8739 We also want to make sure that splitting is done no matter what --
8740 according to POSIX.2, this expands to a list of the positional
8741 parameters no matter what IFS is set to. */
3b34f6e6
CR
8742 /* XXX - what to do when in a context where word splitting is not
8743 performed? Even when IFS is not the default, posix seems to imply
8744 that we behave like unquoted $* ? Maybe we should use PF_NOSPLIT2
8745 here. */
a0c0a00f
CR
8746 /* XXX - bash-4.4/bash-5.0 passing PFLAGS */
8747 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted, pflags);
cce855bc 8748
ac50fbac 8749 tflag |= W_DOLLARAT;
cce855bc
JA
8750 dispose_words (list);
8751 break;
8752
8753 case LBRACE:
0001803f 8754 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
95732b49
JA
8755 quoted_dollar_at_p,
8756 contains_dollar_at);
8757
95732b49
JA
8758 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8759 return (tdesc);
8760 temp = tdesc ? tdesc->word : (char *)0;
cce855bc
JA
8761
8762 /* XXX */
bb70624e 8763 /* Quoted nulls should be removed if there is anything else
cce855bc
JA
8764 in the string. */
8765 /* Note that we saw the quoted null so we can add one back at
8766 the end of this function if there are no other characters
28ef6c31
JA
8767 in the string, discard TEMP, and go on. The exception to
8768 this is when we have "${@}" and $1 is '', since $@ needs
8769 special handling. */
95732b49 8770 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
cce855bc
JA
8771 {
8772 if (had_quoted_null_p)
8773 *had_quoted_null_p = 1;
28ef6c31
JA
8774 if (*quoted_dollar_at_p == 0)
8775 {
8776 free (temp);
95732b49 8777 tdesc->word = temp = (char *)NULL;
28ef6c31
JA
8778 }
8779
cce855bc
JA
8780 }
8781
95732b49 8782 ret = tdesc;
cce855bc
JA
8783 goto return0;
8784
8785 /* Do command or arithmetic substitution. */
8786 case LPAREN:
8787 /* We have to extract the contents of this paren substitution. */
8788 t_index = zindex + 1;
a0c0a00f
CR
8789 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
8790 as arithmetic substitution immediately. */
8791 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
cce855bc
JA
8792 zindex = t_index;
8793
8794 /* For Posix.2-style `$(( ))' arithmetic substitution,
28ef6c31 8795 extract the expression and pass it to the evaluator. */
cce855bc
JA
8796 if (temp && *temp == LPAREN)
8797 {
8798 char *temp2;
8799 temp1 = temp + 1;
8800 temp2 = savestring (temp1);
8801 t_index = strlen (temp2) - 1;
8802
8803 if (temp2[t_index] != RPAREN)
8804 {
8805 free (temp2);
8806 goto comsub;
8807 }
8808
8809 /* Cut off ending `)' */
8810 temp2[t_index] = '\0';
8811
0628567a
JA
8812 if (chk_arithsub (temp2, t_index) == 0)
8813 {
8814 free (temp2);
0001803f
CR
8815#if 0
8816 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
8817#endif
0628567a
JA
8818 goto comsub;
8819 }
8820
cce855bc 8821 /* Expand variables found inside the expression. */
a0c0a00f 8822 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
8823 free (temp2);
8824
8825arithsub:
8826 /* No error messages. */
a0c0a00f 8827 savecmd = this_command_name;
cce855bc
JA
8828 this_command_name = (char *)NULL;
8829 number = evalexp (temp1, &expok);
a0c0a00f 8830 this_command_name = savecmd;
cce855bc
JA
8831 free (temp);
8832 free (temp1);
8833 if (expok == 0)
8834 {
8835 if (interactive_shell == 0 && posixly_correct)
8836 {
8837 last_command_exit_value = EXECUTION_FAILURE;
95732b49 8838 return (&expand_wdesc_fatal);
cce855bc
JA
8839 }
8840 else
95732b49 8841 return (&expand_wdesc_error);
cce855bc
JA
8842 }
8843 temp = itos (number);
8844 break;
8845 }
8846
8847comsub:
b80f6443
JA
8848 if (pflags & PF_NOCOMSUB)
8849 /* we need zindex+1 because string[zindex] == RPAREN */
8850 temp1 = substring (string, *sindex, zindex+1);
8851 else
3185942a
JA
8852 {
8853 tdesc = command_substitute (temp, quoted);
8854 temp1 = tdesc ? tdesc->word : (char *)NULL;
8855 if (tdesc)
8856 dispose_word_desc (tdesc);
8857 }
cce855bc
JA
8858 FREE (temp);
8859 temp = temp1;
8860 break;
8861
8862 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
8863 away in a future bash release. */
8864 case '[':
bb70624e 8865 /* Extract the contents of this arithmetic substitution. */
cce855bc
JA
8866 t_index = zindex + 1;
8867 temp = extract_arithmetic_subst (string, &t_index);
8868 zindex = t_index;
3185942a
JA
8869 if (temp == 0)
8870 {
8871 temp = savestring (string);
8872 if (expanded_something)
8873 *expanded_something = 0;
8874 goto return0;
8875 }
cce855bc
JA
8876
8877 /* Do initial variable expansion. */
a0c0a00f 8878 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
8879
8880 goto arithsub;
8881
8882 default:
8883 /* Find the variable in VARIABLE_LIST. */
8884 temp = (char *)NULL;
8885
8886 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
8887 ;
8888 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
8889
8890 /* If this isn't a variable name, then just output the `$'. */
8891 if (temp1 == 0 || *temp1 == '\0')
8892 {
8893 FREE (temp1);
f73dda09 8894 temp = (char *)xmalloc (2);
cce855bc
JA
8895 temp[0] = '$';
8896 temp[1] = '\0';
8897 if (expanded_something)
8898 *expanded_something = 0;
8899 goto return0;
8900 }
8901
8902 /* If the variable exists, return its value cell. */
8903 var = find_variable (temp1);
8904
7117c2d2 8905 if (var && invisible_p (var) == 0 && var_isset (var))
cce855bc
JA
8906 {
8907#if defined (ARRAY_VARS)
3185942a 8908 if (assoc_p (var) || array_p (var))
cce855bc 8909 {
3185942a
JA
8910 temp = array_p (var) ? array_reference (array_cell (var), 0)
8911 : assoc_reference (assoc_cell (var), "0");
cce855bc 8912 if (temp)
b80f6443
JA
8913 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8914 ? quote_string (temp)
8915 : quote_escapes (temp);
8916 else if (unbound_vars_is_error)
8917 goto unbound_variable;
cce855bc
JA
8918 }
8919 else
8920#endif
b80f6443
JA
8921 {
8922 temp = value_cell (var);
8923
8924 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8925 ? quote_string (temp)
8926 : quote_escapes (temp);
8927 }
8928
cce855bc 8929 free (temp1);
7117c2d2 8930
cce855bc
JA
8931 goto return0;
8932 }
b64a0e1d
CR
8933 else if (var && (invisible_p (var) || var_isset (var) == 0))
8934 temp = (char *)NULL;
a0c0a00f 8935 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
ac50fbac
CR
8936 {
8937 temp = nameref_cell (var);
8938#if defined (ARRAY_VARS)
a0c0a00f 8939 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
8940 {
8941 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8942 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8943 return (tdesc);
8944 ret = tdesc;
8945 goto return0;
8946 }
8947 else
8948#endif
8949 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8950 if (temp && *temp && legal_identifier (temp) == 0)
8951 {
8952 last_command_exit_value = EXECUTION_FAILURE;
8953 report_error (_("%s: invalid variable name for name reference"), temp);
8954 return (&expand_wdesc_error); /* XXX */
8955 }
8956 else
8957 temp = (char *)NULL;
8958 }
cce855bc
JA
8959
8960 temp = (char *)NULL;
8961
b80f6443 8962unbound_variable:
cce855bc 8963 if (unbound_vars_is_error)
0001803f
CR
8964 {
8965 last_command_exit_value = EXECUTION_FAILURE;
8966 err_unboundvar (temp1);
8967 }
cce855bc
JA
8968 else
8969 {
8970 free (temp1);
8971 goto return0;
8972 }
8973
8974 free (temp1);
8975 last_command_exit_value = EXECUTION_FAILURE;
8976 return ((unbound_vars_is_error && interactive_shell == 0)
95732b49
JA
8977 ? &expand_wdesc_fatal
8978 : &expand_wdesc_error);
cce855bc
JA
8979 }
8980
8981 if (string[zindex])
8982 zindex++;
8983
8984return0:
8985 *sindex = zindex;
95732b49
JA
8986
8987 if (ret == 0)
8988 {
8989 ret = alloc_word_desc ();
8990 ret->flags = tflag; /* XXX */
8991 ret->word = temp;
8992 }
8993 return ret;
cce855bc
JA
8994}
8995
a0c0a00f
CR
8996void
8997invalidate_cached_quoted_dollar_at ()
8998{
8999 dispose_words (cached_quoted_dollar_at);
9000 cached_quoted_dollar_at = 0;
9001}
9002
cce855bc
JA
9003/* Make a word list which is the result of parameter and variable
9004 expansion, command substitution, arithmetic substitution, and
9005 quote removal of WORD. Return a pointer to a WORD_LIST which is
9006 the result of the expansion. If WORD contains a null word, the
9007 word list returned is also null.
726f6388 9008
ccc6cda3
JA
9009 QUOTED contains flag values defined in shell.h.
9010
b72432fd
JA
9011 ISEXP is used to tell expand_word_internal that the word should be
9012 treated as the result of an expansion. This has implications for
9013 how IFS characters in the word are treated.
9014
726f6388
JA
9015 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
9016 they point to an integer value which receives information about expansion.
9017 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
9018 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
9019 else zero.
9020
9021 This only does word splitting in the case of $@ expansion. In that
9022 case, we split on ' '. */
9023
9024/* Values for the local variable quoted_state. */
9025#define UNQUOTED 0
9026#define PARTIALLY_QUOTED 1
9027#define WHOLLY_QUOTED 2
9028
9029static WORD_LIST *
b72432fd 9030expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
726f6388 9031 WORD_DESC *word;
b72432fd 9032 int quoted, isexp;
726f6388
JA
9033 int *contains_dollar_at;
9034 int *expanded_something;
9035{
ccc6cda3
JA
9036 WORD_LIST *list;
9037 WORD_DESC *tword;
726f6388
JA
9038
9039 /* The intermediate string that we build while expanding. */
ccc6cda3 9040 char *istring;
726f6388
JA
9041
9042 /* The current size of the above object. */
a0c0a00f 9043 size_t istring_size;
726f6388
JA
9044
9045 /* Index into ISTRING. */
ccc6cda3 9046 int istring_index;
726f6388
JA
9047
9048 /* Temporary string storage. */
ccc6cda3 9049 char *temp, *temp1;
726f6388
JA
9050
9051 /* The text of WORD. */
ccc6cda3 9052 register char *string;
726f6388 9053
7117c2d2
JA
9054 /* The size of STRING. */
9055 size_t string_size;
9056
726f6388 9057 /* The index into STRING. */
ccc6cda3 9058 int sindex;
726f6388
JA
9059
9060 /* This gets 1 if we see a $@ while quoted. */
ccc6cda3 9061 int quoted_dollar_at;
726f6388
JA
9062
9063 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
9064 whether WORD contains no quoting characters, a partially quoted
9065 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
ccc6cda3
JA
9066 int quoted_state;
9067
95732b49 9068 /* State flags */
ccc6cda3 9069 int had_quoted_null;
0b913689 9070 int has_dollar_at, temp_has_dollar_at;
ac50fbac 9071 int split_on_spaces;
28ef6c31 9072 int tflag;
0001803f 9073 int pflags; /* flags passed to param_expand */
a0c0a00f 9074 int mb_cur_max;
726f6388 9075
95732b49
JA
9076 int assignoff; /* If assignment, offset of `=' */
9077
f73dda09 9078 register unsigned char c; /* Current character. */
726f6388 9079 int t_index; /* For calls to string_extract_xxx. */
726f6388 9080
bb70624e 9081 char twochars[2];
b72432fd 9082
7117c2d2
JA
9083 DECLARE_MBSTATE;
9084
a0c0a00f
CR
9085 /* OK, let's see if we can optimize a common idiom: "$@" */
9086 if (STREQ (word->word, "\"$@\"") &&
9087 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
9088 dollar_vars[1]) /* XXX - check IFS here as well? */
9089 {
9090 if (contains_dollar_at)
9091 *contains_dollar_at = 1;
9092 if (expanded_something)
9093 *expanded_something = 1;
9094 if (cached_quoted_dollar_at)
9095 return (copy_word_list (cached_quoted_dollar_at));
9096 list = list_rest_of_args ();
9097 list = quote_list (list);
9098 cached_quoted_dollar_at = copy_word_list (list);
9099 return (list);
9100 }
9101
f73dda09 9102 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
ccc6cda3 9103 istring[istring_index = 0] = '\0';
cce855bc 9104 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
ac50fbac 9105 split_on_spaces = 0;
ccc6cda3
JA
9106 quoted_state = UNQUOTED;
9107
9108 string = word->word;
9109 if (string == 0)
9110 goto finished_with_string;
a0c0a00f
CR
9111 mb_cur_max = MB_CUR_MAX;
9112
95732b49
JA
9113 /* Don't need the string length for the SADD... and COPY_ macros unless
9114 multibyte characters are possible. */
a0c0a00f 9115 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
726f6388
JA
9116
9117 if (contains_dollar_at)
9118 *contains_dollar_at = 0;
9119
95732b49
JA
9120 assignoff = -1;
9121
726f6388
JA
9122 /* Begin the expansion. */
9123
ccc6cda3 9124 for (sindex = 0; ;)
726f6388
JA
9125 {
9126 c = string[sindex];
9127
ac50fbac 9128 /* Case on top-level character. */
726f6388
JA
9129 switch (c)
9130 {
9131 case '\0':
9132 goto finished_with_string;
9133
9134 case CTLESC:
7117c2d2
JA
9135 sindex++;
9136#if HANDLE_MULTIBYTE
a0c0a00f 9137 if (mb_cur_max > 1 && string[sindex])
7117c2d2 9138 {
b80f6443 9139 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
9140 }
9141 else
9142#endif
9143 {
9144 temp = (char *)xmalloc (3);
9145 temp[0] = CTLESC;
9146 temp[1] = c = string[sindex];
9147 temp[2] = '\0';
9148 }
726f6388 9149
cce855bc 9150dollar_add_string:
726f6388
JA
9151 if (string[sindex])
9152 sindex++;
9153
cce855bc
JA
9154add_string:
9155 if (temp)
9156 {
9157 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
9158 temp = (char *)0;
9159 }
9160
9161 break;
726f6388
JA
9162
9163#if defined (PROCESS_SUBSTITUTION)
9164 /* Process substitution. */
9165 case '<':
9166 case '>':
9167 {
a0c0a00f
CR
9168 /* bash-4.4/bash-5.0
9169 XXX - technically this should only be expanded at the start
9170 of a word */
0628567a 9171 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
726f6388 9172 {
bb70624e 9173 sindex--; /* add_character: label increments sindex */
726f6388
JA
9174 goto add_character;
9175 }
9176 else
cce855bc 9177 t_index = sindex + 1; /* skip past both '<' and LPAREN */
726f6388 9178
85b94814 9179 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
ccc6cda3 9180 sindex = t_index;
726f6388
JA
9181
9182 /* If the process substitution specification is `<()', we want to
9183 open the pipe for writing in the child and produce output; if
9184 it is `>()', we want to open the pipe for reading in the child
9185 and consume input. */
ccc6cda3 9186 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
726f6388
JA
9187
9188 FREE (temp1);
9189
9190 goto dollar_add_string;
9191 }
9192#endif /* PROCESS_SUBSTITUTION */
9193
95732b49
JA
9194 case '=':
9195 /* Posix.2 section 3.6.1 says that tildes following `=' in words
9196 which are not assignment statements are not expanded. If the
9197 shell isn't in posix mode, though, we perform tilde expansion
9198 on `likely candidate' unquoted assignment statements (flags
9199 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
9200 contains an unquoted :~ or =~. Something to think about: we
9201 now have a flag that says to perform tilde expansion on arguments
9202 to `assignment builtins' like declare and export that look like
9203 assignment statements. We now do tilde expansion on such words
9204 even in POSIX mode. */
9205 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
17345e5a 9206 {
0001803f 9207 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9208 goto add_ifs_character;
9209 else
9210 goto add_character;
9211 }
95732b49
JA
9212 /* If we're not in posix mode or forcing assignment-statement tilde
9213 expansion, note where the `=' appears in the word and prepare to
9214 do tilde expansion following the first `='. */
9215 if ((word->flags & W_ASSIGNMENT) &&
9216 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9217 assignoff == -1 && sindex > 0)
9218 assignoff = sindex;
9219 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
9220 word->flags |= W_ITILDE;
9221#if 0
9222 else if ((word->flags & W_ASSIGNMENT) &&
9223 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9224 string[sindex+1] == '~')
9225 word->flags |= W_ITILDE;
9226#endif
a0c0a00f
CR
9227
9228 /* XXX - bash-4.4/bash-5.0 */
9229 if (word->flags & W_ASSIGNARG)
9230 word->flags |= W_ASSIGNRHS; /* affects $@ */
9231
0001803f 9232 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9233 goto add_ifs_character;
9234 else
9235 goto add_character;
95732b49
JA
9236
9237 case ':':
9238 if (word->flags & W_NOTILDE)
17345e5a 9239 {
0001803f 9240 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9241 goto add_ifs_character;
9242 else
9243 goto add_character;
9244 }
95732b49
JA
9245
9246 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
9247 string[sindex+1] == '~')
9248 word->flags |= W_ITILDE;
17345e5a 9249
0001803f 9250 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9251 goto add_ifs_character;
9252 else
9253 goto add_character;
95732b49
JA
9254
9255 case '~':
9256 /* If the word isn't supposed to be tilde expanded, or we're not
9257 at the start of a word or after an unquoted : or = in an
a0c0a00f
CR
9258 assignment statement, we don't do tilde expansion. If we don't want
9259 tilde expansion when expanding words to be passed to the arithmetic
9260 evaluator, remove the check for Q_ARITH. */
95732b49
JA
9261 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
9262 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
a0c0a00f 9263 ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0)))
95732b49
JA
9264 {
9265 word->flags &= ~W_ITILDE;
0001803f 9266 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
17345e5a
JA
9267 goto add_ifs_character;
9268 else
9269 goto add_character;
95732b49
JA
9270 }
9271
9272 if (word->flags & W_ASSIGNRHS)
9273 tflag = 2;
9274 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
9275 tflag = 1;
9276 else
9277 tflag = 0;
9278
9279 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
9280
9281 word->flags &= ~W_ITILDE;
9282
9283 if (temp && *temp && t_index > 0)
9284 {
9285 temp1 = bash_tilde_expand (temp, tflag);
0628567a
JA
9286 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
9287 {
9288 FREE (temp);
9289 FREE (temp1);
9290 goto add_character; /* tilde expansion failed */
9291 }
95732b49
JA
9292 free (temp);
9293 temp = temp1;
9294 sindex += t_index;
3185942a 9295 goto add_quoted_string; /* XXX was add_string */
95732b49
JA
9296 }
9297 else
9298 {
9299 FREE (temp);
9300 goto add_character;
9301 }
9302
726f6388 9303 case '$':
726f6388
JA
9304 if (expanded_something)
9305 *expanded_something = 1;
9306
0b913689 9307 temp_has_dollar_at = 0;
0001803f
CR
9308 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
9309 if (word->flags & W_NOSPLIT2)
9310 pflags |= PF_NOSPLIT2;
ac50fbac
CR
9311 if (word->flags & W_ASSIGNRHS)
9312 pflags |= PF_ASSIGNRHS;
a0c0a00f
CR
9313 if (word->flags & W_COMPLETE)
9314 pflags |= PF_COMPLETE;
95732b49 9315 tword = param_expand (string, &sindex, quoted, expanded_something,
0b913689 9316 &temp_has_dollar_at, &quoted_dollar_at,
0001803f 9317 &had_quoted_null, pflags);
0b913689 9318 has_dollar_at += temp_has_dollar_at;
ac50fbac 9319 split_on_spaces += (tword->flags & W_SPLITSPACE);
726f6388 9320
95732b49 9321 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
726f6388 9322 {
cce855bc
JA
9323 free (string);
9324 free (istring);
95732b49
JA
9325 return ((tword == &expand_wdesc_error) ? &expand_word_error
9326 : &expand_word_fatal);
cce855bc
JA
9327 }
9328 if (contains_dollar_at && has_dollar_at)
9329 *contains_dollar_at = 1;
95732b49
JA
9330
9331 if (tword && (tword->flags & W_HASQUOTEDNULL))
9332 had_quoted_null = 1;
9333
ac50fbac 9334 temp = tword ? tword->word : (char *)NULL;
95732b49
JA
9335 dispose_word_desc (tword);
9336
a601c749
CR
9337 /* Kill quoted nulls; we will add them back at the end of
9338 expand_word_internal if nothing else in the string */
9339 if (had_quoted_null && temp && QUOTED_NULL (temp))
9340 {
9341 FREE (temp);
9342 temp = (char *)NULL;
9343 }
9344
cce855bc
JA
9345 goto add_string;
9346 break;
726f6388 9347
cce855bc
JA
9348 case '`': /* Backquoted command substitution. */
9349 {
b80f6443 9350 t_index = sindex++;
726f6388 9351
3185942a 9352 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
95732b49
JA
9353 /* The test of sindex against t_index is to allow bare instances of
9354 ` to pass through, for backwards compatibility. */
9355 if (temp == &extract_string_error || temp == &extract_string_fatal)
9356 {
9357 if (sindex - 1 == t_index)
9358 {
9359 sindex = t_index;
9360 goto add_character;
9361 }
ac50fbac 9362 last_command_exit_value = EXECUTION_FAILURE;
3185942a 9363 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
95732b49
JA
9364 free (string);
9365 free (istring);
9366 return ((temp == &extract_string_error) ? &expand_word_error
9367 : &expand_word_fatal);
9368 }
9369
cce855bc
JA
9370 if (expanded_something)
9371 *expanded_something = 1;
726f6388 9372
b80f6443
JA
9373 if (word->flags & W_NOCOMSUB)
9374 /* sindex + 1 because string[sindex] == '`' */
9375 temp1 = substring (string, t_index, sindex + 1);
9376 else
9377 {
9378 de_backslash (temp);
3185942a
JA
9379 tword = command_substitute (temp, quoted);
9380 temp1 = tword ? tword->word : (char *)NULL;
9381 if (tword)
9382 dispose_word_desc (tword);
b80f6443 9383 }
cce855bc
JA
9384 FREE (temp);
9385 temp = temp1;
9386 goto dollar_add_string;
9387 }
ccc6cda3 9388
cce855bc
JA
9389 case '\\':
9390 if (string[sindex + 1] == '\n')
9391 {
9392 sindex += 2;
9393 continue;
9394 }
726f6388 9395
cce855bc 9396 c = string[++sindex];
726f6388 9397
cce855bc 9398 if (quoted & Q_HERE_DOCUMENT)
28ef6c31 9399 tflag = CBSHDOC;
cce855bc 9400 else if (quoted & Q_DOUBLE_QUOTES)
28ef6c31 9401 tflag = CBSDQUOTE;
cce855bc 9402 else
28ef6c31
JA
9403 tflag = 0;
9404
495aee44
CR
9405 /* From Posix discussion on austin-group list: Backslash escaping
9406 a } in ${...} is removed. Issue 0000221 */
9407 if ((quoted & Q_DOLBRACE) && c == RBRACE)
9408 {
ac50fbac
CR
9409 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9410 }
9411 /* This is the fix for " $@\ " */
9412 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
9413 {
9414 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9415 DEFAULT_ARRAY_SIZE);
9416 istring[istring_index++] = CTLESC;
9417 istring[istring_index++] = '\\';
9418 istring[istring_index] = '\0';
9419
495aee44
CR
9420 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9421 }
9422 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
cce855bc 9423 {
7117c2d2 9424 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
bb70624e
JA
9425 }
9426 else if (c == 0)
9427 {
9428 c = CTLNUL;
9429 sindex--; /* add_character: label increments sindex */
9430 goto add_character;
cce855bc
JA
9431 }
9432 else
bb70624e 9433 {
7117c2d2 9434 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
bb70624e 9435 }
726f6388 9436
bb70624e
JA
9437 sindex++;
9438add_twochars:
9439 /* BEFORE jumping here, we need to increment sindex if appropriate */
9440 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9441 DEFAULT_ARRAY_SIZE);
9442 istring[istring_index++] = twochars[0];
9443 istring[istring_index++] = twochars[1];
9444 istring[istring_index] = '\0';
9445
9446 break;
726f6388 9447
cce855bc 9448 case '"':
a0c0a00f 9449 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
cce855bc 9450 goto add_character;
ccc6cda3
JA
9451
9452 t_index = ++sindex;
a0c0a00f 9453 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
ccc6cda3
JA
9454
9455 /* If the quotes surrounded the entire string, then the
9456 whole word was quoted. */
9457 quoted_state = (t_index == 1 && string[sindex] == '\0')
9458 ? WHOLLY_QUOTED
7117c2d2 9459 : PARTIALLY_QUOTED;
ccc6cda3
JA
9460
9461 if (temp && *temp)
726f6388 9462 {
95732b49
JA
9463 tword = alloc_word_desc ();
9464 tword->word = temp;
9465
a0c0a00f
CR
9466 /* XXX - bash-4.4/bash-5.0 */
9467 if (word->flags & W_ASSIGNARG)
9468 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
9469 if (word->flags & W_COMPLETE)
9470 tword->flags |= W_COMPLETE; /* for command substitutions */
4f747edc
CR
9471 if (word->flags & W_NOCOMSUB)
9472 tword->flags |= W_NOCOMSUB;
9473 if (word->flags & W_NOPROCSUB)
9474 tword->flags |= W_NOPROCSUB;
a0c0a00f 9475
ccc6cda3
JA
9476 temp = (char *)NULL;
9477
0b913689 9478 temp_has_dollar_at = 0; /* XXX */
95732b49 9479 /* Need to get W_HASQUOTEDNULL flag through this function. */
0b913689
CR
9480 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
9481 has_dollar_at += temp_has_dollar_at;
726f6388 9482
ccc6cda3
JA
9483 if (list == &expand_word_error || list == &expand_word_fatal)
9484 {
9485 free (istring);
9486 free (string);
9487 /* expand_word_internal has already freed temp_word->word
9488 for us because of the way it prints error messages. */
9489 tword->word = (char *)NULL;
9490 dispose_word (tword);
9491 return list;
9492 }
726f6388 9493
ccc6cda3 9494 dispose_word (tword);
726f6388 9495
ccc6cda3
JA
9496 /* "$@" (a double-quoted dollar-at) expands into nothing,
9497 not even a NULL word, when there are no positional
9498 parameters. */
a0c0a00f 9499 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
726f6388 9500 {
ccc6cda3
JA
9501 quoted_dollar_at++;
9502 break;
9503 }
9504
9505 /* If we get "$@", we know we have expanded something, so we
9506 need to remember it for the final split on $IFS. This is
9507 a special case; it's the only case where a quoted string
9508 can expand into more than one word. It's going to come back
9509 from the above call to expand_word_internal as a list with
9510 a single word, in which all characters are quoted and
9511 separated by blanks. What we want to do is to turn it back
9512 into a list for the next piece of code. */
9513 if (list)
9514 dequote_list (list);
9515
95732b49 9516 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
ac50fbac 9517 had_quoted_null = 1; /* XXX */
95732b49 9518
a0c0a00f 9519 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
ccc6cda3
JA
9520 {
9521 quoted_dollar_at++;
9522 if (contains_dollar_at)
9523 *contains_dollar_at = 1;
9524 if (expanded_something)
9525 *expanded_something = 1;
9526 }
9527 }
9528 else
9529 {
9530 /* What we have is "". This is a minor optimization. */
f73dda09 9531 FREE (temp);
ccc6cda3
JA
9532 list = (WORD_LIST *)NULL;
9533 }
9534
9535 /* The code above *might* return a list (consider the case of "$@",
9536 where it returns "$1", "$2", etc.). We can't throw away the
9537 rest of the list, and we have to make sure each word gets added
9538 as quoted. We test on tresult->next: if it is non-NULL, we
9539 quote the whole list, save it to a string with string_list, and
9540 add that string. We don't need to quote the results of this
9541 (and it would be wrong, since that would quote the separators
9542 as well), so we go directly to add_string. */
9543 if (list)
9544 {
9545 if (list->next)
9546 {
bc4cd23c
JA
9547 /* Testing quoted_dollar_at makes sure that "$@" is
9548 split correctly when $IFS does not contain a space. */
9549 temp = quoted_dollar_at
a0c0a00f 9550 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
bc4cd23c 9551 : string_list (quote_list (list));
ccc6cda3 9552 dispose_words (list);
726f6388
JA
9553 goto add_string;
9554 }
9555 else
9556 {
ccc6cda3 9557 temp = savestring (list->word->word);
95732b49 9558 tflag = list->word->flags;
ccc6cda3 9559 dispose_words (list);
95732b49 9560
cce855bc
JA
9561 /* If the string is not a quoted null string, we want
9562 to remove any embedded unquoted CTLNUL characters.
9563 We do not want to turn quoted null strings back into
9564 the empty string, though. We do this because we
9565 want to remove any quoted nulls from expansions that
9566 contain other characters. For example, if we have
9567 x"$*"y or "x$*y" and there are no positional parameters,
7117c2d2 9568 the $* should expand into nothing. */
95732b49
JA
9569 /* We use the W_HASQUOTEDNULL flag to differentiate the
9570 cases: a quoted null character as above and when
9571 CTLNUL is contained in the (non-null) expansion
9572 of some variable. We use the had_quoted_null flag to
9573 pass the value through this function to its caller. */
9574 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
cce855bc 9575 remove_quoted_nulls (temp); /* XXX */
726f6388
JA
9576 }
9577 }
ccc6cda3
JA
9578 else
9579 temp = (char *)NULL;
726f6388 9580
ccc6cda3 9581 /* We do not want to add quoted nulls to strings that are only
ac50fbac
CR
9582 partially quoted; we can throw them away. The exception to
9583 this is when we are going to be performing word splitting,
9584 since we have to preserve a null argument if the next character
9585 will cause word splitting. */
495aee44 9586 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
cce855bc 9587 continue;
726f6388 9588
ccc6cda3 9589 add_quoted_string:
726f6388 9590
ccc6cda3
JA
9591 if (temp)
9592 {
9593 temp1 = temp;
9594 temp = quote_string (temp);
9595 free (temp1);
bb70624e 9596 goto add_string;
ccc6cda3
JA
9597 }
9598 else
9599 {
9600 /* Add NULL arg. */
bb70624e
JA
9601 c = CTLNUL;
9602 sindex--; /* add_character: label increments sindex */
9603 goto add_character;
ccc6cda3 9604 }
bb70624e 9605
ccc6cda3 9606 /* break; */
726f6388 9607
ccc6cda3 9608 case '\'':
95732b49 9609 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
ccc6cda3 9610 goto add_character;
726f6388 9611
ccc6cda3
JA
9612 t_index = ++sindex;
9613 temp = string_extract_single_quoted (string, &sindex);
726f6388 9614
ccc6cda3
JA
9615 /* If the entire STRING was surrounded by single quotes,
9616 then the string is wholly quoted. */
9617 quoted_state = (t_index == 1 && string[sindex] == '\0')
9618 ? WHOLLY_QUOTED
7117c2d2 9619 : PARTIALLY_QUOTED;
726f6388 9620
ccc6cda3
JA
9621 /* If all we had was '', it is a null expansion. */
9622 if (*temp == '\0')
9623 {
9624 free (temp);
9625 temp = (char *)NULL;
9626 }
9627 else
7117c2d2 9628 remove_quoted_escapes (temp); /* ??? */
726f6388 9629
ccc6cda3 9630 /* We do not want to add quoted nulls to strings that are only
a0c0a00f
CR
9631 partially quoted; such nulls are discarded. See above for the
9632 exception, which is when the string is going to be split. */
9633 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
ccc6cda3 9634 continue;
726f6388 9635
bb70624e
JA
9636 /* If we have a quoted null expansion, add a quoted NULL to istring. */
9637 if (temp == 0)
9638 {
9639 c = CTLNUL;
9640 sindex--; /* add_character: label increments sindex */
9641 goto add_character;
9642 }
9643 else
9644 goto add_quoted_string;
9645
ccc6cda3 9646 /* break; */
726f6388
JA
9647
9648 default:
726f6388 9649 /* This is the fix for " $@ " */
17345e5a 9650 add_ifs_character:
a0c0a00f 9651 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
726f6388 9652 {
bb70624e
JA
9653 if (string[sindex]) /* from old goto dollar_add_string */
9654 sindex++;
9655 if (c == 0)
9656 {
9657 c = CTLNUL;
9658 goto add_character;
9659 }
9660 else
9661 {
7117c2d2 9662#if HANDLE_MULTIBYTE
a0c0a00f 9663 if (mb_cur_max > 1)
b80f6443
JA
9664 sindex--;
9665
a0c0a00f 9666 if (mb_cur_max > 1)
7117c2d2 9667 {
b80f6443 9668 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
9669 }
9670 else
9671#endif
9672 {
9673 twochars[0] = CTLESC;
9674 twochars[1] = c;
9675 goto add_twochars;
9676 }
bb70624e 9677 }
726f6388
JA
9678 }
9679
7117c2d2
JA
9680 SADD_MBCHAR (temp, string, sindex, string_size);
9681
726f6388 9682 add_character:
ccc6cda3
JA
9683 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
9684 DEFAULT_ARRAY_SIZE);
726f6388
JA
9685 istring[istring_index++] = c;
9686 istring[istring_index] = '\0';
9687
9688 /* Next character. */
9689 sindex++;
9690 }
9691 }
9692
9693finished_with_string:
726f6388
JA
9694 /* OK, we're ready to return. If we have a quoted string, and
9695 quoted_dollar_at is not set, we do no splitting at all; otherwise
9696 we split on ' '. The routines that call this will handle what to
9697 do if nothing has been expanded. */
ccc6cda3
JA
9698
9699 /* Partially and wholly quoted strings which expand to the empty
9700 string are retained as an empty arguments. Unquoted strings
9701 which expand to the empty string are discarded. The single
9702 exception is the case of expanding "$@" when there are no
9703 positional parameters. In that case, we discard the expansion. */
9704
9705 /* Because of how the code that handles "" and '' in partially
9706 quoted strings works, we need to make ISTRING into a QUOTED_NULL
9707 if we saw quoting characters, but the expansion was empty.
9708 "" and '' are tossed away before we get to this point when
9709 processing partially quoted strings. This makes "" and $xxx""
9710 equivalent when xxx is unset. We also look to see whether we
9711 saw a quoted null from a ${} expansion and add one back if we
9712 need to. */
9713
9714 /* If we expand to nothing and there were no single or double quotes
9715 in the word, we throw it away. Otherwise, we return a NULL word.
9716 The single exception is for $@ surrounded by double quotes when
9717 there are no positional parameters. In that case, we also throw
9718 the word away. */
9719
9720 if (*istring == '\0')
9721 {
9722 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
726f6388 9723 {
726f6388
JA
9724 istring[0] = CTLNUL;
9725 istring[1] = '\0';
ccc6cda3 9726 tword = make_bare_word (istring);
95732b49 9727 tword->flags |= W_HASQUOTEDNULL; /* XXX */
ccc6cda3
JA
9728 list = make_word_list (tword, (WORD_LIST *)NULL);
9729 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9730 tword->flags |= W_QUOTED;
726f6388 9731 }
ccc6cda3
JA
9732 /* According to sh, ksh, and Posix.2, if a word expands into nothing
9733 and a double-quoted "$@" appears anywhere in it, then the entire
9734 word is removed. */
a0c0a00f
CR
9735 /* XXX - exception appears to be that quoted null strings result in
9736 null arguments */
ccc6cda3
JA
9737 else if (quoted_state == UNQUOTED || quoted_dollar_at)
9738 list = (WORD_LIST *)NULL;
9739#if 0
9740 else
726f6388 9741 {
ccc6cda3 9742 tword = make_bare_word (istring);
ccc6cda3
JA
9743 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9744 tword->flags |= W_QUOTED;
95732b49 9745 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 9746 }
f73dda09
JA
9747#else
9748 else
9749 list = (WORD_LIST *)NULL;
ccc6cda3
JA
9750#endif
9751 }
9752 else if (word->flags & W_NOSPLIT)
9753 {
9754 tword = make_bare_word (istring);
ccc6cda3
JA
9755 if (word->flags & W_ASSIGNMENT)
9756 tword->flags |= W_ASSIGNMENT; /* XXX */
95732b49
JA
9757 if (word->flags & W_COMPASSIGN)
9758 tword->flags |= W_COMPASSIGN; /* XXX */
b72432fd
JA
9759 if (word->flags & W_NOGLOB)
9760 tword->flags |= W_NOGLOB; /* XXX */
ac50fbac
CR
9761 if (word->flags & W_NOBRACE)
9762 tword->flags |= W_NOBRACE; /* XXX */
95732b49
JA
9763 if (word->flags & W_NOEXPAND)
9764 tword->flags |= W_NOEXPAND; /* XXX */
ccc6cda3 9765 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
28ef6c31 9766 tword->flags |= W_QUOTED;
a601c749 9767 if (had_quoted_null && QUOTED_NULL (istring))
95732b49
JA
9768 tword->flags |= W_HASQUOTEDNULL;
9769 list = make_word_list (tword, (WORD_LIST *)NULL);
ccc6cda3
JA
9770 }
9771 else
9772 {
9773 char *ifs_chars;
3b34f6e6 9774 char *tstring;
ccc6cda3 9775
7117c2d2 9776 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
726f6388 9777
cce855bc
JA
9778 /* If we have $@, we need to split the results no matter what. If
9779 IFS is unset or NULL, string_list_dollar_at has separated the
9780 positional parameters with a space, so we split on space (we have
9781 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
9782 string_list_dollar_at has separated the positional parameters
ac50fbac
CR
9783 with the first character of $IFS, so we split on $IFS. If
9784 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
9785 unset or null, and we want to make sure that we split on spaces
a0c0a00f
CR
9786 regardless of what else has happened to IFS since the expansion,
9787 or we expanded "$@" with IFS null and we need to split the positional
9788 parameters into separate words. */
ac50fbac
CR
9789 if (split_on_spaces)
9790 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
3b34f6e6
CR
9791 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
9792 don't want to split the result (W_NOSPLIT2), and we are not quoted,
9793 we have already separated the arguments with the first character of
9794 $IFS. In this case, we want to return a list with a single word
9795 with the separator possibly replaced with a space (it's what other
9796 shells seem to do).
9797 quoted_dollar_at is internal to this function and is set if we are
9798 passed an argument that is unquoted (quoted == 0) but we encounter a
9799 double-quoted $@ while expanding it. */
9800 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
9801 {
9802 /* Only split and rejoin if we have to */
9803 if (*ifs_chars && *ifs_chars != ' ')
9804 {
9805 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
9806 tstring = string_list (list);
9807 }
9808 else
9809 tstring = istring;
9810 tword = make_bare_word (tstring);
9811 if (tstring != istring)
9812 free (tstring);
9813 goto set_word_flags;
9814 }
a0c0a00f
CR
9815 /* This is the attempt to make $* in an assignment context (a=$*) and
9816 array variables subscripted with * in an assignment context (a=${foo[*]})
9817 behave similarly. It has side effects that, though they increase
9818 compatibility with other shells, are not backwards compatible. */
9819#if 0
9820 else if (has_dollar_at && quoted == 0 && ifs_chars && (word->flags & W_ASSIGNRHS))
9821 {
9822 tword = make_bare_word (istring);
9823 goto set_word_flags;
9824 }
9825#endif
ac50fbac 9826 else if (has_dollar_at && ifs_chars)
cce855bc 9827 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
ccc6cda3
JA
9828 else
9829 {
9830 tword = make_bare_word (istring);
3b34f6e6 9831set_word_flags:
ccc6cda3
JA
9832 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
9833 tword->flags |= W_QUOTED;
9834 if (word->flags & W_ASSIGNMENT)
9835 tword->flags |= W_ASSIGNMENT;
95732b49
JA
9836 if (word->flags & W_COMPASSIGN)
9837 tword->flags |= W_COMPASSIGN;
b72432fd
JA
9838 if (word->flags & W_NOGLOB)
9839 tword->flags |= W_NOGLOB;
ac50fbac
CR
9840 if (word->flags & W_NOBRACE)
9841 tword->flags |= W_NOBRACE;
95732b49
JA
9842 if (word->flags & W_NOEXPAND)
9843 tword->flags |= W_NOEXPAND;
a601c749 9844 if (had_quoted_null && QUOTED_NULL (istring))
95732b49
JA
9845 tword->flags |= W_HASQUOTEDNULL; /* XXX */
9846 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 9847 }
726f6388 9848 }
726f6388 9849
ccc6cda3
JA
9850 free (istring);
9851 return (list);
726f6388
JA
9852}
9853
9854/* **************************************************************** */
9855/* */
9856/* Functions for Quote Removal */
9857/* */
9858/* **************************************************************** */
9859
9860/* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
7117c2d2 9861 backslash quoting rules for within double quotes or a here document. */
726f6388
JA
9862char *
9863string_quote_removal (string, quoted)
9864 char *string;
9865 int quoted;
9866{
7117c2d2
JA
9867 size_t slen;
9868 char *r, *result_string, *temp, *send;
f73dda09
JA
9869 int sindex, tindex, dquote;
9870 unsigned char c;
7117c2d2 9871 DECLARE_MBSTATE;
726f6388
JA
9872
9873 /* The result can be no longer than the original string. */
7117c2d2
JA
9874 slen = strlen (string);
9875 send = string + slen;
9876
9877 r = result_string = (char *)xmalloc (slen + 1);
726f6388 9878
ccc6cda3 9879 for (dquote = sindex = 0; c = string[sindex];)
726f6388
JA
9880 {
9881 switch (c)
9882 {
9883 case '\\':
9884 c = string[++sindex];
3185942a
JA
9885 if (c == 0)
9886 {
9887 *r++ = '\\';
9888 break;
9889 }
28ef6c31 9890 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
726f6388 9891 *r++ = '\\';
ccc6cda3 9892 /* FALLTHROUGH */
726f6388
JA
9893
9894 default:
7117c2d2 9895 SCOPY_CHAR_M (r, string, send, sindex);
726f6388
JA
9896 break;
9897
9898 case '\'':
ccc6cda3 9899 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
726f6388
JA
9900 {
9901 *r++ = c;
9902 sindex++;
ccc6cda3 9903 break;
726f6388 9904 }
ccc6cda3
JA
9905 tindex = sindex + 1;
9906 temp = string_extract_single_quoted (string, &tindex);
9907 if (temp)
726f6388 9908 {
ccc6cda3
JA
9909 strcpy (r, temp);
9910 r += strlen (r);
9911 free (temp);
726f6388 9912 }
ccc6cda3 9913 sindex = tindex;
726f6388
JA
9914 break;
9915
9916 case '"':
9917 dquote = 1 - dquote;
9918 sindex++;
9919 break;
9920 }
9921 }
9922 *r = '\0';
9923 return (result_string);
9924}
9925
ccc6cda3
JA
9926#if 0
9927/* UNUSED */
726f6388
JA
9928/* Perform quote removal on word WORD. This allocates and returns a new
9929 WORD_DESC *. */
9930WORD_DESC *
9931word_quote_removal (word, quoted)
9932 WORD_DESC *word;
9933 int quoted;
9934{
9935 WORD_DESC *w;
9936 char *t;
9937
9938 t = string_quote_removal (word->word, quoted);
95732b49
JA
9939 w = alloc_word_desc ();
9940 w->word = t ? t : savestring ("");
726f6388
JA
9941 return (w);
9942}
9943
9944/* Perform quote removal on all words in LIST. If QUOTED is non-zero,
9945 the members of the list are treated as if they are surrounded by
9946 double quotes. Return a new list, or NULL if LIST is NULL. */
9947WORD_LIST *
9948word_list_quote_removal (list, quoted)
9949 WORD_LIST *list;
9950 int quoted;
9951{
95732b49 9952 WORD_LIST *result, *t, *tresult, *e;
726f6388 9953
ccc6cda3 9954 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 9955 {
7117c2d2 9956 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
95732b49 9957#if 0
726f6388 9958 result = (WORD_LIST *) list_append (result, tresult);
95732b49
JA
9959#else
9960 if (result == 0)
9961 result = e = tresult;
9962 else
9963 {
9964 e->next = tresult;
9965 while (e->next)
9966 e = e->next;
9967 }
9968#endif
726f6388
JA
9969 }
9970 return (result);
9971}
ccc6cda3 9972#endif
726f6388 9973
726f6388
JA
9974/*******************************************
9975 * *
9976 * Functions to perform word splitting *
9977 * *
9978 *******************************************/
9979
7117c2d2
JA
9980void
9981setifs (v)
9982 SHELL_VAR *v;
b72432fd 9983{
7117c2d2
JA
9984 char *t;
9985 unsigned char uc;
9986
9987 ifs_var = v;
95732b49 9988 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
b72432fd 9989
ac50fbac
CR
9990 ifs_is_set = ifs_var != 0;
9991 ifs_is_null = ifs_is_set && (*ifs_value == 0);
9992
95732b49
JA
9993 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9994 handle multibyte chars in IFS */
7117c2d2
JA
9995 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9996 for (t = ifs_value ; t && *t; t++)
9997 {
9998 uc = *t;
9999 ifs_cmap[uc] = 1;
10000 }
10001
95732b49
JA
10002#if defined (HANDLE_MULTIBYTE)
10003 if (ifs_value == 0)
10004 {
10005 ifs_firstc[0] = '\0';
10006 ifs_firstc_len = 1;
10007 }
10008 else
10009 {
10010 size_t ifs_len;
10011 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
10012 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
10013 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
10014 {
10015 ifs_firstc[0] = ifs_value[0];
10016 ifs_firstc[1] = '\0';
10017 ifs_firstc_len = 1;
10018 }
10019 else
10020 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
10021 }
10022#else
7117c2d2 10023 ifs_firstc = ifs_value ? *ifs_value : 0;
95732b49 10024#endif
7117c2d2
JA
10025}
10026
10027char *
10028getifs ()
10029{
10030 return ifs_value;
b72432fd
JA
10031}
10032
726f6388
JA
10033/* This splits a single word into a WORD LIST on $IFS, but only if the word
10034 is not quoted. list_string () performs quote removal for us, even if we
10035 don't do any splitting. */
10036WORD_LIST *
7117c2d2 10037word_split (w, ifs_chars)
726f6388 10038 WORD_DESC *w;
7117c2d2 10039 char *ifs_chars;
726f6388
JA
10040{
10041 WORD_LIST *result;
10042
10043 if (w)
10044 {
7117c2d2 10045 char *xifs;
726f6388 10046
7117c2d2
JA
10047 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
10048 result = list_string (w->word, xifs, w->flags & W_QUOTED);
726f6388
JA
10049 }
10050 else
10051 result = (WORD_LIST *)NULL;
ccc6cda3 10052
726f6388
JA
10053 return (result);
10054}
10055
10056/* Perform word splitting on LIST and return the RESULT. It is possible
10057 to return (WORD_LIST *)NULL. */
10058static WORD_LIST *
10059word_list_split (list)
10060 WORD_LIST *list;
10061{
95732b49 10062 WORD_LIST *result, *t, *tresult, *e;
726f6388 10063
ccc6cda3 10064 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 10065 {
7117c2d2 10066 tresult = word_split (t->word, ifs_value);
95732b49
JA
10067 if (result == 0)
10068 result = e = tresult;
10069 else
10070 {
10071 e->next = tresult;
10072 while (e->next)
10073 e = e->next;
10074 }
726f6388
JA
10075 }
10076 return (result);
10077}
10078
10079/**************************************************
10080 * *
cce855bc 10081 * Functions to expand an entire WORD_LIST *
726f6388
JA
10082 * *
10083 **************************************************/
10084
b80f6443
JA
10085/* Do any word-expansion-specific cleanup and jump to top_level */
10086static void
10087exp_jump_to_top_level (v)
10088 int v;
10089{
3185942a
JA
10090 set_pipestatus_from_exit (last_command_exit_value);
10091
b80f6443
JA
10092 /* Cleanup code goes here. */
10093 expand_no_split_dollar_star = 0; /* XXX */
10094 expanding_redir = 0;
3185942a 10095 assigning_in_environment = 0;
b80f6443 10096
f1be666c
JA
10097 if (parse_and_execute_level == 0)
10098 top_level_cleanup (); /* from sig.c */
10099
b80f6443
JA
10100 jump_to_top_level (v);
10101}
10102
cce855bc
JA
10103/* Put NLIST (which is a WORD_LIST * of only one element) at the front of
10104 ELIST, and set ELIST to the new list. */
10105#define PREPEND_LIST(nlist, elist) \
10106 do { nlist->next = elist; elist = nlist; } while (0)
10107
726f6388
JA
10108/* Separate out any initial variable assignments from TLIST. If set -k has
10109 been executed, remove all assignment statements from TLIST. Initial
10110 variable assignments and other environment assignments are placed
bb70624e 10111 on SUBST_ASSIGN_VARLIST. */
726f6388
JA
10112static WORD_LIST *
10113separate_out_assignments (tlist)
10114 WORD_LIST *tlist;
10115{
10116 register WORD_LIST *vp, *lp;
10117
0001803f 10118 if (tlist == 0)
726f6388
JA
10119 return ((WORD_LIST *)NULL);
10120
bb70624e
JA
10121 if (subst_assign_varlist)
10122 dispose_words (subst_assign_varlist); /* Clean up after previous error */
b72432fd 10123
bb70624e 10124 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
10125 vp = lp = tlist;
10126
10127 /* Separate out variable assignments at the start of the command.
10128 Loop invariant: vp->next == lp
10129 Loop postcondition:
7117c2d2
JA
10130 lp = list of words left after assignment statements skipped
10131 tlist = original list of words
726f6388 10132 */
ccc6cda3 10133 while (lp && (lp->word->flags & W_ASSIGNMENT))
726f6388
JA
10134 {
10135 vp = lp;
10136 lp = lp->next;
10137 }
10138
bb70624e
JA
10139 /* If lp != tlist, we have some initial assignment statements.
10140 We make SUBST_ASSIGN_VARLIST point to the list of assignment
10141 words and TLIST point to the remaining words. */
726f6388
JA
10142 if (lp != tlist)
10143 {
bb70624e 10144 subst_assign_varlist = tlist;
726f6388
JA
10145 /* ASSERT(vp->next == lp); */
10146 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
10147 tlist = lp; /* remainder of word list */
10148 }
10149
10150 /* vp == end of variable list */
10151 /* tlist == remainder of original word list without variable assignments */
10152 if (!tlist)
10153 /* All the words in tlist were assignment statements */
10154 return ((WORD_LIST *)NULL);
10155
10156 /* ASSERT(tlist != NULL); */
ccc6cda3 10157 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
726f6388
JA
10158
10159 /* If the -k option is in effect, we need to go through the remaining
bb70624e
JA
10160 words, separate out the assignment words, and place them on
10161 SUBST_ASSIGN_VARLIST. */
726f6388
JA
10162 if (place_keywords_in_env)
10163 {
10164 WORD_LIST *tp; /* tp == running pointer into tlist */
10165
10166 tp = tlist;
10167 lp = tlist->next;
10168
10169 /* Loop Invariant: tp->next == lp */
10170 /* Loop postcondition: tlist == word list without assignment statements */
10171 while (lp)
10172 {
ccc6cda3 10173 if (lp->word->flags & W_ASSIGNMENT)
726f6388
JA
10174 {
10175 /* Found an assignment statement, add this word to end of
bb70624e
JA
10176 subst_assign_varlist (vp). */
10177 if (!subst_assign_varlist)
10178 subst_assign_varlist = vp = lp;
726f6388
JA
10179 else
10180 {
10181 vp->next = lp;
10182 vp = lp;
10183 }
10184
10185 /* Remove the word pointed to by LP from TLIST. */
10186 tp->next = lp->next;
10187 /* ASSERT(vp == lp); */
10188 lp->next = (WORD_LIST *)NULL;
10189 lp = tp->next;
10190 }
10191 else
10192 {
10193 tp = lp;
10194 lp = lp->next;
10195 }
10196 }
10197 }
10198 return (tlist);
10199}
10200
cce855bc
JA
10201#define WEXP_VARASSIGN 0x001
10202#define WEXP_BRACEEXP 0x002
10203#define WEXP_TILDEEXP 0x004
10204#define WEXP_PARAMEXP 0x008
10205#define WEXP_PATHEXP 0x010
10206
10207/* All of the expansions, including variable assignments at the start of
10208 the list. */
10209#define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10210
10211/* All of the expansions except variable assignments at the start of
10212 the list. */
10213#define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10214
10215/* All of the `shell expansions': brace expansion, tilde expansion, parameter
10216 expansion, command substitution, arithmetic expansion, word splitting, and
10217 quote removal. */
10218#define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
10219
726f6388
JA
10220/* Take the list of words in LIST and do the various substitutions. Return
10221 a new list of words which is the expanded list, and without things like
10222 variable assignments. */
10223
10224WORD_LIST *
10225expand_words (list)
10226 WORD_LIST *list;
10227{
cce855bc 10228 return (expand_word_list_internal (list, WEXP_ALL));
726f6388
JA
10229}
10230
10231/* Same as expand_words (), but doesn't hack variable or environment
10232 variables. */
10233WORD_LIST *
10234expand_words_no_vars (list)
10235 WORD_LIST *list;
10236{
cce855bc 10237 return (expand_word_list_internal (list, WEXP_NOVARS));
726f6388
JA
10238}
10239
cce855bc
JA
10240WORD_LIST *
10241expand_words_shellexp (list)
726f6388 10242 WORD_LIST *list;
726f6388 10243{
cce855bc
JA
10244 return (expand_word_list_internal (list, WEXP_SHELLEXP));
10245}
726f6388 10246
cce855bc
JA
10247static WORD_LIST *
10248glob_expand_word_list (tlist, eflags)
10249 WORD_LIST *tlist;
10250 int eflags;
10251{
10252 char **glob_array, *temp_string;
10253 register int glob_index;
10254 WORD_LIST *glob_list, *output_list, *disposables, *next;
10255 WORD_DESC *tword;
726f6388 10256
cce855bc
JA
10257 output_list = disposables = (WORD_LIST *)NULL;
10258 glob_array = (char **)NULL;
10259 while (tlist)
10260 {
10261 /* For each word, either globbing is attempted or the word is
10262 added to orig_list. If globbing succeeds, the results are
10263 added to orig_list and the word (tlist) is added to the list
10264 of disposable words. If globbing fails and failed glob
10265 expansions are left unchanged (the shell default), the
10266 original word is added to orig_list. If globbing fails and
10267 failed glob expansions are removed, the original word is
10268 added to the list of disposable words. orig_list ends up
7117c2d2 10269 in reverse order and requires a call to REVERSE_LIST to
cce855bc
JA
10270 be set right. After all words are examined, the disposable
10271 words are freed. */
10272 next = tlist->next;
726f6388 10273
cce855bc 10274 /* If the word isn't an assignment and contains an unquoted
28ef6c31 10275 pattern matching character, then glob it. */
b72432fd 10276 if ((tlist->word->flags & W_NOGLOB) == 0 &&
cce855bc 10277 unquoted_glob_pattern_p (tlist->word->word))
726f6388 10278 {
cce855bc
JA
10279 glob_array = shell_glob_filename (tlist->word->word);
10280
10281 /* Handle error cases.
10282 I don't think we should report errors like "No such file
10283 or directory". However, I would like to report errors
10284 like "Read failed". */
10285
b80f6443 10286 if (glob_array == 0 || GLOB_FAILED (glob_array))
726f6388 10287 {
bb70624e 10288 glob_array = (char **)xmalloc (sizeof (char *));
cce855bc
JA
10289 glob_array[0] = (char *)NULL;
10290 }
10291
10292 /* Dequote the current word in case we have to use it. */
10293 if (glob_array[0] == NULL)
10294 {
10295 temp_string = dequote_string (tlist->word->word);
10296 free (tlist->word->word);
10297 tlist->word->word = temp_string;
10298 }
10299
10300 /* Make the array into a word list. */
10301 glob_list = (WORD_LIST *)NULL;
10302 for (glob_index = 0; glob_array[glob_index]; glob_index++)
10303 {
10304 tword = make_bare_word (glob_array[glob_index]);
cce855bc
JA
10305 glob_list = make_word_list (tword, glob_list);
10306 }
10307
10308 if (glob_list)
10309 {
10310 output_list = (WORD_LIST *)list_append (glob_list, output_list);
10311 PREPEND_LIST (tlist, disposables);
10312 }
b80f6443
JA
10313 else if (fail_glob_expansion != 0)
10314 {
ac50fbac 10315 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 10316 report_error (_("no match: %s"), tlist->word->word);
f1be666c 10317 exp_jump_to_top_level (DISCARD);
b80f6443 10318 }
cce855bc
JA
10319 else if (allow_null_glob_expansion == 0)
10320 {
10321 /* Failed glob expressions are left unchanged. */
10322 PREPEND_LIST (tlist, output_list);
10323 }
10324 else
10325 {
10326 /* Failed glob expressions are removed. */
10327 PREPEND_LIST (tlist, disposables);
726f6388 10328 }
726f6388 10329 }
cce855bc
JA
10330 else
10331 {
10332 /* Dequote the string. */
10333 temp_string = dequote_string (tlist->word->word);
10334 free (tlist->word->word);
10335 tlist->word->word = temp_string;
10336 PREPEND_LIST (tlist, output_list);
10337 }
10338
7117c2d2 10339 strvec_dispose (glob_array);
cce855bc
JA
10340 glob_array = (char **)NULL;
10341
10342 tlist = next;
726f6388
JA
10343 }
10344
cce855bc
JA
10345 if (disposables)
10346 dispose_words (disposables);
10347
10348 if (output_list)
10349 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10350
10351 return (output_list);
10352}
726f6388
JA
10353
10354#if defined (BRACE_EXPANSION)
cce855bc
JA
10355static WORD_LIST *
10356brace_expand_word_list (tlist, eflags)
10357 WORD_LIST *tlist;
10358 int eflags;
10359{
10360 register char **expansions;
10361 char *temp_string;
10362 WORD_LIST *disposables, *output_list, *next;
10363 WORD_DESC *w;
10364 int eindex;
10365
10366 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
726f6388 10367 {
cce855bc 10368 next = tlist->next;
726f6388 10369
ac50fbac
CR
10370 if (tlist->word->flags & W_NOBRACE)
10371 {
10372/*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
10373 PREPEND_LIST (tlist, output_list);
10374 continue;
10375 }
10376
0001803f
CR
10377 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10378 {
10379/*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
10380 PREPEND_LIST (tlist, output_list);
10381 continue;
10382 }
ac50fbac 10383
cce855bc
JA
10384 /* Only do brace expansion if the word has a brace character. If
10385 not, just add the word list element to BRACES and continue. In
10386 the common case, at least when running shell scripts, this will
0001803f 10387 degenerate to a bunch of calls to `mbschr', and then what is
cce855bc 10388 basically a reversal of TLIST into BRACES, which is corrected
7117c2d2 10389 by a call to REVERSE_LIST () on BRACES when the end of TLIST
cce855bc 10390 is reached. */
0001803f 10391 if (mbschr (tlist->word->word, LBRACE))
726f6388 10392 {
cce855bc 10393 expansions = brace_expand (tlist->word->word);
726f6388 10394
cce855bc 10395 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
726f6388 10396 {
ac50fbac
CR
10397 w = alloc_word_desc ();
10398 w->word = temp_string;
10399
cce855bc
JA
10400 /* If brace expansion didn't change the word, preserve
10401 the flags. We may want to preserve the flags
10402 unconditionally someday -- XXX */
10403 if (STREQ (temp_string, tlist->word->word))
10404 w->flags = tlist->word->flags;
ac50fbac
CR
10405 else
10406 w = make_word_flags (w, temp_string);
10407
cce855bc 10408 output_list = make_word_list (w, output_list);
726f6388 10409 }
cce855bc 10410 free (expansions);
726f6388 10411
cce855bc
JA
10412 /* Add TLIST to the list of words to be freed after brace
10413 expansion has been performed. */
10414 PREPEND_LIST (tlist, disposables);
10415 }
10416 else
10417 PREPEND_LIST (tlist, output_list);
726f6388 10418 }
cce855bc
JA
10419
10420 if (disposables)
10421 dispose_words (disposables);
10422
10423 if (output_list)
10424 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10425
10426 return (output_list);
10427}
10428#endif
10429
3185942a
JA
10430#if defined (ARRAY_VARS)
10431/* Take WORD, a compound associative array assignment, and internally run
10432 'declare -A w', where W is the variable name portion of WORD. */
10433static int
a0c0a00f 10434make_internal_declare (word, option, cmd)
3185942a
JA
10435 char *word;
10436 char *option;
a0c0a00f 10437 char *cmd;
3185942a 10438{
a0c0a00f 10439 int t, r;
3185942a
JA
10440 WORD_LIST *wl;
10441 WORD_DESC *w;
10442
10443 w = make_word (word);
10444
10445 t = assignment (w->word, 0);
a0c0a00f
CR
10446 if (w->word[t] == '=')
10447 {
10448 w->word[t] = '\0';
10449 if (w->word[t - 1] == '+') /* cut off any append op */
10450 w->word[t - 1] = '\0';
10451 }
3185942a
JA
10452
10453 wl = make_word_list (w, (WORD_LIST *)NULL);
10454 wl = make_word_list (make_word (option), wl);
10455
a0c0a00f
CR
10456 r = declare_builtin (wl);
10457
10458 dispose_words (wl);
10459 return r;
3185942a
JA
10460}
10461#endif
10462
cce855bc
JA
10463static WORD_LIST *
10464shell_expand_word_list (tlist, eflags)
10465 WORD_LIST *tlist;
10466 int eflags;
10467{
a0c0a00f 10468 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
cce855bc
JA
10469 int expanded_something, has_dollar_at;
10470 char *temp_string;
726f6388 10471
726f6388 10472 /* We do tilde expansion all the time. This is what 1003.2 says. */
cce855bc 10473 new_list = (WORD_LIST *)NULL;
a0c0a00f
CR
10474 for (wcmd = tlist; wcmd; wcmd = wcmd->next)
10475 if (wcmd->word->flags & W_ASSNBLTIN)
10476 break;
10477
cce855bc 10478 for (orig_list = tlist; tlist; tlist = next)
726f6388 10479 {
ccc6cda3 10480 temp_string = tlist->word->word;
726f6388
JA
10481
10482 next = tlist->next;
10483
95732b49
JA
10484#if defined (ARRAY_VARS)
10485 /* If this is a compound array assignment to a builtin that accepts
10486 such assignments (e.g., `declare'), take the assignment and perform
10487 it separately, handling the semantics of declarations inside shell
10488 functions. This avoids the double-evaluation of such arguments,
10489 because `declare' does some evaluation of compound assignments on
10490 its own. */
10491 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10492 {
10493 int t;
a0c0a00f 10494 char opts[16], opti;
ac50fbac
CR
10495
10496 opti = 0;
10497 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
10498 opts[opti++] = '-';
10499
10500 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
10501 {
10502 opts[opti++] = 'g';
10503 opts[opti++] = 'A';
10504 }
10505 else if (tlist->word->flags & W_ASSIGNASSOC)
10506 opts[opti++] = 'A';
10507 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
10508 {
10509 opts[opti++] = 'g';
10510 opts[opti++] = 'a';
10511 }
10512 else if (tlist->word->flags & W_ASSIGNARRAY)
10513 opts[opti++] = 'a';
10514 else if (tlist->word->flags & W_ASSNGLOBAL)
10515 opts[opti++] = 'g';
95732b49 10516
a0c0a00f
CR
10517 /* If we have special handling note the integer attribute and others
10518 that transform the value upon assignment. What we do is take all
10519 of the option arguments and scan through them looking for options
10520 that cause such transformations, and add them to the `opts' array. */
10521/* if (opti > 0) */
10522 {
10523 char omap[128];
10524 int oind;
10525 WORD_LIST *l;
10526
10527 memset (omap, '\0', sizeof (omap));
10528 for (l = orig_list->next; l != tlist; l = l->next)
10529 {
10530 if (l->word->word[0] != '-')
10531 break; /* non-option argument */
10532 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
10533 break; /* -- signals end of options */
10534 for (oind = 1; l->word->word[oind]; oind++)
10535 switch (l->word->word[oind])
10536 {
10537 case 'i':
10538 case 'l':
10539 case 'u':
10540 case 'c':
10541 omap[l->word->word[oind]] = 1;
10542 if (opti == 0)
10543 opts[opti++] = '-';
10544 break;
10545 default:
10546 break;
10547 }
10548 }
10549
10550 for (oind = 0; oind < sizeof (omap); oind++)
10551 if (omap[oind])
10552 opts[opti++] = oind;
10553 }
ac50fbac
CR
10554
10555 opts[opti] = '\0';
10556 if (opti > 0)
30a978b7 10557 {
a0c0a00f 10558 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
30a978b7
CR
10559 if (t != EXECUTION_SUCCESS)
10560 {
10561 last_command_exit_value = t;
10562 exp_jump_to_top_level (DISCARD);
10563 }
10564 }
3185942a 10565
495aee44 10566 t = do_word_assignment (tlist->word, 0);
95732b49
JA
10567 if (t == 0)
10568 {
10569 last_command_exit_value = EXECUTION_FAILURE;
10570 exp_jump_to_top_level (DISCARD);
10571 }
10572
10573 /* Now transform the word as ksh93 appears to do and go on */
10574 t = assignment (tlist->word->word, 0);
10575 tlist->word->word[t] = '\0';
a0c0a00f
CR
10576 if (tlist->word->word[t - 1] == '+')
10577 tlist->word->word[t - 1] = '\0'; /* cut off append op */
ac50fbac 10578 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
726f6388 10579 }
95732b49 10580#endif
726f6388 10581
ccc6cda3 10582 expanded_something = 0;
726f6388 10583 expanded = expand_word_internal
b72432fd 10584 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
726f6388
JA
10585
10586 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
10587 {
10588 /* By convention, each time this error is returned,
10589 tlist->word->word has already been freed. */
10590 tlist->word->word = (char *)NULL;
ccc6cda3 10591
726f6388
JA
10592 /* Dispose our copy of the original list. */
10593 dispose_words (orig_list);
d166f048 10594 /* Dispose the new list we're building. */
726f6388
JA
10595 dispose_words (new_list);
10596
28ef6c31 10597 last_command_exit_value = EXECUTION_FAILURE;
726f6388 10598 if (expanded == &expand_word_error)
b80f6443 10599 exp_jump_to_top_level (DISCARD);
726f6388 10600 else
b80f6443 10601 exp_jump_to_top_level (FORCE_EOF);
726f6388
JA
10602 }
10603
ccc6cda3
JA
10604 /* Don't split words marked W_NOSPLIT. */
10605 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
726f6388 10606 {
ccc6cda3 10607 temp_list = word_list_split (expanded);
726f6388
JA
10608 dispose_words (expanded);
10609 }
10610 else
10611 {
10612 /* If no parameter expansion, command substitution, process
10613 substitution, or arithmetic substitution took place, then
10614 do not do word splitting. We still have to remove quoted
10615 null characters from the result. */
10616 word_list_remove_quoted_nulls (expanded);
ccc6cda3 10617 temp_list = expanded;
726f6388
JA
10618 }
10619
ccc6cda3
JA
10620 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
10621 new_list = (WORD_LIST *)list_append (expanded, new_list);
726f6388
JA
10622 }
10623
cce855bc
JA
10624 if (orig_list)
10625 dispose_words (orig_list);
726f6388 10626
726f6388 10627 if (new_list)
cce855bc 10628 new_list = REVERSE_LIST (new_list, WORD_LIST *);
726f6388 10629
cce855bc
JA
10630 return (new_list);
10631}
726f6388 10632
cce855bc
JA
10633/* The workhorse for expand_words () and expand_words_no_vars ().
10634 First arg is LIST, a WORD_LIST of words.
b72432fd
JA
10635 Second arg EFLAGS is a flags word controlling which expansions are
10636 performed.
726f6388 10637
cce855bc
JA
10638 This does all of the substitutions: brace expansion, tilde expansion,
10639 parameter expansion, command substitution, arithmetic expansion,
10640 process substitution, word splitting, and pathname expansion, according
10641 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
10642 set, or for which no expansion is done, do not undergo word splitting.
ac50fbac
CR
10643 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
10644 with W_NOBRACE set do not undergo brace expansion (see
10645 brace_expand_word_list above). */
cce855bc
JA
10646static WORD_LIST *
10647expand_word_list_internal (list, eflags)
10648 WORD_LIST *list;
10649 int eflags;
10650{
10651 WORD_LIST *new_list, *temp_list;
10652 int tint;
a0c0a00f 10653 char *savecmd;
726f6388 10654
ac50fbac 10655 tempenv_assign_error = 0;
cce855bc
JA
10656 if (list == 0)
10657 return ((WORD_LIST *)NULL);
726f6388 10658
bb70624e 10659 garglist = new_list = copy_word_list (list);
cce855bc
JA
10660 if (eflags & WEXP_VARASSIGN)
10661 {
bb70624e 10662 garglist = new_list = separate_out_assignments (new_list);
cce855bc
JA
10663 if (new_list == 0)
10664 {
bb70624e 10665 if (subst_assign_varlist)
cce855bc
JA
10666 {
10667 /* All the words were variable assignments, so they are placed
10668 into the shell's environment. */
bb70624e 10669 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
cce855bc 10670 {
a0c0a00f 10671 savecmd = this_command_name;
cce855bc 10672 this_command_name = (char *)NULL; /* no arithmetic errors */
495aee44 10673 tint = do_word_assignment (temp_list->word, 0);
a0c0a00f 10674 this_command_name = savecmd;
cce855bc
JA
10675 /* Variable assignment errors in non-interactive shells
10676 running in Posix.2 mode cause the shell to exit. */
28ef6c31 10677 if (tint == 0)
ccc6cda3 10678 {
cce855bc 10679 last_command_exit_value = EXECUTION_FAILURE;
28ef6c31 10680 if (interactive_shell == 0 && posixly_correct)
b80f6443 10681 exp_jump_to_top_level (FORCE_EOF);
28ef6c31 10682 else
b80f6443 10683 exp_jump_to_top_level (DISCARD);
ccc6cda3 10684 }
726f6388 10685 }
bb70624e
JA
10686 dispose_words (subst_assign_varlist);
10687 subst_assign_varlist = (WORD_LIST *)NULL;
cce855bc
JA
10688 }
10689 return ((WORD_LIST *)NULL);
10690 }
10691 }
726f6388 10692
cce855bc
JA
10693 /* Begin expanding the words that remain. The expansions take place on
10694 things that aren't really variable assignments. */
726f6388 10695
cce855bc
JA
10696#if defined (BRACE_EXPANSION)
10697 /* Do brace expansion on this word if there are any brace characters
10698 in the string. */
10699 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
10700 new_list = brace_expand_word_list (new_list, eflags);
10701#endif /* BRACE_EXPANSION */
726f6388 10702
cce855bc
JA
10703 /* Perform the `normal' shell expansions: tilde expansion, parameter and
10704 variable substitution, command substitution, arithmetic expansion,
10705 and word splitting. */
10706 new_list = shell_expand_word_list (new_list, eflags);
726f6388 10707
cce855bc
JA
10708 /* Okay, we're almost done. Now let's just do some filename
10709 globbing. */
10710 if (new_list)
10711 {
10712 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
10713 /* Glob expand the word list unless globbing has been disabled. */
10714 new_list = glob_expand_word_list (new_list, eflags);
726f6388 10715 else
cce855bc
JA
10716 /* Dequote the words, because we're not performing globbing. */
10717 new_list = dequote_list (new_list);
726f6388
JA
10718 }
10719
bb70624e 10720 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
726f6388 10721 {
95732b49 10722 sh_wassign_func_t *assign_func;
495aee44 10723 int is_special_builtin, is_builtin_or_func;
726f6388
JA
10724
10725 /* If the remainder of the words expand to nothing, Posix.2 requires
10726 that the variable and environment assignments affect the shell's
10727 environment. */
95732b49 10728 assign_func = new_list ? assign_in_env : do_word_assignment;
b80f6443 10729 tempenv_assign_error = 0;
726f6388 10730
495aee44
CR
10731 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
10732 /* Posix says that special builtins exit if a variable assignment error
10733 occurs in an assignment preceding it. */
10734 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
10735
bb70624e 10736 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
726f6388 10737 {
a0c0a00f 10738 savecmd = this_command_name;
ccc6cda3 10739 this_command_name = (char *)NULL;
3185942a 10740 assigning_in_environment = (assign_func == assign_in_env);
495aee44 10741 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
3185942a 10742 assigning_in_environment = 0;
a0c0a00f 10743 this_command_name = savecmd;
ccc6cda3
JA
10744 /* Variable assignment errors in non-interactive shells running
10745 in Posix.2 mode cause the shell to exit. */
b80f6443 10746 if (tint == 0)
ccc6cda3 10747 {
95732b49 10748 if (assign_func == do_word_assignment)
b80f6443
JA
10749 {
10750 last_command_exit_value = EXECUTION_FAILURE;
495aee44 10751 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
b80f6443
JA
10752 exp_jump_to_top_level (FORCE_EOF);
10753 else
10754 exp_jump_to_top_level (DISCARD);
10755 }
28ef6c31 10756 else
b80f6443 10757 tempenv_assign_error++;
ccc6cda3 10758 }
726f6388 10759 }
726f6388 10760
bb70624e
JA
10761 dispose_words (subst_assign_varlist);
10762 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
10763 }
10764
cce855bc 10765 return (new_list);
ccc6cda3 10766}