]> git.ipfire.org Git - thirdparty/bash.git/blame - subst.c
Bash-4.4 patch 2
[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
7117c2d2
JA
2828 the `read' builtin. This is never called with SEPARATORS != $IFS;
2829 it should be simplified.
2830
726f6388
JA
2831 XXX - this function is very similar to list_string; they should be
2832 combined - XXX */
2833char *
2834get_word_from_string (stringp, separators, endptr)
2835 char **stringp, *separators, **endptr;
2836{
2837 register char *s;
2838 char *current_word;
3185942a 2839 int sindex, sh_style_split, whitesep, xflags;
95732b49 2840 size_t slen;
726f6388
JA
2841
2842 if (!stringp || !*stringp || !**stringp)
2843 return ((char *)NULL);
ccc6cda3 2844
7117c2d2
JA
2845 sh_style_split = separators && separators[0] == ' ' &&
2846 separators[1] == '\t' &&
2847 separators[2] == '\n' &&
2848 separators[3] == '\0';
3185942a
JA
2849 for (xflags = 0, s = ifs_value; s && *s; s++)
2850 {
2851 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2852 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2853 }
726f6388 2854
3185942a 2855 s = *stringp;
95732b49
JA
2856 slen = 0;
2857
726f6388
JA
2858 /* Remove sequences of whitespace at the beginning of STRING, as
2859 long as those characters appear in IFS. */
2860 if (sh_style_split || !separators || !*separators)
2861 {
7117c2d2 2862 for (; *s && spctabnl (*s) && isifs (*s); s++);
726f6388
JA
2863
2864 /* If the string is nothing but whitespace, update it and return. */
2865 if (!*s)
2866 {
2867 *stringp = s;
2868 if (endptr)
2869 *endptr = s;
2870 return ((char *)NULL);
2871 }
2872 }
2873
2874 /* OK, S points to a word that does not begin with white space.
2875 Now extract a word, stopping at a separator, save a pointer to
2876 the first character after the word, then skip sequences of spc,
2877 tab, or nl as long as they are separators.
ccc6cda3 2878
726f6388
JA
2879 This obeys the field splitting rules in Posix.2. */
2880 sindex = 0;
95732b49
JA
2881 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2882 unless multibyte chars are possible. */
a0c0a00f 2883 slen = (MB_CUR_MAX > 1) ? STRLEN (s) : 1;
3185942a 2884 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
726f6388
JA
2885
2886 /* Set ENDPTR to the first character after the end of the word. */
2887 if (endptr)
2888 *endptr = s + sindex;
2889
28ef6c31
JA
2890 /* Note whether or not the separator is IFS whitespace, used later. */
2891 whitesep = s[sindex] && spctabnl (s[sindex]);
2892
726f6388
JA
2893 /* Move past the current separator character. */
2894 if (s[sindex])
95732b49
JA
2895 {
2896 DECLARE_MBSTATE;
2897 ADVANCE_CHAR (s, slen, sindex);
2898 }
726f6388
JA
2899
2900 /* Now skip sequences of space, tab, or newline characters if they are
2901 in the list of separators. */
7117c2d2 2902 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
726f6388
JA
2903 sindex++;
2904
28ef6c31
JA
2905 /* If the first separator was IFS whitespace and the current character is
2906 a non-whitespace IFS character, it should be part of the current field
2907 delimiter, not a separate delimiter that would result in an empty field.
2908 Look at POSIX.2, 3.6.5, (3)(b). */
7117c2d2 2909 if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex]))
95732b49
JA
2910 {
2911 sindex++;
2912 /* An IFS character that is not IFS white space, along with any adjacent
2913 IFS white space, shall delimit a field. */
2914 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2915 sindex++;
2916 }
28ef6c31 2917
726f6388
JA
2918 /* Update STRING to point to the next field. */
2919 *stringp = s + sindex;
2920 return (current_word);
2921}
2922
2923/* Remove IFS white space at the end of STRING. Start at the end
2924 of the string and walk backwards until the beginning of the string
2925 or we find a character that's not IFS white space and not CTLESC.
2926 Only let CTLESC escape a white space character if SAW_ESCAPE is
2927 non-zero. */
2928char *
2929strip_trailing_ifs_whitespace (string, separators, saw_escape)
2930 char *string, *separators;
2931 int saw_escape;
2932{
2933 char *s;
ccc6cda3 2934
726f6388 2935 s = string + STRLEN (string) - 1;
7117c2d2 2936 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
726f6388
JA
2937 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2938 s--;
2939 *++s = '\0';
2940 return string;
2941}
2942
bb70624e
JA
2943#if 0
2944/* UNUSED */
2945/* Split STRING into words at whitespace. Obeys shell-style quoting with
2946 backslashes, single and double quotes. */
ccc6cda3
JA
2947WORD_LIST *
2948list_string_with_quotes (string)
2949 char *string;
2950{
2951 WORD_LIST *list;
2952 char *token, *s;
7117c2d2 2953 size_t s_len;
ccc6cda3
JA
2954 int c, i, tokstart, len;
2955
2956 for (s = string; s && *s && spctabnl (*s); s++)
2957 ;
2958 if (s == 0 || *s == 0)
2959 return ((WORD_LIST *)NULL);
2960
7117c2d2 2961 s_len = strlen (s);
ccc6cda3
JA
2962 tokstart = i = 0;
2963 list = (WORD_LIST *)NULL;
2964 while (1)
2965 {
2966 c = s[i];
2967 if (c == '\\')
2968 {
2969 i++;
2970 if (s[i])
2971 i++;
2972 }
2973 else if (c == '\'')
a0c0a00f 2974 i = skip_single_quoted (s, s_len, ++i, 0);
ccc6cda3 2975 else if (c == '"')
a0c0a00f 2976 i = skip_double_quoted (s, s_len, ++i, 0);
ccc6cda3
JA
2977 else if (c == 0 || spctabnl (c))
2978 {
2979 /* We have found the end of a token. Make a word out of it and
2980 add it to the word list. */
bb70624e 2981 token = substring (s, tokstart, i);
cce855bc 2982 list = add_string_to_list (token, list);
ccc6cda3
JA
2983 free (token);
2984 while (spctabnl (s[i]))
2985 i++;
2986 if (s[i])
2987 tokstart = i;
2988 else
2989 break;
2990 }
2991 else
2992 i++; /* normal character */
2993 }
2994 return (REVERSE_LIST (list, WORD_LIST *));
2995}
bb70624e 2996#endif
d166f048 2997
cce855bc
JA
2998/********************************************************/
2999/* */
3000/* Functions to perform assignment statements */
3001/* */
3002/********************************************************/
d166f048 3003
95732b49
JA
3004#if defined (ARRAY_VARS)
3005static SHELL_VAR *
3006do_compound_assignment (name, value, flags)
3007 char *name, *value;
3008 int flags;
3009{
3010 SHELL_VAR *v;
ac50fbac 3011 int mklocal, mkassoc, mkglobal;
0628567a 3012 WORD_LIST *list;
95732b49
JA
3013
3014 mklocal = flags & ASS_MKLOCAL;
3185942a 3015 mkassoc = flags & ASS_MKASSOC;
ac50fbac 3016 mkglobal = flags & ASS_MKGLOBAL;
95732b49
JA
3017
3018 if (mklocal && variable_context)
3019 {
3020 v = find_variable (name);
a0c0a00f
CR
3021 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3022 {
3023 if (v && readonly_p (v))
3024 err_readonly (name);
3025 return (v); /* XXX */
3026 }
3185942a
JA
3027 list = expand_compound_array_assignment (v, value, flags);
3028 if (mkassoc)
3029 v = make_local_assoc_variable (name);
3030 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
ac50fbac
CR
3031 v = make_local_array_variable (name, 0);
3032 if (v)
3033 assign_compound_array_list (v, list, flags);
a0c0a00f
CR
3034 if (list)
3035 dispose_words (list);
ac50fbac
CR
3036 }
3037 /* In a function but forcing assignment in global context */
3038 else if (mkglobal && variable_context)
3039 {
3040 v = find_global_variable (name);
a0c0a00f
CR
3041 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3042 {
3043 if (v && readonly_p (v))
3044 err_readonly (name);
3045 return (v); /* XXX */
3046 }
ac50fbac
CR
3047 list = expand_compound_array_assignment (v, value, flags);
3048 if (v == 0 && mkassoc)
3049 v = make_new_assoc_variable (name);
3050 else if (v && mkassoc && assoc_p (v) == 0)
3051 v = convert_var_to_assoc (v);
3052 else if (v == 0)
3053 v = make_new_array_variable (name);
3054 else if (v && mkassoc == 0 && array_p (v) == 0)
3055 v = convert_var_to_array (v);
3056 if (v)
3057 assign_compound_array_list (v, list, flags);
a0c0a00f
CR
3058 if (list)
3059 dispose_words (list);
95732b49
JA
3060 }
3061 else
a0c0a00f
CR
3062 {
3063 v = assign_array_from_string (name, value, flags);
3064 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3065 {
3066 if (v && readonly_p (v))
3067 err_readonly (name);
3068 return (v); /* XXX */
3069 }
3070 }
95732b49
JA
3071
3072 return (v);
3073}
3074#endif
3075
726f6388
JA
3076/* Given STRING, an assignment string, get the value of the right side
3077 of the `=', and bind it to the left side. If EXPAND is true, then
3078 perform parameter expansion, command substitution, and arithmetic
3079 expansion on the right-hand side. Perform tilde expansion in any
3080 case. Do not perform word splitting on the result of expansion. */
3081static int
95732b49
JA
3082do_assignment_internal (word, expand)
3083 const WORD_DESC *word;
726f6388
JA
3084 int expand;
3085{
495aee44
CR
3086 int offset, appendop, assign_list, aflags, retval;
3087 char *name, *value, *temp;
ccc6cda3
JA
3088 SHELL_VAR *entry;
3089#if defined (ARRAY_VARS)
3090 char *t;
b80f6443 3091 int ni;
ccc6cda3 3092#endif
95732b49 3093 const char *string;
ccc6cda3 3094
95732b49
JA
3095 if (word == 0 || word->word == 0)
3096 return 0;
3097
3098 appendop = assign_list = aflags = 0;
3099 string = word->word;
b80f6443 3100 offset = assignment (string, 0);
ccc6cda3
JA
3101 name = savestring (string);
3102 value = (char *)NULL;
726f6388
JA
3103
3104 if (name[offset] == '=')
3105 {
95732b49
JA
3106 if (name[offset - 1] == '+')
3107 {
3108 appendop = 1;
3109 name[offset - 1] = '\0';
3110 }
3111
3112 name[offset] = 0; /* might need this set later */
726f6388
JA
3113 temp = name + offset + 1;
3114
ccc6cda3 3115#if defined (ARRAY_VARS)
95732b49 3116 if (expand && (word->flags & W_COMPASSIGN))
726f6388 3117 {
ccc6cda3 3118 assign_list = ni = 1;
95732b49 3119 value = extract_array_assignment_list (temp, &ni);
ccc6cda3
JA
3120 }
3121 else
3122#endif
ccc6cda3 3123 if (expand && temp[0])
95732b49 3124 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
726f6388
JA
3125 else
3126 value = savestring (temp);
3127 }
3128
3129 if (value == 0)
d166f048 3130 {
f73dda09 3131 value = (char *)xmalloc (1);
d166f048
JA
3132 value[0] = '\0';
3133 }
726f6388 3134
726f6388 3135 if (echo_command_at_execute)
95732b49
JA
3136 {
3137 if (appendop)
3138 name[offset - 1] = '+';
3139 xtrace_print_assignment (name, value, assign_list, 1);
3140 if (appendop)
3141 name[offset - 1] = '\0';
3142 }
726f6388 3143
d166f048 3144#define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
ccc6cda3 3145
95732b49
JA
3146 if (appendop)
3147 aflags |= ASS_APPEND;
3148
ccc6cda3 3149#if defined (ARRAY_VARS)
0001803f 3150 if (t = mbschr (name, '[')) /*]*/
ccc6cda3
JA
3151 {
3152 if (assign_list)
3153 {
b80f6443 3154 report_error (_("%s: cannot assign list to array member"), name);
ccc6cda3
JA
3155 ASSIGN_RETURN (0);
3156 }
95732b49 3157 entry = assign_array_element (name, value, aflags);
ccc6cda3 3158 if (entry == 0)
28ef6c31 3159 ASSIGN_RETURN (0);
ccc6cda3
JA
3160 }
3161 else if (assign_list)
95732b49 3162 {
6d41b715 3163 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
95732b49 3164 aflags |= ASS_MKLOCAL;
ac50fbac
CR
3165 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
3166 aflags |= ASS_MKGLOBAL;
3185942a
JA
3167 if (word->flags & W_ASSIGNASSOC)
3168 aflags |= ASS_MKASSOC;
95732b49
JA
3169 entry = do_compound_assignment (name, value, aflags);
3170 }
ccc6cda3
JA
3171 else
3172#endif /* ARRAY_VARS */
95732b49 3173 entry = bind_variable (name, value, aflags);
ccc6cda3 3174
726f6388
JA
3175 stupidly_hack_special_variables (name);
3176
3185942a
JA
3177 /* Return 1 if the assignment seems to have been performed correctly. */
3178 if (entry == 0 || readonly_p (entry))
3179 retval = 0; /* assignment failure */
3180 else if (noassign_p (entry))
3181 {
3182 last_command_exit_value = EXECUTION_FAILURE;
3183 retval = 1; /* error status, but not assignment failure */
3184 }
3185 else
3186 retval = 1;
3187
3188 if (entry && retval != 0 && noassign_p (entry) == 0)
3189 VUNSETATTR (entry, att_invisible);
3190
3191 ASSIGN_RETURN (retval);
726f6388
JA
3192}
3193
3194/* Perform the assignment statement in STRING, and expand the
95732b49 3195 right side by doing tilde, command and parameter expansion. */
ccc6cda3 3196int
726f6388 3197do_assignment (string)
95732b49 3198 char *string;
726f6388 3199{
95732b49
JA
3200 WORD_DESC td;
3201
3202 td.flags = W_ASSIGNMENT;
3203 td.word = string;
3204
3205 return do_assignment_internal (&td, 1);
3206}
3207
3208int
495aee44 3209do_word_assignment (word, flags)
95732b49 3210 WORD_DESC *word;
495aee44 3211 int flags;
95732b49
JA
3212{
3213 return do_assignment_internal (word, 1);
726f6388
JA
3214}
3215
3216/* Given STRING, an assignment string, get the value of the right side
95732b49
JA
3217 of the `=', and bind it to the left side. Do not perform any word
3218 expansions on the right hand side. */
ccc6cda3 3219int
726f6388 3220do_assignment_no_expand (string)
95732b49 3221 char *string;
726f6388 3222{
95732b49
JA
3223 WORD_DESC td;
3224
3225 td.flags = W_ASSIGNMENT;
3226 td.word = string;
3227
3228 return (do_assignment_internal (&td, 0));
726f6388
JA
3229}
3230
cce855bc
JA
3231/***************************************************
3232 * *
3233 * Functions to manage the positional parameters *
3234 * *
3235 ***************************************************/
726f6388
JA
3236
3237/* Return the word list that corresponds to `$*'. */
3238WORD_LIST *
3239list_rest_of_args ()
3240{
ccc6cda3 3241 register WORD_LIST *list, *args;
726f6388
JA
3242 int i;
3243
3244 /* Break out of the loop as soon as one of the dollar variables is null. */
ccc6cda3
JA
3245 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
3246 list = make_word_list (make_bare_word (dollar_vars[i]), list);
3247
3248 for (args = rest_of_args; args; args = args->next)
3249 list = make_word_list (make_bare_word (args->word->word), list);
726f6388 3250
726f6388
JA
3251 return (REVERSE_LIST (list, WORD_LIST *));
3252}
3253
ccc6cda3
JA
3254int
3255number_of_args ()
3256{
3257 register WORD_LIST *list;
3258 int n;
3259
3260 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
3261 ;
3262 for (list = rest_of_args; list; list = list->next)
3263 n++;
3264 return n;
3265}
3266
cce855bc
JA
3267/* Return the value of a positional parameter. This handles values > 10. */
3268char *
3269get_dollar_var_value (ind)
7117c2d2 3270 intmax_t ind;
cce855bc
JA
3271{
3272 char *temp;
3273 WORD_LIST *p;
3274
3275 if (ind < 10)
3276 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
3277 else /* We want something like ${11} */
3278 {
3279 ind -= 10;
3280 for (p = rest_of_args; p && ind--; p = p->next)
28ef6c31 3281 ;
cce855bc
JA
3282 temp = p ? savestring (p->word->word) : (char *)NULL;
3283 }
3284 return (temp);
3285}
3286
726f6388
JA
3287/* Make a single large string out of the dollar digit variables,
3288 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
3289 case of "$*" with respect to IFS. */
3290char *
3291string_rest_of_args (dollar_star)
3292 int dollar_star;
3293{
ccc6cda3 3294 register WORD_LIST *list;
726f6388
JA
3295 char *string;
3296
ccc6cda3 3297 list = list_rest_of_args ();
726f6388
JA
3298 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
3299 dispose_words (list);
3300 return (string);
3301}
3302
cce855bc
JA
3303/* Return a string containing the positional parameters from START to
3304 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
7117c2d2
JA
3305 which only makes a difference if QUOTED is non-zero. If QUOTED includes
3306 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3307 no quoting chars are added. */
cce855bc
JA
3308static char *
3309pos_params (string, start, end, quoted)
3310 char *string;
3311 int start, end, quoted;
726f6388 3312{
cce855bc
JA
3313 WORD_LIST *save, *params, *h, *t;
3314 char *ret;
3315 int i;
726f6388 3316
bb70624e
JA
3317 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3318 if (start == end)
3319 return ((char *)NULL);
3320
cce855bc 3321 save = params = list_rest_of_args ();
a0c0a00f 3322 if (save == 0 && start > 0)
cce855bc
JA
3323 return ((char *)NULL);
3324
3185942a
JA
3325 if (start == 0) /* handle ${@:0[:x]} specially */
3326 {
3327 t = make_word_list (make_word (dollar_vars[0]), params);
3328 save = params = t;
3329 }
3330
0001803f 3331 for (i = start ? 1 : 0; params && i < start; i++)
cce855bc
JA
3332 params = params->next;
3333 if (params == 0)
a0c0a00f
CR
3334 {
3335 dispose_words (save);
3336 return ((char *)NULL);
3337 }
cce855bc 3338 for (h = t = params; params && i < end; i++)
d166f048 3339 {
cce855bc
JA
3340 t = params;
3341 params = params->next;
d166f048 3342 }
726f6388 3343
cce855bc 3344 t->next = (WORD_LIST *)NULL;
3185942a
JA
3345
3346 ret = string_list_pos_params (string[0], h, quoted);
3347
bb70624e
JA
3348 if (t != params)
3349 t->next = params;
726f6388 3350
cce855bc
JA
3351 dispose_words (save);
3352 return (ret);
3353}
3354
3355/******************************************************************/
3356/* */
3357/* Functions to expand strings to strings or WORD_LISTs */
3358/* */
3359/******************************************************************/
3360
3361#if defined (PROCESS_SUBSTITUTION)
95732b49 3362#define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
cce855bc 3363#else
95732b49 3364#define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
cce855bc
JA
3365#endif
3366
3367/* If there are any characters in STRING that require full expansion,
3368 then call FUNC to expand STRING; otherwise just perform quote
3369 removal if necessary. This returns a new string. */
3370static char *
f73dda09 3371expand_string_if_necessary (string, quoted, func)
cce855bc
JA
3372 char *string;
3373 int quoted;
f73dda09 3374 EXPFUNC *func;
cce855bc
JA
3375{
3376 WORD_LIST *list;
7117c2d2 3377 size_t slen;
cce855bc
JA
3378 int i, saw_quote;
3379 char *ret;
7117c2d2 3380 DECLARE_MBSTATE;
cce855bc 3381
95732b49
JA
3382 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3383 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
7117c2d2
JA
3384 i = saw_quote = 0;
3385 while (string[i])
cce855bc
JA
3386 {
3387 if (EXP_CHAR (string[i]))
3388 break;
3389 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3390 saw_quote = 1;
7117c2d2 3391 ADVANCE_CHAR (string, slen, i);
cce855bc
JA
3392 }
3393
3394 if (string[i])
3395 {
3396 list = (*func) (string, quoted);
3397 if (list)
3398 {
3399 ret = string_list (list);
3400 dispose_words (list);
3401 }
3402 else
3403 ret = (char *)NULL;
3404 }
3405 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3406 ret = string_quote_removal (string, quoted);
3407 else
3408 ret = savestring (string);
7117c2d2 3409
cce855bc
JA
3410 return ret;
3411}
3412
3413static inline char *
f73dda09 3414expand_string_to_string_internal (string, quoted, func)
cce855bc
JA
3415 char *string;
3416 int quoted;
f73dda09 3417 EXPFUNC *func;
cce855bc
JA
3418{
3419 WORD_LIST *list;
3420 char *ret;
3421
3422 if (string == 0 || *string == '\0')
3423 return ((char *)NULL);
3424
3425 list = (*func) (string, quoted);
3426 if (list)
3427 {
3428 ret = string_list (list);
3429 dispose_words (list);
3430 }
3431 else
3432 ret = (char *)NULL;
3433
3434 return (ret);
3435}
3436
f73dda09
JA
3437char *
3438expand_string_to_string (string, quoted)
3439 char *string;
3440 int quoted;
3441{
3442 return (expand_string_to_string_internal (string, quoted, expand_string));
3443}
3444
3445char *
3446expand_string_unsplit_to_string (string, quoted)
3447 char *string;
3448 int quoted;
3449{
3450 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3451}
3452
95732b49
JA
3453char *
3454expand_assignment_string_to_string (string, quoted)
3455 char *string;
3456 int quoted;
3457{
3458 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3459}
3460
0628567a
JA
3461char *
3462expand_arith_string (string, quoted)
3463 char *string;
3185942a 3464 int quoted;
0628567a 3465{
ac50fbac
CR
3466 WORD_DESC td;
3467 WORD_LIST *list, *tlist;
3468 size_t slen;
3469 int i, saw_quote;
3470 char *ret;
3471 DECLARE_MBSTATE;
3472
3473 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3474 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3475 i = saw_quote = 0;
3476 while (string[i])
3477 {
3478 if (EXP_CHAR (string[i]))
3479 break;
3480 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3481 saw_quote = 1;
3482 ADVANCE_CHAR (string, slen, i);
3483 }
3484
3485 if (string[i])
3486 {
3487 /* This is expanded version of expand_string_internal as it's called by
3488 expand_string_leave_quoted */
3489 td.flags = W_NOPROCSUB; /* don't want process substitution */
3490 td.word = savestring (string);
3491 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3492 /* This takes care of the calls from expand_string_leave_quoted and
3493 expand_string */
3494 if (list)
3495 {
3496 tlist = word_list_split (list);
3497 dispose_words (list);
3498 list = tlist;
3499 if (list)
3500 dequote_list (list);
3501 }
3502 /* This comes from expand_string_if_necessary */
3503 if (list)
3504 {
3505 ret = string_list (list);
3506 dispose_words (list);
3507 }
3508 else
3509 ret = (char *)NULL;
3510 FREE (td.word);
3511 }
a0c0a00f
CR
3512 else if (saw_quote && (quoted & Q_ARITH))
3513 ret = string_quote_removal (string, quoted);
ac50fbac
CR
3514 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3515 ret = string_quote_removal (string, quoted);
3516 else
3517 ret = savestring (string);
3518
3519 return ret;
0628567a
JA
3520}
3521
cce855bc
JA
3522#if defined (COND_COMMAND)
3523/* Just remove backslashes in STRING. Returns a new string. */
3524char *
3525remove_backslashes (string)
3526 char *string;
3527{
3528 char *r, *ret, *s;
3529
f73dda09 3530 r = ret = (char *)xmalloc (strlen (string) + 1);
cce855bc
JA
3531 for (s = string; s && *s; )
3532 {
3533 if (*s == '\\')
28ef6c31 3534 s++;
cce855bc 3535 if (*s == 0)
28ef6c31 3536 break;
cce855bc
JA
3537 *r++ = *s++;
3538 }
3539 *r = '\0';
3540 return ret;
3541}
3542
3543/* This needs better error handling. */
3544/* Expand W for use as an argument to a unary or binary operator in a
f1be666c 3545 [[...]] expression. If SPECIAL is 1, this is the rhs argument
cce855bc 3546 to the != or == operator, and should be treated as a pattern. In
f1be666c
JA
3547 this case, we quote the string specially for the globbing code. If
3548 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3549 be quoted appropriately for regcomp/regexec. The caller is responsible
3550 for removing the backslashes if the unquoted word is needed later. */
cce855bc
JA
3551char *
3552cond_expand_word (w, special)
3553 WORD_DESC *w;
3554 int special;
3555{
3556 char *r, *p;
3557 WORD_LIST *l;
f1be666c 3558 int qflags;
cce855bc
JA
3559
3560 if (w->word == 0 || w->word[0] == '\0')
3561 return ((char *)NULL);
3562
3b34f6e6 3563 expand_no_split_dollar_star = 1;
0001803f 3564 w->flags |= W_NOSPLIT2;
b72432fd 3565 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3b34f6e6 3566 expand_no_split_dollar_star = 0;
cce855bc
JA
3567 if (l)
3568 {
ac50fbac 3569 if (special == 0) /* LHS */
cce855bc
JA
3570 {
3571 dequote_list (l);
3572 r = string_list (l);
3573 }
3574 else
28ef6c31 3575 {
ac50fbac
CR
3576 /* Need to figure out whether or not we should call dequote_escapes
3577 or a new dequote_ctlnul function here, and under what
3578 circumstances. */
f1be666c
JA
3579 qflags = QGLOB_CVTNULL;
3580 if (special == 2)
3581 qflags |= QGLOB_REGEXP;
a0c0a00f 3582 word_list_remove_quoted_nulls (l);
28ef6c31 3583 p = string_list (l);
f1be666c 3584 r = quote_string_for_globbing (p, qflags);
28ef6c31
JA
3585 free (p);
3586 }
cce855bc
JA
3587 dispose_words (l);
3588 }
3589 else
3590 r = (char *)NULL;
3591
3592 return r;
3593}
3594#endif
3595
3596/* Call expand_word_internal to expand W and handle error returns.
3597 A convenience function for functions that don't want to handle
3598 any errors or free any memory before aborting. */
3599static WORD_LIST *
b72432fd 3600call_expand_word_internal (w, q, i, c, e)
cce855bc 3601 WORD_DESC *w;
b72432fd 3602 int q, i, *c, *e;
cce855bc
JA
3603{
3604 WORD_LIST *result;
3605
b72432fd 3606 result = expand_word_internal (w, q, i, c, e);
bb70624e 3607 if (result == &expand_word_error || result == &expand_word_fatal)
cce855bc
JA
3608 {
3609 /* By convention, each time this error is returned, w->word has
bb70624e
JA
3610 already been freed (it sometimes may not be in the fatal case,
3611 but that doesn't result in a memory leak because we're going
3612 to exit in most cases). */
cce855bc 3613 w->word = (char *)NULL;
28ef6c31 3614 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 3615 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
cce855bc 3616 /* NOTREACHED */
ac50fbac 3617 return (NULL);
cce855bc 3618 }
cce855bc
JA
3619 else
3620 return (result);
3621}
3622
3623/* Perform parameter expansion, command substitution, and arithmetic
ac50fbac
CR
3624 expansion on STRING, as if it were a word. Leave the result quoted.
3625 Since this does not perform word splitting, it leaves quoted nulls
3626 in the result. */
cce855bc
JA
3627static WORD_LIST *
3628expand_string_internal (string, quoted)
3629 char *string;
3630 int quoted;
3631{
3632 WORD_DESC td;
3633 WORD_LIST *tresult;
3634
3635 if (string == 0 || *string == 0)
3636 return ((WORD_LIST *)NULL);
3637
28ef6c31
JA
3638 td.flags = 0;
3639 td.word = savestring (string);
3640
b72432fd 3641 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31
JA
3642
3643 FREE (td.word);
cce855bc 3644 return (tresult);
726f6388
JA
3645}
3646
3647/* Expand STRING by performing parameter expansion, command substitution,
3648 and arithmetic expansion. Dequote the resulting WORD_LIST before
3649 returning it, but do not perform word splitting. The call to
3650 remove_quoted_nulls () is in here because word splitting normally
3651 takes care of quote removal. */
3652WORD_LIST *
3653expand_string_unsplit (string, quoted)
3654 char *string;
3655 int quoted;
3656{
3657 WORD_LIST *value;
3658
28ef6c31 3659 if (string == 0 || *string == '\0')
726f6388
JA
3660 return ((WORD_LIST *)NULL);
3661
28ef6c31 3662 expand_no_split_dollar_star = 1;
726f6388 3663 value = expand_string_internal (string, quoted);
28ef6c31
JA
3664 expand_no_split_dollar_star = 0;
3665
726f6388
JA
3666 if (value)
3667 {
3668 if (value->word)
95732b49
JA
3669 {
3670 remove_quoted_nulls (value->word->word);
3671 value->word->flags &= ~W_HASQUOTEDNULL;
3672 }
3673 dequote_list (value);
3674 }
3675 return (value);
3676}
3677
3678/* Expand the rhs of an assignment statement */
3679WORD_LIST *
3680expand_string_assignment (string, quoted)
3681 char *string;
3682 int quoted;
3683{
3684 WORD_DESC td;
3685 WORD_LIST *value;
3686
3687 if (string == 0 || *string == '\0')
3688 return ((WORD_LIST *)NULL);
3689
3690 expand_no_split_dollar_star = 1;
3691
3692 td.flags = W_ASSIGNRHS;
3693 td.word = savestring (string);
3694 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3695 FREE (td.word);
3696
3697 expand_no_split_dollar_star = 0;
3698
3699 if (value)
3700 {
3701 if (value->word)
3702 {
3703 remove_quoted_nulls (value->word->word);
3704 value->word->flags &= ~W_HASQUOTEDNULL;
3705 }
726f6388
JA
3706 dequote_list (value);
3707 }
3708 return (value);
3709}
3710
bb70624e
JA
3711
3712/* Expand one of the PS? prompt strings. This is a sort of combination of
3713 expand_string_unsplit and expand_string_internal, but returns the
3714 passed string when an error occurs. Might want to trap other calls
3715 to jump_to_top_level here so we don't endlessly loop. */
3716WORD_LIST *
f1be666c 3717expand_prompt_string (string, quoted, wflags)
bb70624e
JA
3718 char *string;
3719 int quoted;
f1be666c 3720 int wflags;
bb70624e
JA
3721{
3722 WORD_LIST *value;
3723 WORD_DESC td;
3724
3725 if (string == 0 || *string == 0)
3726 return ((WORD_LIST *)NULL);
3727
f1be666c 3728 td.flags = wflags;
bb70624e 3729 td.word = savestring (string);
28ef6c31
JA
3730
3731 no_longjmp_on_fatal_error = 1;
bb70624e 3732 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31
JA
3733 no_longjmp_on_fatal_error = 0;
3734
bb70624e
JA
3735 if (value == &expand_word_error || value == &expand_word_fatal)
3736 {
3737 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3738 return value;
3739 }
3740 FREE (td.word);
3741 if (value)
3742 {
3743 if (value->word)
95732b49
JA
3744 {
3745 remove_quoted_nulls (value->word->word);
3746 value->word->flags &= ~W_HASQUOTEDNULL;
3747 }
bb70624e
JA
3748 dequote_list (value);
3749 }
3750 return (value);
3751}
3752
726f6388
JA
3753/* Expand STRING just as if you were expanding a word, but do not dequote
3754 the resultant WORD_LIST. This is called only from within this file,
3755 and is used to correctly preserve quoted characters when expanding
3756 things like ${1+"$@"}. This does parameter expansion, command
b72432fd 3757 substitution, arithmetic expansion, and word splitting. */
726f6388
JA
3758static WORD_LIST *
3759expand_string_leave_quoted (string, quoted)
3760 char *string;
3761 int quoted;
3762{
3763 WORD_LIST *tlist;
3764 WORD_LIST *tresult;
3765
ccc6cda3 3766 if (string == 0 || *string == '\0')
726f6388
JA
3767 return ((WORD_LIST *)NULL);
3768
3769 tlist = expand_string_internal (string, quoted);
3770
3771 if (tlist)
3772 {
3773 tresult = word_list_split (tlist);
3774 dispose_words (tlist);
3775 return (tresult);
3776 }
3777 return ((WORD_LIST *)NULL);
3778}
3779
ccc6cda3
JA
3780/* This does not perform word splitting or dequote the WORD_LIST
3781 it returns. */
3782static WORD_LIST *
a0c0a00f 3783expand_string_for_rhs (string, quoted, dollar_at_p, expanded_p)
ccc6cda3 3784 char *string;
a0c0a00f 3785 int quoted, *dollar_at_p, *expanded_p;
ccc6cda3
JA
3786{
3787 WORD_DESC td;
3788 WORD_LIST *tresult;
3789
3790 if (string == 0 || *string == '\0')
3791 return (WORD_LIST *)NULL;
3792
a0c0a00f 3793 expand_no_split_dollar_star = 1;
aeb26a67 3794 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
ccc6cda3 3795 td.word = string;
a0c0a00f
CR
3796 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, expanded_p);
3797 expand_no_split_dollar_star = 0;
3798
ccc6cda3
JA
3799 return (tresult);
3800}
3801
726f6388
JA
3802/* Expand STRING just as if you were expanding a word. This also returns
3803 a list of words. Note that filename globbing is *NOT* done for word
3804 or string expansion, just when the shell is expanding a command. This
3805 does parameter expansion, command substitution, arithmetic expansion,
3806 and word splitting. Dequote the resultant WORD_LIST before returning. */
3807WORD_LIST *
3808expand_string (string, quoted)
3809 char *string;
3810 int quoted;
3811{
3812 WORD_LIST *result;
3813
28ef6c31 3814 if (string == 0 || *string == '\0')
726f6388
JA
3815 return ((WORD_LIST *)NULL);
3816
3817 result = expand_string_leave_quoted (string, quoted);
ccc6cda3 3818 return (result ? dequote_list (result) : result);
726f6388
JA
3819}
3820
3821/***************************************************
3822 * *
3823 * Functions to handle quoting chars *
3824 * *
3825 ***************************************************/
3826
cce855bc
JA
3827/* Conventions:
3828
3829 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3830 The parser passes CTLNUL as CTLESC CTLNUL. */
3831
cce855bc
JA
3832/* Quote escape characters in string s, but no other characters. This is
3833 used to protect CTLESC and CTLNUL in variable values from the rest of
3185942a
JA
3834 the word expansion process after the variable is expanded (word splitting
3835 and filename generation). If IFS is null, we quote spaces as well, just
3836 in case we split on spaces later (in the case of unquoted $@, we will
3837 eventually attempt to split the entire word on spaces). Corresponding
3838 code exists in dequote_escapes. Even if we don't end up splitting on
3839 spaces, quoting spaces is not a problem. This should never be called on
3840 a string that is quoted with single or double quotes or part of a here
3841 document (effectively double-quoted). */
f73dda09 3842char *
cce855bc
JA
3843quote_escapes (string)
3844 char *string;
3845{
3846 register char *s, *t;
7117c2d2
JA
3847 size_t slen;
3848 char *result, *send;
3185942a 3849 int quote_spaces, skip_ctlesc, skip_ctlnul;
7117c2d2 3850 DECLARE_MBSTATE;
cce855bc 3851
7117c2d2
JA
3852 slen = strlen (string);
3853 send = string + slen;
3854
f1be666c 3855 quote_spaces = (ifs_value && *ifs_value == 0);
3185942a
JA
3856
3857 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3858 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3859
7117c2d2
JA
3860 t = result = (char *)xmalloc ((slen * 2) + 1);
3861 s = string;
3862
3863 while (*s)
cce855bc 3864 {
3185942a 3865 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
cce855bc 3866 *t++ = CTLESC;
7117c2d2 3867 COPY_CHAR_P (t, s, send);
cce855bc
JA
3868 }
3869 *t = '\0';
ac50fbac 3870
cce855bc
JA
3871 return (result);
3872}
3873
3874static WORD_LIST *
3875list_quote_escapes (list)
3876 WORD_LIST *list;
3877{
3878 register WORD_LIST *w;
3879 char *t;
3880
3881 for (w = list; w; w = w->next)
3882 {
3883 t = w->word->word;
3884 w->word->word = quote_escapes (t);
3885 free (t);
3886 }
3887 return list;
3888}
3889
7117c2d2
JA
3890/* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3891
3892 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3893 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3894 data stream pass through properly.
3895
3896 We need to remove doubled CTLESC characters inside quoted strings before
3897 quoting the entire string, so we do not double the number of CTLESC
3898 characters.
3899
3900 Also used by parts of the pattern substitution code. */
3185942a 3901char *
cce855bc
JA
3902dequote_escapes (string)
3903 char *string;
3904{
3185942a 3905 register char *s, *t, *s1;
7117c2d2
JA
3906 size_t slen;
3907 char *result, *send;
f1be666c 3908 int quote_spaces;
7117c2d2 3909 DECLARE_MBSTATE;
cce855bc 3910
7117c2d2
JA
3911 if (string == 0)
3912 return string;
3913
3914 slen = strlen (string);
3915 send = string + slen;
3916
3917 t = result = (char *)xmalloc (slen + 1);
7117c2d2
JA
3918
3919 if (strchr (string, CTLESC) == 0)
3185942a 3920 return (strcpy (result, string));
7117c2d2 3921
f1be666c 3922 quote_spaces = (ifs_value && *ifs_value == 0);
3185942a
JA
3923
3924 s = string;
7117c2d2 3925 while (*s)
cce855bc 3926 {
f1be666c 3927 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
cce855bc
JA
3928 {
3929 s++;
3930 if (*s == '\0')
3931 break;
3932 }
7117c2d2 3933 COPY_CHAR_P (t, s, send);
cce855bc
JA
3934 }
3935 *t = '\0';
ac50fbac 3936
cce855bc
JA
3937 return result;
3938}
726f6388 3939
ac50fbac
CR
3940static WORD_LIST *
3941list_dequote_escapes (list)
3942 WORD_LIST *list;
3943{
3944 register WORD_LIST *w;
3945 char *t;
3946
3947 for (w = list; w; w = w->next)
3948 {
3949 t = w->word->word;
3950 w->word->word = dequote_escapes (t);
3951 free (t);
3952 }
3953 return list;
3954}
3955
0628567a
JA
3956/* Return a new string with the quoted representation of character C.
3957 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3958 set in any resultant WORD_DESC where this value is the word. */
726f6388
JA
3959static char *
3960make_quoted_char (c)
3961 int c;
3962{
3963 char *temp;
3964
f73dda09 3965 temp = (char *)xmalloc (3);
726f6388
JA
3966 if (c == 0)
3967 {
3968 temp[0] = CTLNUL;
3969 temp[1] = '\0';
3970 }
3971 else
3972 {
3973 temp[0] = CTLESC;
3974 temp[1] = c;
3975 temp[2] = '\0';
3976 }
3977 return (temp);
3978}
3979
0628567a
JA
3980/* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3981 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3982 this value is the word. */
ccc6cda3 3983char *
726f6388
JA
3984quote_string (string)
3985 char *string;
3986{
ccc6cda3 3987 register char *t;
7117c2d2
JA
3988 size_t slen;
3989 char *result, *send;
726f6388 3990
ccc6cda3 3991 if (*string == 0)
726f6388 3992 {
f73dda09 3993 result = (char *)xmalloc (2);
726f6388
JA
3994 result[0] = CTLNUL;
3995 result[1] = '\0';
3996 }
3997 else
3998 {
7117c2d2 3999 DECLARE_MBSTATE;
726f6388 4000
7117c2d2
JA
4001 slen = strlen (string);
4002 send = string + slen;
4003
4004 result = (char *)xmalloc ((slen * 2) + 1);
4005
4006 for (t = result; string < send; )
726f6388
JA
4007 {
4008 *t++ = CTLESC;
7117c2d2 4009 COPY_CHAR_P (t, string, send);
726f6388
JA
4010 }
4011 *t = '\0';
4012 }
4013 return (result);
4014}
4015
0628567a 4016/* De-quote quoted characters in STRING. */
726f6388
JA
4017char *
4018dequote_string (string)
4019 char *string;
4020{
7117c2d2
JA
4021 register char *s, *t;
4022 size_t slen;
4023 char *result, *send;
4024 DECLARE_MBSTATE;
726f6388 4025
a0c0a00f
CR
4026#if defined (DEBUG)
4027 if (string[0] == CTLESC && string[1] == 0)
4028 internal_inform ("dequote_string: string with bare CTLESC");
4029#endif
4030
7117c2d2
JA
4031 slen = strlen (string);
4032
4033 t = result = (char *)xmalloc (slen + 1);
726f6388
JA
4034
4035 if (QUOTED_NULL (string))
4036 {
4037 result[0] = '\0';
4038 return (result);
4039 }
4040
a0c0a00f
CR
4041 /* A string consisting of only a single CTLESC should pass through unchanged */
4042 if (string[0] == CTLESC && string[1] == 0)
4043 {
4044 result[0] = CTLESC;
4045 result[1] = '\0';
4046 return (result);
4047 }
4048
726f6388
JA
4049 /* If no character in the string can be quoted, don't bother examining
4050 each character. Just return a copy of the string passed to us. */
7117c2d2
JA
4051 if (strchr (string, CTLESC) == NULL)
4052 return (strcpy (result, string));
726f6388 4053
7117c2d2
JA
4054 send = string + slen;
4055 s = string;
4056 while (*s)
726f6388 4057 {
7117c2d2 4058 if (*s == CTLESC)
726f6388 4059 {
7117c2d2
JA
4060 s++;
4061 if (*s == '\0')
726f6388
JA
4062 break;
4063 }
7117c2d2 4064 COPY_CHAR_P (t, s, send);
726f6388
JA
4065 }
4066
4067 *t = '\0';
4068 return (result);
4069}
4070
4071/* Quote the entire WORD_LIST list. */
ccc6cda3 4072static WORD_LIST *
726f6388
JA
4073quote_list (list)
4074 WORD_LIST *list;
4075{
4076 register WORD_LIST *w;
ccc6cda3 4077 char *t;
726f6388
JA
4078
4079 for (w = list; w; w = w->next)
4080 {
ccc6cda3 4081 t = w->word->word;
726f6388 4082 w->word->word = quote_string (t);
3185942a
JA
4083 if (*t == 0)
4084 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
ccc6cda3 4085 w->word->flags |= W_QUOTED;
3185942a 4086 free (t);
726f6388 4087 }
ccc6cda3 4088 return list;
726f6388
JA
4089}
4090
0628567a
JA
4091/* De-quote quoted characters in each word in LIST. */
4092WORD_LIST *
7117c2d2
JA
4093dequote_list (list)
4094 WORD_LIST *list;
4095{
4096 register char *s;
4097 register WORD_LIST *tlist;
4098
4099 for (tlist = list; tlist; tlist = tlist->next)
4100 {
4101 s = dequote_string (tlist->word->word);
3185942a
JA
4102 if (QUOTED_NULL (tlist->word->word))
4103 tlist->word->flags &= ~W_HASQUOTEDNULL;
7117c2d2
JA
4104 free (tlist->word->word);
4105 tlist->word->word = s;
4106 }
4107 return list;
4108}
4109
4110/* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
4111 string. */
3185942a 4112char *
7117c2d2
JA
4113remove_quoted_escapes (string)
4114 char *string;
4115{
4116 char *t;
4117
4118 if (string)
4119 {
4120 t = dequote_escapes (string);
4121 strcpy (string, t);
4122 free (t);
4123 }
4124
4125 return (string);
4126}
4127
cce855bc
JA
4128/* Perform quoted null character removal on STRING. We don't allow any
4129 quoted null characters in the middle or at the ends of strings because
4130 of how expand_word_internal works. remove_quoted_nulls () turns
4131 STRING into an empty string iff it only consists of a quoted null,
4132 and removes all unquoted CTLNUL characters. */
3185942a 4133char *
cce855bc
JA
4134remove_quoted_nulls (string)
4135 char *string;
4136{
7117c2d2
JA
4137 register size_t slen;
4138 register int i, j, prev_i;
4139 DECLARE_MBSTATE;
4140
4141 if (strchr (string, CTLNUL) == 0) /* XXX */
4142 return string; /* XXX */
4143
4144 slen = strlen (string);
4145 i = j = 0;
4146
4147 while (i < slen)
4148 {
4149 if (string[i] == CTLESC)
b80f6443
JA
4150 {
4151 /* Old code had j++, but we cannot assume that i == j at this
4152 point -- what if a CTLNUL has already been removed from the
4153 string? We don't want to drop the CTLESC or recopy characters
4154 that we've already copied down. */
4155 i++; string[j++] = CTLESC;
4156 if (i == slen)
4157 break;
4158 }
7117c2d2 4159 else if (string[i] == CTLNUL)
a48a8ac3
CR
4160 {
4161 i++;
4162 continue;
4163 }
7117c2d2
JA
4164
4165 prev_i = i;
4166 ADVANCE_CHAR (string, slen, i);
4167 if (j < prev_i)
cce855bc 4168 {
7117c2d2 4169 do string[j++] = string[prev_i++]; while (prev_i < i);
cce855bc 4170 }
7117c2d2
JA
4171 else
4172 j = i;
cce855bc 4173 }
7117c2d2
JA
4174 string[j] = '\0';
4175
4176 return (string);
cce855bc
JA
4177}
4178
4179/* Perform quoted null character removal on each element of LIST.
4180 This modifies LIST. */
4181void
4182word_list_remove_quoted_nulls (list)
4183 WORD_LIST *list;
4184{
4185 register WORD_LIST *t;
4186
4187 for (t = list; t; t = t->next)
95732b49
JA
4188 {
4189 remove_quoted_nulls (t->word->word);
4190 t->word->flags &= ~W_HASQUOTEDNULL;
4191 }
cce855bc
JA
4192}
4193
4194/* **************************************************************** */
4195/* */
4196/* Functions for Matching and Removing Patterns */
4197/* */
4198/* **************************************************************** */
4199
b80f6443
JA
4200#if defined (HANDLE_MULTIBYTE)
4201#if 0 /* Currently unused */
4202static unsigned char *
4203mb_getcharlens (string, len)
4204 char *string;
4205 int len;
4206{
4207 int i, offset, last;
4208 unsigned char *ret;
4209 char *p;
4210 DECLARE_MBSTATE;
4211
4212 i = offset = 0;
4213 last = 0;
4214 ret = (unsigned char *)xmalloc (len);
4215 memset (ret, 0, len);
4216 while (string[last])
4217 {
4218 ADVANCE_CHAR (string, len, offset);
4219 ret[last] = offset - last;
4220 last = offset;
4221 }
4222 return ret;
4223}
4224#endif
4225#endif
4226
cce855bc
JA
4227/* Remove the portion of PARAM matched by PATTERN according to OP, where OP
4228 can have one of 4 values:
4229 RP_LONG_LEFT remove longest matching portion at start of PARAM
726f6388
JA
4230 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
4231 RP_LONG_RIGHT remove longest matching portion at end of PARAM
4232 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
4233*/
4234
4235#define RP_LONG_LEFT 1
4236#define RP_SHORT_LEFT 2
4237#define RP_LONG_RIGHT 3
4238#define RP_SHORT_RIGHT 4
4239
495aee44 4240/* Returns its first argument if nothing matched; new memory otherwise */
726f6388 4241static char *
b80f6443 4242remove_upattern (param, pattern, op)
726f6388
JA
4243 char *param, *pattern;
4244 int op;
4245{
a0c0a00f 4246 register size_t len;
ccc6cda3 4247 register char *end;
726f6388
JA
4248 register char *p, *ret, c;
4249
ccc6cda3
JA
4250 len = STRLEN (param);
4251 end = param + len;
726f6388
JA
4252
4253 switch (op)
4254 {
4255 case RP_LONG_LEFT: /* remove longest match at start */
4256 for (p = end; p >= param; p--)
4257 {
4258 c = *p; *p = '\0';
f73dda09 4259 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
726f6388
JA
4260 {
4261 *p = c;
4262 return (savestring (p));
4263 }
4264 *p = c;
b80f6443 4265
726f6388
JA
4266 }
4267 break;
4268
4269 case RP_SHORT_LEFT: /* remove shortest match at start */
4270 for (p = param; p <= end; p++)
4271 {
4272 c = *p; *p = '\0';
f73dda09 4273 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
726f6388
JA
4274 {
4275 *p = c;
4276 return (savestring (p));
4277 }
4278 *p = c;
4279 }
4280 break;
4281
ccc6cda3
JA
4282 case RP_LONG_RIGHT: /* remove longest match at end */
4283 for (p = param; p <= end; p++)
4284 {
f73dda09 4285 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
ccc6cda3
JA
4286 {
4287 c = *p; *p = '\0';
4288 ret = savestring (param);
4289 *p = c;
4290 return (ret);
4291 }
4292 }
4293 break;
4294
4295 case RP_SHORT_RIGHT: /* remove shortest match at end */
4296 for (p = end; p >= param; p--)
4297 {
f73dda09 4298 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
ccc6cda3
JA
4299 {
4300 c = *p; *p = '\0';
4301 ret = savestring (param);
4302 *p = c;
4303 return (ret);
4304 }
4305 }
4306 break;
4307 }
b80f6443 4308
495aee44 4309 return (param); /* no match, return original string */
ccc6cda3
JA
4310}
4311
b80f6443 4312#if defined (HANDLE_MULTIBYTE)
495aee44 4313/* Returns its first argument if nothing matched; new memory otherwise */
b80f6443
JA
4314static wchar_t *
4315remove_wpattern (wparam, wstrlen, wpattern, op)
4316 wchar_t *wparam;
4317 size_t wstrlen;
4318 wchar_t *wpattern;
4319 int op;
4320{
0628567a
JA
4321 wchar_t wc, *ret;
4322 int n;
b80f6443
JA
4323
4324 switch (op)
4325 {
4326 case RP_LONG_LEFT: /* remove longest match at start */
4327 for (n = wstrlen; n >= 0; n--)
4328 {
4329 wc = wparam[n]; wparam[n] = L'\0';
4330 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4331 {
4332 wparam[n] = wc;
4333 return (wcsdup (wparam + n));
4334 }
4335 wparam[n] = wc;
4336 }
4337 break;
4338
4339 case RP_SHORT_LEFT: /* remove shortest match at start */
4340 for (n = 0; n <= wstrlen; n++)
4341 {
4342 wc = wparam[n]; wparam[n] = L'\0';
4343 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4344 {
4345 wparam[n] = wc;
4346 return (wcsdup (wparam + n));
4347 }
4348 wparam[n] = wc;
4349 }
4350 break;
4351
4352 case RP_LONG_RIGHT: /* remove longest match at end */
4353 for (n = 0; n <= wstrlen; n++)
4354 {
4355 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4356 {
4357 wc = wparam[n]; wparam[n] = L'\0';
4358 ret = wcsdup (wparam);
4359 wparam[n] = wc;
4360 return (ret);
4361 }
4362 }
4363 break;
4364
4365 case RP_SHORT_RIGHT: /* remove shortest match at end */
4366 for (n = wstrlen; n >= 0; n--)
4367 {
4368 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4369 {
4370 wc = wparam[n]; wparam[n] = L'\0';
4371 ret = wcsdup (wparam);
4372 wparam[n] = wc;
4373 return (ret);
4374 }
4375 }
4376 break;
4377 }
4378
495aee44 4379 return (wparam); /* no match, return original string */
b80f6443
JA
4380}
4381#endif /* HANDLE_MULTIBYTE */
4382
4383static char *
4384remove_pattern (param, pattern, op)
4385 char *param, *pattern;
4386 int op;
4387{
495aee44
CR
4388 char *xret;
4389
b80f6443
JA
4390 if (param == NULL)
4391 return (param);
4392 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4393 return (savestring (param));
4394
4395#if defined (HANDLE_MULTIBYTE)
4396 if (MB_CUR_MAX > 1)
4397 {
4398 wchar_t *ret, *oret;
4399 size_t n;
4400 wchar_t *wparam, *wpattern;
4401 mbstate_t ps;
b80f6443
JA
4402
4403 n = xdupmbstowcs (&wpattern, NULL, pattern);
4404 if (n == (size_t)-1)
495aee44
CR
4405 {
4406 xret = remove_upattern (param, pattern, op);
4407 return ((xret == param) ? savestring (param) : xret);
4408 }
b80f6443 4409 n = xdupmbstowcs (&wparam, NULL, param);
ac50fbac 4410
b80f6443
JA
4411 if (n == (size_t)-1)
4412 {
4413 free (wpattern);
495aee44
CR
4414 xret = remove_upattern (param, pattern, op);
4415 return ((xret == param) ? savestring (param) : xret);
b80f6443
JA
4416 }
4417 oret = ret = remove_wpattern (wparam, n, wpattern, op);
495aee44
CR
4418 /* Don't bother to convert wparam back to multibyte string if nothing
4419 matched; just return copy of original string */
4420 if (ret == wparam)
4421 {
4422 free (wparam);
4423 free (wpattern);
4424 return (savestring (param));
4425 }
b80f6443
JA
4426
4427 free (wparam);
4428 free (wpattern);
4429
4430 n = strlen (param);
0628567a 4431 xret = (char *)xmalloc (n + 1);
b80f6443
JA
4432 memset (&ps, '\0', sizeof (mbstate_t));
4433 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4434 xret[n] = '\0'; /* just to make sure */
4435 free (oret);
4436 return xret;
4437 }
4438 else
4439#endif
ccc6cda3 4440 {
495aee44
CR
4441 xret = remove_upattern (param, pattern, op);
4442 return ((xret == param) ? savestring (param) : xret);
ccc6cda3
JA
4443 }
4444}
4445
4446/* Match PAT anywhere in STRING and return the match boundaries.
4447 This returns 1 in case of a successful match, 0 otherwise. SP
4448 and EP are pointers into the string where the match begins and
4449 ends, respectively. MTYPE controls what kind of match is attempted.
4450 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4451 of the string, respectively. The longest match is returned. */
4452static int
b80f6443 4453match_upattern (string, pat, mtype, sp, ep)
ccc6cda3
JA
4454 char *string, *pat;
4455 int mtype;
4456 char **sp, **ep;
4457{
a0c0a00f
CR
4458 int c, mlen;
4459 size_t len;
95732b49 4460 register char *p, *p1, *npat;
ccc6cda3 4461 char *end;
495aee44 4462 int n1;
ccc6cda3 4463
95732b49
JA
4464 /* If the pattern doesn't match anywhere in the string, go ahead and
4465 short-circuit right away. A minor optimization, saves a bunch of
4466 unnecessary calls to strmatch (up to N calls for a string of N
4467 characters) if the match is unsuccessful. To preserve the semantics
4468 of the substring matches below, we make sure that the pattern has
4469 `*' as first and last character, making a new pattern if necessary. */
4470 /* XXX - check this later if I ever implement `**' with special meaning,
4471 since this will potentially result in `**' at the beginning or end */
4472 len = STRLEN (pat);
0001803f 4473 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
95732b49 4474 {
a0c0a00f
CR
4475 int unescaped_backslash;
4476 char *pp;
4477
0628567a 4478 p = npat = (char *)xmalloc (len + 3);
95732b49 4479 p1 = pat;
0001803f 4480 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
95732b49
JA
4481 *p++ = '*';
4482 while (*p1)
4483 *p++ = *p1++;
a0c0a00f
CR
4484#if 1
4485 /* Need to also handle a pattern that ends with an unescaped backslash.
4486 For right now, we ignore it because the pattern matching code will
4487 fail the match anyway */
4488 /* If the pattern ends with a `*' we leave it alone if it's preceded by
4489 an even number of backslashes, but if it's escaped by a backslash
4490 we need to add another `*'. */
4491 if (p1[-1] == '*' && (unescaped_backslash = p1[-2] == '\\'))
4492 {
4493 pp = p1 - 3;
4494 while (pp >= pat && *pp-- == '\\')
4495 unescaped_backslash = 1 - unescaped_backslash;
4496 if (unescaped_backslash)
4497 *p++ = '*';
4498 }
4499 else if (p1[-1] != '*')
95732b49 4500 *p++ = '*';
a0c0a00f
CR
4501#else
4502 if (p1[-1] != '*' || p1[-2] == '\\')
4503 *p++ = '*';
4504#endif
95732b49
JA
4505 *p = '\0';
4506 }
4507 else
4508 npat = pat;
a0c0a00f 4509 c = strmatch (npat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
95732b49
JA
4510 if (npat != pat)
4511 free (npat);
4512 if (c == FNM_NOMATCH)
4513 return (0);
4514
b80f6443
JA
4515 len = STRLEN (string);
4516 end = string + len;
ccc6cda3 4517
495aee44
CR
4518 mlen = umatchlen (pat, len);
4519
ccc6cda3
JA
4520 switch (mtype)
4521 {
4522 case MATCH_ANY:
4523 for (p = string; p <= end; p++)
4524 {
a0c0a00f 4525 if (match_pattern_char (pat, p, FNMATCH_IGNCASE))
ccc6cda3 4526 {
495aee44
CR
4527 p1 = (mlen == -1) ? end : p + mlen;
4528 /* p1 - p = length of portion of string to be considered
4529 p = current position in string
4530 mlen = number of characters consumed by match (-1 for entire string)
4531 end = end of string
4532 we want to break immediately if the potential match len
4533 is greater than the number of characters remaining in the
4534 string
4535 */
4536 if (p1 > end)
4537 break;
4538 for ( ; p1 >= p; p1--)
ccc6cda3
JA
4539 {
4540 c = *p1; *p1 = '\0';
a0c0a00f 4541 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
ccc6cda3
JA
4542 {
4543 *p1 = c;
4544 *sp = p;
4545 *ep = p1;
4546 return 1;
4547 }
4548 *p1 = c;
495aee44
CR
4549#if 1
4550 /* If MLEN != -1, we have a fixed length pattern. */
4551 if (mlen != -1)
4552 break;
4553#endif
ccc6cda3
JA
4554 }
4555 }
4556 }
b80f6443 4557
ccc6cda3
JA
4558 return (0);
4559
4560 case MATCH_BEG:
a0c0a00f 4561 if (match_pattern_char (pat, string, FNMATCH_IGNCASE) == 0)
28ef6c31 4562 return (0);
b80f6443 4563
495aee44 4564 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
ccc6cda3
JA
4565 {
4566 c = *p; *p = '\0';
a0c0a00f 4567 if (strmatch (pat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
ccc6cda3
JA
4568 {
4569 *p = c;
4570 *sp = string;
4571 *ep = p;
4572 return 1;
4573 }
4574 *p = c;
495aee44
CR
4575 /* If MLEN != -1, we have a fixed length pattern. */
4576 if (mlen != -1)
4577 break;
ccc6cda3 4578 }
b80f6443 4579
ccc6cda3 4580 return (0);
726f6388 4581
ccc6cda3 4582 case MATCH_END:
495aee44 4583 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
b80f6443 4584 {
a0c0a00f 4585 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4586 {
4587 *sp = p;
4588 *ep = end;
4589 return 1;
4590 }
495aee44
CR
4591 /* If MLEN != -1, we have a fixed length pattern. */
4592 if (mlen != -1)
4593 break;
b80f6443
JA
4594 }
4595
4596 return (0);
4597 }
4598
4599 return (0);
4600}
4601
4602#if defined (HANDLE_MULTIBYTE)
a0c0a00f
CR
4603
4604#define WFOLD(c) (match_ignore_case && iswupper (c) ? towlower (c) : (c))
4605
b80f6443
JA
4606/* Match WPAT anywhere in WSTRING and return the match boundaries.
4607 This returns 1 in case of a successful match, 0 otherwise. Wide
4608 character version. */
4609static int
4610match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4611 wchar_t *wstring;
4612 char **indices;
4613 size_t wstrlen;
4614 wchar_t *wpat;
4615 int mtype;
4616 char **sp, **ep;
4617{
95732b49 4618 wchar_t wc, *wp, *nwpat, *wp1;
495aee44
CR
4619 size_t len;
4620 int mlen;
4621 int n, n1, n2, simple;
4622
4623 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4624#if defined (EXTENDED_GLOB)
4625 if (extended_glob)
91717ba3 4626 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
b80f6443
JA
4627#endif
4628
95732b49
JA
4629 /* If the pattern doesn't match anywhere in the string, go ahead and
4630 short-circuit right away. A minor optimization, saves a bunch of
4631 unnecessary calls to strmatch (up to N calls for a string of N
4632 characters) if the match is unsuccessful. To preserve the semantics
4633 of the substring matches below, we make sure that the pattern has
4634 `*' as first and last character, making a new pattern if necessary. */
95732b49 4635 len = wcslen (wpat);
0001803f 4636 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
95732b49 4637 {
a0c0a00f
CR
4638 int unescaped_backslash;
4639 wchar_t *wpp;
4640
0628567a 4641 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
95732b49 4642 wp1 = wpat;
0001803f 4643 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
95732b49
JA
4644 *wp++ = L'*';
4645 while (*wp1 != L'\0')
4646 *wp++ = *wp1++;
a0c0a00f
CR
4647#if 1
4648 /* See comments above in match_upattern. */
4649 if (wp1[-1] == L'*' && (unescaped_backslash = wp1[-2] == L'\\'))
4650 {
4651 wpp = wp1 - 3;
4652 while (wpp >= wpat && *wpp-- == L'\\')
4653 unescaped_backslash = 1 - unescaped_backslash;
4654 if (unescaped_backslash)
4655 *wp++ = L'*';
4656 }
4657 else if (wp1[-1] != L'*')
4658 *wp++ = L'*';
4659#else
95732b49
JA
4660 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4661 *wp++ = L'*';
a0c0a00f 4662#endif
95732b49
JA
4663 *wp = '\0';
4664 }
4665 else
4666 nwpat = wpat;
a0c0a00f 4667 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
95732b49
JA
4668 if (nwpat != wpat)
4669 free (nwpat);
4670 if (len == FNM_NOMATCH)
4671 return (0);
4672
495aee44
CR
4673 mlen = wmatchlen (wpat, wstrlen);
4674
4675/* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
b80f6443
JA
4676 switch (mtype)
4677 {
4678 case MATCH_ANY:
4679 for (n = 0; n <= wstrlen; n++)
4680 {
a0c0a00f 4681 n2 = simple ? (WFOLD(*wpat) == WFOLD(wstring[n])) : match_pattern_wchar (wpat, wstring + n, FNMATCH_IGNCASE);
495aee44 4682 if (n2)
b80f6443 4683 {
495aee44
CR
4684 n1 = (mlen == -1) ? wstrlen : n + mlen;
4685 if (n1 > wstrlen)
4686 break;
4687
4688 for ( ; n1 >= n; n1--)
b80f6443
JA
4689 {
4690 wc = wstring[n1]; wstring[n1] = L'\0';
a0c0a00f 4691 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4692 {
4693 wstring[n1] = wc;
4694 *sp = indices[n];
4695 *ep = indices[n1];
4696 return 1;
4697 }
4698 wstring[n1] = wc;
495aee44
CR
4699 /* If MLEN != -1, we have a fixed length pattern. */
4700 if (mlen != -1)
4701 break;
b80f6443
JA
4702 }
4703 }
4704 }
4705
4706 return (0);
4707
4708 case MATCH_BEG:
a0c0a00f 4709 if (match_pattern_wchar (wpat, wstring, FNMATCH_IGNCASE) == 0)
b80f6443
JA
4710 return (0);
4711
495aee44 4712 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
b80f6443
JA
4713 {
4714 wc = wstring[n]; wstring[n] = L'\0';
a0c0a00f 4715 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4716 {
4717 wstring[n] = wc;
4718 *sp = indices[0];
4719 *ep = indices[n];
4720 return 1;
4721 }
4722 wstring[n] = wc;
495aee44
CR
4723 /* If MLEN != -1, we have a fixed length pattern. */
4724 if (mlen != -1)
4725 break;
b80f6443
JA
4726 }
4727
4728 return (0);
4729
4730 case MATCH_END:
495aee44 4731 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
b80f6443 4732 {
a0c0a00f 4733 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4734 {
4735 *sp = indices[n];
4736 *ep = indices[wstrlen];
4737 return 1;
4738 }
495aee44
CR
4739 /* If MLEN != -1, we have a fixed length pattern. */
4740 if (mlen != -1)
4741 break;
b80f6443
JA
4742 }
4743
ccc6cda3 4744 return (0);
726f6388 4745 }
ccc6cda3
JA
4746
4747 return (0);
726f6388 4748}
a0c0a00f 4749#undef WFOLD
b80f6443
JA
4750#endif /* HANDLE_MULTIBYTE */
4751
4752static int
4753match_pattern (string, pat, mtype, sp, ep)
4754 char *string, *pat;
4755 int mtype;
4756 char **sp, **ep;
4757{
4758#if defined (HANDLE_MULTIBYTE)
4759 int ret;
4760 size_t n;
4761 wchar_t *wstring, *wpat;
4762 char **indices;
495aee44 4763 size_t slen, plen, mslen, mplen;
b80f6443
JA
4764#endif
4765
a0c0a00f 4766 if (string == 0 || pat == 0 || *pat == 0)
b80f6443
JA
4767 return (0);
4768
4769#if defined (HANDLE_MULTIBYTE)
4770 if (MB_CUR_MAX > 1)
4771 {
495aee44 4772 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
495aee44
CR
4773 return (match_upattern (string, pat, mtype, sp, ep));
4774
b80f6443
JA
4775 n = xdupmbstowcs (&wpat, NULL, pat);
4776 if (n == (size_t)-1)
4777 return (match_upattern (string, pat, mtype, sp, ep));
4778 n = xdupmbstowcs (&wstring, &indices, string);
4779 if (n == (size_t)-1)
4780 {
4781 free (wpat);
4782 return (match_upattern (string, pat, mtype, sp, ep));
4783 }
4784 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4785
4786 free (wpat);
4787 free (wstring);
4788 free (indices);
4789
4790 return (ret);
4791 }
4792 else
4793#endif
4794 return (match_upattern (string, pat, mtype, sp, ep));
4795}
726f6388 4796
cce855bc
JA
4797static int
4798getpatspec (c, value)
4799 int c;
4800 char *value;
4801{
4802 if (c == '#')
4803 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4804 else /* c == '%' */
4805 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4806}
4807
4808/* Posix.2 says that the WORD should be run through tilde expansion,
4809 parameter expansion, command substitution and arithmetic expansion.
4810 This leaves the result quoted, so quote_string_for_globbing () has
f73dda09 4811 to be called to fix it up for strmatch (). If QUOTED is non-zero,
cce855bc
JA
4812 it means that the entire expression was enclosed in double quotes.
4813 This means that quoting characters in the pattern do not make any
4814 special pattern characters quoted. For example, the `*' in the
4815 following retains its special meaning: "${foo#'*'}". */
4816static char *
4817getpattern (value, quoted, expandpat)
4818 char *value;
4819 int quoted, expandpat;
4820{
4821 char *pat, *tword;
4822 WORD_LIST *l;
0628567a 4823#if 0
cce855bc 4824 int i;
0628567a 4825#endif
7117c2d2
JA
4826 /* There is a problem here: how to handle single or double quotes in the
4827 pattern string when the whole expression is between double quotes?
4828 POSIX.2 says that enclosing double quotes do not cause the pattern to
4829 be quoted, but does that leave us a problem with @ and array[@] and their
4830 expansions inside a pattern? */
4831#if 0
cce855bc
JA
4832 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4833 {
4834 i = 0;
a0c0a00f 4835 pat = string_extract_double_quoted (tword, &i, SX_STRIPDQ);
cce855bc
JA
4836 free (tword);
4837 tword = pat;
4838 }
7117c2d2 4839#endif
cce855bc 4840
7117c2d2
JA
4841 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4842 word splitting. */
95732b49 4843 l = *value ? expand_string_for_rhs (value,
7117c2d2 4844 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
cce855bc 4845 (int *)NULL, (int *)NULL)
cce855bc 4846 : (WORD_LIST *)0;
cce855bc
JA
4847 pat = string_list (l);
4848 dispose_words (l);
4849 if (pat)
4850 {
4851 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4852 free (pat);
4853 pat = tword;
4854 }
4855 return (pat);
4856}
4857
7117c2d2 4858#if 0
cce855bc
JA
4859/* Handle removing a pattern from a string as a result of ${name%[%]value}
4860 or ${name#[#]value}. */
4861static char *
7117c2d2
JA
4862variable_remove_pattern (value, pattern, patspec, quoted)
4863 char *value, *pattern;
4864 int patspec, quoted;
cce855bc 4865{
7117c2d2 4866 char *tword;
cce855bc 4867
7117c2d2 4868 tword = remove_pattern (value, pattern, patspec);
cce855bc 4869
cce855bc
JA
4870 return (tword);
4871}
a0c0a00f
CR
4872#endif
4873
4874static char *
4875list_remove_pattern (list, pattern, patspec, itype, quoted)
4876 WORD_LIST *list;
4877 char *pattern;
4878 int patspec, itype, quoted;
4879{
4880 WORD_LIST *new, *l;
4881 WORD_DESC *w;
4882 char *tword;
4883
4884 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4885 {
4886 tword = remove_pattern (l->word->word, pattern, patspec);
4887 w = alloc_word_desc ();
4888 w->word = tword ? tword : savestring ("");
4889 new = make_word_list (w, new);
4890 }
4891
4892 l = REVERSE_LIST (new, WORD_LIST *);
4893 tword = string_list_pos_params (itype, l, quoted);
4894 dispose_words (l);
4895
4896 return (tword);
4897}
4898
4899static char *
4900parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4901 int itype;
4902 char *pattern;
4903 int patspec, quoted;
4904{
4905 char *ret;
4906 WORD_LIST *list;
4907
4908 list = list_rest_of_args ();
4909 if (list == 0)
4910 return ((char *)NULL);
4911 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4912 dispose_words (list);
4913 return (ret);
4914}
4915
4916#if defined (ARRAY_VARS)
4917static char *
4918array_remove_pattern (var, pattern, patspec, varname, quoted)
4919 SHELL_VAR *var;
4920 char *pattern;
4921 int patspec;
4922 char *varname; /* so we can figure out how it's indexed */
4923 int quoted;
4924{
4925 ARRAY *a;
4926 HASH_TABLE *h;
4927 int itype;
4928 char *ret;
4929 WORD_LIST *list;
4930 SHELL_VAR *v;
4931
4932 /* compute itype from varname here */
4933 v = array_variable_part (varname, &ret, 0);
4934
4935 /* XXX */
4936 if (v && invisible_p (v))
4937 return ((char *)NULL);
4938
4939 itype = ret[0];
4940
4941 a = (v && array_p (v)) ? array_cell (v) : 0;
4942 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4943
4944 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4945 if (list == 0)
4946 return ((char *)NULL);
4947 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4948 dispose_words (list);
4949
4950 return ret;
4951}
4952#endif /* ARRAY_VARS */
4953
4954static char *
4955parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4956 char *varname, *value;
4957 int ind;
4958 char *patstr;
4959 int rtype, quoted, flags;
4960{
4961 int vtype, patspec, starsub;
4962 char *temp1, *val, *pattern;
4963 SHELL_VAR *v;
4964
4965 if (value == 0)
4966 return ((char *)NULL);
4967
4968 this_command_name = varname;
4969
4970 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4971 if (vtype == -1)
4972 return ((char *)NULL);
4973
4974 starsub = vtype & VT_STARSUB;
4975 vtype &= ~VT_STARSUB;
4976
4977 patspec = getpatspec (rtype, patstr);
4978 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4979 patstr++;
4980
4981 /* Need to pass getpattern newly-allocated memory in case of expansion --
4982 the expansion code will free the passed string on an error. */
4983 temp1 = savestring (patstr);
4984 pattern = getpattern (temp1, quoted, 1);
4985 free (temp1);
4986
4987 temp1 = (char *)NULL; /* shut up gcc */
4988 switch (vtype)
4989 {
4990 case VT_VARIABLE:
4991 case VT_ARRAYMEMBER:
4992 temp1 = remove_pattern (val, pattern, patspec);
4993 if (vtype == VT_VARIABLE)
4994 FREE (val);
4995 if (temp1)
4996 {
4997 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
4998 ? quote_string (temp1)
4999 : quote_escapes (temp1);
5000 free (temp1);
5001 temp1 = val;
5002 }
5003 break;
5004#if defined (ARRAY_VARS)
5005 case VT_ARRAYVAR:
5006 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
5007 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5008 {
5009 val = quote_escapes (temp1);
5010 free (temp1);
5011 temp1 = val;
5012 }
5013 break;
5014#endif
5015 case VT_POSPARMS:
5016 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
5017 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5018 {
5019 val = quote_escapes (temp1);
5020 free (temp1);
5021 temp1 = val;
5022 }
5023 break;
5024 }
5025
5026 FREE (pattern);
5027 return temp1;
5028}
5029
5030static char *
5031string_var_assignment (v, s)
5032 SHELL_VAR *v;
5033 char *s;
5034{
5035 char flags[MAX_ATTRIBUTES], *ret, *val;
5036 int i;
5037
5038 val = sh_quote_reusable (s, 0);
5039 i = var_attribute_string (v, 0, flags);
5040 ret = (char *)xmalloc (i + strlen (val) + strlen (v->name) + 16 + MAX_ATTRIBUTES);
5041 if (i > 0)
5042 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
5043 else
5044 sprintf (ret, "%s=%s", v->name, val);
5045 free (val);
5046 return ret;
5047}
5048
5049#if defined (ARRAY_VARS)
5050static char *
5051array_var_assignment (v, itype, quoted)
5052 SHELL_VAR *v;
5053 int itype, quoted;
5054{
5055 char *ret, *val, flags[MAX_ATTRIBUTES];
5056 int i;
5057
5058 if (v == 0)
5059 return (char *)NULL;
5060 val = array_p (v) ? array_to_assign (array_cell (v), 0)
5061 : assoc_to_assign (assoc_cell (v), 0);
5062 if (val == 0)
5063 {
5064 val = (char *)xmalloc (3);
5065 val[0] = '(';
5066 val[1] = ')';
5067 val[2] = 0;
5068 }
5069 else
5070 {
5071 ret = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (val) : quote_escapes (val);
5072 free (val);
5073 val = ret;
5074 }
5075 i = var_attribute_string (v, 0, flags);
5076 ret = (char *)xmalloc (i + strlen (val) + strlen (v->name) + 16);
5077 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
5078 free (val);
5079 return ret;
5080}
5081#endif
5082
5083static char *
5084pos_params_assignment (list, itype, quoted)
5085 WORD_LIST *list;
5086 int itype;
5087 int quoted;
5088{
5089 char *temp, *ret;
5090
5091 /* first, we transform the list to quote each word. */
5092 temp = list_transform ('Q', (SHELL_VAR *)0, list, itype, quoted);
5093 ret = (char *)xmalloc (strlen (temp) + 8);
5094 strcpy (ret, "set -- ");
5095 strcpy (ret + 7, temp);
5096 free (temp);
5097 return ret;
5098}
5099
5100static char *
5101string_transform (xc, v, s)
5102 int xc;
5103 SHELL_VAR *v;
5104 char *s;
5105{
5106 char *ret, flags[MAX_ATTRIBUTES];
5107 int i;
5108
5109 if (((xc == 'A' || xc == 'a') && v == 0) || (xc != 'a' && s == 0))
5110 return (char *)NULL;
5111
5112 switch (xc)
5113 {
5114 /* Transformations that interrogate the variable */
5115 case 'a':
5116 i = var_attribute_string (v, 0, flags);
5117 ret = (i > 0) ? savestring (flags) : (char *)NULL;
5118 break;
5119 case 'A':
5120 ret = string_var_assignment (v, s);
5121 break;
5122 /* Transformations that modify the variable's value */
5123 case 'E':
5124 ret = ansiexpand (s, 0, strlen (s), (int *)0);
5125 break;
5126 case 'P':
5127 ret = decode_prompt_string (s);
5128 break;
5129 case 'Q':
5130 ret = sh_quote_reusable (s, 0);
5131 break;
5132 default:
5133 ret = (char *)NULL;
5134 break;
5135 }
5136 return ret;
5137}
cce855bc
JA
5138
5139static char *
a0c0a00f
CR
5140list_transform (xc, v, list, itype, quoted)
5141 int xc;
5142 SHELL_VAR *v;
cce855bc 5143 WORD_LIST *list;
a0c0a00f 5144 int itype, quoted;
cce855bc
JA
5145{
5146 WORD_LIST *new, *l;
5147 WORD_DESC *w;
5148 char *tword;
5149
5150 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
5151 {
a0c0a00f 5152 tword = string_transform (xc, v, l->word->word);
95732b49 5153 w = alloc_word_desc ();
a0c0a00f 5154 w->word = tword ? tword : savestring (""); /* XXX */
cce855bc
JA
5155 new = make_word_list (w, new);
5156 }
5157
5158 l = REVERSE_LIST (new, WORD_LIST *);
3185942a 5159 tword = string_list_pos_params (itype, l, quoted);
cce855bc 5160 dispose_words (l);
3185942a 5161
cce855bc
JA
5162 return (tword);
5163}
5164
5165static char *
a0c0a00f
CR
5166parameter_list_transform (xc, itype, quoted)
5167 int xc;
7117c2d2 5168 int itype;
a0c0a00f 5169 int quoted;
cce855bc 5170{
7117c2d2 5171 char *ret;
cce855bc
JA
5172 WORD_LIST *list;
5173
cce855bc 5174 list = list_rest_of_args ();
7117c2d2
JA
5175 if (list == 0)
5176 return ((char *)NULL);
a0c0a00f
CR
5177 if (xc == 'A')
5178 return (pos_params_assignment (list, itype, quoted));
5179 ret = list_transform (xc, (SHELL_VAR *)0, list, itype, quoted);
cce855bc 5180 dispose_words (list);
cce855bc
JA
5181 return (ret);
5182}
5183
5184#if defined (ARRAY_VARS)
5185static char *
a0c0a00f
CR
5186array_transform (xc, var, varname, quoted)
5187 int xc;
3185942a 5188 SHELL_VAR *var;
7117c2d2
JA
5189 char *varname; /* so we can figure out how it's indexed */
5190 int quoted;
cce855bc 5191{
3185942a
JA
5192 ARRAY *a;
5193 HASH_TABLE *h;
7117c2d2
JA
5194 int itype;
5195 char *ret;
5196 WORD_LIST *list;
5197 SHELL_VAR *v;
cce855bc 5198
7117c2d2
JA
5199 /* compute itype from varname here */
5200 v = array_variable_part (varname, &ret, 0);
ac50fbac
CR
5201
5202 /* XXX */
a0c0a00f 5203 if (v && invisible_p (v))
ac50fbac
CR
5204 return ((char *)NULL);
5205
7117c2d2
JA
5206 itype = ret[0];
5207
a0c0a00f
CR
5208 if (xc == 'A')
5209 return (array_var_assignment (v, itype, quoted));
5210
3185942a
JA
5211 a = (v && array_p (v)) ? array_cell (v) : 0;
5212 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
5213
5214 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
7117c2d2
JA
5215 if (list == 0)
5216 return ((char *)NULL);
a0c0a00f 5217 ret = list_transform (xc, v, list, itype, quoted);
7117c2d2
JA
5218 dispose_words (list);
5219
5220 return ret;
5221}
5222#endif /* ARRAY_VARS */
5223
5224static char *
a0c0a00f 5225parameter_brace_transform (varname, value, ind, xform, rtype, quoted, flags)
495aee44
CR
5226 char *varname, *value;
5227 int ind;
a0c0a00f 5228 char *xform;
495aee44 5229 int rtype, quoted, flags;
7117c2d2 5230{
a0c0a00f
CR
5231 int vtype, xc;
5232 char *temp1, *val;
7117c2d2
JA
5233 SHELL_VAR *v;
5234
a0c0a00f
CR
5235 xc = xform[0];
5236 if (value == 0 && xc != 'A' && xc != 'a')
7117c2d2
JA
5237 return ((char *)NULL);
5238
5239 this_command_name = varname;
5240
495aee44 5241 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7117c2d2 5242 if (vtype == -1)
cce855bc
JA
5243 return ((char *)NULL);
5244
a0c0a00f
CR
5245 /* check for valid values of xc */
5246 switch (xc)
5247 {
5248 case 'a': /* expand to a string with just attributes */
5249 case 'A': /* expand as an assignment statement with attributes */
5250 case 'E': /* expand like $'...' */
5251 case 'P': /* expand like prompt string */
5252 case 'Q': /* quote reusably */
5253 break;
5254 default:
5255 return &expand_param_error;
5256 }
cce855bc 5257
7117c2d2 5258 temp1 = (char *)NULL; /* shut up gcc */
a0c0a00f 5259 switch (vtype & ~VT_STARSUB)
cce855bc 5260 {
7117c2d2
JA
5261 case VT_VARIABLE:
5262 case VT_ARRAYMEMBER:
a0c0a00f 5263 temp1 = string_transform (xc, v, val);
7117c2d2
JA
5264 if (vtype == VT_VARIABLE)
5265 FREE (val);
5266 if (temp1)
28ef6c31 5267 {
3185942a
JA
5268 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5269 ? quote_string (temp1)
5270 : quote_escapes (temp1);
7117c2d2
JA
5271 free (temp1);
5272 temp1 = val;
28ef6c31 5273 }
7117c2d2
JA
5274 break;
5275#if defined (ARRAY_VARS)
5276 case VT_ARRAYVAR:
a0c0a00f 5277 temp1 = array_transform (xc, v, varname, quoted);
7117c2d2 5278 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
cce855bc 5279 {
7117c2d2
JA
5280 val = quote_escapes (temp1);
5281 free (temp1);
5282 temp1 = val;
cce855bc 5283 }
7117c2d2
JA
5284 break;
5285#endif
5286 case VT_POSPARMS:
a0c0a00f 5287 temp1 = parameter_list_transform (xc, varname[0], quoted);
7117c2d2
JA
5288 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5289 {
5290 val = quote_escapes (temp1);
5291 free (temp1);
5292 temp1 = val;
5293 }
5294 break;
cce855bc
JA
5295 }
5296
7117c2d2 5297 return temp1;
a0c0a00f 5298}
cce855bc 5299
726f6388
JA
5300/*******************************************
5301 * *
5302 * Functions to expand WORD_DESCs *
5303 * *
5304 *******************************************/
5305
5306/* Expand WORD, performing word splitting on the result. This does
5307 parameter expansion, command substitution, arithmetic expansion,
5308 word splitting, and quote removal. */
5309
5310WORD_LIST *
5311expand_word (word, quoted)
5312 WORD_DESC *word;
5313 int quoted;
5314{
5315 WORD_LIST *result, *tresult;
5316
b72432fd 5317 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
726f6388
JA
5318 result = word_list_split (tresult);
5319 dispose_words (tresult);
ccc6cda3 5320 return (result ? dequote_list (result) : result);
726f6388
JA
5321}
5322
5323/* Expand WORD, but do not perform word splitting on the result. This
5324 does parameter expansion, command substitution, arithmetic expansion,
5325 and quote removal. */
5326WORD_LIST *
28ef6c31 5327expand_word_unsplit (word, quoted)
726f6388
JA
5328 WORD_DESC *word;
5329 int quoted;
5330{
5331 WORD_LIST *result;
5332
28ef6c31 5333 expand_no_split_dollar_star = 1;
0001803f
CR
5334#if defined (HANDLE_MULTIBYTE)
5335 if (ifs_firstc[0] == 0)
5336#else
5337 if (ifs_firstc == 0)
5338#endif
5339 word->flags |= W_NOSPLIT;
1cc06898 5340 word->flags |= W_NOSPLIT2;
b72432fd 5341 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31 5342 expand_no_split_dollar_star = 0;
7117c2d2 5343
ccc6cda3 5344 return (result ? dequote_list (result) : result);
726f6388
JA
5345}
5346
5347/* Perform shell expansions on WORD, but do not perform word splitting or
0001803f
CR
5348 quote removal on the result. Virtually identical to expand_word_unsplit;
5349 could be combined if implementations don't diverge. */
726f6388
JA
5350WORD_LIST *
5351expand_word_leave_quoted (word, quoted)
5352 WORD_DESC *word;
5353 int quoted;
5354{
0001803f
CR
5355 WORD_LIST *result;
5356
5357 expand_no_split_dollar_star = 1;
5358#if defined (HANDLE_MULTIBYTE)
5359 if (ifs_firstc[0] == 0)
5360#else
5361 if (ifs_firstc == 0)
5362#endif
5363 word->flags |= W_NOSPLIT;
5364 word->flags |= W_NOSPLIT2;
5365 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
5366 expand_no_split_dollar_star = 0;
5367
5368 return result;
726f6388
JA
5369}
5370
726f6388
JA
5371#if defined (PROCESS_SUBSTITUTION)
5372
cce855bc
JA
5373/*****************************************************************/
5374/* */
5375/* Hacking Process Substitution */
5376/* */
5377/*****************************************************************/
726f6388 5378
726f6388
JA
5379#if !defined (HAVE_DEV_FD)
5380/* Named pipes must be removed explicitly with `unlink'. This keeps a list
5381 of FIFOs the shell has open. unlink_fifo_list will walk the list and
5382 unlink all of them. add_fifo_list adds the name of an open FIFO to the
5383 list. NFIFO is a count of the number of FIFOs in the list. */
5384#define FIFO_INCR 20
5385
f73dda09
JA
5386struct temp_fifo {
5387 char *file;
5388 pid_t proc;
5389};
5390
5391static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
ccc6cda3
JA
5392static int nfifo;
5393static int fifo_list_size;
726f6388 5394
a0c0a00f
CR
5395void
5396clear_fifo_list ()
5397{
5398}
5399
495aee44
CR
5400char *
5401copy_fifo_list (sizep)
5402 int *sizep;
5403{
5404 if (sizep)
5405 *sizep = 0;
5406 return (char *)NULL;
5407}
5408
726f6388
JA
5409static void
5410add_fifo_list (pathname)
5411 char *pathname;
5412{
5413 if (nfifo >= fifo_list_size - 1)
5414 {
5415 fifo_list_size += FIFO_INCR;
f73dda09
JA
5416 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
5417 fifo_list_size * sizeof (struct temp_fifo));
726f6388
JA
5418 }
5419
f73dda09
JA
5420 fifo_list[nfifo].file = savestring (pathname);
5421 nfifo++;
726f6388
JA
5422}
5423
495aee44
CR
5424void
5425unlink_fifo (i)
5426 int i;
5427{
5428 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
5429 {
5430 unlink (fifo_list[i].file);
5431 free (fifo_list[i].file);
5432 fifo_list[i].file = (char *)NULL;
5433 fifo_list[i].proc = -1;
5434 }
5435}
5436
726f6388
JA
5437void
5438unlink_fifo_list ()
5439{
f73dda09
JA
5440 int saved, i, j;
5441
ccc6cda3 5442 if (nfifo == 0)
726f6388
JA
5443 return;
5444
f73dda09 5445 for (i = saved = 0; i < nfifo; i++)
726f6388 5446 {
f73dda09
JA
5447 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
5448 {
7117c2d2
JA
5449 unlink (fifo_list[i].file);
5450 free (fifo_list[i].file);
5451 fifo_list[i].file = (char *)NULL;
5452 fifo_list[i].proc = -1;
f73dda09
JA
5453 }
5454 else
7117c2d2 5455 saved++;
f73dda09
JA
5456 }
5457
5458 /* If we didn't remove some of the FIFOs, compact the list. */
5459 if (saved)
5460 {
5461 for (i = j = 0; i < nfifo; i++)
5462 if (fifo_list[i].file)
5463 {
5464 fifo_list[j].file = fifo_list[i].file;
5465 fifo_list[j].proc = fifo_list[i].proc;
5466 j++;
5467 }
5468 nfifo = j;
726f6388 5469 }
f73dda09
JA
5470 else
5471 nfifo = 0;
726f6388
JA
5472}
5473
495aee44
CR
5474/* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
5475 from some point in the past, and close all open FIFOs in fifo_list
5476 that are not marked as active in LIST. If LIST is NULL, close
5477 everything in fifo_list. LSIZE is the number of elements in LIST, in
5478 case it's larger than fifo_list_size (size of fifo_list). */
5479void
5480close_new_fifos (list, lsize)
5481 char *list;
5482 int lsize;
5483{
5484 int i;
5485
5486 if (list == 0)
5487 {
5488 unlink_fifo_list ();
5489 return;
5490 }
5491
5492 for (i = 0; i < lsize; i++)
5493 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
5494 unlink_fifo (i);
5495
5496 for (i = lsize; i < fifo_list_size; i++)
5497 unlink_fifo (i);
5498}
5499
f1be666c
JA
5500int
5501fifos_pending ()
5502{
5503 return nfifo;
5504}
5505
495aee44
CR
5506int
5507num_fifos ()
5508{
5509 return nfifo;
5510}
5511
726f6388
JA
5512static char *
5513make_named_pipe ()
5514{
5515 char *tname;
5516
3185942a 5517 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
726f6388
JA
5518 if (mkfifo (tname, 0600) < 0)
5519 {
5520 free (tname);
5521 return ((char *)NULL);
5522 }
5523
5524 add_fifo_list (tname);
5525 return (tname);
5526}
5527
726f6388
JA
5528#else /* HAVE_DEV_FD */
5529
5530/* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
5531 has open to children. NFDS is a count of the number of bits currently
5532 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
5533 of open files. */
5534static char *dev_fd_list = (char *)NULL;
ccc6cda3 5535static int nfds;
726f6388
JA
5536static int totfds; /* The highest possible number of open files. */
5537
a0c0a00f
CR
5538void
5539clear_fifo (i)
5540 int i;
5541{
5542 if (dev_fd_list[i])
5543 {
5544 dev_fd_list[i] = 0;
5545 nfds--;
5546 }
5547}
5548
5549void
5550clear_fifo_list ()
5551{
5552 register int i;
5553
5554 if (nfds == 0)
5555 return;
5556
5557 for (i = 0; nfds && i < totfds; i++)
5558 clear_fifo (i);
5559
5560 nfds = 0;
5561}
5562
495aee44
CR
5563char *
5564copy_fifo_list (sizep)
5565 int *sizep;
5566{
5567 char *ret;
5568
5569 if (nfds == 0 || totfds == 0)
5570 {
5571 if (sizep)
5572 *sizep = 0;
5573 return (char *)NULL;
5574 }
5575
5576 if (sizep)
5577 *sizep = totfds;
5578 ret = (char *)xmalloc (totfds);
5579 return (memcpy (ret, dev_fd_list, totfds));
5580}
5581
726f6388
JA
5582static void
5583add_fifo_list (fd)
5584 int fd;
5585{
495aee44 5586 if (dev_fd_list == 0 || fd >= totfds)
726f6388
JA
5587 {
5588 int ofds;
5589
5590 ofds = totfds;
5591 totfds = getdtablesize ();
5592 if (totfds < 0 || totfds > 256)
5593 totfds = 256;
3185942a 5594 if (fd >= totfds)
726f6388
JA
5595 totfds = fd + 2;
5596
f73dda09 5597 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
7117c2d2 5598 memset (dev_fd_list + ofds, '\0', totfds - ofds);
726f6388
JA
5599 }
5600
5601 dev_fd_list[fd] = 1;
5602 nfds++;
5603}
5604
f1be666c
JA
5605int
5606fifos_pending ()
5607{
5608 return 0; /* used for cleanup; not needed with /dev/fd */
5609}
5610
495aee44
CR
5611int
5612num_fifos ()
5613{
5614 return nfds;
5615}
5616
5617void
5618unlink_fifo (fd)
5619 int fd;
5620{
5621 if (dev_fd_list[fd])
5622 {
5623 close (fd);
5624 dev_fd_list[fd] = 0;
5625 nfds--;
5626 }
5627}
5628
726f6388
JA
5629void
5630unlink_fifo_list ()
5631{
5632 register int i;
5633
ccc6cda3 5634 if (nfds == 0)
726f6388
JA
5635 return;
5636
5637 for (i = 0; nfds && i < totfds; i++)
495aee44 5638 unlink_fifo (i);
726f6388
JA
5639
5640 nfds = 0;
5641}
5642
495aee44
CR
5643/* Take LIST, which is a snapshot copy of dev_fd_list from some point in
5644 the past, and close all open fds in dev_fd_list that are not marked
5645 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
5646 LSIZE is the number of elements in LIST, in case it's larger than
5647 totfds (size of dev_fd_list). */
5648void
5649close_new_fifos (list, lsize)
5650 char *list;
5651 int lsize;
5652{
5653 int i;
5654
5655 if (list == 0)
5656 {
5657 unlink_fifo_list ();
5658 return;
5659 }
5660
5661 for (i = 0; i < lsize; i++)
5662 if (list[i] == 0 && i < totfds && dev_fd_list[i])
5663 unlink_fifo (i);
5664
5665 for (i = lsize; i < totfds; i++)
5666 unlink_fifo (i);
5667}
5668
726f6388
JA
5669#if defined (NOTDEF)
5670print_dev_fd_list ()
5671{
5672 register int i;
5673
f73dda09 5674 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
726f6388
JA
5675 fflush (stderr);
5676
5677 for (i = 0; i < totfds; i++)
5678 {
5679 if (dev_fd_list[i])
5680 fprintf (stderr, " %d", i);
5681 }
5682 fprintf (stderr, "\n");
5683}
5684#endif /* NOTDEF */
5685
5686static char *
5687make_dev_fd_filename (fd)
5688 int fd;
5689{
f73dda09 5690 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
726f6388 5691
17345e5a 5692 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
bb70624e
JA
5693
5694 strcpy (ret, DEV_FD_PREFIX);
5695 p = inttostr (fd, intbuf, sizeof (intbuf));
5696 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5697
726f6388
JA
5698 add_fifo_list (fd);
5699 return (ret);
5700}
5701
5702#endif /* HAVE_DEV_FD */
5703
5704/* Return a filename that will open a connection to the process defined by
5705 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5706 a filename in /dev/fd corresponding to a descriptor that is one of the
5707 ends of the pipe. If not defined, we use named pipes on systems that have
5708 them. Systems without /dev/fd and named pipes are out of luck.
5709
5710 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5711 use the read end of the pipe and dup that file descriptor to fd 0 in
5712 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5713 writing or use the write end of the pipe in the child, and dup that
5714 file descriptor to fd 1 in the child. The parent does the opposite. */
5715
5716static char *
5717process_substitute (string, open_for_read_in_child)
5718 char *string;
5719 int open_for_read_in_child;
5720{
5721 char *pathname;
5722 int fd, result;
5723 pid_t old_pid, pid;
5724#if defined (HAVE_DEV_FD)
5725 int parent_pipe_fd, child_pipe_fd;
5726 int fildes[2];
5727#endif /* HAVE_DEV_FD */
5728#if defined (JOB_CONTROL)
5729 pid_t old_pipeline_pgrp;
ccc6cda3 5730#endif
726f6388 5731
cce855bc 5732 if (!string || !*string || wordexp_only)
726f6388
JA
5733 return ((char *)NULL);
5734
5735#if !defined (HAVE_DEV_FD)
5736 pathname = make_named_pipe ();
5737#else /* HAVE_DEV_FD */
5738 if (pipe (fildes) < 0)
5739 {
a0c0a00f 5740 sys_error ("%s", _("cannot make pipe for process substitution"));
726f6388
JA
5741 return ((char *)NULL);
5742 }
5743 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5744 the pipe in the parent, otherwise the read end. */
5745 parent_pipe_fd = fildes[open_for_read_in_child];
5746 child_pipe_fd = fildes[1 - open_for_read_in_child];
d166f048
JA
5747 /* Move the parent end of the pipe to some high file descriptor, to
5748 avoid clashes with FDs used by the script. */
5749 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5750
726f6388
JA
5751 pathname = make_dev_fd_filename (parent_pipe_fd);
5752#endif /* HAVE_DEV_FD */
5753
3185942a 5754 if (pathname == 0)
726f6388 5755 {
a0c0a00f 5756 sys_error ("%s", _("cannot make pipe for process substitution"));
726f6388
JA
5757 return ((char *)NULL);
5758 }
5759
5760 old_pid = last_made_pid;
5761
5762#if defined (JOB_CONTROL)
5763 old_pipeline_pgrp = pipeline_pgrp;
a0c0a00f
CR
5764 if (pipeline_pgrp == 0 || (subshell_environment & (SUBSHELL_PIPE|SUBSHELL_FORK|SUBSHELL_ASYNC)) == 0)
5765 pipeline_pgrp = shell_pgrp;
ccc6cda3 5766 save_pipeline (1);
ccc6cda3
JA
5767#endif /* JOB_CONTROL */
5768
726f6388
JA
5769 pid = make_child ((char *)NULL, 1);
5770 if (pid == 0)
5771 {
ccc6cda3 5772 reset_terminating_signals (); /* XXX */
b80f6443 5773 free_pushed_string_input ();
726f6388 5774 /* Cancel traps, in trap.c. */
495aee44 5775 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
a0c0a00f 5776 QUIT; /* catch any interrupts we got post-fork */
726f6388 5777 setup_async_signals ();
3185942a 5778 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
a0c0a00f
CR
5779
5780 /* if we're expanding a redirection, we shouldn't have access to the
5781 temporary environment, but commands in the subshell should have
5782 access to their own temporary environment. */
5783 if (expanding_redir)
5784 flush_temporary_env ();
726f6388 5785 }
ccc6cda3
JA
5786
5787#if defined (JOB_CONTROL)
726f6388
JA
5788 set_sigchld_handler ();
5789 stop_making_children ();
3185942a 5790 /* XXX - should we only do this in the parent? (as in command subst) */
726f6388 5791 pipeline_pgrp = old_pipeline_pgrp;
a0c0a00f
CR
5792#else
5793 stop_making_children ();
ccc6cda3 5794#endif /* JOB_CONTROL */
726f6388
JA
5795
5796 if (pid < 0)
5797 {
a0c0a00f 5798 sys_error ("%s", _("cannot make child for process substitution"));
726f6388
JA
5799 free (pathname);
5800#if defined (HAVE_DEV_FD)
5801 close (parent_pipe_fd);
5802 close (child_pipe_fd);
5803#endif /* HAVE_DEV_FD */
5804 return ((char *)NULL);
5805 }
5806
5807 if (pid > 0)
5808 {
ccc6cda3 5809#if defined (JOB_CONTROL)
a0c0a00f
CR
5810 if (last_procsub_child)
5811 {
5812 discard_pipeline (last_procsub_child);
5813 last_procsub_child = (PROCESS *)NULL;
5814 }
5815 last_procsub_child = restore_pipeline (0);
ccc6cda3
JA
5816#endif
5817
f73dda09
JA
5818#if !defined (HAVE_DEV_FD)
5819 fifo_list[nfifo-1].proc = pid;
5820#endif
5821
726f6388
JA
5822 last_made_pid = old_pid;
5823
5824#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5825 close_pgrp_pipe ();
5826#endif /* JOB_CONTROL && PGRP_PIPE */
5827
5828#if defined (HAVE_DEV_FD)
5829 close (child_pipe_fd);
5830#endif /* HAVE_DEV_FD */
5831
5832 return (pathname);
5833 }
5834
5835 set_sigint_handler ();
5836
5837#if defined (JOB_CONTROL)
5838 set_job_control (0);
5839#endif /* JOB_CONTROL */
5840
5841#if !defined (HAVE_DEV_FD)
5842 /* Open the named pipe in the child. */
ac50fbac 5843 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
726f6388
JA
5844 if (fd < 0)
5845 {
b80f6443
JA
5846 /* Two separate strings for ease of translation. */
5847 if (open_for_read_in_child)
5848 sys_error (_("cannot open named pipe %s for reading"), pathname);
5849 else
5850 sys_error (_("cannot open named pipe %s for writing"), pathname);
5851
726f6388
JA
5852 exit (127);
5853 }
bb70624e
JA
5854 if (open_for_read_in_child)
5855 {
28ef6c31 5856 if (sh_unset_nodelay_mode (fd) < 0)
bb70624e 5857 {
3185942a 5858 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
bb70624e
JA
5859 exit (127);
5860 }
5861 }
726f6388
JA
5862#else /* HAVE_DEV_FD */
5863 fd = child_pipe_fd;
5864#endif /* HAVE_DEV_FD */
5865
a0c0a00f
CR
5866 /* Discard buffered stdio output before replacing the underlying file
5867 descriptor. */
5868 if (open_for_read_in_child == 0)
5869 fpurge (stdout);
5870
726f6388
JA
5871 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5872 {
b80f6443 5873 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
ccc6cda3 5874 open_for_read_in_child ? 0 : 1);
726f6388
JA
5875 exit (127);
5876 }
5877
f73dda09
JA
5878 if (fd != (open_for_read_in_child ? 0 : 1))
5879 close (fd);
726f6388
JA
5880
5881 /* Need to close any files that this process has open to pipes inherited
5882 from its parent. */
5883 if (current_fds_to_close)
5884 {
5885 close_fd_bitmap (current_fds_to_close);
5886 current_fds_to_close = (struct fd_bitmap *)NULL;
5887 }
5888
5889#if defined (HAVE_DEV_FD)
5890 /* Make sure we close the parent's end of the pipe and clear the slot
5891 in the fd list so it is not closed later, if reallocated by, for
5892 instance, pipe(2). */
5893 close (parent_pipe_fd);
5894 dev_fd_list[parent_pipe_fd] = 0;
5895#endif /* HAVE_DEV_FD */
5896
8dea6e87 5897 /* subshells shouldn't have this flag, which controls using the temporary
a0c0a00f
CR
5898 environment for variable lookups. We have already flushed the temporary
5899 environment above in the case we're expanding a redirection, so processes
5900 executed by this command need to be able to set it independently of their
5901 parent. */
8dea6e87
CR
5902 expanding_redir = 0;
5903
a0c0a00f 5904 subshell_level++;
d166f048 5905 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
a0c0a00f 5906 subshell_level--;
726f6388
JA
5907
5908#if !defined (HAVE_DEV_FD)
5909 /* Make sure we close the named pipe in the child before we exit. */
5910 close (open_for_read_in_child ? 0 : 1);
5911#endif /* !HAVE_DEV_FD */
5912
ac50fbac
CR
5913 last_command_exit_value = result;
5914 result = run_exit_trap ();
726f6388
JA
5915 exit (result);
5916 /*NOTREACHED*/
5917}
5918#endif /* PROCESS_SUBSTITUTION */
5919
cce855bc
JA
5920/***********************************/
5921/* */
5922/* Command Substitution */
5923/* */
5924/***********************************/
5925
d166f048 5926static char *
3185942a 5927read_comsub (fd, quoted, rflag)
d166f048 5928 int fd, quoted;
3185942a 5929 int *rflag;
d166f048 5930{
3185942a
JA
5931 char *istring, buf[128], *bufp, *s;
5932 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
f73dda09 5933 ssize_t bufn;
280bd77d 5934 int nullbyte;
d166f048
JA
5935
5936 istring = (char *)NULL;
3185942a
JA
5937 istring_index = istring_size = bufn = tflag = 0;
5938
5939 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5940 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
d166f048 5941
280bd77d
CR
5942 nullbyte = 0;
5943
3185942a
JA
5944 /* Read the output of the command through the pipe. This may need to be
5945 changed to understand multibyte characters in the future. */
d166f048
JA
5946 while (1)
5947 {
5948 if (fd < 0)
28ef6c31 5949 break;
d166f048
JA
5950 if (--bufn <= 0)
5951 {
bb70624e 5952 bufn = zread (fd, buf, sizeof (buf));
d166f048
JA
5953 if (bufn <= 0)
5954 break;
5955 bufp = buf;
5956 }
5957 c = *bufp++;
5958
28ef6c31
JA
5959 if (c == 0)
5960 {
a0c0a00f 5961#if 1
280bd77d
CR
5962 if (nullbyte == 0)
5963 {
5964 internal_warning ("%s", _("command substitution: ignored null byte in input"));
5965 nullbyte = 1;
5966 }
28ef6c31
JA
5967#endif
5968 continue;
5969 }
5970
d166f048
JA
5971 /* Add the character to ISTRING, possibly after resizing it. */
5972 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5973
f1be666c
JA
5974 /* This is essentially quote_string inline */
5975 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5976 istring[istring_index++] = CTLESC;
5977 /* Escape CTLESC and CTLNUL in the output to protect those characters
5978 from the rest of the word expansions (word splitting and globbing.)
5979 This is essentially quote_escapes inline. */
3185942a
JA
5980 else if (skip_ctlesc == 0 && c == CTLESC)
5981 {
5982 tflag |= W_HASCTLESC;
5983 istring[istring_index++] = CTLESC;
5984 }
5985 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
d166f048
JA
5986 istring[istring_index++] = CTLESC;
5987
5988 istring[istring_index++] = c;
28ef6c31
JA
5989
5990#if 0
5991#if defined (__CYGWIN__)
5992 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5993 {
5994 istring_index--;
5995 istring[istring_index - 1] = '\n';
5996 }
5997#endif
5998#endif
d166f048
JA
5999 }
6000
6001 if (istring)
6002 istring[istring_index] = '\0';
6003
6004 /* If we read no output, just return now and save ourselves some
6005 trouble. */
6006 if (istring_index == 0)
6007 {
6008 FREE (istring);
3185942a
JA
6009 if (rflag)
6010 *rflag = tflag;
d166f048
JA
6011 return (char *)NULL;
6012 }
6013
6014 /* Strip trailing newlines from the output of the command. */
6015 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6016 {
6017 while (istring_index > 0)
6018 {
6019 if (istring[istring_index - 1] == '\n')
6020 {
6021 --istring_index;
6022
6023 /* If the newline was quoted, remove the quoting char. */
6024 if (istring[istring_index - 1] == CTLESC)
6025 --istring_index;
6026 }
6027 else
6028 break;
6029 }
6030 istring[istring_index] = '\0';
6031 }
6032 else
6033 strip_trailing (istring, istring_index - 1, 1);
6034
3185942a
JA
6035 if (rflag)
6036 *rflag = tflag;
d166f048
JA
6037 return istring;
6038}
6039
3185942a
JA
6040/* Perform command substitution on STRING. This returns a WORD_DESC * with the
6041 contained string possibly quoted. */
6042WORD_DESC *
726f6388
JA
6043command_substitute (string, quoted)
6044 char *string;
6045 int quoted;
6046{
95732b49 6047 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
a0c0a00f 6048 char *istring, *s;
3185942a
JA
6049 int result, fildes[2], function_value, pflags, rc, tflag;
6050 WORD_DESC *ret;
726f6388 6051
ccc6cda3 6052 istring = (char *)NULL;
726f6388
JA
6053
6054 /* Don't fork () if there is no need to. In the case of no command to
6055 run, just return NULL. */
a0c0a00f
CR
6056#if 1
6057 for (s = string; s && *s && (shellblank (*s) || *s == '\n'); s++)
6058 ;
6059 if (s == 0 || *s == 0)
6060 return ((WORD_DESC *)NULL);
6061#else
726f6388 6062 if (!string || !*string || (string[0] == '\n' && !string[1]))
3185942a 6063 return ((WORD_DESC *)NULL);
a0c0a00f 6064#endif
726f6388 6065
cce855bc
JA
6066 if (wordexp_only && read_but_dont_execute)
6067 {
0001803f 6068 last_command_exit_value = EX_WEXPCOMSUB;
cce855bc
JA
6069 jump_to_top_level (EXITPROG);
6070 }
6071
bb70624e
JA
6072 /* We're making the assumption here that the command substitution will
6073 eventually run a command from the file system. Since we'll run
6074 maybe_make_export_env in this subshell before executing that command,
6075 the parent shell and any other shells it starts will have to remake
6076 the environment. If we make it before we fork, other shells won't
6077 have to. Don't bother if we have any temporary variable assignments,
6078 though, because the export environment will be remade after this
6079 command completes anyway, but do it if all the words to be expanded
6080 are variable assignments. */
6081 if (subst_assign_varlist == 0 || garglist == 0)
6082 maybe_make_export_env (); /* XXX */
6083
b80f6443 6084 /* Flags to pass to parse_and_execute() */
0001803f 6085 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
b80f6443 6086
726f6388
JA
6087 /* Pipe the output of executing STRING into the current shell. */
6088 if (pipe (fildes) < 0)
6089 {
a0c0a00f 6090 sys_error ("%s", _("cannot make pipe for command substitution"));
726f6388
JA
6091 goto error_exit;
6092 }
6093
6094 old_pid = last_made_pid;
6095#if defined (JOB_CONTROL)
ccc6cda3 6096 old_pipeline_pgrp = pipeline_pgrp;
28ef6c31
JA
6097 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
6098 if ((subshell_environment & SUBSHELL_PIPE) == 0)
6099 pipeline_pgrp = shell_pgrp;
ccc6cda3 6100 cleanup_the_pipeline ();
95732b49 6101#endif /* JOB_CONTROL */
726f6388 6102
95732b49 6103 old_async_pid = last_asynchronous_pid;
95732b49 6104 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
95732b49
JA
6105 last_asynchronous_pid = old_async_pid;
6106
726f6388 6107 if (pid == 0)
495aee44
CR
6108 {
6109 /* Reset the signal handlers in the child, but don't free the
6110 trap strings. Set a flag noting that we have to free the
6111 trap strings if we run trap to change a signal disposition. */
6112 reset_signal_handlers ();
a0c0a00f
CR
6113 if (ISINTERRUPT)
6114 {
6115 kill (getpid (), SIGINT);
6116 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6117 }
6118 QUIT; /* catch any interrupts we got post-fork */
495aee44
CR
6119 subshell_environment |= SUBSHELL_RESETTRAP;
6120 }
ccc6cda3
JA
6121
6122#if defined (JOB_CONTROL)
3185942a 6123 /* XXX DO THIS ONLY IN PARENT ? XXX */
ccc6cda3
JA
6124 set_sigchld_handler ();
6125 stop_making_children ();
f1be666c
JA
6126 if (pid != 0)
6127 pipeline_pgrp = old_pipeline_pgrp;
f73dda09
JA
6128#else
6129 stop_making_children ();
ccc6cda3 6130#endif /* JOB_CONTROL */
726f6388
JA
6131
6132 if (pid < 0)
6133 {
b80f6443 6134 sys_error (_("cannot make child for command substitution"));
726f6388
JA
6135 error_exit:
6136
ac50fbac
CR
6137 last_made_pid = old_pid;
6138
726f6388
JA
6139 FREE (istring);
6140 close (fildes[0]);
6141 close (fildes[1]);
3185942a 6142 return ((WORD_DESC *)NULL);
726f6388
JA
6143 }
6144
6145 if (pid == 0)
6146 {
a0c0a00f
CR
6147 /* The currently executing shell is not interactive. */
6148 interactive = 0;
6149
726f6388 6150 set_sigint_handler (); /* XXX */
28ef6c31 6151
b80f6443
JA
6152 free_pushed_string_input ();
6153
a0c0a00f
CR
6154 /* Discard buffered stdio output before replacing the underlying file
6155 descriptor. */
6156 fpurge (stdout);
6157
726f6388
JA
6158 if (dup2 (fildes[1], 1) < 0)
6159 {
a0c0a00f 6160 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
726f6388
JA
6161 exit (EXECUTION_FAILURE);
6162 }
6163
6164 /* If standard output is closed in the parent shell
6165 (such as after `exec >&-'), file descriptor 1 will be
6166 the lowest available file descriptor, and end up in
6167 fildes[0]. This can happen for stdin and stderr as well,
6168 but stdout is more important -- it will cause no output
6169 to be generated from this command. */
6170 if ((fildes[1] != fileno (stdin)) &&
6171 (fildes[1] != fileno (stdout)) &&
6172 (fildes[1] != fileno (stderr)))
6173 close (fildes[1]);
6174
6175 if ((fildes[0] != fileno (stdin)) &&
6176 (fildes[0] != fileno (stdout)) &&
6177 (fildes[0] != fileno (stderr)))
6178 close (fildes[0]);
6179
495aee44
CR
6180#ifdef __CYGWIN__
6181 /* Let stdio know the fd may have changed from text to binary mode, and
6182 make sure to preserve stdout line buffering. */
6183 freopen (NULL, "w", stdout);
6184 sh_setlinebuf (stdout);
6185#endif /* __CYGWIN__ */
6186
ccc6cda3 6187 /* This is a subshell environment. */
28ef6c31 6188 subshell_environment |= SUBSHELL_COMSUB;
ccc6cda3 6189
a0c0a00f
CR
6190 /* Many shells do not appear to inherit the -v option for command
6191 substitutions. */
6192 change_flag ('v', FLAG_OFF);
6193
6194 /* When inherit_errexit option is not enabled, command substitution does
6195 not inherit the -e flag. It is enabled when Posix mode is enabled */
6196 if (inherit_errexit == 0)
ac50fbac
CR
6197 {
6198 builtin_ignoring_errexit = 0;
6199 change_flag ('e', FLAG_OFF);
ac50fbac 6200 }
a0c0a00f
CR
6201 set_shellopts ();
6202
6203 /* If we are expanding a redirection, we can dispose of any temporary
6204 environment we received, since redirections are not supposed to have
6205 access to the temporary environment. We will have to see whether this
6206 affects temporary environments supplied to `eval', but the temporary
6207 environment gets copied to builtin_env at some point. */
6208 if (expanding_redir)
6209 {
6210 flush_temporary_env ();
6211 expanding_redir = 0;
6212 }
726f6388
JA
6213
6214 remove_quoted_escapes (string);
6215
ccc6cda3 6216 startup_state = 2; /* see if we can avoid a fork */
726f6388
JA
6217 /* Give command substitution a place to jump back to on failure,
6218 so we don't go back up to main (). */
ac50fbac 6219 result = setjmp_nosigs (top_level);
726f6388 6220
bb70624e
JA
6221 /* If we're running a command substitution inside a shell function,
6222 trap `return' so we don't return from the function in the subshell
6223 and go off to never-never land. */
6224 if (result == 0 && return_catch_flag)
ac50fbac 6225 function_value = setjmp_nosigs (return_catch);
bb70624e
JA
6226 else
6227 function_value = 0;
6228
b80f6443
JA
6229 if (result == ERREXIT)
6230 rc = last_command_exit_value;
6231 else if (result == EXITPROG)
6232 rc = last_command_exit_value;
726f6388 6233 else if (result)
b80f6443 6234 rc = EXECUTION_FAILURE;
bb70624e 6235 else if (function_value)
b80f6443 6236 rc = return_catch_value;
726f6388 6237 else
b80f6443
JA
6238 {
6239 subshell_level++;
6240 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
6241 subshell_level--;
6242 }
6243
6244 last_command_exit_value = rc;
6245 rc = run_exit_trap ();
f1be666c
JA
6246#if defined (PROCESS_SUBSTITUTION)
6247 unlink_fifo_list ();
6248#endif
b80f6443 6249 exit (rc);
726f6388
JA
6250 }
6251 else
6252 {
726f6388
JA
6253#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6254 close_pgrp_pipe ();
6255#endif /* JOB_CONTROL && PGRP_PIPE */
6256
6257 close (fildes[1]);
6258
3185942a
JA
6259 tflag = 0;
6260 istring = read_comsub (fildes[0], quoted, &tflag);
ccc6cda3 6261
726f6388
JA
6262 close (fildes[0]);
6263
b72432fd 6264 current_command_subst_pid = pid;
726f6388
JA
6265 last_command_exit_value = wait_for (pid);
6266 last_command_subst_pid = pid;
6267 last_made_pid = old_pid;
6268
6269#if defined (JOB_CONTROL)
6270 /* If last_command_exit_value > 128, then the substituted command
6271 was terminated by a signal. If that signal was SIGINT, then send
6272 SIGINT to ourselves. This will break out of loops, for instance. */
b80f6443 6273 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
726f6388
JA
6274 kill (getpid (), SIGINT);
6275
6276 /* wait_for gives the terminal back to shell_pgrp. If some other
cce855bc
JA
6277 process group should have it, give it away to that group here.
6278 pipeline_pgrp is non-zero only while we are constructing a
ac50fbac 6279 pipeline, so what we are concerned about is whether or not that
cce855bc 6280 pipeline was started in the background. A pipeline started in
a0c0a00f
CR
6281 the background should never get the tty back here. We duplicate
6282 the conditions that wait_for tests to make sure we only give
6283 the terminal back to pipeline_pgrp under the conditions that wait_for
6284 gave it to shell_pgrp. If wait_for doesn't mess with the terminal
6285 pgrp, we should not either. */
6286 if (interactive && pipeline_pgrp != (pid_t)0 && running_in_background == 0 &&
6287 (subshell_environment & (SUBSHELL_ASYNC|SUBSHELL_PIPE)) == 0)
28ef6c31 6288 give_terminal_to (pipeline_pgrp, 0);
726f6388
JA
6289#endif /* JOB_CONTROL */
6290
3185942a
JA
6291 ret = alloc_word_desc ();
6292 ret->word = istring;
6293 ret->flags = tflag;
6294
6295 return ret;
726f6388
JA
6296 }
6297}
6298
6299/********************************************************
6300 * *
6301 * Utility functions for parameter expansion *
6302 * *
6303 ********************************************************/
6304
ccc6cda3 6305#if defined (ARRAY_VARS)
ccc6cda3 6306
f73dda09 6307static arrayind_t
ccc6cda3
JA
6308array_length_reference (s)
6309 char *s;
6310{
f73dda09
JA
6311 int len;
6312 arrayind_t ind;
3185942a 6313 char *akey;
f73dda09 6314 char *t, c;
ccc6cda3 6315 ARRAY *array;
495aee44 6316 HASH_TABLE *h;
ccc6cda3
JA
6317 SHELL_VAR *var;
6318
6319 var = array_variable_part (s, &t, &len);
726f6388 6320
ccc6cda3
JA
6321 /* If unbound variables should generate an error, report one and return
6322 failure. */
ac50fbac 6323 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
726f6388 6324 {
f73dda09 6325 c = *--t;
ccc6cda3 6326 *t = '\0';
0001803f 6327 last_command_exit_value = EXECUTION_FAILURE;
7117c2d2 6328 err_unboundvar (s);
f73dda09 6329 *t = c;
ccc6cda3 6330 return (-1);
726f6388 6331 }
ac50fbac 6332 else if (var == 0 || invisible_p (var))
ccc6cda3 6333 return 0;
726f6388 6334
28ef6c31
JA
6335 /* We support a couple of expansions for variables that are not arrays.
6336 We'll return the length of the value for v[0], and 1 for v[@] or
6337 v[*]. Return 0 for everything else. */
6338
6339 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
495aee44 6340 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
726f6388 6341
ccc6cda3 6342 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
ccc6cda3 6343 {
3185942a 6344 if (assoc_p (var))
495aee44 6345 return (h ? assoc_num_elements (h) : 0);
3185942a 6346 else if (array_p (var))
495aee44 6347 return (array ? array_num_elements (array) : 0);
3185942a 6348 else
495aee44 6349 return (var_isset (var) ? 1 : 0);
ccc6cda3 6350 }
ccc6cda3 6351
3185942a
JA
6352 if (assoc_p (var))
6353 {
6354 t[len - 1] = '\0';
6355 akey = expand_assignment_string_to_string (t, 0); /* [ */
6356 t[len - 1] = ']';
6357 if (akey == 0 || *akey == 0)
6358 {
6359 err_badarraysub (t);
ac50fbac 6360 FREE (akey);
3185942a
JA
6361 return (-1);
6362 }
6363 t = assoc_reference (assoc_cell (var), akey);
ac50fbac 6364 free (akey);
3185942a 6365 }
28ef6c31 6366 else
3185942a 6367 {
ac50fbac
CR
6368 ind = array_expand_index (var, t, len);
6369 /* negative subscripts to indexed arrays count back from end */
6370 if (var && array_p (var) && ind < 0)
6371 ind = array_max_index (array_cell (var)) + 1 + ind;
3185942a
JA
6372 if (ind < 0)
6373 {
6374 err_badarraysub (t);
6375 return (-1);
6376 }
6377 if (array_p (var))
6378 t = array_reference (array, ind);
6379 else
6380 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6381 }
28ef6c31 6382
f1be666c 6383 len = MB_STRLEN (t);
ccc6cda3 6384 return (len);
726f6388 6385}
ccc6cda3 6386#endif /* ARRAY_VARS */
726f6388
JA
6387
6388static int
6389valid_brace_expansion_word (name, var_is_special)
6390 char *name;
6391 int var_is_special;
6392{
f73dda09 6393 if (DIGIT (*name) && all_digits (name))
726f6388
JA
6394 return 1;
6395 else if (var_is_special)
6396 return 1;
ccc6cda3 6397#if defined (ARRAY_VARS)
a0c0a00f 6398 else if (valid_array_reference (name, 0))
ccc6cda3
JA
6399 return 1;
6400#endif /* ARRAY_VARS */
726f6388
JA
6401 else if (legal_identifier (name))
6402 return 1;
6403 else
6404 return 0;
6405}
ccc6cda3 6406
b80f6443
JA
6407static int
6408chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
6409 char *name;
6410 int quoted;
6411 int *quoted_dollar_atp, *contains_dollar_at;
6412{
6413 char *temp1;
6414
6415 if (name == 0)
6416 {
6417 if (quoted_dollar_atp)
6418 *quoted_dollar_atp = 0;
6419 if (contains_dollar_at)
6420 *contains_dollar_at = 0;
6421 return 0;
6422 }
6423
6424 /* check for $@ and $* */
6425 if (name[0] == '@' && name[1] == 0)
6426 {
6427 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6428 *quoted_dollar_atp = 1;
6429 if (contains_dollar_at)
6430 *contains_dollar_at = 1;
6431 return 1;
6432 }
6433 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6434 {
6435 if (contains_dollar_at)
6436 *contains_dollar_at = 1;
6437 return 1;
6438 }
6439
6440 /* Now check for ${array[@]} and ${array[*]} */
6441#if defined (ARRAY_VARS)
a0c0a00f 6442 else if (valid_array_reference (name, 0))
b80f6443 6443 {
0001803f 6444 temp1 = mbschr (name, '[');
b80f6443
JA
6445 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
6446 {
6447 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6448 *quoted_dollar_atp = 1;
6449 if (contains_dollar_at)
6450 *contains_dollar_at = 1;
6451 return 1;
6452 } /* [ */
6453 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6454 which should result in separate words even when IFS is unset. */
6455 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
6456 {
6457 if (contains_dollar_at)
6458 *contains_dollar_at = 1;
6459 return 1;
6460 }
6461 }
6462#endif
6463 return 0;
6464}
6465
726f6388
JA
6466/* Parameter expand NAME, and return a new string which is the expansion,
6467 or NULL if there was no expansion.
6468 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6469 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6470 NAME was found inside of a double-quoted expression. */
95732b49 6471static WORD_DESC *
495aee44 6472parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
726f6388 6473 char *name;
89a92869 6474 int var_is_special, quoted, pflags;
495aee44 6475 arrayind_t *indp;
726f6388 6476{
95732b49 6477 WORD_DESC *ret;
ccc6cda3 6478 char *temp, *tt;
7117c2d2 6479 intmax_t arg_index;
ccc6cda3 6480 SHELL_VAR *var;
f1be666c 6481 int atype, rflags;
495aee44 6482 arrayind_t ind;
726f6388 6483
95732b49
JA
6484 ret = 0;
6485 temp = 0;
f1be666c 6486 rflags = 0;
95732b49 6487
495aee44
CR
6488 if (indp)
6489 *indp = INTMAX_MIN;
6490
95732b49 6491 /* Handle multiple digit arguments, as in ${11}. */
f73dda09 6492 if (legal_number (name, &arg_index))
7117c2d2
JA
6493 {
6494 tt = get_dollar_var_value (arg_index);
b80f6443
JA
6495 if (tt)
6496 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6497 ? quote_string (tt)
6498 : quote_escapes (tt);
6499 else
6500 temp = (char *)NULL;
7117c2d2
JA
6501 FREE (tt);
6502 }
726f6388
JA
6503 else if (var_is_special) /* ${@} */
6504 {
cce855bc 6505 int sindex;
f73dda09 6506 tt = (char *)xmalloc (2 + strlen (name));
cce855bc 6507 tt[sindex = 0] = '$';
726f6388 6508 strcpy (tt + 1, name);
7117c2d2 6509
95732b49 6510 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
89a92869 6511 (int *)NULL, (int *)NULL, pflags);
cce855bc 6512 free (tt);
726f6388 6513 }
ccc6cda3 6514#if defined (ARRAY_VARS)
a0c0a00f 6515 else if (valid_array_reference (name, 0))
ccc6cda3 6516 {
ac50fbac 6517expand_arrayref:
ac50fbac 6518 if (pflags & PF_ASSIGNRHS)
a0c0a00f
CR
6519 {
6520 var = array_variable_part (name, &tt, (int *)0);
6521 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
6522 {
6523 /* Only treat as double quoted if array variable */
6524 if (var && (array_p (var) || assoc_p (var)))
6525 /* XXX - bash-4.4/bash-5.0 pass AV_ASSIGNRHS */
6526 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6527 else
6528 temp = array_value (name, quoted, 0, &atype, &ind);
6529 }
ac50fbac
CR
6530 else
6531 temp = array_value (name, quoted, 0, &atype, &ind);
a0c0a00f 6532 }
ac50fbac
CR
6533 else
6534 temp = array_value (name, quoted, 0, &atype, &ind);
7117c2d2 6535 if (atype == 0 && temp)
495aee44
CR
6536 {
6537 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6538 ? quote_string (temp)
6539 : quote_escapes (temp);
6540 rflags |= W_ARRAYIND;
6541 if (indp)
6542 *indp = ind;
6543 }
f1be666c
JA
6544 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6545 rflags |= W_HASQUOTEDNULL;
ccc6cda3
JA
6546 }
6547#endif
6548 else if (var = find_variable (name))
6549 {
7117c2d2 6550 if (var_isset (var) && invisible_p (var) == 0)
28ef6c31 6551 {
ccc6cda3 6552#if defined (ARRAY_VARS)
3185942a
JA
6553 if (assoc_p (var))
6554 temp = assoc_reference (assoc_cell (var), "0");
6555 else if (array_p (var))
6556 temp = array_reference (array_cell (var), 0);
6557 else
6558 temp = value_cell (var);
ccc6cda3
JA
6559#else
6560 temp = value_cell (var);
6561#endif
6562
6563 if (temp)
b80f6443
JA
6564 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6565 ? quote_string (temp)
6566 : quote_escapes (temp);
28ef6c31 6567 }
ccc6cda3
JA
6568 else
6569 temp = (char *)NULL;
6570 }
a0c0a00f 6571 else if (var = find_variable_last_nameref (name, 0))
ac50fbac
CR
6572 {
6573 temp = nameref_cell (var);
6574#if defined (ARRAY_VARS)
6575 /* Handle expanding nameref whose value is x[n] */
a0c0a00f 6576 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
6577 {
6578 name = temp;
6579 goto expand_arrayref;
6580 }
6581 else
6582#endif
6583 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6584 if (temp && *temp && legal_identifier (temp) == 0)
6585 {
6586 last_command_exit_value = EXECUTION_FAILURE;
6587 report_error (_("%s: invalid variable name for name reference"), temp);
6588 temp = &expand_param_error;
6589 }
6590 else
6591 temp = (char *)NULL;
6592 }
726f6388 6593 else
ccc6cda3 6594 temp = (char *)NULL;
726f6388 6595
95732b49
JA
6596 if (ret == 0)
6597 {
6598 ret = alloc_word_desc ();
6599 ret->word = temp;
f1be666c 6600 ret->flags |= rflags;
95732b49
JA
6601 }
6602 return ret;
726f6388
JA
6603}
6604
ac50fbac
CR
6605static char *
6606parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
ccc6cda3 6607 char *name;
ac50fbac 6608 int var_is_special, quoted, find_nameref;
ccc6cda3
JA
6609{
6610 char *temp, *t;
95732b49 6611 WORD_DESC *w;
ac50fbac
CR
6612 SHELL_VAR *v;
6613
a0c0a00f 6614 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
ac50fbac
CR
6615 nameref_p (v) && (t = nameref_cell (v)) && *t)
6616 return (savestring (t));
ccc6cda3 6617
ac50fbac
CR
6618 /* If var_is_special == 0, and name is not an array reference, this does
6619 more expansion than necessary. It should really look up the variable's
6620 value and not try to expand it. */
495aee44 6621 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
95732b49 6622 t = w->word;
b80f6443
JA
6623 /* Have to dequote here if necessary */
6624 if (t)
6625 {
6626 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6627 ? dequote_string (t)
6628 : dequote_escapes (t);
6629 free (t);
6630 t = temp;
6631 }
95732b49
JA
6632 dispose_word_desc (w);
6633
ac50fbac
CR
6634 return t;
6635}
6636
6637/* Expand an indirect reference to a variable: ${!NAME} expands to the
6638 value of the variable whose name is the value of NAME. */
6639static WORD_DESC *
6640parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
6641 char *name;
6642 int var_is_special, quoted;
6643 int *quoted_dollar_atp, *contains_dollar_at;
6644{
6645 char *temp, *t;
6646 WORD_DESC *w;
6647 SHELL_VAR *v;
6648
6649 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6650 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6651 bash performs an indirect lookup on foo[0] and expands the result;
6652 ksh93 expands bar[0]. We could do that here -- there are enough usable
6653 primitives to do that -- but do not at this point. */
a0c0a00f 6654 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
ac50fbac
CR
6655 {
6656 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6657 {
6658 w = alloc_word_desc ();
6659 w->word = savestring (t);
6660 w->flags = 0;
6661 return w;
6662 }
6663 }
6664
6665 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6666
b80f6443 6667 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
ccc6cda3 6668 if (t == 0)
95732b49
JA
6669 return (WORD_DESC *)NULL;
6670
a0c0a00f
CR
6671 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
6672 {
6673 report_error (_("%s: bad substitution"), t);
6674 free (t);
6675 w = alloc_word_desc ();
6676 w->word = &expand_param_error;
6677 w->flags = 0;
6678 return (w);
6679 }
6680
495aee44 6681 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
ccc6cda3 6682 free (t);
95732b49
JA
6683
6684 return w;
ccc6cda3
JA
6685}
6686
726f6388
JA
6687/* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
6688 depending on the value of C, the separating character. C can be one of
ccc6cda3
JA
6689 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
6690 between double quotes. */
95732b49 6691static WORD_DESC *
a0c0a00f 6692parameter_brace_expand_rhs (name, value, c, quoted, pflags, qdollaratp, hasdollarat)
726f6388 6693 char *name, *value;
a0c0a00f 6694 int c, quoted, pflags, *qdollaratp, *hasdollarat;
726f6388 6695{
95732b49 6696 WORD_DESC *w;
726f6388 6697 WORD_LIST *l;
a0c0a00f
CR
6698 char *t, *t1, *temp, *vname;
6699 int l_hasdollat, sindex;
726f6388 6700
a0c0a00f 6701/*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
ccc6cda3
JA
6702 /* If the entire expression is between double quotes, we want to treat
6703 the value as a double-quoted string, with the exception that we strip
3185942a 6704 embedded unescaped double quotes (for sh backwards compatibility). */
95732b49 6705 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
726f6388 6706 {
a0c0a00f
CR
6707 sindex = 0;
6708 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
726f6388 6709 }
95732b49
JA
6710 else
6711 temp = value;
ccc6cda3 6712
95732b49 6713 w = alloc_word_desc ();
a0c0a00f 6714 l_hasdollat = 0;
ccc6cda3 6715 /* XXX was 0 not quoted */
a0c0a00f 6716 l = *temp ? expand_string_for_rhs (temp, quoted, &l_hasdollat, (int *)NULL)
ccc6cda3
JA
6717 : (WORD_LIST *)0;
6718 if (hasdollarat)
a0c0a00f 6719 *hasdollarat = l_hasdollat || (l && l->next);
95732b49
JA
6720 if (temp != value)
6721 free (temp);
726f6388
JA
6722 if (l)
6723 {
a0c0a00f
CR
6724 /* If l->next is not null, we know that TEMP contained "$@", since that
6725 is the only expansion that creates more than one word. */
6726 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
6727 {
6728/*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
6729 *qdollaratp = 1;
6730 }
6731
ccc6cda3 6732 /* The expansion of TEMP returned something. We need to treat things
a0c0a00f
CR
6733 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
6734 the individual words have already been quoted. We need to turn them
b80f6443
JA
6735 into a string with the words separated by the first character of
6736 $IFS without any additional quoting, so string_list_dollar_at won't
a0c0a00f
CR
6737 do the right thing. If IFS is null, we want "$@" to split into
6738 separate arguments, not be concatenated, so we use string_list_internal
6739 and mark the word to be split on spaces later. We use
6740 string_list_dollar_star for "$@" otherwise. */
6741 if (l->next && ifs_is_null)
6742 {
6743 temp = string_list_internal (l, " ");
6744 w->flags |= W_SPLITSPACE;
6745 }
6746 else
6747 temp = (l_hasdollat || l->next) ? string_list_dollar_star (l) : string_list (l);
b80f6443 6748
cd110fdf
CR
6749 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
6750 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
6751 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
6752 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6753 (which is more paranoia than anything else), we need to return the
6754 quoted null string and set the flags to indicate it. */
ac50fbac 6755 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
cd110fdf
CR
6756 {
6757 w->flags |= W_HASQUOTEDNULL;
a0c0a00f
CR
6758/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
6759 /* If we return a quoted null with L_HASDOLLARAT, we either have a
6760 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
6761 parameters or a quoted expansion of "$@" with $1 == ''. In either
6762 case, we don't want to enable special handling of $@. */
6763 if (qdollaratp && l_hasdollat)
6764 *qdollaratp = 0;
cd110fdf 6765 }
726f6388
JA
6766 dispose_words (l);
6767 }
a0c0a00f 6768 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
726f6388 6769 {
a0c0a00f
CR
6770 /* Posix interp 221 changed the rules on this. The idea is that
6771 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
6772 when foo and xxx are unset. The problem is that it's not in any
6773 way backwards compatible and few other shells do it. We're eventually
6774 going to try and split the difference (heh) a little bit here. */
6775 /* l_hasdollat == 1 means we saw a quoted dollar at. */
6776
ccc6cda3
JA
6777 /* The brace expansion occurred between double quotes and there was
6778 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
7117c2d2 6779 it does not expand to anything. In this case, we want to return
a0c0a00f 6780 a quoted empty string. Posix interp 888 */
0628567a 6781 temp = make_quoted_char ('\0');
95732b49 6782 w->flags |= W_HASQUOTEDNULL;
a0c0a00f 6783/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
726f6388
JA
6784 }
6785 else
6786 temp = (char *)NULL;
6787
6788 if (c == '-' || c == '+')
95732b49
JA
6789 {
6790 w->word = temp;
6791 return w;
6792 }
726f6388
JA
6793
6794 /* c == '=' */
ccc6cda3 6795 t = temp ? savestring (temp) : savestring ("");
726f6388
JA
6796 t1 = dequote_string (t);
6797 free (t);
a0c0a00f
CR
6798
6799 /* bash-4.4/5.0 */
6800 vname = name;
6801 if (*name == '!' &&
6802 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
6803 {
6804 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
6805 if (vname == 0 || *vname == 0)
6806 {
6807 report_error (_("%s: invalid indirect expansion"), name);
6808 free (vname);
6809 dispose_word (w);
6810 return &expand_wdesc_error;
6811 }
6812 if (legal_identifier (vname) == 0)
6813 {
6814 report_error (_("%s: invalid variable name"), vname);
6815 free (vname);
6816 dispose_word (w);
6817 return &expand_wdesc_error;
6818 }
6819 }
6820
b80f6443 6821#if defined (ARRAY_VARS)
a0c0a00f
CR
6822 if (valid_array_reference (vname, 0))
6823 assign_array_element (vname, t1, 0);
b80f6443
JA
6824 else
6825#endif /* ARRAY_VARS */
a0c0a00f
CR
6826 bind_variable (vname, t1, 0);
6827
6828 stupidly_hack_special_variables (vname);
6829
6830 if (vname != name)
6831 free (vname);
95732b49 6832
495aee44
CR
6833 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6834 free (temp);
6835
6836 w->word = t1;
95732b49 6837 return w;
726f6388
JA
6838}
6839
6840/* Deal with the right hand side of a ${name:?value} expansion in the case
6841 that NAME is null or not set. If VALUE is non-null it is expanded and
6842 used as the error message to print, otherwise a standard message is
6843 printed. */
6844static void
6845parameter_brace_expand_error (name, value)
6846 char *name, *value;
6847{
ccc6cda3
JA
6848 WORD_LIST *l;
6849 char *temp;
6850
ac50fbac 6851 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
726f6388
JA
6852 if (value && *value)
6853 {
95732b49 6854 l = expand_string (value, 0);
ccc6cda3
JA
6855 temp = string_list (l);
6856 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6857 FREE (temp);
726f6388
JA
6858 dispose_words (l);
6859 }
6860 else
b80f6443 6861 report_error (_("%s: parameter null or not set"), name);
726f6388
JA
6862
6863 /* Free the data we have allocated during this expansion, since we
6864 are about to longjmp out. */
6865 free (name);
6866 FREE (value);
6867}
6868
6869/* Return 1 if NAME is something for which parameter_brace_expand_length is
6870 OK to do. */
6871static int
6872valid_length_expression (name)
6873 char *name;
6874{
28ef6c31 6875 return (name[1] == '\0' || /* ${#} */
f73dda09
JA
6876 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6877 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
ccc6cda3 6878#if defined (ARRAY_VARS)
a0c0a00f 6879 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
ccc6cda3 6880#endif
726f6388
JA
6881 legal_identifier (name + 1)); /* ${#PS1} */
6882}
6883
6884/* Handle the parameter brace expansion that requires us to return the
6885 length of a parameter. */
7117c2d2 6886static intmax_t
726f6388
JA
6887parameter_brace_expand_length (name)
6888 char *name;
6889{
ccc6cda3 6890 char *t, *newname;
7117c2d2 6891 intmax_t number, arg_index;
ccc6cda3
JA
6892 WORD_LIST *list;
6893#if defined (ARRAY_VARS)
6894 SHELL_VAR *var;
6895#endif
6896
6897 if (name[1] == '\0') /* ${#} */
6898 number = number_of_args ();
a0c0a00f 6899 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
cce855bc 6900 number = number_of_args ();
f73dda09 6901 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
cce855bc
JA
6902 {
6903 /* Take the lengths of some of the shell's special parameters. */
6904 switch (name[1])
6905 {
6906 case '-':
6907 t = which_set_flags ();
6908 break;
6909 case '?':
6910 t = itos (last_command_exit_value);
6911 break;
6912 case '$':
6913 t = itos (dollar_dollar_pid);
6914 break;
6915 case '!':
6916 if (last_asynchronous_pid == NO_PID)
495aee44 6917 t = (char *)NULL; /* XXX - error if set -u set? */
cce855bc 6918 else
f73dda09 6919 t = itos (last_asynchronous_pid);
cce855bc
JA
6920 break;
6921 case '#':
6922 t = itos (number_of_args ());
6923 break;
6924 }
6925 number = STRLEN (t);
6926 FREE (t);
6927 }
ccc6cda3 6928#if defined (ARRAY_VARS)
a0c0a00f 6929 else if (valid_array_reference (name + 1, 0))
ccc6cda3
JA
6930 number = array_length_reference (name + 1);
6931#endif /* ARRAY_VARS */
cce855bc 6932 else
ccc6cda3
JA
6933 {
6934 number = 0;
6935
f73dda09 6936 if (legal_number (name + 1, &arg_index)) /* ${#1} */
ccc6cda3 6937 {
f73dda09 6938 t = get_dollar_var_value (arg_index);
495aee44
CR
6939 if (t == 0 && unbound_vars_is_error)
6940 return INTMAX_MIN;
eb873671 6941 number = MB_STRLEN (t);
ccc6cda3
JA
6942 FREE (t);
6943 }
6944#if defined (ARRAY_VARS)
3185942a 6945 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
ccc6cda3 6946 {
3185942a
JA
6947 if (assoc_p (var))
6948 t = assoc_reference (assoc_cell (var), "0");
6949 else
6950 t = array_reference (array_cell (var), 0);
495aee44
CR
6951 if (t == 0 && unbound_vars_is_error)
6952 return INTMAX_MIN;
eb873671 6953 number = MB_STRLEN (t);
ccc6cda3
JA
6954 }
6955#endif
6956 else /* ${#PS1} */
6957 {
6958 newname = savestring (name);
6959 newname[0] = '$';
6960 list = expand_string (newname, Q_DOUBLE_QUOTES);
6961 t = list ? string_list (list) : (char *)NULL;
6962 free (newname);
6963 if (list)
6964 dispose_words (list);
6965
495aee44 6966 number = t ? MB_STRLEN (t) : 0;
ccc6cda3
JA
6967 FREE (t);
6968 }
6969 }
ccc6cda3
JA
6970
6971 return (number);
6972}
6973
28ef6c31
JA
6974/* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6975 so we do some ad-hoc parsing of an arithmetic expression to find
6976 the first DELIM, instead of using strchr(3). Two rules:
6977 1. If the substring contains a `(', read until closing `)'.
6978 2. If the substring contains a `?', read past one `:' for each `?'.
a0c0a00f 6979 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
28ef6c31
JA
6980*/
6981
6982static char *
6983skiparith (substr, delim)
6984 char *substr;
6985 int delim;
6986{
a0c0a00f
CR
6987 int i;
6988 char delims[2];
28ef6c31 6989
a0c0a00f
CR
6990 delims[0] = delim;
6991 delims[1] = '\0';
7117c2d2 6992
a0c0a00f 6993 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
7117c2d2 6994 return (substr + i);
28ef6c31
JA
6995}
6996
ccc6cda3
JA
6997/* Verify and limit the start and end of the desired substring. If
6998 VTYPE == 0, a regular shell variable is being used; if it is 1,
cce855bc 6999 then the positional parameters are being used; if it is 2, then
e8ce775d
JA
7000 VALUE is really a pointer to an array variable that should be used.
7001 Return value is 1 if both values were OK, 0 if there was a problem
7002 with an invalid expression, or -1 if the values were out of range. */
ccc6cda3 7003static int
3185942a
JA
7004verify_substring_values (v, value, substr, vtype, e1p, e2p)
7005 SHELL_VAR *v;
ccc6cda3 7006 char *value, *substr;
f73dda09 7007 int vtype;
7117c2d2 7008 intmax_t *e1p, *e2p;
ccc6cda3 7009{
bb70624e 7010 char *t, *temp1, *temp2;
f73dda09
JA
7011 arrayind_t len;
7012 int expok;
ccc6cda3
JA
7013#if defined (ARRAY_VARS)
7014 ARRAY *a;
3185942a 7015 HASH_TABLE *h;
ccc6cda3
JA
7016#endif
7017
28ef6c31
JA
7018 /* duplicate behavior of strchr(3) */
7019 t = skiparith (substr, ':');
7020 if (*t && *t == ':')
7117c2d2 7021 *t = '\0';
28ef6c31
JA
7022 else
7023 t = (char *)0;
f73dda09 7024
0628567a 7025 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
d166f048 7026 *e1p = evalexp (temp1, &expok);
ccc6cda3 7027 free (temp1);
d166f048
JA
7028 if (expok == 0)
7029 return (0);
ccc6cda3 7030
f73dda09 7031 len = -1; /* paranoia */
ccc6cda3
JA
7032 switch (vtype)
7033 {
7034 case VT_VARIABLE:
d166f048 7035 case VT_ARRAYMEMBER:
eb873671 7036 len = MB_STRLEN (value);
ccc6cda3
JA
7037 break;
7038 case VT_POSPARMS:
7039 len = number_of_args () + 1;
3185942a
JA
7040 if (*e1p == 0)
7041 len++; /* add one arg if counting from $0 */
ccc6cda3
JA
7042 break;
7043#if defined (ARRAY_VARS)
7044 case VT_ARRAYVAR:
eb873671 7045 /* For arrays, the first value deals with array indices. Negative
3185942a
JA
7046 offsets count from one past the array's maximum index. Associative
7047 arrays treat the number of elements as the maximum index. */
7048 if (assoc_p (v))
7049 {
7050 h = assoc_cell (v);
7051 len = assoc_num_elements (h) + (*e1p < 0);
7052 }
7053 else
7054 {
7055 a = (ARRAY *)value;
7056 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7057 }
ccc6cda3
JA
7058 break;
7059#endif
7060 }
7061
f73dda09
JA
7062 if (len == -1) /* paranoia */
7063 return -1;
7064
ccc6cda3
JA
7065 if (*e1p < 0) /* negative offsets count from end */
7066 *e1p += len;
7067
eb873671 7068 if (*e1p > len || *e1p < 0)
e8ce775d 7069 return (-1);
d166f048 7070
b80f6443
JA
7071#if defined (ARRAY_VARS)
7072 /* For arrays, the second offset deals with the number of elements. */
7073 if (vtype == VT_ARRAYVAR)
3185942a 7074 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
b80f6443
JA
7075#endif
7076
ccc6cda3
JA
7077 if (t)
7078 {
7079 t++;
bb70624e 7080 temp2 = savestring (t);
0628567a 7081 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
bb70624e 7082 free (temp2);
ccc6cda3 7083 t[-1] = ':';
d166f048 7084 *e2p = evalexp (temp1, &expok);
ccc6cda3 7085 free (temp1);
d166f048 7086 if (expok == 0)
28ef6c31 7087 return (0);
ac50fbac 7088#if 1
495aee44 7089 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
ac50fbac
CR
7090#else
7091 /* bash-4.3: allow positional parameter length < 0 to count backwards
7092 from end of positional parameters */
7093 if (vtype == VT_ARRAYVAR && *e2p < 0)
7094#endif
28ef6c31 7095 {
b80f6443 7096 internal_error (_("%s: substring expression < 0"), t);
ccc6cda3 7097 return (0);
28ef6c31 7098 }
b80f6443
JA
7099#if defined (ARRAY_VARS)
7100 /* In order to deal with sparse arrays, push the intelligence about how
7101 to deal with the number of elements desired down to the array-
7102 specific functions. */
7103 if (vtype != VT_ARRAYVAR)
7104#endif
7105 {
495aee44
CR
7106 if (*e2p < 0)
7107 {
7108 *e2p += len;
7109 if (*e2p < 0 || *e2p < *e1p)
7110 {
7111 internal_error (_("%s: substring expression < 0"), t);
7112 return (0);
7113 }
7114 }
7115 else
7116 *e2p += *e1p; /* want E2 chars starting at E1 */
b80f6443
JA
7117 if (*e2p > len)
7118 *e2p = len;
7119 }
ccc6cda3
JA
7120 }
7121 else
7122 *e2p = len;
7123
7124 return (1);
7125}
7126
ccc6cda3 7127/* Return the type of variable specified by VARNAME (simple variable,
cce855bc 7128 positional param, or array variable). Also return the value specified
7117c2d2 7129 by VARNAME (value of a variable or a reference to an array element).
495aee44
CR
7130 QUOTED is the standard description of quoting state, using Q_* defines.
7131 FLAGS is currently a set of flags to pass to array_value. If IND is
7132 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7133 passed to array_value so the array index is not computed again.
7117c2d2
JA
7134 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7135 characters in the value are quoted with CTLESC and takes appropriate
7136 steps. For convenience, *VALP is set to the dequoted VALUE. */
ccc6cda3 7137static int
495aee44 7138get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
ccc6cda3 7139 char *varname, *value;
495aee44
CR
7140 arrayind_t ind;
7141 int quoted, flags;
ccc6cda3
JA
7142 SHELL_VAR **varp;
7143 char **valp;
7144{
ac50fbac
CR
7145 int vtype, want_indir;
7146 char *temp, *vname;
7147 WORD_DESC *wd;
ccc6cda3 7148 SHELL_VAR *v;
495aee44 7149 arrayind_t lind;
ccc6cda3 7150
ac50fbac
CR
7151 want_indir = *varname == '!' &&
7152 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7153 || VALID_INDIR_PARAM (varname[1]));
7154 if (want_indir)
7155 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
a0c0a00f 7156 /* XXX - what if vname == 0 || *vname == 0 ? */
ac50fbac
CR
7157 else
7158 vname = varname;
a0c0a00f
CR
7159
7160 if (vname == 0)
7161 {
7162 vtype = VT_VARIABLE;
7163 *varp = (SHELL_VAR *)NULL;
7164 *valp = (char *)NULL;
7165 return (vtype);
7166 }
7167
7117c2d2 7168 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
a0c0a00f 7169 vtype = STR_DOLLAR_AT_STAR (vname);
ac50fbac 7170 if (vtype == VT_POSPARMS && vname[0] == '*')
b80f6443 7171 vtype |= VT_STARSUB;
ccc6cda3
JA
7172 *varp = (SHELL_VAR *)NULL;
7173
7174#if defined (ARRAY_VARS)
a0c0a00f 7175 if (valid_array_reference (vname, 0))
ccc6cda3 7176 {
ac50fbac 7177 v = array_variable_part (vname, &temp, (int *)0);
495aee44
CR
7178 /* If we want to signal array_value to use an already-computed index,
7179 set LIND to that index */
7180 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
ac50fbac
CR
7181 if (v && invisible_p (v))
7182 {
7183 vtype = VT_ARRAYMEMBER;
7184 *varp = (SHELL_VAR *)NULL;
7185 *valp = (char *)NULL;
7186 }
3185942a 7187 if (v && (array_p (v) || assoc_p (v)))
f73dda09
JA
7188 { /* [ */
7189 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
ccc6cda3 7190 {
ac50fbac 7191 /* Callers have to differentiate between indexed and associative */
ccc6cda3 7192 vtype = VT_ARRAYVAR;
b80f6443
JA
7193 if (temp[0] == '*')
7194 vtype |= VT_STARSUB;
3185942a 7195 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
ccc6cda3
JA
7196 }
7197 else
7198 {
d166f048 7199 vtype = VT_ARRAYMEMBER;
ac50fbac 7200 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
ccc6cda3
JA
7201 }
7202 *varp = v;
7203 }
95732b49
JA
7204 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
7205 {
7206 vtype = VT_VARIABLE;
7207 *varp = v;
7208 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7209 *valp = dequote_string (value);
7210 else
7211 *valp = dequote_escapes (value);
7212 }
ccc6cda3 7213 else
3185942a
JA
7214 {
7215 vtype = VT_ARRAYMEMBER;
7216 *varp = v;
ac50fbac 7217 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
3185942a 7218 }
ccc6cda3 7219 }
ac50fbac 7220 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
ccc6cda3 7221 {
7117c2d2 7222 vtype = VT_ARRAYMEMBER;
ccc6cda3 7223 *varp = v;
3185942a 7224 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
ccc6cda3
JA
7225 }
7226 else
7227#endif
b80f6443
JA
7228 {
7229 if (value && vtype == VT_VARIABLE)
7230 {
a0c0a00f 7231 *varp = find_variable (vname);
b80f6443
JA
7232 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7233 *valp = dequote_string (value);
7234 else
7235 *valp = dequote_escapes (value);
7236 }
7237 else
7238 *valp = value;
7239 }
ccc6cda3 7240
ac50fbac
CR
7241 if (want_indir)
7242 free (vname);
7243
ccc6cda3
JA
7244 return vtype;
7245}
7246
cce855bc
JA
7247/******************************************************/
7248/* */
7249/* Functions to extract substrings of variable values */
7250/* */
7251/******************************************************/
7252
b80f6443
JA
7253#if defined (HANDLE_MULTIBYTE)
7254/* Character-oriented rather than strictly byte-oriented substrings. S and
7255 E, rather being strict indices into STRING, indicate character (possibly
7256 multibyte character) positions that require calculation.
7257 Used by the ${param:offset[:length]} expansion. */
7258static char *
7259mb_substring (string, s, e)
7260 char *string;
7261 int s, e;
7262{
7263 char *tt;
a0c0a00f
CR
7264 int start, stop, i;
7265 size_t slen;
b80f6443
JA
7266 DECLARE_MBSTATE;
7267
7268 start = 0;
95732b49
JA
7269 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
7270 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
b80f6443
JA
7271
7272 i = s;
7273 while (string[start] && i--)
7274 ADVANCE_CHAR (string, slen, start);
7275 stop = start;
7276 i = e - s;
7277 while (string[stop] && i--)
7278 ADVANCE_CHAR (string, slen, stop);
7279 tt = substring (string, start, stop);
7280 return tt;
7281}
7282#endif
7283
ccc6cda3
JA
7284/* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
7285 is `@', use the positional parameters; otherwise, use the value of
7286 VARNAME. If VARNAME is an array variable, use the array elements. */
7287
7288static char *
495aee44
CR
7289parameter_brace_substring (varname, value, ind, substr, quoted, flags)
7290 char *varname, *value;
7291 int ind;
7292 char *substr;
7293 int quoted, flags;
ccc6cda3 7294{
7117c2d2 7295 intmax_t e1, e2;
b80f6443 7296 int vtype, r, starsub;
0628567a 7297 char *temp, *val, *tt, *oname;
ccc6cda3
JA
7298 SHELL_VAR *v;
7299
a0c0a00f 7300 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
ccc6cda3
JA
7301 return ((char *)NULL);
7302
0628567a 7303 oname = this_command_name;
ccc6cda3
JA
7304 this_command_name = varname;
7305
495aee44 7306 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3 7307 if (vtype == -1)
0628567a
JA
7308 {
7309 this_command_name = oname;
7310 return ((char *)NULL);
7311 }
ccc6cda3 7312
b80f6443
JA
7313 starsub = vtype & VT_STARSUB;
7314 vtype &= ~VT_STARSUB;
7315
3185942a 7316 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
0628567a 7317 this_command_name = oname;
e8ce775d 7318 if (r <= 0)
ac50fbac
CR
7319 {
7320 if (vtype == VT_VARIABLE)
7321 FREE (val);
7322 return ((r == 0) ? &expand_param_error : (char *)NULL);
7323 }
ccc6cda3
JA
7324
7325 switch (vtype)
7326 {
7327 case VT_VARIABLE:
d166f048 7328 case VT_ARRAYMEMBER:
b80f6443
JA
7329#if defined (HANDLE_MULTIBYTE)
7330 if (MB_CUR_MAX > 1)
7331 tt = mb_substring (val, e1, e2);
7332 else
7333#endif
7117c2d2 7334 tt = substring (val, e1, e2);
b80f6443 7335
7117c2d2
JA
7336 if (vtype == VT_VARIABLE)
7337 FREE (val);
7338 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7339 temp = quote_string (tt);
7340 else
7341 temp = tt ? quote_escapes (tt) : (char *)NULL;
7342 FREE (tt);
ccc6cda3
JA
7343 break;
7344 case VT_POSPARMS:
7117c2d2
JA
7345 tt = pos_params (varname, e1, e2, quoted);
7346 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
7347 {
7348 temp = tt ? quote_escapes (tt) : (char *)NULL;
7349 FREE (tt);
7350 }
7351 else
7352 temp = tt;
ccc6cda3
JA
7353 break;
7354#if defined (ARRAY_VARS)
7355 case VT_ARRAYVAR:
3185942a
JA
7356 if (assoc_p (v))
7357 /* we convert to list and take first e2 elements starting at e1th
7358 element -- officially undefined for now */
7359 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
7360 else
b80f6443
JA
7361 /* We want E2 to be the number of elements desired (arrays can be sparse,
7362 so verify_substring_values just returns the numbers specified and we
7363 rely on array_subrange to understand how to deal with them). */
3185942a 7364 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
f1be666c
JA
7365 /* array_subrange now calls array_quote_escapes as appropriate, so the
7366 caller no longer needs to. */
ccc6cda3
JA
7367 break;
7368#endif
f73dda09
JA
7369 default:
7370 temp = (char *)NULL;
ccc6cda3
JA
7371 }
7372
7373 return temp;
7374}
7375
cce855bc
JA
7376/****************************************************************/
7377/* */
7378/* Functions to perform pattern substitution on variable values */
7379/* */
7380/****************************************************************/
7381
a0c0a00f 7382#if 0 /* Unused */
495aee44
CR
7383static int
7384shouldexp_replacement (s)
7385 char *s;
7386{
7387 register char *p;
7388
7389 for (p = s; p && *p; p++)
7390 {
7391 if (*p == '\\')
7392 p++;
7393 else if (*p == '&')
7394 return 1;
7395 }
7396 return 0;
7397}
a0c0a00f 7398#endif
495aee44 7399
ccc6cda3
JA
7400char *
7401pat_subst (string, pat, rep, mflags)
7402 char *string, *pat, *rep;
7403 int mflags;
7404{
495aee44 7405 char *ret, *s, *e, *str, *rstr, *mstr;
a0c0a00f
CR
7406 int rptr, mtype, rxpand, mlen;
7407 size_t rsize, l, replen, rslen;
495aee44
CR
7408
7409 if (string == 0)
7410 return (savestring (""));
ccc6cda3 7411
b72432fd
JA
7412 mtype = mflags & MATCH_TYPEMASK;
7413
495aee44
CR
7414#if 0 /* bash-4.2 ? */
7415 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
7416#else
7417 rxpand = 0;
7418#endif
7419
b72432fd
JA
7420 /* Special cases:
7421 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
7422 * with REP and return the result.
7423 * 2. A null pattern with mtype == MATCH_END means to append REP to
7424 * STRING and return the result.
a0c0a00f
CR
7425 * 3. A null STRING with a matching pattern means to append REP to
7426 * STRING and return the result.
495aee44 7427 * These don't understand or process `&' in the replacement string.
b72432fd
JA
7428 */
7429 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
7430 {
7431 replen = STRLEN (rep);
495aee44 7432 l = STRLEN (string);
f73dda09 7433 ret = (char *)xmalloc (replen + l + 2);
bb70624e
JA
7434 if (replen == 0)
7435 strcpy (ret, string);
7436 else if (mtype == MATCH_BEG)
b72432fd
JA
7437 {
7438 strcpy (ret, rep);
7439 strcpy (ret + replen, string);
7440 }
7441 else
7442 {
7443 strcpy (ret, string);
7444 strcpy (ret + l, rep);
7445 }
7446 return (ret);
7447 }
a0c0a00f
CR
7448 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
7449 {
7450 replen = STRLEN (rep);
7451 ret = (char *)xmalloc (replen + 1);
7452 if (replen == 0)
7453 ret[0] = '\0';
7454 else
7455 strcpy (ret, rep);
7456 return (ret);
7457 }
b72432fd 7458
f73dda09 7459 ret = (char *)xmalloc (rsize = 64);
ccc6cda3
JA
7460 ret[0] = '\0';
7461
a0c0a00f 7462 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
ccc6cda3
JA
7463 {
7464 if (match_pattern (str, pat, mtype, &s, &e) == 0)
7465 break;
7466 l = s - str;
495aee44 7467
a0c0a00f 7468 if (rep && rxpand)
495aee44 7469 {
a0c0a00f
CR
7470 int x;
7471 mlen = e - s;
7472 mstr = xmalloc (mlen + 1);
495aee44
CR
7473 for (x = 0; x < mlen; x++)
7474 mstr[x] = s[x];
a0c0a00f
CR
7475 mstr[mlen] = '\0';
7476 rstr = strcreplace (rep, '&', mstr, 0);
7477 free (mstr);
7478 rslen = strlen (rstr);
495aee44
CR
7479 }
7480 else
a0c0a00f
CR
7481 {
7482 rstr = rep;
7483 rslen = replen;
7484 }
495aee44
CR
7485
7486 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
ccc6cda3
JA
7487
7488 /* OK, now copy the leading unmatched portion of the string (from
7489 str to s) to ret starting at rptr (the current offset). Then copy
28ef6c31
JA
7490 the replacement string at ret + rptr + (s - str). Increment
7491 rptr (if necessary) and str and go on. */
ccc6cda3
JA
7492 if (l)
7493 {
7494 strncpy (ret + rptr, str, l);
7495 rptr += l;
7496 }
7497 if (replen)
7498 {
495aee44
CR
7499 strncpy (ret + rptr, rstr, rslen);
7500 rptr += rslen;
ccc6cda3
JA
7501 }
7502 str = e; /* e == end of match */
b80f6443 7503
495aee44
CR
7504 if (rstr != rep)
7505 free (rstr);
7506
ccc6cda3 7507 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
28ef6c31 7508 break;
b80f6443
JA
7509
7510 if (s == e)
0001803f
CR
7511 {
7512 /* On a zero-length match, make sure we copy one character, since
7513 we increment one character to avoid infinite recursion. */
7514 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
7515 ret[rptr++] = *str++;
7516 e++; /* avoid infinite recursion on zero-length match */
7517 }
ccc6cda3
JA
7518 }
7519
7520 /* Now copy the unmatched portion of the input string */
495aee44 7521 if (str && *str)
d166f048
JA
7522 {
7523 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
7524 strcpy (ret + rptr, str);
7525 }
ccc6cda3
JA
7526 else
7527 ret[rptr] = '\0';
7528
7529 return ret;
7530}
7531
7532/* Do pattern match and replacement on the positional parameters. */
7533static char *
7534pos_params_pat_subst (string, pat, rep, mflags)
7535 char *string, *pat, *rep;
7536 int mflags;
7537{
7538 WORD_LIST *save, *params;
7539 WORD_DESC *w;
0628567a 7540 char *ret;
3185942a 7541 int pchar, qflags;
ccc6cda3
JA
7542
7543 save = params = list_rest_of_args ();
7544 if (save == 0)
7545 return ((char *)NULL);
7546
7547 for ( ; params; params = params->next)
7548 {
7549 ret = pat_subst (params->word->word, pat, rep, mflags);
95732b49
JA
7550 w = alloc_word_desc ();
7551 w->word = ret ? ret : savestring ("");
ccc6cda3
JA
7552 dispose_word (params->word);
7553 params->word = w;
ccc6cda3
JA
7554 }
7555
3185942a
JA
7556 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7557 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7558
3185942a 7559 ret = string_list_pos_params (pchar, save, qflags);
3185942a 7560
ccc6cda3
JA
7561 dispose_words (save);
7562
7563 return (ret);
7564}
7565
cce855bc
JA
7566/* Perform pattern substitution on VALUE, which is the expansion of
7567 VARNAME. PATSUB is an expression supplying the pattern to match
7568 and the string to substitute. QUOTED is a flags word containing
7569 the type of quoting currently in effect. */
ccc6cda3 7570static char *
a0c0a00f 7571parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
495aee44
CR
7572 char *varname, *value;
7573 int ind;
7574 char *patsub;
a0c0a00f 7575 int quoted, pflags, flags;
ccc6cda3 7576{
3185942a 7577 int vtype, mflags, starsub, delim;
7117c2d2 7578 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
ccc6cda3
JA
7579 SHELL_VAR *v;
7580
7581 if (value == 0)
7582 return ((char *)NULL);
7583
7584 this_command_name = varname;
7585
495aee44 7586 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3
JA
7587 if (vtype == -1)
7588 return ((char *)NULL);
7589
b80f6443
JA
7590 starsub = vtype & VT_STARSUB;
7591 vtype &= ~VT_STARSUB;
7592
ccc6cda3 7593 mflags = 0;
ac50fbac
CR
7594 /* PATSUB is never NULL when this is called. */
7595 if (*patsub == '/')
f1be666c
JA
7596 {
7597 mflags |= MATCH_GLOBREP;
7598 patsub++;
7599 }
7117c2d2
JA
7600
7601 /* Malloc this because expand_string_if_necessary or one of the expansion
7602 functions in its call chain may free it on a substitution error. */
bb70624e 7603 lpatsub = savestring (patsub);
ccc6cda3
JA
7604
7605 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7606 mflags |= MATCH_QUOTED;
7607
b80f6443
JA
7608 if (starsub)
7609 mflags |= MATCH_STARSUB;
7610
a0c0a00f
CR
7611 if (pflags & PF_ASSIGNRHS)
7612 mflags |= MATCH_ASSIGNRHS;
7613
0628567a
JA
7614 /* If the pattern starts with a `/', make sure we skip over it when looking
7615 for the replacement delimiter. */
3185942a
JA
7616 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
7617 if (lpatsub[delim] == '/')
7618 {
7619 lpatsub[delim] = 0;
7620 rep = lpatsub + delim + 1;
7621 }
7622 else
7623 rep = (char *)NULL;
ccc6cda3
JA
7624
7625 if (rep && *rep == '\0')
7626 rep = (char *)NULL;
7627
b80f6443
JA
7628 /* Perform the same expansions on the pattern as performed by the
7629 pattern removal expansions. */
7630 pat = getpattern (lpatsub, quoted, 1);
bb70624e 7631
ccc6cda3 7632 if (rep)
d166f048 7633 {
ac50fbac
CR
7634 /* We want to perform quote removal on the expanded replacement even if
7635 the entire expansion is double-quoted because the parser and string
7636 extraction functions treated quotes in the replacement string as
7637 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
7638 if (shell_compatibility_level > 42)
7639 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
7640 /* This is the bash-4.2 code. */
7641 else if ((mflags & MATCH_QUOTED) == 0)
f73dda09 7642 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
d166f048 7643 else
f73dda09 7644 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
d166f048 7645 }
ccc6cda3 7646
0628567a 7647 /* ksh93 doesn't allow the match specifier to be a part of the expanded
f1be666c
JA
7648 pattern. This is an extension. Make sure we don't anchor the pattern
7649 at the beginning or end of the string if we're doing global replacement,
7650 though. */
ccc6cda3 7651 p = pat;
f1be666c
JA
7652 if (mflags & MATCH_GLOBREP)
7653 mflags |= MATCH_ANY;
0628567a 7654 else if (pat && pat[0] == '#')
ccc6cda3
JA
7655 {
7656 mflags |= MATCH_BEG;
7657 p++;
7658 }
d166f048 7659 else if (pat && pat[0] == '%')
ccc6cda3
JA
7660 {
7661 mflags |= MATCH_END;
7662 p++;
7663 }
7664 else
7665 mflags |= MATCH_ANY;
7666
cce855bc
JA
7667 /* OK, we now want to substitute REP for PAT in VAL. If
7668 flags & MATCH_GLOBREP is non-zero, the substitution is done
7669 everywhere, otherwise only the first occurrence of PAT is
7117c2d2
JA
7670 replaced. The pattern matching code doesn't understand
7671 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
7672 values passed in (VT_VARIABLE) so the pattern substitution
7673 code works right. We need to requote special chars after
7674 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
7675 other cases if QUOTED == 0, since the posparams and arrays
7676 indexed by * or @ do special things when QUOTED != 0. */
7677
ccc6cda3
JA
7678 switch (vtype)
7679 {
7680 case VT_VARIABLE:
d166f048 7681 case VT_ARRAYMEMBER:
ccc6cda3 7682 temp = pat_subst (val, p, rep, mflags);
7117c2d2
JA
7683 if (vtype == VT_VARIABLE)
7684 FREE (val);
7685 if (temp)
7686 {
3185942a 7687 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7117c2d2
JA
7688 free (temp);
7689 temp = tt;
7690 }
ccc6cda3
JA
7691 break;
7692 case VT_POSPARMS:
7693 temp = pos_params_pat_subst (val, p, rep, mflags);
7117c2d2
JA
7694 if (temp && (mflags & MATCH_QUOTED) == 0)
7695 {
7696 tt = quote_escapes (temp);
7697 free (temp);
7698 temp = tt;
7699 }
ccc6cda3
JA
7700 break;
7701#if defined (ARRAY_VARS)
7702 case VT_ARRAYVAR:
3185942a
JA
7703 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
7704 : array_patsub (array_cell (v), p, rep, mflags);
7705 /* Don't call quote_escapes anymore; array_patsub calls
7706 array_quote_escapes as appropriate before adding the
7707 space separators; ditto for assoc_patsub. */
7708 break;
7709#endif
7710 }
7711
7712 FREE (pat);
7713 FREE (rep);
7714 free (lpatsub);
7715
7716 return temp;
7717}
7718
7719/****************************************************************/
7720/* */
7721/* Functions to perform case modification on variable values */
7722/* */
7723/****************************************************************/
7724
7725/* Do case modification on the positional parameters. */
7726
7727static char *
7728pos_params_modcase (string, pat, modop, mflags)
7729 char *string, *pat;
7730 int modop;
7731 int mflags;
7732{
7733 WORD_LIST *save, *params;
7734 WORD_DESC *w;
7735 char *ret;
7736 int pchar, qflags;
7737
7738 save = params = list_rest_of_args ();
7739 if (save == 0)
7740 return ((char *)NULL);
7741
7742 for ( ; params; params = params->next)
7743 {
7744 ret = sh_modcase (params->word->word, pat, modop);
7745 w = alloc_word_desc ();
7746 w->word = ret ? ret : savestring ("");
7747 dispose_word (params->word);
7748 params->word = w;
7749 }
7750
7751 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7752 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7753
7754 ret = string_list_pos_params (pchar, save, qflags);
7755 dispose_words (save);
7756
7757 return (ret);
7758}
7759
7760/* Perform case modification on VALUE, which is the expansion of
7761 VARNAME. MODSPEC is an expression supplying the type of modification
7762 to perform. QUOTED is a flags word containing the type of quoting
7763 currently in effect. */
7764static char *
495aee44 7765parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
3185942a 7766 char *varname, *value;
495aee44 7767 int ind, modspec;
3185942a 7768 char *patspec;
495aee44 7769 int quoted, flags;
3185942a
JA
7770{
7771 int vtype, starsub, modop, mflags, x;
7772 char *val, *temp, *pat, *p, *lpat, *tt;
7773 SHELL_VAR *v;
7774
7775 if (value == 0)
7776 return ((char *)NULL);
7777
7778 this_command_name = varname;
7779
495aee44 7780 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
3185942a
JA
7781 if (vtype == -1)
7782 return ((char *)NULL);
7783
7784 starsub = vtype & VT_STARSUB;
7785 vtype &= ~VT_STARSUB;
7786
7787 modop = 0;
7788 mflags = 0;
7789 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7790 mflags |= MATCH_QUOTED;
7791 if (starsub)
7792 mflags |= MATCH_STARSUB;
7793
7794 p = patspec;
7795 if (modspec == '^')
7796 {
7797 x = p && p[0] == modspec;
17345e5a 7798 modop = x ? CASE_UPPER : CASE_UPFIRST;
3185942a
JA
7799 p += x;
7800 }
7801 else if (modspec == ',')
7802 {
7803 x = p && p[0] == modspec;
17345e5a 7804 modop = x ? CASE_LOWER : CASE_LOWFIRST;
3185942a
JA
7805 p += x;
7806 }
7807 else if (modspec == '~')
7808 {
7809 x = p && p[0] == modspec;
7810 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7811 p += x;
7812 }
7813
7814 lpat = p ? savestring (p) : 0;
7815 /* Perform the same expansions on the pattern as performed by the
7816 pattern removal expansions. FOR LATER */
7817 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7818
7819 /* OK, now we do the case modification. */
7820 switch (vtype)
7821 {
7822 case VT_VARIABLE:
7823 case VT_ARRAYMEMBER:
7824 temp = sh_modcase (val, pat, modop);
7825 if (vtype == VT_VARIABLE)
7826 FREE (val);
7827 if (temp)
7828 {
7829 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7830 free (temp);
7831 temp = tt;
7832 }
7833 break;
7834
7835 case VT_POSPARMS:
7836 temp = pos_params_modcase (val, pat, modop, mflags);
7837 if (temp && (mflags & MATCH_QUOTED) == 0)
7117c2d2
JA
7838 {
7839 tt = quote_escapes (temp);
7840 free (temp);
7841 temp = tt;
7842 }
3185942a
JA
7843 break;
7844
7845#if defined (ARRAY_VARS)
7846 case VT_ARRAYVAR:
7847 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7848 : array_modcase (array_cell (v), pat, modop, mflags);
7849 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7850 as appropriate before adding the space separators; ditto for
7851 assoc_modcase. */
ccc6cda3
JA
7852 break;
7853#endif
7854 }
7855
7856 FREE (pat);
3185942a 7857 free (lpat);
ccc6cda3
JA
7858
7859 return temp;
7860}
7861
0628567a
JA
7862/* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7863 any occur, this must be a nested command substitution, so return 0.
7864 Otherwise, return 1. A valid arithmetic expression must always have a
7865 ( before a matching ), so any cases where there are more right parens
7866 means that this must not be an arithmetic expression, though the parser
7867 will not accept it without a balanced total number of parens. */
7868static int
7869chk_arithsub (s, len)
7870 const char *s;
7871 int len;
7872{
7873 int i, count;
7874 DECLARE_MBSTATE;
7875
7876 i = count = 0;
7877 while (i < len)
7878 {
0001803f 7879 if (s[i] == LPAREN)
0628567a 7880 count++;
0001803f 7881 else if (s[i] == RPAREN)
0628567a
JA
7882 {
7883 count--;
7884 if (count < 0)
7885 return 0;
7886 }
7887
7888 switch (s[i])
7889 {
7890 default:
7891 ADVANCE_CHAR (s, len, i);
7892 break;
7893
7894 case '\\':
7895 i++;
7896 if (s[i])
7897 ADVANCE_CHAR (s, len, i);
7898 break;
7899
7900 case '\'':
a0c0a00f 7901 i = skip_single_quoted (s, len, ++i, 0);
0628567a
JA
7902 break;
7903
7904 case '"':
a0c0a00f 7905 i = skip_double_quoted ((char *)s, len, ++i, 0);
0628567a
JA
7906 break;
7907 }
7908 }
7909
7910 return (count == 0);
7911}
7912
cce855bc
JA
7913/****************************************************************/
7914/* */
7915/* Functions to perform parameter expansion on a string */
7916/* */
7917/****************************************************************/
7918
3185942a 7919/* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
95732b49 7920static WORD_DESC *
0001803f 7921parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
ccc6cda3 7922 char *string;
a0c0a00f 7923 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
ccc6cda3
JA
7924{
7925 int check_nullness, var_is_set, var_is_null, var_is_special;
3185942a 7926 int want_substring, want_indir, want_patsub, want_casemod;
ccc6cda3 7927 char *name, *value, *temp, *temp1;
95732b49 7928 WORD_DESC *tdesc, *ret;
a0c0a00f 7929 int t_index, sindex, c, tflag, modspec, all_element_arrayref;
7117c2d2 7930 intmax_t number;
495aee44 7931 arrayind_t ind;
ccc6cda3 7932
3185942a 7933 temp = temp1 = value = (char *)NULL;
ccc6cda3 7934 var_is_set = var_is_null = var_is_special = check_nullness = 0;
3185942a 7935 want_substring = want_indir = want_patsub = want_casemod = 0;
ccc6cda3 7936
a0c0a00f
CR
7937 all_element_arrayref = 0;
7938
cce855bc
JA
7939 sindex = *indexp;
7940 t_index = ++sindex;
0628567a
JA
7941 /* ${#var} doesn't have any of the other parameter expansions on it. */
7942 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
3185942a 7943 name = string_extract (string, &t_index, "}", SX_VARNAME);
0628567a 7944 else
3185942a
JA
7945#if defined (CASEMOD_EXPANSIONS)
7946 /* To enable case-toggling expansions using the `~' operator character
7947 change the 1 to 0. */
7948# if defined (CASEMOD_CAPCASE)
a0c0a00f 7949 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
3185942a 7950# else
a0c0a00f 7951 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
3185942a
JA
7952# endif /* CASEMOD_CAPCASE */
7953#else
a0c0a00f 7954 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
3185942a 7955#endif /* CASEMOD_EXPANSIONS */
cce855bc 7956
a0c0a00f
CR
7957 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
7958 the cleanest code ever. */
7959 if (*name == 0 && sindex == t_index && string[sindex] == '@')
7960 {
7961 name = (char *)xrealloc (name, 2);
7962 name[0] = '@';
7963 name[1] = '\0';
7964 t_index++;
7965 }
7966 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == '}')
7967 {
7968 name = (char *)xrealloc (name, t_index - sindex + 2);
7969 name[t_index - sindex] = '@';
7970 name[t_index - sindex + 1] = '\0';
7971 t_index++;
7972 }
7973
95732b49
JA
7974 ret = 0;
7975 tflag = 0;
7976
495aee44
CR
7977 ind = INTMAX_MIN;
7978
cce855bc
JA
7979 /* If the name really consists of a special variable, then make sure
7980 that we have the entire name. We don't allow indirect references
a0c0a00f
CR
7981 to special variables except `#', `?', `@' and `*'. This clause is
7982 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
7983 general. */
495aee44 7984 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
a0c0a00f 7985 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
495aee44 7986 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
ccc6cda3
JA
7987 {
7988 t_index++;
a0c0a00f 7989 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
ac50fbac 7990 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
ccc6cda3
JA
7991 *name = string[sindex];
7992 if (string[sindex] == '!')
7993 {
28ef6c31
JA
7994 /* indirect reference of $#, $?, $@, or $* */
7995 name[1] = string[sindex + 1];
7996 strcpy (name + 2, temp1);
ccc6cda3 7997 }
cce855bc 7998 else
ccc6cda3
JA
7999 strcpy (name + 1, temp1);
8000 free (temp1);
8001 }
8002 sindex = t_index;
8003
8004 /* Find out what character ended the variable name. Then
8005 do the appropriate thing. */
8006 if (c = string[sindex])
8007 sindex++;
8008
8009 /* If c is followed by one of the valid parameter expansion
8010 characters, move past it as normal. If not, assume that
8011 a substring specification is being given, and do not move
8012 past it. */
28ef6c31 8013 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
ccc6cda3
JA
8014 {
8015 check_nullness++;
8016 if (c = string[sindex])
8017 sindex++;
8018 }
cce855bc 8019 else if (c == ':' && string[sindex] != RBRACE)
ccc6cda3 8020 want_substring = 1;
ac50fbac 8021 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
ccc6cda3 8022 want_patsub = 1;
3185942a
JA
8023#if defined (CASEMOD_EXPANSIONS)
8024 else if (c == '^' || c == ',' || c == '~')
8025 {
8026 modspec = c;
8027 want_casemod = 1;
8028 }
8029#endif
ccc6cda3 8030
cce855bc
JA
8031 /* Catch the valid and invalid brace expressions that made it through the
8032 tests above. */
8033 /* ${#-} is a valid expansion and means to take the length of $-.
8034 Similarly for ${#?} and ${##}... */
8035 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
28ef6c31 8036 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
cce855bc 8037 {
f73dda09 8038 name = (char *)xrealloc (name, 3);
cce855bc
JA
8039 name[1] = c;
8040 name[2] = '\0';
8041 c = string[sindex++];
8042 }
8043
8044 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8045 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8046 member (c, "%:=+/") && string[sindex] == RBRACE)
8047 {
8048 temp = (char *)NULL;
a0c0a00f 8049 goto bad_substitution; /* XXX - substitution error */
cce855bc
JA
8050 }
8051
8052 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8053 either a variable name, one of the positional parameters or a special
8054 variable that expands to one of the positional parameters. */
8055 want_indir = *name == '!' &&
f73dda09 8056 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7117c2d2 8057 || VALID_INDIR_PARAM (name[1]));
ccc6cda3
JA
8058
8059 /* Determine the value of this variable. */
8060
cce855bc 8061 /* Check for special variables, directly referenced. */
bb70624e 8062 if (SPECIAL_VAR (name, want_indir))
ccc6cda3
JA
8063 var_is_special++;
8064
cce855bc
JA
8065 /* Check for special expansion things, like the length of a parameter */
8066 if (*name == '#' && name[1])
ccc6cda3 8067 {
cce855bc 8068 /* If we are not pointing at the character just after the
28ef6c31
JA
8069 closing brace, then we haven't gotten all of the name.
8070 Since it begins with a special character, this is a bad
8071 substitution. Also check NAME for validity before trying
8072 to go on. */
cce855bc 8073 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
ccc6cda3
JA
8074 {
8075 temp = (char *)NULL;
a0c0a00f 8076 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8077 }
8078
8079 number = parameter_brace_expand_length (name);
495aee44
CR
8080 if (number == INTMAX_MIN && unbound_vars_is_error)
8081 {
8082 last_command_exit_value = EXECUTION_FAILURE;
8083 err_unboundvar (name+1);
8084 free (name);
8085 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8086 }
ccc6cda3
JA
8087 free (name);
8088
8089 *indexp = sindex;
95732b49
JA
8090 if (number < 0)
8091 return (&expand_wdesc_error);
8092 else
8093 {
8094 ret = alloc_word_desc ();
8095 ret->word = itos (number);
8096 return ret;
8097 }
ccc6cda3
JA
8098 }
8099
8100 /* ${@} is identical to $@. */
8101 if (name[0] == '@' && name[1] == '\0')
8102 {
8103 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8104 *quoted_dollar_atp = 1;
8105
8106 if (contains_dollar_at)
8107 *contains_dollar_at = 1;
ac50fbac
CR
8108
8109 tflag |= W_DOLLARAT;
ccc6cda3
JA
8110 }
8111
b80f6443 8112 /* Process ${!PREFIX*} expansion. */
bb70624e
JA
8113 if (want_indir && string[sindex - 1] == RBRACE &&
8114 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
f73dda09 8115 legal_variable_starter ((unsigned char) name[1]))
bb70624e
JA
8116 {
8117 char **x;
8118 WORD_LIST *xlist;
8119
8120 temp1 = savestring (name + 1);
8121 number = strlen (temp1);
8122 temp1[number - 1] = '\0';
8123 x = all_variables_matching_prefix (temp1);
7117c2d2 8124 xlist = strvec_to_word_list (x, 0, 0);
28ef6c31
JA
8125 if (string[sindex - 2] == '*')
8126 temp = string_list_dollar_star (xlist);
8127 else
8128 {
a0c0a00f 8129 temp = string_list_dollar_at (xlist, quoted, 0);
28ef6c31
JA
8130 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8131 *quoted_dollar_atp = 1;
8132 if (contains_dollar_at)
8133 *contains_dollar_at = 1;
ac50fbac
CR
8134
8135 tflag |= W_DOLLARAT;
28ef6c31 8136 }
bb70624e 8137 free (x);
89a92869 8138 dispose_words (xlist);
bb70624e
JA
8139 free (temp1);
8140 *indexp = sindex;
95732b49 8141
ac50fbac
CR
8142 free (name);
8143
95732b49
JA
8144 ret = alloc_word_desc ();
8145 ret->word = temp;
ac50fbac 8146 ret->flags = tflag; /* XXX */
95732b49 8147 return ret;
bb70624e 8148 }
b80f6443
JA
8149
8150#if defined (ARRAY_VARS)
8151 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
8152 if (want_indir && string[sindex - 1] == RBRACE &&
a0c0a00f 8153 string[sindex - 2] == ']' && valid_array_reference (name+1, 0))
b80f6443
JA
8154 {
8155 char *x, *x1;
8156
8157 temp1 = savestring (name + 1);
8158 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
8159 FREE (x);
8160 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
8161 {
3185942a 8162 temp = array_keys (temp1, quoted); /* handles assoc vars too */
b80f6443
JA
8163 if (x1[0] == '@')
8164 {
8165 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8166 *quoted_dollar_atp = 1;
8167 if (contains_dollar_at)
8168 *contains_dollar_at = 1;
ac50fbac
CR
8169
8170 tflag |= W_DOLLARAT;
b80f6443
JA
8171 }
8172
a0c0a00f 8173 free (name);
b80f6443
JA
8174 free (temp1);
8175 *indexp = sindex;
95732b49
JA
8176
8177 ret = alloc_word_desc ();
8178 ret->word = temp;
ac50fbac 8179 ret->flags = tflag; /* XXX */
95732b49 8180 return ret;
b80f6443
JA
8181 }
8182
8183 free (temp1);
8184 }
8185#endif /* ARRAY_VARS */
bb70624e 8186
ccc6cda3
JA
8187 /* Make sure that NAME is valid before trying to go on. */
8188 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
8189 var_is_special) == 0)
8190 {
8191 temp = (char *)NULL;
a0c0a00f 8192 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8193 }
8194
8195 if (want_indir)
1a1f8b54
CR
8196 {
8197 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
a0c0a00f
CR
8198 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8199 {
8200 temp = (char *)NULL;
8201 goto bad_substitution;
8202 }
1a1f8b54
CR
8203 /* Turn off the W_ARRAYIND flag because there is no way for this function
8204 to return the index we're supposed to be using. */
8205 if (tdesc && tdesc->flags)
8206 tdesc->flags &= ~W_ARRAYIND;
8207 }
95732b49 8208 else
ac50fbac 8209 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
95732b49
JA
8210
8211 if (tdesc)
8212 {
8213 temp = tdesc->word;
8214 tflag = tdesc->flags;
8215 dispose_word_desc (tdesc);
8216 }
ccc6cda3 8217 else
95732b49 8218 temp = (char *)0;
ccc6cda3 8219
ac50fbac
CR
8220 if (temp == &expand_param_error || temp == &expand_param_fatal)
8221 {
8222 FREE (name);
8223 FREE (value);
8224 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8225 }
8226
ccc6cda3 8227#if defined (ARRAY_VARS)
a0c0a00f
CR
8228 if (valid_array_reference (name, 0))
8229 {
8230 int qflags;
8231 char *t;
8232
8233 qflags = quoted;
8234 /* If in a context where word splitting will not take place, treat as
8235 if double-quoted. Has effects with $* and ${array[*]} */
8236 if (pflags & PF_ASSIGNRHS)
8237 qflags |= Q_DOUBLE_QUOTES;
8238 chk_atstar (name, qflags, quoted_dollar_atp, contains_dollar_at);
8239 /* We duplicate a little code here */
8240 t = mbschr (name, '[');
8241 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == ']')
8242 all_element_arrayref = 1;
8243 }
ccc6cda3
JA
8244#endif
8245
8246 var_is_set = temp != (char *)0;
8247 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
ac50fbac
CR
8248 /* XXX - this may not need to be restricted to special variables */
8249 if (check_nullness)
8250 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
ccc6cda3
JA
8251
8252 /* Get the rest of the stuff inside the braces. */
cce855bc 8253 if (c && c != RBRACE)
ccc6cda3
JA
8254 {
8255 /* Extract the contents of the ${ ... } expansion
28ef6c31 8256 according to the Posix.2 rules. */
49ed961b 8257 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
cce855bc 8258 if (string[sindex] == RBRACE)
28ef6c31 8259 sindex++;
ccc6cda3 8260 else
a0c0a00f 8261 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8262 }
8263 else
8264 value = (char *)NULL;
726f6388 8265
ccc6cda3
JA
8266 *indexp = sindex;
8267
495aee44
CR
8268 /* All the cases where an expansion can possibly generate an unbound
8269 variable error. */
8270 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
8271 {
a0c0a00f 8272 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
495aee44
CR
8273 {
8274 last_command_exit_value = EXECUTION_FAILURE;
8275 err_unboundvar (name);
8276 FREE (value);
8277 FREE (temp);
8278 free (name);
8279 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8280 }
8281 }
8282
ccc6cda3
JA
8283 /* If this is a substring spec, process it and add the result. */
8284 if (want_substring)
726f6388 8285 {
495aee44 8286 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8287 FREE (name);
8288 FREE (value);
8289 FREE (temp);
95732b49
JA
8290
8291 if (temp1 == &expand_param_error)
8292 return (&expand_wdesc_error);
8293 else if (temp1 == &expand_param_fatal)
8294 return (&expand_wdesc_fatal);
8295
8296 ret = alloc_word_desc ();
8297 ret->word = temp1;
ac50fbac
CR
8298 /* We test quoted_dollar_atp because we want variants with double-quoted
8299 "$@" to take a different code path. In fact, we make sure at the end
8300 of expand_word_internal that we're only looking at these flags if
8301 quoted_dollar_at == 0. */
8302 if (temp1 &&
8303 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8304 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
0628567a 8305 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
95732b49 8306 return ret;
726f6388 8307 }
ccc6cda3 8308 else if (want_patsub)
726f6388 8309 {
a0c0a00f 8310 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8311 FREE (name);
8312 FREE (value);
8313 FREE (temp);
95732b49
JA
8314
8315 if (temp1 == &expand_param_error)
8316 return (&expand_wdesc_error);
8317 else if (temp1 == &expand_param_fatal)
8318 return (&expand_wdesc_fatal);
8319
8320 ret = alloc_word_desc ();
8321 ret->word = temp1;
ac50fbac
CR
8322 if (temp1 &&
8323 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8324 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a 8325 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
95732b49 8326 return ret;
ccc6cda3 8327 }
3185942a
JA
8328#if defined (CASEMOD_EXPANSIONS)
8329 else if (want_casemod)
8330 {
495aee44 8331 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
3185942a
JA
8332 FREE (name);
8333 FREE (value);
8334 FREE (temp);
8335
8336 if (temp1 == &expand_param_error)
8337 return (&expand_wdesc_error);
8338 else if (temp1 == &expand_param_fatal)
8339 return (&expand_wdesc_fatal);
8340
8341 ret = alloc_word_desc ();
8342 ret->word = temp1;
ac50fbac
CR
8343 if (temp1 &&
8344 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8345 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a
JA
8346 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8347 return ret;
8348 }
8349#endif
726f6388 8350
ccc6cda3
JA
8351 /* Do the right thing based on which character ended the variable name. */
8352 switch (c)
8353 {
8354 default:
8355 case '\0':
a0c0a00f 8356bad_substitution:
98043138 8357 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 8358 report_error (_("%s: bad substitution"), string ? string : "??");
ccc6cda3
JA
8359 FREE (value);
8360 FREE (temp);
8361 free (name);
a0c0a00f
CR
8362 if (shell_compatibility_level <= 43)
8363 return &expand_wdesc_error;
8364 else
8365 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
ccc6cda3 8366
cce855bc 8367 case RBRACE:
ccc6cda3 8368 break;
726f6388 8369
a0c0a00f
CR
8370 case '@':
8371 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8372 free (temp);
8373 free (value);
8374 free (name);
8375 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
8376 {
8377 last_command_exit_value = EXECUTION_FAILURE;
8378 report_error (_("%s: bad substitution"), string ? string : "??");
8379 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8380 }
8381
8382 ret = alloc_word_desc ();
8383 ret->word = temp1;
8384 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8385 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8386 return ret;
8387
ccc6cda3
JA
8388 case '#': /* ${param#[#]pattern} */
8389 case '%': /* ${param%[%]pattern} */
8390 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
28ef6c31
JA
8391 {
8392 FREE (value);
ccc6cda3 8393 break;
28ef6c31 8394 }
495aee44 8395 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8396 free (temp);
8397 free (value);
495aee44 8398 free (name);
3185942a
JA
8399
8400 ret = alloc_word_desc ();
8401 ret->word = temp1;
8402 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8403 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8404 return ret;
ccc6cda3
JA
8405
8406 case '-':
8407 case '=':
8408 case '?':
8409 case '+':
8410 if (var_is_set && var_is_null == 0)
28ef6c31
JA
8411 {
8412 /* If the operator is `+', we don't want the value of the named
8413 variable for anything, just the value of the right hand side. */
ccc6cda3
JA
8414 if (c == '+')
8415 {
28ef6c31
JA
8416 /* XXX -- if we're double-quoted and the named variable is "$@",
8417 we want to turn off any special handling of "$@" --
8418 we're not using it, so whatever is on the rhs applies. */
8419 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8420 *quoted_dollar_atp = 0;
8421 if (contains_dollar_at)
8422 *contains_dollar_at = 0;
8423
ccc6cda3
JA
8424 FREE (temp);
8425 if (value)
28ef6c31 8426 {
495aee44
CR
8427 /* From Posix discussion on austin-group list. Issue 221
8428 requires that backslashes escaping `}' inside
8429 double-quoted ${...} be removed. */
8430 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8431 quoted |= Q_DOLBRACE;
95732b49
JA
8432 ret = parameter_brace_expand_rhs (name, value, c,
8433 quoted,
a0c0a00f 8434 pflags,
95732b49
JA
8435 quoted_dollar_atp,
8436 contains_dollar_at);
8437 /* XXX - fix up later, esp. noting presence of
8438 W_HASQUOTEDNULL in ret->flags */
ccc6cda3
JA
8439 free (value);
8440 }
8441 else
28ef6c31 8442 temp = (char *)NULL;
ccc6cda3
JA
8443 }
8444 else
8445 {
8446 FREE (value);
8447 }
8448 /* Otherwise do nothing; just use the value in TEMP. */
726f6388 8449 }
ccc6cda3 8450 else /* VAR not set or VAR is NULL. */
28ef6c31 8451 {
ccc6cda3
JA
8452 FREE (temp);
8453 temp = (char *)NULL;
8454 if (c == '=' && var_is_special)
8455 {
ac50fbac 8456 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 8457 report_error (_("$%s: cannot assign in this way"), name);
ccc6cda3
JA
8458 free (name);
8459 free (value);
95732b49 8460 return &expand_wdesc_error;
ccc6cda3
JA
8461 }
8462 else if (c == '?')
8463 {
8464 parameter_brace_expand_error (name, value);
95732b49 8465 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
ccc6cda3
JA
8466 }
8467 else if (c != '+')
28ef6c31
JA
8468 {
8469 /* XXX -- if we're double-quoted and the named variable is "$@",
8470 we want to turn off any special handling of "$@" --
8471 we're not using it, so whatever is on the rhs applies. */
8472 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8473 *quoted_dollar_atp = 0;
8474 if (contains_dollar_at)
8475 *contains_dollar_at = 0;
8476
495aee44
CR
8477 /* From Posix discussion on austin-group list. Issue 221 requires
8478 that backslashes escaping `}' inside double-quoted ${...} be
8479 removed. */
8480 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8481 quoted |= Q_DOLBRACE;
a0c0a00f 8482 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
95732b49
JA
8483 quoted_dollar_atp,
8484 contains_dollar_at);
8485 /* XXX - fix up later, esp. noting presence of
8486 W_HASQUOTEDNULL in tdesc->flags */
28ef6c31 8487 }
ccc6cda3 8488 free (value);
726f6388 8489 }
28ef6c31 8490
ccc6cda3 8491 break;
726f6388 8492 }
ccc6cda3 8493 free (name);
95732b49
JA
8494
8495 if (ret == 0)
8496 {
8497 ret = alloc_word_desc ();
8498 ret->flags = tflag;
8499 ret->word = temp;
8500 }
8501 return (ret);
726f6388
JA
8502}
8503
cce855bc
JA
8504/* Expand a single ${xxx} expansion. The braces are optional. When
8505 the braces are used, parameter_brace_expand() does the work,
8506 possibly calling param_expand recursively. */
95732b49 8507static WORD_DESC *
cce855bc
JA
8508param_expand (string, sindex, quoted, expanded_something,
8509 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
8510 pflags)
8511 char *string;
8512 int *sindex, quoted, *expanded_something, *contains_dollar_at;
8513 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
8514{
a0c0a00f 8515 char *temp, *temp1, uerror[3], *savecmd;
f73dda09
JA
8516 int zindex, t_index, expok;
8517 unsigned char c;
7117c2d2 8518 intmax_t number;
cce855bc 8519 SHELL_VAR *var;
f73dda09 8520 WORD_LIST *list;
95732b49
JA
8521 WORD_DESC *tdesc, *ret;
8522 int tflag;
cce855bc 8523
a0c0a00f 8524/*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
cce855bc
JA
8525 zindex = *sindex;
8526 c = string[++zindex];
8527
8528 temp = (char *)NULL;
95732b49
JA
8529 ret = tdesc = (WORD_DESC *)NULL;
8530 tflag = 0;
cce855bc
JA
8531
8532 /* Do simple cases first. Switch on what follows '$'. */
8533 switch (c)
8534 {
8535 /* $0 .. $9? */
8536 case '0':
8537 case '1':
8538 case '2':
8539 case '3':
8540 case '4':
8541 case '5':
8542 case '6':
8543 case '7':
8544 case '8':
8545 case '9':
f73dda09 8546 temp1 = dollar_vars[TODIGIT (c)];
cce855bc
JA
8547 if (unbound_vars_is_error && temp1 == (char *)NULL)
8548 {
7117c2d2
JA
8549 uerror[0] = '$';
8550 uerror[1] = c;
8551 uerror[2] = '\0';
cce855bc 8552 last_command_exit_value = EXECUTION_FAILURE;
0001803f 8553 err_unboundvar (uerror);
95732b49 8554 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc 8555 }
b80f6443
JA
8556 if (temp1)
8557 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8558 ? quote_string (temp1)
8559 : quote_escapes (temp1);
8560 else
8561 temp = (char *)NULL;
95732b49 8562
cce855bc
JA
8563 break;
8564
8565 /* $$ -- pid of the invoking shell. */
8566 case '$':
8567 temp = itos (dollar_dollar_pid);
8568 break;
8569
8570 /* $# -- number of positional parameters. */
8571 case '#':
8572 temp = itos (number_of_args ());
8573 break;
8574
8575 /* $? -- return value of the last synchronous command. */
8576 case '?':
8577 temp = itos (last_command_exit_value);
8578 break;
8579
8580 /* $- -- flags supplied to the shell on invocation or by `set'. */
8581 case '-':
8582 temp = which_set_flags ();
8583 break;
8584
8585 /* $! -- Pid of the last asynchronous command. */
8586 case '!':
8587 /* If no asynchronous pids have been created, expand to nothing.
8588 If `set -u' has been executed, and no async processes have
8589 been created, this is an expansion error. */
8590 if (last_asynchronous_pid == NO_PID)
8591 {
8592 if (expanded_something)
8593 *expanded_something = 0;
8594 temp = (char *)NULL;
8595 if (unbound_vars_is_error)
8596 {
7117c2d2
JA
8597 uerror[0] = '$';
8598 uerror[1] = c;
8599 uerror[2] = '\0';
cce855bc 8600 last_command_exit_value = EXECUTION_FAILURE;
0001803f 8601 err_unboundvar (uerror);
95732b49 8602 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc
JA
8603 }
8604 }
8605 else
f73dda09 8606 temp = itos (last_asynchronous_pid);
cce855bc
JA
8607 break;
8608
8609 /* The only difference between this and $@ is when the arg is quoted. */
8610 case '*': /* `$*' */
8611 list = list_rest_of_args ();
8612
89a92869
CR
8613#if 0
8614 /* According to austin-group posix proposal by Geoff Clare in
8615 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8616
8617 "The shell shall write a message to standard error and
8618 immediately exit when it tries to expand an unset parameter
8619 other than the '@' and '*' special parameters."
8620 */
8621
8622 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
8623 {
8624 uerror[0] = '$';
8625 uerror[1] = '*';
8626 uerror[2] = '\0';
3185942a 8627 last_command_exit_value = EXECUTION_FAILURE;
89a92869 8628 err_unboundvar (uerror);
3185942a
JA
8629 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8630 }
89a92869 8631#endif
3185942a 8632
cce855bc
JA
8633 /* If there are no command-line arguments, this should just
8634 disappear if there are other characters in the expansion,
8635 even if it's quoted. */
8636 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
8637 temp = (char *)NULL;
0001803f 8638 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
cce855bc
JA
8639 {
8640 /* If we have "$*" we want to make a string of the positional
8641 parameters, separated by the first character of $IFS, and
8642 quote the whole string, including the separators. If IFS
8643 is unset, the parameters are separated by ' '; if $IFS is
8644 null, the parameters are concatenated. */
0001803f 8645 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
495aee44
CR
8646 if (temp)
8647 {
a0c0a00f 8648 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
495aee44
CR
8649 if (*temp == 0)
8650 tflag |= W_HASQUOTEDNULL;
a0c0a00f
CR
8651 if (temp != temp1)
8652 free (temp);
495aee44
CR
8653 temp = temp1;
8654 }
cce855bc
JA
8655 }
8656 else
28ef6c31 8657 {
95732b49
JA
8658 /* We check whether or not we're eventually going to split $* here,
8659 for example when IFS is empty and we are processing the rhs of
8660 an assignment statement. In that case, we don't separate the
8661 arguments at all. Otherwise, if the $* is not quoted it is
8662 identical to $@ */
95732b49
JA
8663# if defined (HANDLE_MULTIBYTE)
8664 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
8665# else
8666 if (expand_no_split_dollar_star && ifs_firstc == 0)
8667# endif
8668 temp = string_list_dollar_star (list);
8669 else
ac50fbac 8670 {
a0c0a00f 8671 temp = string_list_dollar_at (list, quoted, 0);
ac50fbac
CR
8672 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
8673 tflag |= W_SPLITSPACE;
a0c0a00f
CR
8674 /* If we're not quoted but we still don't want word splitting, make
8675 we quote the IFS characters to protect them from splitting (e.g.,
8676 when $@ is in the string as well). */
8677 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
8678 {
8679 temp1 = quote_string (temp);
8680 free (temp);
8681 temp = temp1;
8682 }
ac50fbac
CR
8683 }
8684
28ef6c31
JA
8685 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
8686 *contains_dollar_at = 1;
8687 }
cce855bc
JA
8688
8689 dispose_words (list);
8690 break;
8691
8692 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
8693 means that we have to turn quoting off after we split into
8694 the individually quoted arguments so that the final split
8695 on the first character of $IFS is still done. */
8696 case '@': /* `$@' */
8697 list = list_rest_of_args ();
8698
89a92869
CR
8699#if 0
8700 /* According to austin-group posix proposal by Geoff Clare in
8701 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8702
8703 "The shell shall write a message to standard error and
8704 immediately exit when it tries to expand an unset parameter
8705 other than the '@' and '*' special parameters."
8706 */
8707
8708 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
8709 {
8710 uerror[0] = '$';
8711 uerror[1] = '@';
8712 uerror[2] = '\0';
3185942a 8713 last_command_exit_value = EXECUTION_FAILURE;
89a92869 8714 err_unboundvar (uerror);
3185942a
JA
8715 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8716 }
89a92869 8717#endif
3185942a 8718
cce855bc
JA
8719 /* We want to flag the fact that we saw this. We can't turn
8720 off quoting entirely, because other characters in the
8721 string might need it (consider "\"$@\""), but we need some
8722 way to signal that the final split on the first character
8723 of $IFS should be done, even though QUOTED is 1. */
0001803f 8724 /* XXX - should this test include Q_PATQUOTE? */
cce855bc
JA
8725 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8726 *quoted_dollar_at_p = 1;
8727 if (contains_dollar_at)
8728 *contains_dollar_at = 1;
8729
8730 /* We want to separate the positional parameters with the first
8731 character of $IFS in case $IFS is something other than a space.
8732 We also want to make sure that splitting is done no matter what --
8733 according to POSIX.2, this expands to a list of the positional
8734 parameters no matter what IFS is set to. */
3b34f6e6
CR
8735 /* XXX - what to do when in a context where word splitting is not
8736 performed? Even when IFS is not the default, posix seems to imply
8737 that we behave like unquoted $* ? Maybe we should use PF_NOSPLIT2
8738 here. */
a0c0a00f
CR
8739 /* XXX - bash-4.4/bash-5.0 passing PFLAGS */
8740 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted, pflags);
cce855bc 8741
ac50fbac 8742 tflag |= W_DOLLARAT;
cce855bc
JA
8743 dispose_words (list);
8744 break;
8745
8746 case LBRACE:
0001803f 8747 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
95732b49
JA
8748 quoted_dollar_at_p,
8749 contains_dollar_at);
8750
95732b49
JA
8751 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8752 return (tdesc);
8753 temp = tdesc ? tdesc->word : (char *)0;
cce855bc
JA
8754
8755 /* XXX */
bb70624e 8756 /* Quoted nulls should be removed if there is anything else
cce855bc
JA
8757 in the string. */
8758 /* Note that we saw the quoted null so we can add one back at
8759 the end of this function if there are no other characters
28ef6c31
JA
8760 in the string, discard TEMP, and go on. The exception to
8761 this is when we have "${@}" and $1 is '', since $@ needs
8762 special handling. */
95732b49 8763 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
cce855bc
JA
8764 {
8765 if (had_quoted_null_p)
8766 *had_quoted_null_p = 1;
28ef6c31
JA
8767 if (*quoted_dollar_at_p == 0)
8768 {
8769 free (temp);
95732b49 8770 tdesc->word = temp = (char *)NULL;
28ef6c31
JA
8771 }
8772
cce855bc
JA
8773 }
8774
95732b49 8775 ret = tdesc;
cce855bc
JA
8776 goto return0;
8777
8778 /* Do command or arithmetic substitution. */
8779 case LPAREN:
8780 /* We have to extract the contents of this paren substitution. */
8781 t_index = zindex + 1;
a0c0a00f
CR
8782 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
8783 as arithmetic substitution immediately. */
8784 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
cce855bc
JA
8785 zindex = t_index;
8786
8787 /* For Posix.2-style `$(( ))' arithmetic substitution,
28ef6c31 8788 extract the expression and pass it to the evaluator. */
cce855bc
JA
8789 if (temp && *temp == LPAREN)
8790 {
8791 char *temp2;
8792 temp1 = temp + 1;
8793 temp2 = savestring (temp1);
8794 t_index = strlen (temp2) - 1;
8795
8796 if (temp2[t_index] != RPAREN)
8797 {
8798 free (temp2);
8799 goto comsub;
8800 }
8801
8802 /* Cut off ending `)' */
8803 temp2[t_index] = '\0';
8804
0628567a
JA
8805 if (chk_arithsub (temp2, t_index) == 0)
8806 {
8807 free (temp2);
0001803f
CR
8808#if 0
8809 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
8810#endif
0628567a
JA
8811 goto comsub;
8812 }
8813
cce855bc 8814 /* Expand variables found inside the expression. */
a0c0a00f 8815 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
8816 free (temp2);
8817
8818arithsub:
8819 /* No error messages. */
a0c0a00f 8820 savecmd = this_command_name;
cce855bc
JA
8821 this_command_name = (char *)NULL;
8822 number = evalexp (temp1, &expok);
a0c0a00f 8823 this_command_name = savecmd;
cce855bc
JA
8824 free (temp);
8825 free (temp1);
8826 if (expok == 0)
8827 {
8828 if (interactive_shell == 0 && posixly_correct)
8829 {
8830 last_command_exit_value = EXECUTION_FAILURE;
95732b49 8831 return (&expand_wdesc_fatal);
cce855bc
JA
8832 }
8833 else
95732b49 8834 return (&expand_wdesc_error);
cce855bc
JA
8835 }
8836 temp = itos (number);
8837 break;
8838 }
8839
8840comsub:
b80f6443
JA
8841 if (pflags & PF_NOCOMSUB)
8842 /* we need zindex+1 because string[zindex] == RPAREN */
8843 temp1 = substring (string, *sindex, zindex+1);
8844 else
3185942a
JA
8845 {
8846 tdesc = command_substitute (temp, quoted);
8847 temp1 = tdesc ? tdesc->word : (char *)NULL;
8848 if (tdesc)
8849 dispose_word_desc (tdesc);
8850 }
cce855bc
JA
8851 FREE (temp);
8852 temp = temp1;
8853 break;
8854
8855 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
8856 away in a future bash release. */
8857 case '[':
bb70624e 8858 /* Extract the contents of this arithmetic substitution. */
cce855bc
JA
8859 t_index = zindex + 1;
8860 temp = extract_arithmetic_subst (string, &t_index);
8861 zindex = t_index;
3185942a
JA
8862 if (temp == 0)
8863 {
8864 temp = savestring (string);
8865 if (expanded_something)
8866 *expanded_something = 0;
8867 goto return0;
8868 }
cce855bc
JA
8869
8870 /* Do initial variable expansion. */
a0c0a00f 8871 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
8872
8873 goto arithsub;
8874
8875 default:
8876 /* Find the variable in VARIABLE_LIST. */
8877 temp = (char *)NULL;
8878
8879 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
8880 ;
8881 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
8882
8883 /* If this isn't a variable name, then just output the `$'. */
8884 if (temp1 == 0 || *temp1 == '\0')
8885 {
8886 FREE (temp1);
f73dda09 8887 temp = (char *)xmalloc (2);
cce855bc
JA
8888 temp[0] = '$';
8889 temp[1] = '\0';
8890 if (expanded_something)
8891 *expanded_something = 0;
8892 goto return0;
8893 }
8894
8895 /* If the variable exists, return its value cell. */
8896 var = find_variable (temp1);
8897
7117c2d2 8898 if (var && invisible_p (var) == 0 && var_isset (var))
cce855bc
JA
8899 {
8900#if defined (ARRAY_VARS)
3185942a 8901 if (assoc_p (var) || array_p (var))
cce855bc 8902 {
3185942a
JA
8903 temp = array_p (var) ? array_reference (array_cell (var), 0)
8904 : assoc_reference (assoc_cell (var), "0");
cce855bc 8905 if (temp)
b80f6443
JA
8906 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8907 ? quote_string (temp)
8908 : quote_escapes (temp);
8909 else if (unbound_vars_is_error)
8910 goto unbound_variable;
cce855bc
JA
8911 }
8912 else
8913#endif
b80f6443
JA
8914 {
8915 temp = value_cell (var);
8916
8917 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8918 ? quote_string (temp)
8919 : quote_escapes (temp);
8920 }
8921
cce855bc 8922 free (temp1);
7117c2d2 8923
cce855bc
JA
8924 goto return0;
8925 }
b64a0e1d
CR
8926 else if (var && (invisible_p (var) || var_isset (var) == 0))
8927 temp = (char *)NULL;
a0c0a00f 8928 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
ac50fbac
CR
8929 {
8930 temp = nameref_cell (var);
8931#if defined (ARRAY_VARS)
a0c0a00f 8932 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
8933 {
8934 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8935 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8936 return (tdesc);
8937 ret = tdesc;
8938 goto return0;
8939 }
8940 else
8941#endif
8942 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8943 if (temp && *temp && legal_identifier (temp) == 0)
8944 {
8945 last_command_exit_value = EXECUTION_FAILURE;
8946 report_error (_("%s: invalid variable name for name reference"), temp);
8947 return (&expand_wdesc_error); /* XXX */
8948 }
8949 else
8950 temp = (char *)NULL;
8951 }
cce855bc
JA
8952
8953 temp = (char *)NULL;
8954
b80f6443 8955unbound_variable:
cce855bc 8956 if (unbound_vars_is_error)
0001803f
CR
8957 {
8958 last_command_exit_value = EXECUTION_FAILURE;
8959 err_unboundvar (temp1);
8960 }
cce855bc
JA
8961 else
8962 {
8963 free (temp1);
8964 goto return0;
8965 }
8966
8967 free (temp1);
8968 last_command_exit_value = EXECUTION_FAILURE;
8969 return ((unbound_vars_is_error && interactive_shell == 0)
95732b49
JA
8970 ? &expand_wdesc_fatal
8971 : &expand_wdesc_error);
cce855bc
JA
8972 }
8973
8974 if (string[zindex])
8975 zindex++;
8976
8977return0:
8978 *sindex = zindex;
95732b49
JA
8979
8980 if (ret == 0)
8981 {
8982 ret = alloc_word_desc ();
8983 ret->flags = tflag; /* XXX */
8984 ret->word = temp;
8985 }
8986 return ret;
cce855bc
JA
8987}
8988
a0c0a00f
CR
8989void
8990invalidate_cached_quoted_dollar_at ()
8991{
8992 dispose_words (cached_quoted_dollar_at);
8993 cached_quoted_dollar_at = 0;
8994}
8995
cce855bc
JA
8996/* Make a word list which is the result of parameter and variable
8997 expansion, command substitution, arithmetic substitution, and
8998 quote removal of WORD. Return a pointer to a WORD_LIST which is
8999 the result of the expansion. If WORD contains a null word, the
9000 word list returned is also null.
726f6388 9001
ccc6cda3
JA
9002 QUOTED contains flag values defined in shell.h.
9003
b72432fd
JA
9004 ISEXP is used to tell expand_word_internal that the word should be
9005 treated as the result of an expansion. This has implications for
9006 how IFS characters in the word are treated.
9007
726f6388
JA
9008 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
9009 they point to an integer value which receives information about expansion.
9010 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
9011 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
9012 else zero.
9013
9014 This only does word splitting in the case of $@ expansion. In that
9015 case, we split on ' '. */
9016
9017/* Values for the local variable quoted_state. */
9018#define UNQUOTED 0
9019#define PARTIALLY_QUOTED 1
9020#define WHOLLY_QUOTED 2
9021
9022static WORD_LIST *
b72432fd 9023expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
726f6388 9024 WORD_DESC *word;
b72432fd 9025 int quoted, isexp;
726f6388
JA
9026 int *contains_dollar_at;
9027 int *expanded_something;
9028{
ccc6cda3
JA
9029 WORD_LIST *list;
9030 WORD_DESC *tword;
726f6388
JA
9031
9032 /* The intermediate string that we build while expanding. */
ccc6cda3 9033 char *istring;
726f6388
JA
9034
9035 /* The current size of the above object. */
a0c0a00f 9036 size_t istring_size;
726f6388
JA
9037
9038 /* Index into ISTRING. */
ccc6cda3 9039 int istring_index;
726f6388
JA
9040
9041 /* Temporary string storage. */
ccc6cda3 9042 char *temp, *temp1;
726f6388
JA
9043
9044 /* The text of WORD. */
ccc6cda3 9045 register char *string;
726f6388 9046
7117c2d2
JA
9047 /* The size of STRING. */
9048 size_t string_size;
9049
726f6388 9050 /* The index into STRING. */
ccc6cda3 9051 int sindex;
726f6388
JA
9052
9053 /* This gets 1 if we see a $@ while quoted. */
ccc6cda3 9054 int quoted_dollar_at;
726f6388
JA
9055
9056 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
9057 whether WORD contains no quoting characters, a partially quoted
9058 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
ccc6cda3
JA
9059 int quoted_state;
9060
95732b49 9061 /* State flags */
ccc6cda3 9062 int had_quoted_null;
0b913689 9063 int has_dollar_at, temp_has_dollar_at;
ac50fbac 9064 int split_on_spaces;
28ef6c31 9065 int tflag;
0001803f 9066 int pflags; /* flags passed to param_expand */
a0c0a00f 9067 int mb_cur_max;
726f6388 9068
95732b49
JA
9069 int assignoff; /* If assignment, offset of `=' */
9070
f73dda09 9071 register unsigned char c; /* Current character. */
726f6388 9072 int t_index; /* For calls to string_extract_xxx. */
726f6388 9073
bb70624e 9074 char twochars[2];
b72432fd 9075
7117c2d2
JA
9076 DECLARE_MBSTATE;
9077
a0c0a00f
CR
9078 /* OK, let's see if we can optimize a common idiom: "$@" */
9079 if (STREQ (word->word, "\"$@\"") &&
9080 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
9081 dollar_vars[1]) /* XXX - check IFS here as well? */
9082 {
9083 if (contains_dollar_at)
9084 *contains_dollar_at = 1;
9085 if (expanded_something)
9086 *expanded_something = 1;
9087 if (cached_quoted_dollar_at)
9088 return (copy_word_list (cached_quoted_dollar_at));
9089 list = list_rest_of_args ();
9090 list = quote_list (list);
9091 cached_quoted_dollar_at = copy_word_list (list);
9092 return (list);
9093 }
9094
f73dda09 9095 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
ccc6cda3 9096 istring[istring_index = 0] = '\0';
cce855bc 9097 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
ac50fbac 9098 split_on_spaces = 0;
ccc6cda3
JA
9099 quoted_state = UNQUOTED;
9100
9101 string = word->word;
9102 if (string == 0)
9103 goto finished_with_string;
a0c0a00f
CR
9104 mb_cur_max = MB_CUR_MAX;
9105
95732b49
JA
9106 /* Don't need the string length for the SADD... and COPY_ macros unless
9107 multibyte characters are possible. */
a0c0a00f 9108 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
726f6388
JA
9109
9110 if (contains_dollar_at)
9111 *contains_dollar_at = 0;
9112
95732b49
JA
9113 assignoff = -1;
9114
726f6388
JA
9115 /* Begin the expansion. */
9116
ccc6cda3 9117 for (sindex = 0; ;)
726f6388
JA
9118 {
9119 c = string[sindex];
9120
ac50fbac 9121 /* Case on top-level character. */
726f6388
JA
9122 switch (c)
9123 {
9124 case '\0':
9125 goto finished_with_string;
9126
9127 case CTLESC:
7117c2d2
JA
9128 sindex++;
9129#if HANDLE_MULTIBYTE
a0c0a00f 9130 if (mb_cur_max > 1 && string[sindex])
7117c2d2 9131 {
b80f6443 9132 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
9133 }
9134 else
9135#endif
9136 {
9137 temp = (char *)xmalloc (3);
9138 temp[0] = CTLESC;
9139 temp[1] = c = string[sindex];
9140 temp[2] = '\0';
9141 }
726f6388 9142
cce855bc 9143dollar_add_string:
726f6388
JA
9144 if (string[sindex])
9145 sindex++;
9146
cce855bc
JA
9147add_string:
9148 if (temp)
9149 {
9150 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
9151 temp = (char *)0;
9152 }
9153
9154 break;
726f6388
JA
9155
9156#if defined (PROCESS_SUBSTITUTION)
9157 /* Process substitution. */
9158 case '<':
9159 case '>':
9160 {
a0c0a00f
CR
9161 /* bash-4.4/bash-5.0
9162 XXX - technically this should only be expanded at the start
9163 of a word */
0628567a 9164 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
726f6388 9165 {
bb70624e 9166 sindex--; /* add_character: label increments sindex */
726f6388
JA
9167 goto add_character;
9168 }
9169 else
cce855bc 9170 t_index = sindex + 1; /* skip past both '<' and LPAREN */
726f6388 9171
85b94814 9172 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
ccc6cda3 9173 sindex = t_index;
726f6388
JA
9174
9175 /* If the process substitution specification is `<()', we want to
9176 open the pipe for writing in the child and produce output; if
9177 it is `>()', we want to open the pipe for reading in the child
9178 and consume input. */
ccc6cda3 9179 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
726f6388
JA
9180
9181 FREE (temp1);
9182
9183 goto dollar_add_string;
9184 }
9185#endif /* PROCESS_SUBSTITUTION */
9186
95732b49
JA
9187 case '=':
9188 /* Posix.2 section 3.6.1 says that tildes following `=' in words
9189 which are not assignment statements are not expanded. If the
9190 shell isn't in posix mode, though, we perform tilde expansion
9191 on `likely candidate' unquoted assignment statements (flags
9192 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
9193 contains an unquoted :~ or =~. Something to think about: we
9194 now have a flag that says to perform tilde expansion on arguments
9195 to `assignment builtins' like declare and export that look like
9196 assignment statements. We now do tilde expansion on such words
9197 even in POSIX mode. */
9198 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
17345e5a 9199 {
0001803f 9200 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9201 goto add_ifs_character;
9202 else
9203 goto add_character;
9204 }
95732b49
JA
9205 /* If we're not in posix mode or forcing assignment-statement tilde
9206 expansion, note where the `=' appears in the word and prepare to
9207 do tilde expansion following the first `='. */
9208 if ((word->flags & W_ASSIGNMENT) &&
9209 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9210 assignoff == -1 && sindex > 0)
9211 assignoff = sindex;
9212 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
9213 word->flags |= W_ITILDE;
9214#if 0
9215 else if ((word->flags & W_ASSIGNMENT) &&
9216 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9217 string[sindex+1] == '~')
9218 word->flags |= W_ITILDE;
9219#endif
a0c0a00f
CR
9220
9221 /* XXX - bash-4.4/bash-5.0 */
9222 if (word->flags & W_ASSIGNARG)
9223 word->flags |= W_ASSIGNRHS; /* affects $@ */
9224
0001803f 9225 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9226 goto add_ifs_character;
9227 else
9228 goto add_character;
95732b49
JA
9229
9230 case ':':
9231 if (word->flags & W_NOTILDE)
17345e5a 9232 {
0001803f 9233 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9234 goto add_ifs_character;
9235 else
9236 goto add_character;
9237 }
95732b49
JA
9238
9239 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
9240 string[sindex+1] == '~')
9241 word->flags |= W_ITILDE;
17345e5a 9242
0001803f 9243 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9244 goto add_ifs_character;
9245 else
9246 goto add_character;
95732b49
JA
9247
9248 case '~':
9249 /* If the word isn't supposed to be tilde expanded, or we're not
9250 at the start of a word or after an unquoted : or = in an
a0c0a00f
CR
9251 assignment statement, we don't do tilde expansion. If we don't want
9252 tilde expansion when expanding words to be passed to the arithmetic
9253 evaluator, remove the check for Q_ARITH. */
95732b49
JA
9254 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
9255 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
a0c0a00f 9256 ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0)))
95732b49
JA
9257 {
9258 word->flags &= ~W_ITILDE;
0001803f 9259 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
17345e5a
JA
9260 goto add_ifs_character;
9261 else
9262 goto add_character;
95732b49
JA
9263 }
9264
9265 if (word->flags & W_ASSIGNRHS)
9266 tflag = 2;
9267 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
9268 tflag = 1;
9269 else
9270 tflag = 0;
9271
9272 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
9273
9274 word->flags &= ~W_ITILDE;
9275
9276 if (temp && *temp && t_index > 0)
9277 {
9278 temp1 = bash_tilde_expand (temp, tflag);
0628567a
JA
9279 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
9280 {
9281 FREE (temp);
9282 FREE (temp1);
9283 goto add_character; /* tilde expansion failed */
9284 }
95732b49
JA
9285 free (temp);
9286 temp = temp1;
9287 sindex += t_index;
3185942a 9288 goto add_quoted_string; /* XXX was add_string */
95732b49
JA
9289 }
9290 else
9291 {
9292 FREE (temp);
9293 goto add_character;
9294 }
9295
726f6388 9296 case '$':
726f6388
JA
9297 if (expanded_something)
9298 *expanded_something = 1;
9299
0b913689 9300 temp_has_dollar_at = 0;
0001803f
CR
9301 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
9302 if (word->flags & W_NOSPLIT2)
9303 pflags |= PF_NOSPLIT2;
ac50fbac
CR
9304 if (word->flags & W_ASSIGNRHS)
9305 pflags |= PF_ASSIGNRHS;
a0c0a00f
CR
9306 if (word->flags & W_COMPLETE)
9307 pflags |= PF_COMPLETE;
95732b49 9308 tword = param_expand (string, &sindex, quoted, expanded_something,
0b913689 9309 &temp_has_dollar_at, &quoted_dollar_at,
0001803f 9310 &had_quoted_null, pflags);
0b913689 9311 has_dollar_at += temp_has_dollar_at;
ac50fbac 9312 split_on_spaces += (tword->flags & W_SPLITSPACE);
726f6388 9313
95732b49 9314 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
726f6388 9315 {
cce855bc
JA
9316 free (string);
9317 free (istring);
95732b49
JA
9318 return ((tword == &expand_wdesc_error) ? &expand_word_error
9319 : &expand_word_fatal);
cce855bc
JA
9320 }
9321 if (contains_dollar_at && has_dollar_at)
9322 *contains_dollar_at = 1;
95732b49
JA
9323
9324 if (tword && (tword->flags & W_HASQUOTEDNULL))
9325 had_quoted_null = 1;
9326
ac50fbac 9327 temp = tword ? tword->word : (char *)NULL;
95732b49
JA
9328 dispose_word_desc (tword);
9329
a601c749
CR
9330 /* Kill quoted nulls; we will add them back at the end of
9331 expand_word_internal if nothing else in the string */
9332 if (had_quoted_null && temp && QUOTED_NULL (temp))
9333 {
9334 FREE (temp);
9335 temp = (char *)NULL;
9336 }
9337
cce855bc
JA
9338 goto add_string;
9339 break;
726f6388 9340
cce855bc
JA
9341 case '`': /* Backquoted command substitution. */
9342 {
b80f6443 9343 t_index = sindex++;
726f6388 9344
3185942a 9345 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
95732b49
JA
9346 /* The test of sindex against t_index is to allow bare instances of
9347 ` to pass through, for backwards compatibility. */
9348 if (temp == &extract_string_error || temp == &extract_string_fatal)
9349 {
9350 if (sindex - 1 == t_index)
9351 {
9352 sindex = t_index;
9353 goto add_character;
9354 }
ac50fbac 9355 last_command_exit_value = EXECUTION_FAILURE;
3185942a 9356 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
95732b49
JA
9357 free (string);
9358 free (istring);
9359 return ((temp == &extract_string_error) ? &expand_word_error
9360 : &expand_word_fatal);
9361 }
9362
cce855bc
JA
9363 if (expanded_something)
9364 *expanded_something = 1;
726f6388 9365
b80f6443
JA
9366 if (word->flags & W_NOCOMSUB)
9367 /* sindex + 1 because string[sindex] == '`' */
9368 temp1 = substring (string, t_index, sindex + 1);
9369 else
9370 {
9371 de_backslash (temp);
3185942a
JA
9372 tword = command_substitute (temp, quoted);
9373 temp1 = tword ? tword->word : (char *)NULL;
9374 if (tword)
9375 dispose_word_desc (tword);
b80f6443 9376 }
cce855bc
JA
9377 FREE (temp);
9378 temp = temp1;
9379 goto dollar_add_string;
9380 }
ccc6cda3 9381
cce855bc
JA
9382 case '\\':
9383 if (string[sindex + 1] == '\n')
9384 {
9385 sindex += 2;
9386 continue;
9387 }
726f6388 9388
cce855bc 9389 c = string[++sindex];
726f6388 9390
cce855bc 9391 if (quoted & Q_HERE_DOCUMENT)
28ef6c31 9392 tflag = CBSHDOC;
cce855bc 9393 else if (quoted & Q_DOUBLE_QUOTES)
28ef6c31 9394 tflag = CBSDQUOTE;
cce855bc 9395 else
28ef6c31
JA
9396 tflag = 0;
9397
495aee44
CR
9398 /* From Posix discussion on austin-group list: Backslash escaping
9399 a } in ${...} is removed. Issue 0000221 */
9400 if ((quoted & Q_DOLBRACE) && c == RBRACE)
9401 {
ac50fbac
CR
9402 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9403 }
9404 /* This is the fix for " $@\ " */
9405 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
9406 {
9407 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9408 DEFAULT_ARRAY_SIZE);
9409 istring[istring_index++] = CTLESC;
9410 istring[istring_index++] = '\\';
9411 istring[istring_index] = '\0';
9412
495aee44
CR
9413 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9414 }
9415 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
cce855bc 9416 {
7117c2d2 9417 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
bb70624e
JA
9418 }
9419 else if (c == 0)
9420 {
9421 c = CTLNUL;
9422 sindex--; /* add_character: label increments sindex */
9423 goto add_character;
cce855bc
JA
9424 }
9425 else
bb70624e 9426 {
7117c2d2 9427 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
bb70624e 9428 }
726f6388 9429
bb70624e
JA
9430 sindex++;
9431add_twochars:
9432 /* BEFORE jumping here, we need to increment sindex if appropriate */
9433 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9434 DEFAULT_ARRAY_SIZE);
9435 istring[istring_index++] = twochars[0];
9436 istring[istring_index++] = twochars[1];
9437 istring[istring_index] = '\0';
9438
9439 break;
726f6388 9440
cce855bc 9441 case '"':
a0c0a00f 9442 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
cce855bc 9443 goto add_character;
ccc6cda3
JA
9444
9445 t_index = ++sindex;
a0c0a00f 9446 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
ccc6cda3
JA
9447
9448 /* If the quotes surrounded the entire string, then the
9449 whole word was quoted. */
9450 quoted_state = (t_index == 1 && string[sindex] == '\0')
9451 ? WHOLLY_QUOTED
7117c2d2 9452 : PARTIALLY_QUOTED;
ccc6cda3
JA
9453
9454 if (temp && *temp)
726f6388 9455 {
95732b49
JA
9456 tword = alloc_word_desc ();
9457 tword->word = temp;
9458
a0c0a00f
CR
9459 /* XXX - bash-4.4/bash-5.0 */
9460 if (word->flags & W_ASSIGNARG)
9461 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
9462 if (word->flags & W_COMPLETE)
9463 tword->flags |= W_COMPLETE; /* for command substitutions */
9464
ccc6cda3
JA
9465 temp = (char *)NULL;
9466
0b913689 9467 temp_has_dollar_at = 0; /* XXX */
95732b49 9468 /* Need to get W_HASQUOTEDNULL flag through this function. */
0b913689
CR
9469 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
9470 has_dollar_at += temp_has_dollar_at;
726f6388 9471
ccc6cda3
JA
9472 if (list == &expand_word_error || list == &expand_word_fatal)
9473 {
9474 free (istring);
9475 free (string);
9476 /* expand_word_internal has already freed temp_word->word
9477 for us because of the way it prints error messages. */
9478 tword->word = (char *)NULL;
9479 dispose_word (tword);
9480 return list;
9481 }
726f6388 9482
ccc6cda3 9483 dispose_word (tword);
726f6388 9484
ccc6cda3
JA
9485 /* "$@" (a double-quoted dollar-at) expands into nothing,
9486 not even a NULL word, when there are no positional
9487 parameters. */
a0c0a00f 9488 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
726f6388 9489 {
ccc6cda3
JA
9490 quoted_dollar_at++;
9491 break;
9492 }
9493
9494 /* If we get "$@", we know we have expanded something, so we
9495 need to remember it for the final split on $IFS. This is
9496 a special case; it's the only case where a quoted string
9497 can expand into more than one word. It's going to come back
9498 from the above call to expand_word_internal as a list with
9499 a single word, in which all characters are quoted and
9500 separated by blanks. What we want to do is to turn it back
9501 into a list for the next piece of code. */
9502 if (list)
9503 dequote_list (list);
9504
95732b49 9505 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
ac50fbac 9506 had_quoted_null = 1; /* XXX */
95732b49 9507
a0c0a00f 9508 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
ccc6cda3
JA
9509 {
9510 quoted_dollar_at++;
9511 if (contains_dollar_at)
9512 *contains_dollar_at = 1;
9513 if (expanded_something)
9514 *expanded_something = 1;
9515 }
9516 }
9517 else
9518 {
9519 /* What we have is "". This is a minor optimization. */
f73dda09 9520 FREE (temp);
ccc6cda3
JA
9521 list = (WORD_LIST *)NULL;
9522 }
9523
9524 /* The code above *might* return a list (consider the case of "$@",
9525 where it returns "$1", "$2", etc.). We can't throw away the
9526 rest of the list, and we have to make sure each word gets added
9527 as quoted. We test on tresult->next: if it is non-NULL, we
9528 quote the whole list, save it to a string with string_list, and
9529 add that string. We don't need to quote the results of this
9530 (and it would be wrong, since that would quote the separators
9531 as well), so we go directly to add_string. */
9532 if (list)
9533 {
9534 if (list->next)
9535 {
bc4cd23c
JA
9536 /* Testing quoted_dollar_at makes sure that "$@" is
9537 split correctly when $IFS does not contain a space. */
9538 temp = quoted_dollar_at
a0c0a00f 9539 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
bc4cd23c 9540 : string_list (quote_list (list));
ccc6cda3 9541 dispose_words (list);
726f6388
JA
9542 goto add_string;
9543 }
9544 else
9545 {
ccc6cda3 9546 temp = savestring (list->word->word);
95732b49 9547 tflag = list->word->flags;
ccc6cda3 9548 dispose_words (list);
95732b49 9549
cce855bc
JA
9550 /* If the string is not a quoted null string, we want
9551 to remove any embedded unquoted CTLNUL characters.
9552 We do not want to turn quoted null strings back into
9553 the empty string, though. We do this because we
9554 want to remove any quoted nulls from expansions that
9555 contain other characters. For example, if we have
9556 x"$*"y or "x$*y" and there are no positional parameters,
7117c2d2 9557 the $* should expand into nothing. */
95732b49
JA
9558 /* We use the W_HASQUOTEDNULL flag to differentiate the
9559 cases: a quoted null character as above and when
9560 CTLNUL is contained in the (non-null) expansion
9561 of some variable. We use the had_quoted_null flag to
9562 pass the value through this function to its caller. */
9563 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
cce855bc 9564 remove_quoted_nulls (temp); /* XXX */
726f6388
JA
9565 }
9566 }
ccc6cda3
JA
9567 else
9568 temp = (char *)NULL;
726f6388 9569
ccc6cda3 9570 /* We do not want to add quoted nulls to strings that are only
ac50fbac
CR
9571 partially quoted; we can throw them away. The exception to
9572 this is when we are going to be performing word splitting,
9573 since we have to preserve a null argument if the next character
9574 will cause word splitting. */
495aee44 9575 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
cce855bc 9576 continue;
726f6388 9577
ccc6cda3 9578 add_quoted_string:
726f6388 9579
ccc6cda3
JA
9580 if (temp)
9581 {
9582 temp1 = temp;
9583 temp = quote_string (temp);
9584 free (temp1);
bb70624e 9585 goto add_string;
ccc6cda3
JA
9586 }
9587 else
9588 {
9589 /* Add NULL arg. */
bb70624e
JA
9590 c = CTLNUL;
9591 sindex--; /* add_character: label increments sindex */
9592 goto add_character;
ccc6cda3 9593 }
bb70624e 9594
ccc6cda3 9595 /* break; */
726f6388 9596
ccc6cda3 9597 case '\'':
95732b49 9598 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
ccc6cda3 9599 goto add_character;
726f6388 9600
ccc6cda3
JA
9601 t_index = ++sindex;
9602 temp = string_extract_single_quoted (string, &sindex);
726f6388 9603
ccc6cda3
JA
9604 /* If the entire STRING was surrounded by single quotes,
9605 then the string is wholly quoted. */
9606 quoted_state = (t_index == 1 && string[sindex] == '\0')
9607 ? WHOLLY_QUOTED
7117c2d2 9608 : PARTIALLY_QUOTED;
726f6388 9609
ccc6cda3
JA
9610 /* If all we had was '', it is a null expansion. */
9611 if (*temp == '\0')
9612 {
9613 free (temp);
9614 temp = (char *)NULL;
9615 }
9616 else
7117c2d2 9617 remove_quoted_escapes (temp); /* ??? */
726f6388 9618
ccc6cda3 9619 /* We do not want to add quoted nulls to strings that are only
a0c0a00f
CR
9620 partially quoted; such nulls are discarded. See above for the
9621 exception, which is when the string is going to be split. */
9622 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
ccc6cda3 9623 continue;
726f6388 9624
bb70624e
JA
9625 /* If we have a quoted null expansion, add a quoted NULL to istring. */
9626 if (temp == 0)
9627 {
9628 c = CTLNUL;
9629 sindex--; /* add_character: label increments sindex */
9630 goto add_character;
9631 }
9632 else
9633 goto add_quoted_string;
9634
ccc6cda3 9635 /* break; */
726f6388
JA
9636
9637 default:
726f6388 9638 /* This is the fix for " $@ " */
17345e5a 9639 add_ifs_character:
a0c0a00f 9640 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
726f6388 9641 {
bb70624e
JA
9642 if (string[sindex]) /* from old goto dollar_add_string */
9643 sindex++;
9644 if (c == 0)
9645 {
9646 c = CTLNUL;
9647 goto add_character;
9648 }
9649 else
9650 {
7117c2d2 9651#if HANDLE_MULTIBYTE
a0c0a00f 9652 if (mb_cur_max > 1)
b80f6443
JA
9653 sindex--;
9654
a0c0a00f 9655 if (mb_cur_max > 1)
7117c2d2 9656 {
b80f6443 9657 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
9658 }
9659 else
9660#endif
9661 {
9662 twochars[0] = CTLESC;
9663 twochars[1] = c;
9664 goto add_twochars;
9665 }
bb70624e 9666 }
726f6388
JA
9667 }
9668
7117c2d2
JA
9669 SADD_MBCHAR (temp, string, sindex, string_size);
9670
726f6388 9671 add_character:
ccc6cda3
JA
9672 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
9673 DEFAULT_ARRAY_SIZE);
726f6388
JA
9674 istring[istring_index++] = c;
9675 istring[istring_index] = '\0';
9676
9677 /* Next character. */
9678 sindex++;
9679 }
9680 }
9681
9682finished_with_string:
726f6388
JA
9683 /* OK, we're ready to return. If we have a quoted string, and
9684 quoted_dollar_at is not set, we do no splitting at all; otherwise
9685 we split on ' '. The routines that call this will handle what to
9686 do if nothing has been expanded. */
ccc6cda3
JA
9687
9688 /* Partially and wholly quoted strings which expand to the empty
9689 string are retained as an empty arguments. Unquoted strings
9690 which expand to the empty string are discarded. The single
9691 exception is the case of expanding "$@" when there are no
9692 positional parameters. In that case, we discard the expansion. */
9693
9694 /* Because of how the code that handles "" and '' in partially
9695 quoted strings works, we need to make ISTRING into a QUOTED_NULL
9696 if we saw quoting characters, but the expansion was empty.
9697 "" and '' are tossed away before we get to this point when
9698 processing partially quoted strings. This makes "" and $xxx""
9699 equivalent when xxx is unset. We also look to see whether we
9700 saw a quoted null from a ${} expansion and add one back if we
9701 need to. */
9702
9703 /* If we expand to nothing and there were no single or double quotes
9704 in the word, we throw it away. Otherwise, we return a NULL word.
9705 The single exception is for $@ surrounded by double quotes when
9706 there are no positional parameters. In that case, we also throw
9707 the word away. */
9708
9709 if (*istring == '\0')
9710 {
9711 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
726f6388 9712 {
726f6388
JA
9713 istring[0] = CTLNUL;
9714 istring[1] = '\0';
ccc6cda3 9715 tword = make_bare_word (istring);
95732b49 9716 tword->flags |= W_HASQUOTEDNULL; /* XXX */
ccc6cda3
JA
9717 list = make_word_list (tword, (WORD_LIST *)NULL);
9718 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9719 tword->flags |= W_QUOTED;
726f6388 9720 }
ccc6cda3
JA
9721 /* According to sh, ksh, and Posix.2, if a word expands into nothing
9722 and a double-quoted "$@" appears anywhere in it, then the entire
9723 word is removed. */
a0c0a00f
CR
9724 /* XXX - exception appears to be that quoted null strings result in
9725 null arguments */
ccc6cda3
JA
9726 else if (quoted_state == UNQUOTED || quoted_dollar_at)
9727 list = (WORD_LIST *)NULL;
9728#if 0
9729 else
726f6388 9730 {
ccc6cda3 9731 tword = make_bare_word (istring);
ccc6cda3
JA
9732 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9733 tword->flags |= W_QUOTED;
95732b49 9734 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 9735 }
f73dda09
JA
9736#else
9737 else
9738 list = (WORD_LIST *)NULL;
ccc6cda3
JA
9739#endif
9740 }
9741 else if (word->flags & W_NOSPLIT)
9742 {
9743 tword = make_bare_word (istring);
ccc6cda3
JA
9744 if (word->flags & W_ASSIGNMENT)
9745 tword->flags |= W_ASSIGNMENT; /* XXX */
95732b49
JA
9746 if (word->flags & W_COMPASSIGN)
9747 tword->flags |= W_COMPASSIGN; /* XXX */
b72432fd
JA
9748 if (word->flags & W_NOGLOB)
9749 tword->flags |= W_NOGLOB; /* XXX */
ac50fbac
CR
9750 if (word->flags & W_NOBRACE)
9751 tword->flags |= W_NOBRACE; /* XXX */
95732b49
JA
9752 if (word->flags & W_NOEXPAND)
9753 tword->flags |= W_NOEXPAND; /* XXX */
ccc6cda3 9754 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
28ef6c31 9755 tword->flags |= W_QUOTED;
a601c749 9756 if (had_quoted_null && QUOTED_NULL (istring))
95732b49
JA
9757 tword->flags |= W_HASQUOTEDNULL;
9758 list = make_word_list (tword, (WORD_LIST *)NULL);
ccc6cda3
JA
9759 }
9760 else
9761 {
9762 char *ifs_chars;
3b34f6e6 9763 char *tstring;
ccc6cda3 9764
7117c2d2 9765 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
726f6388 9766
cce855bc
JA
9767 /* If we have $@, we need to split the results no matter what. If
9768 IFS is unset or NULL, string_list_dollar_at has separated the
9769 positional parameters with a space, so we split on space (we have
9770 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
9771 string_list_dollar_at has separated the positional parameters
ac50fbac
CR
9772 with the first character of $IFS, so we split on $IFS. If
9773 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
9774 unset or null, and we want to make sure that we split on spaces
a0c0a00f
CR
9775 regardless of what else has happened to IFS since the expansion,
9776 or we expanded "$@" with IFS null and we need to split the positional
9777 parameters into separate words. */
ac50fbac
CR
9778 if (split_on_spaces)
9779 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
3b34f6e6
CR
9780 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
9781 don't want to split the result (W_NOSPLIT2), and we are not quoted,
9782 we have already separated the arguments with the first character of
9783 $IFS. In this case, we want to return a list with a single word
9784 with the separator possibly replaced with a space (it's what other
9785 shells seem to do).
9786 quoted_dollar_at is internal to this function and is set if we are
9787 passed an argument that is unquoted (quoted == 0) but we encounter a
9788 double-quoted $@ while expanding it. */
9789 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
9790 {
9791 /* Only split and rejoin if we have to */
9792 if (*ifs_chars && *ifs_chars != ' ')
9793 {
9794 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
9795 tstring = string_list (list);
9796 }
9797 else
9798 tstring = istring;
9799 tword = make_bare_word (tstring);
9800 if (tstring != istring)
9801 free (tstring);
9802 goto set_word_flags;
9803 }
a0c0a00f
CR
9804 /* This is the attempt to make $* in an assignment context (a=$*) and
9805 array variables subscripted with * in an assignment context (a=${foo[*]})
9806 behave similarly. It has side effects that, though they increase
9807 compatibility with other shells, are not backwards compatible. */
9808#if 0
9809 else if (has_dollar_at && quoted == 0 && ifs_chars && (word->flags & W_ASSIGNRHS))
9810 {
9811 tword = make_bare_word (istring);
9812 goto set_word_flags;
9813 }
9814#endif
ac50fbac 9815 else if (has_dollar_at && ifs_chars)
cce855bc 9816 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
ccc6cda3
JA
9817 else
9818 {
9819 tword = make_bare_word (istring);
3b34f6e6 9820set_word_flags:
ccc6cda3
JA
9821 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
9822 tword->flags |= W_QUOTED;
9823 if (word->flags & W_ASSIGNMENT)
9824 tword->flags |= W_ASSIGNMENT;
95732b49
JA
9825 if (word->flags & W_COMPASSIGN)
9826 tword->flags |= W_COMPASSIGN;
b72432fd
JA
9827 if (word->flags & W_NOGLOB)
9828 tword->flags |= W_NOGLOB;
ac50fbac
CR
9829 if (word->flags & W_NOBRACE)
9830 tword->flags |= W_NOBRACE;
95732b49
JA
9831 if (word->flags & W_NOEXPAND)
9832 tword->flags |= W_NOEXPAND;
a601c749 9833 if (had_quoted_null && QUOTED_NULL (istring))
95732b49
JA
9834 tword->flags |= W_HASQUOTEDNULL; /* XXX */
9835 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 9836 }
726f6388 9837 }
726f6388 9838
ccc6cda3
JA
9839 free (istring);
9840 return (list);
726f6388
JA
9841}
9842
9843/* **************************************************************** */
9844/* */
9845/* Functions for Quote Removal */
9846/* */
9847/* **************************************************************** */
9848
9849/* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
7117c2d2 9850 backslash quoting rules for within double quotes or a here document. */
726f6388
JA
9851char *
9852string_quote_removal (string, quoted)
9853 char *string;
9854 int quoted;
9855{
7117c2d2
JA
9856 size_t slen;
9857 char *r, *result_string, *temp, *send;
f73dda09
JA
9858 int sindex, tindex, dquote;
9859 unsigned char c;
7117c2d2 9860 DECLARE_MBSTATE;
726f6388
JA
9861
9862 /* The result can be no longer than the original string. */
7117c2d2
JA
9863 slen = strlen (string);
9864 send = string + slen;
9865
9866 r = result_string = (char *)xmalloc (slen + 1);
726f6388 9867
ccc6cda3 9868 for (dquote = sindex = 0; c = string[sindex];)
726f6388
JA
9869 {
9870 switch (c)
9871 {
9872 case '\\':
9873 c = string[++sindex];
3185942a
JA
9874 if (c == 0)
9875 {
9876 *r++ = '\\';
9877 break;
9878 }
28ef6c31 9879 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
726f6388 9880 *r++ = '\\';
ccc6cda3 9881 /* FALLTHROUGH */
726f6388
JA
9882
9883 default:
7117c2d2 9884 SCOPY_CHAR_M (r, string, send, sindex);
726f6388
JA
9885 break;
9886
9887 case '\'':
ccc6cda3 9888 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
726f6388
JA
9889 {
9890 *r++ = c;
9891 sindex++;
ccc6cda3 9892 break;
726f6388 9893 }
ccc6cda3
JA
9894 tindex = sindex + 1;
9895 temp = string_extract_single_quoted (string, &tindex);
9896 if (temp)
726f6388 9897 {
ccc6cda3
JA
9898 strcpy (r, temp);
9899 r += strlen (r);
9900 free (temp);
726f6388 9901 }
ccc6cda3 9902 sindex = tindex;
726f6388
JA
9903 break;
9904
9905 case '"':
9906 dquote = 1 - dquote;
9907 sindex++;
9908 break;
9909 }
9910 }
9911 *r = '\0';
9912 return (result_string);
9913}
9914
ccc6cda3
JA
9915#if 0
9916/* UNUSED */
726f6388
JA
9917/* Perform quote removal on word WORD. This allocates and returns a new
9918 WORD_DESC *. */
9919WORD_DESC *
9920word_quote_removal (word, quoted)
9921 WORD_DESC *word;
9922 int quoted;
9923{
9924 WORD_DESC *w;
9925 char *t;
9926
9927 t = string_quote_removal (word->word, quoted);
95732b49
JA
9928 w = alloc_word_desc ();
9929 w->word = t ? t : savestring ("");
726f6388
JA
9930 return (w);
9931}
9932
9933/* Perform quote removal on all words in LIST. If QUOTED is non-zero,
9934 the members of the list are treated as if they are surrounded by
9935 double quotes. Return a new list, or NULL if LIST is NULL. */
9936WORD_LIST *
9937word_list_quote_removal (list, quoted)
9938 WORD_LIST *list;
9939 int quoted;
9940{
95732b49 9941 WORD_LIST *result, *t, *tresult, *e;
726f6388 9942
ccc6cda3 9943 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 9944 {
7117c2d2 9945 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
95732b49 9946#if 0
726f6388 9947 result = (WORD_LIST *) list_append (result, tresult);
95732b49
JA
9948#else
9949 if (result == 0)
9950 result = e = tresult;
9951 else
9952 {
9953 e->next = tresult;
9954 while (e->next)
9955 e = e->next;
9956 }
9957#endif
726f6388
JA
9958 }
9959 return (result);
9960}
ccc6cda3 9961#endif
726f6388 9962
726f6388
JA
9963/*******************************************
9964 * *
9965 * Functions to perform word splitting *
9966 * *
9967 *******************************************/
9968
7117c2d2
JA
9969void
9970setifs (v)
9971 SHELL_VAR *v;
b72432fd 9972{
7117c2d2
JA
9973 char *t;
9974 unsigned char uc;
9975
9976 ifs_var = v;
95732b49 9977 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
b72432fd 9978
ac50fbac
CR
9979 ifs_is_set = ifs_var != 0;
9980 ifs_is_null = ifs_is_set && (*ifs_value == 0);
9981
95732b49
JA
9982 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9983 handle multibyte chars in IFS */
7117c2d2
JA
9984 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9985 for (t = ifs_value ; t && *t; t++)
9986 {
9987 uc = *t;
9988 ifs_cmap[uc] = 1;
9989 }
9990
95732b49
JA
9991#if defined (HANDLE_MULTIBYTE)
9992 if (ifs_value == 0)
9993 {
9994 ifs_firstc[0] = '\0';
9995 ifs_firstc_len = 1;
9996 }
9997 else
9998 {
9999 size_t ifs_len;
10000 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
10001 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
10002 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
10003 {
10004 ifs_firstc[0] = ifs_value[0];
10005 ifs_firstc[1] = '\0';
10006 ifs_firstc_len = 1;
10007 }
10008 else
10009 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
10010 }
10011#else
7117c2d2 10012 ifs_firstc = ifs_value ? *ifs_value : 0;
95732b49 10013#endif
7117c2d2
JA
10014}
10015
10016char *
10017getifs ()
10018{
10019 return ifs_value;
b72432fd
JA
10020}
10021
726f6388
JA
10022/* This splits a single word into a WORD LIST on $IFS, but only if the word
10023 is not quoted. list_string () performs quote removal for us, even if we
10024 don't do any splitting. */
10025WORD_LIST *
7117c2d2 10026word_split (w, ifs_chars)
726f6388 10027 WORD_DESC *w;
7117c2d2 10028 char *ifs_chars;
726f6388
JA
10029{
10030 WORD_LIST *result;
10031
10032 if (w)
10033 {
7117c2d2 10034 char *xifs;
726f6388 10035
7117c2d2
JA
10036 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
10037 result = list_string (w->word, xifs, w->flags & W_QUOTED);
726f6388
JA
10038 }
10039 else
10040 result = (WORD_LIST *)NULL;
ccc6cda3 10041
726f6388
JA
10042 return (result);
10043}
10044
10045/* Perform word splitting on LIST and return the RESULT. It is possible
10046 to return (WORD_LIST *)NULL. */
10047static WORD_LIST *
10048word_list_split (list)
10049 WORD_LIST *list;
10050{
95732b49 10051 WORD_LIST *result, *t, *tresult, *e;
726f6388 10052
ccc6cda3 10053 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 10054 {
7117c2d2 10055 tresult = word_split (t->word, ifs_value);
95732b49
JA
10056 if (result == 0)
10057 result = e = tresult;
10058 else
10059 {
10060 e->next = tresult;
10061 while (e->next)
10062 e = e->next;
10063 }
726f6388
JA
10064 }
10065 return (result);
10066}
10067
10068/**************************************************
10069 * *
cce855bc 10070 * Functions to expand an entire WORD_LIST *
726f6388
JA
10071 * *
10072 **************************************************/
10073
b80f6443
JA
10074/* Do any word-expansion-specific cleanup and jump to top_level */
10075static void
10076exp_jump_to_top_level (v)
10077 int v;
10078{
3185942a
JA
10079 set_pipestatus_from_exit (last_command_exit_value);
10080
b80f6443
JA
10081 /* Cleanup code goes here. */
10082 expand_no_split_dollar_star = 0; /* XXX */
10083 expanding_redir = 0;
3185942a 10084 assigning_in_environment = 0;
b80f6443 10085
f1be666c
JA
10086 if (parse_and_execute_level == 0)
10087 top_level_cleanup (); /* from sig.c */
10088
b80f6443
JA
10089 jump_to_top_level (v);
10090}
10091
cce855bc
JA
10092/* Put NLIST (which is a WORD_LIST * of only one element) at the front of
10093 ELIST, and set ELIST to the new list. */
10094#define PREPEND_LIST(nlist, elist) \
10095 do { nlist->next = elist; elist = nlist; } while (0)
10096
726f6388
JA
10097/* Separate out any initial variable assignments from TLIST. If set -k has
10098 been executed, remove all assignment statements from TLIST. Initial
10099 variable assignments and other environment assignments are placed
bb70624e 10100 on SUBST_ASSIGN_VARLIST. */
726f6388
JA
10101static WORD_LIST *
10102separate_out_assignments (tlist)
10103 WORD_LIST *tlist;
10104{
10105 register WORD_LIST *vp, *lp;
10106
0001803f 10107 if (tlist == 0)
726f6388
JA
10108 return ((WORD_LIST *)NULL);
10109
bb70624e
JA
10110 if (subst_assign_varlist)
10111 dispose_words (subst_assign_varlist); /* Clean up after previous error */
b72432fd 10112
bb70624e 10113 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
10114 vp = lp = tlist;
10115
10116 /* Separate out variable assignments at the start of the command.
10117 Loop invariant: vp->next == lp
10118 Loop postcondition:
7117c2d2
JA
10119 lp = list of words left after assignment statements skipped
10120 tlist = original list of words
726f6388 10121 */
ccc6cda3 10122 while (lp && (lp->word->flags & W_ASSIGNMENT))
726f6388
JA
10123 {
10124 vp = lp;
10125 lp = lp->next;
10126 }
10127
bb70624e
JA
10128 /* If lp != tlist, we have some initial assignment statements.
10129 We make SUBST_ASSIGN_VARLIST point to the list of assignment
10130 words and TLIST point to the remaining words. */
726f6388
JA
10131 if (lp != tlist)
10132 {
bb70624e 10133 subst_assign_varlist = tlist;
726f6388
JA
10134 /* ASSERT(vp->next == lp); */
10135 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
10136 tlist = lp; /* remainder of word list */
10137 }
10138
10139 /* vp == end of variable list */
10140 /* tlist == remainder of original word list without variable assignments */
10141 if (!tlist)
10142 /* All the words in tlist were assignment statements */
10143 return ((WORD_LIST *)NULL);
10144
10145 /* ASSERT(tlist != NULL); */
ccc6cda3 10146 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
726f6388
JA
10147
10148 /* If the -k option is in effect, we need to go through the remaining
bb70624e
JA
10149 words, separate out the assignment words, and place them on
10150 SUBST_ASSIGN_VARLIST. */
726f6388
JA
10151 if (place_keywords_in_env)
10152 {
10153 WORD_LIST *tp; /* tp == running pointer into tlist */
10154
10155 tp = tlist;
10156 lp = tlist->next;
10157
10158 /* Loop Invariant: tp->next == lp */
10159 /* Loop postcondition: tlist == word list without assignment statements */
10160 while (lp)
10161 {
ccc6cda3 10162 if (lp->word->flags & W_ASSIGNMENT)
726f6388
JA
10163 {
10164 /* Found an assignment statement, add this word to end of
bb70624e
JA
10165 subst_assign_varlist (vp). */
10166 if (!subst_assign_varlist)
10167 subst_assign_varlist = vp = lp;
726f6388
JA
10168 else
10169 {
10170 vp->next = lp;
10171 vp = lp;
10172 }
10173
10174 /* Remove the word pointed to by LP from TLIST. */
10175 tp->next = lp->next;
10176 /* ASSERT(vp == lp); */
10177 lp->next = (WORD_LIST *)NULL;
10178 lp = tp->next;
10179 }
10180 else
10181 {
10182 tp = lp;
10183 lp = lp->next;
10184 }
10185 }
10186 }
10187 return (tlist);
10188}
10189
cce855bc
JA
10190#define WEXP_VARASSIGN 0x001
10191#define WEXP_BRACEEXP 0x002
10192#define WEXP_TILDEEXP 0x004
10193#define WEXP_PARAMEXP 0x008
10194#define WEXP_PATHEXP 0x010
10195
10196/* All of the expansions, including variable assignments at the start of
10197 the list. */
10198#define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10199
10200/* All of the expansions except variable assignments at the start of
10201 the list. */
10202#define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10203
10204/* All of the `shell expansions': brace expansion, tilde expansion, parameter
10205 expansion, command substitution, arithmetic expansion, word splitting, and
10206 quote removal. */
10207#define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
10208
726f6388
JA
10209/* Take the list of words in LIST and do the various substitutions. Return
10210 a new list of words which is the expanded list, and without things like
10211 variable assignments. */
10212
10213WORD_LIST *
10214expand_words (list)
10215 WORD_LIST *list;
10216{
cce855bc 10217 return (expand_word_list_internal (list, WEXP_ALL));
726f6388
JA
10218}
10219
10220/* Same as expand_words (), but doesn't hack variable or environment
10221 variables. */
10222WORD_LIST *
10223expand_words_no_vars (list)
10224 WORD_LIST *list;
10225{
cce855bc 10226 return (expand_word_list_internal (list, WEXP_NOVARS));
726f6388
JA
10227}
10228
cce855bc
JA
10229WORD_LIST *
10230expand_words_shellexp (list)
726f6388 10231 WORD_LIST *list;
726f6388 10232{
cce855bc
JA
10233 return (expand_word_list_internal (list, WEXP_SHELLEXP));
10234}
726f6388 10235
cce855bc
JA
10236static WORD_LIST *
10237glob_expand_word_list (tlist, eflags)
10238 WORD_LIST *tlist;
10239 int eflags;
10240{
10241 char **glob_array, *temp_string;
10242 register int glob_index;
10243 WORD_LIST *glob_list, *output_list, *disposables, *next;
10244 WORD_DESC *tword;
726f6388 10245
cce855bc
JA
10246 output_list = disposables = (WORD_LIST *)NULL;
10247 glob_array = (char **)NULL;
10248 while (tlist)
10249 {
10250 /* For each word, either globbing is attempted or the word is
10251 added to orig_list. If globbing succeeds, the results are
10252 added to orig_list and the word (tlist) is added to the list
10253 of disposable words. If globbing fails and failed glob
10254 expansions are left unchanged (the shell default), the
10255 original word is added to orig_list. If globbing fails and
10256 failed glob expansions are removed, the original word is
10257 added to the list of disposable words. orig_list ends up
7117c2d2 10258 in reverse order and requires a call to REVERSE_LIST to
cce855bc
JA
10259 be set right. After all words are examined, the disposable
10260 words are freed. */
10261 next = tlist->next;
726f6388 10262
cce855bc 10263 /* If the word isn't an assignment and contains an unquoted
28ef6c31 10264 pattern matching character, then glob it. */
b72432fd 10265 if ((tlist->word->flags & W_NOGLOB) == 0 &&
cce855bc 10266 unquoted_glob_pattern_p (tlist->word->word))
726f6388 10267 {
cce855bc
JA
10268 glob_array = shell_glob_filename (tlist->word->word);
10269
10270 /* Handle error cases.
10271 I don't think we should report errors like "No such file
10272 or directory". However, I would like to report errors
10273 like "Read failed". */
10274
b80f6443 10275 if (glob_array == 0 || GLOB_FAILED (glob_array))
726f6388 10276 {
bb70624e 10277 glob_array = (char **)xmalloc (sizeof (char *));
cce855bc
JA
10278 glob_array[0] = (char *)NULL;
10279 }
10280
10281 /* Dequote the current word in case we have to use it. */
10282 if (glob_array[0] == NULL)
10283 {
10284 temp_string = dequote_string (tlist->word->word);
10285 free (tlist->word->word);
10286 tlist->word->word = temp_string;
10287 }
10288
10289 /* Make the array into a word list. */
10290 glob_list = (WORD_LIST *)NULL;
10291 for (glob_index = 0; glob_array[glob_index]; glob_index++)
10292 {
10293 tword = make_bare_word (glob_array[glob_index]);
cce855bc
JA
10294 glob_list = make_word_list (tword, glob_list);
10295 }
10296
10297 if (glob_list)
10298 {
10299 output_list = (WORD_LIST *)list_append (glob_list, output_list);
10300 PREPEND_LIST (tlist, disposables);
10301 }
b80f6443
JA
10302 else if (fail_glob_expansion != 0)
10303 {
ac50fbac 10304 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 10305 report_error (_("no match: %s"), tlist->word->word);
f1be666c 10306 exp_jump_to_top_level (DISCARD);
b80f6443 10307 }
cce855bc
JA
10308 else if (allow_null_glob_expansion == 0)
10309 {
10310 /* Failed glob expressions are left unchanged. */
10311 PREPEND_LIST (tlist, output_list);
10312 }
10313 else
10314 {
10315 /* Failed glob expressions are removed. */
10316 PREPEND_LIST (tlist, disposables);
726f6388 10317 }
726f6388 10318 }
cce855bc
JA
10319 else
10320 {
10321 /* Dequote the string. */
10322 temp_string = dequote_string (tlist->word->word);
10323 free (tlist->word->word);
10324 tlist->word->word = temp_string;
10325 PREPEND_LIST (tlist, output_list);
10326 }
10327
7117c2d2 10328 strvec_dispose (glob_array);
cce855bc
JA
10329 glob_array = (char **)NULL;
10330
10331 tlist = next;
726f6388
JA
10332 }
10333
cce855bc
JA
10334 if (disposables)
10335 dispose_words (disposables);
10336
10337 if (output_list)
10338 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10339
10340 return (output_list);
10341}
726f6388
JA
10342
10343#if defined (BRACE_EXPANSION)
cce855bc
JA
10344static WORD_LIST *
10345brace_expand_word_list (tlist, eflags)
10346 WORD_LIST *tlist;
10347 int eflags;
10348{
10349 register char **expansions;
10350 char *temp_string;
10351 WORD_LIST *disposables, *output_list, *next;
10352 WORD_DESC *w;
10353 int eindex;
10354
10355 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
726f6388 10356 {
cce855bc 10357 next = tlist->next;
726f6388 10358
ac50fbac
CR
10359 if (tlist->word->flags & W_NOBRACE)
10360 {
10361/*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
10362 PREPEND_LIST (tlist, output_list);
10363 continue;
10364 }
10365
0001803f
CR
10366 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10367 {
10368/*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
10369 PREPEND_LIST (tlist, output_list);
10370 continue;
10371 }
ac50fbac 10372
cce855bc
JA
10373 /* Only do brace expansion if the word has a brace character. If
10374 not, just add the word list element to BRACES and continue. In
10375 the common case, at least when running shell scripts, this will
0001803f 10376 degenerate to a bunch of calls to `mbschr', and then what is
cce855bc 10377 basically a reversal of TLIST into BRACES, which is corrected
7117c2d2 10378 by a call to REVERSE_LIST () on BRACES when the end of TLIST
cce855bc 10379 is reached. */
0001803f 10380 if (mbschr (tlist->word->word, LBRACE))
726f6388 10381 {
cce855bc 10382 expansions = brace_expand (tlist->word->word);
726f6388 10383
cce855bc 10384 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
726f6388 10385 {
ac50fbac
CR
10386 w = alloc_word_desc ();
10387 w->word = temp_string;
10388
cce855bc
JA
10389 /* If brace expansion didn't change the word, preserve
10390 the flags. We may want to preserve the flags
10391 unconditionally someday -- XXX */
10392 if (STREQ (temp_string, tlist->word->word))
10393 w->flags = tlist->word->flags;
ac50fbac
CR
10394 else
10395 w = make_word_flags (w, temp_string);
10396
cce855bc 10397 output_list = make_word_list (w, output_list);
726f6388 10398 }
cce855bc 10399 free (expansions);
726f6388 10400
cce855bc
JA
10401 /* Add TLIST to the list of words to be freed after brace
10402 expansion has been performed. */
10403 PREPEND_LIST (tlist, disposables);
10404 }
10405 else
10406 PREPEND_LIST (tlist, output_list);
726f6388 10407 }
cce855bc
JA
10408
10409 if (disposables)
10410 dispose_words (disposables);
10411
10412 if (output_list)
10413 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10414
10415 return (output_list);
10416}
10417#endif
10418
3185942a
JA
10419#if defined (ARRAY_VARS)
10420/* Take WORD, a compound associative array assignment, and internally run
10421 'declare -A w', where W is the variable name portion of WORD. */
10422static int
a0c0a00f 10423make_internal_declare (word, option, cmd)
3185942a
JA
10424 char *word;
10425 char *option;
a0c0a00f 10426 char *cmd;
3185942a 10427{
a0c0a00f 10428 int t, r;
3185942a
JA
10429 WORD_LIST *wl;
10430 WORD_DESC *w;
10431
10432 w = make_word (word);
10433
10434 t = assignment (w->word, 0);
a0c0a00f
CR
10435 if (w->word[t] == '=')
10436 {
10437 w->word[t] = '\0';
10438 if (w->word[t - 1] == '+') /* cut off any append op */
10439 w->word[t - 1] = '\0';
10440 }
3185942a
JA
10441
10442 wl = make_word_list (w, (WORD_LIST *)NULL);
10443 wl = make_word_list (make_word (option), wl);
10444
a0c0a00f
CR
10445 r = declare_builtin (wl);
10446
10447 dispose_words (wl);
10448 return r;
3185942a
JA
10449}
10450#endif
10451
cce855bc
JA
10452static WORD_LIST *
10453shell_expand_word_list (tlist, eflags)
10454 WORD_LIST *tlist;
10455 int eflags;
10456{
a0c0a00f 10457 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
cce855bc
JA
10458 int expanded_something, has_dollar_at;
10459 char *temp_string;
726f6388 10460
726f6388 10461 /* We do tilde expansion all the time. This is what 1003.2 says. */
cce855bc 10462 new_list = (WORD_LIST *)NULL;
a0c0a00f
CR
10463 for (wcmd = tlist; wcmd; wcmd = wcmd->next)
10464 if (wcmd->word->flags & W_ASSNBLTIN)
10465 break;
10466
cce855bc 10467 for (orig_list = tlist; tlist; tlist = next)
726f6388 10468 {
ccc6cda3 10469 temp_string = tlist->word->word;
726f6388
JA
10470
10471 next = tlist->next;
10472
95732b49
JA
10473#if defined (ARRAY_VARS)
10474 /* If this is a compound array assignment to a builtin that accepts
10475 such assignments (e.g., `declare'), take the assignment and perform
10476 it separately, handling the semantics of declarations inside shell
10477 functions. This avoids the double-evaluation of such arguments,
10478 because `declare' does some evaluation of compound assignments on
10479 its own. */
10480 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10481 {
10482 int t;
a0c0a00f 10483 char opts[16], opti;
ac50fbac
CR
10484
10485 opti = 0;
10486 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
10487 opts[opti++] = '-';
10488
10489 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
10490 {
10491 opts[opti++] = 'g';
10492 opts[opti++] = 'A';
10493 }
10494 else if (tlist->word->flags & W_ASSIGNASSOC)
10495 opts[opti++] = 'A';
10496 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
10497 {
10498 opts[opti++] = 'g';
10499 opts[opti++] = 'a';
10500 }
10501 else if (tlist->word->flags & W_ASSIGNARRAY)
10502 opts[opti++] = 'a';
10503 else if (tlist->word->flags & W_ASSNGLOBAL)
10504 opts[opti++] = 'g';
95732b49 10505
a0c0a00f
CR
10506 /* If we have special handling note the integer attribute and others
10507 that transform the value upon assignment. What we do is take all
10508 of the option arguments and scan through them looking for options
10509 that cause such transformations, and add them to the `opts' array. */
10510/* if (opti > 0) */
10511 {
10512 char omap[128];
10513 int oind;
10514 WORD_LIST *l;
10515
10516 memset (omap, '\0', sizeof (omap));
10517 for (l = orig_list->next; l != tlist; l = l->next)
10518 {
10519 if (l->word->word[0] != '-')
10520 break; /* non-option argument */
10521 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
10522 break; /* -- signals end of options */
10523 for (oind = 1; l->word->word[oind]; oind++)
10524 switch (l->word->word[oind])
10525 {
10526 case 'i':
10527 case 'l':
10528 case 'u':
10529 case 'c':
10530 omap[l->word->word[oind]] = 1;
10531 if (opti == 0)
10532 opts[opti++] = '-';
10533 break;
10534 default:
10535 break;
10536 }
10537 }
10538
10539 for (oind = 0; oind < sizeof (omap); oind++)
10540 if (omap[oind])
10541 opts[opti++] = oind;
10542 }
ac50fbac
CR
10543
10544 opts[opti] = '\0';
10545 if (opti > 0)
30a978b7 10546 {
a0c0a00f 10547 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
30a978b7
CR
10548 if (t != EXECUTION_SUCCESS)
10549 {
10550 last_command_exit_value = t;
10551 exp_jump_to_top_level (DISCARD);
10552 }
10553 }
3185942a 10554
495aee44 10555 t = do_word_assignment (tlist->word, 0);
95732b49
JA
10556 if (t == 0)
10557 {
10558 last_command_exit_value = EXECUTION_FAILURE;
10559 exp_jump_to_top_level (DISCARD);
10560 }
10561
10562 /* Now transform the word as ksh93 appears to do and go on */
10563 t = assignment (tlist->word->word, 0);
10564 tlist->word->word[t] = '\0';
a0c0a00f
CR
10565 if (tlist->word->word[t - 1] == '+')
10566 tlist->word->word[t - 1] = '\0'; /* cut off append op */
ac50fbac 10567 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
726f6388 10568 }
95732b49 10569#endif
726f6388 10570
ccc6cda3 10571 expanded_something = 0;
726f6388 10572 expanded = expand_word_internal
b72432fd 10573 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
726f6388
JA
10574
10575 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
10576 {
10577 /* By convention, each time this error is returned,
10578 tlist->word->word has already been freed. */
10579 tlist->word->word = (char *)NULL;
ccc6cda3 10580
726f6388
JA
10581 /* Dispose our copy of the original list. */
10582 dispose_words (orig_list);
d166f048 10583 /* Dispose the new list we're building. */
726f6388
JA
10584 dispose_words (new_list);
10585
28ef6c31 10586 last_command_exit_value = EXECUTION_FAILURE;
726f6388 10587 if (expanded == &expand_word_error)
b80f6443 10588 exp_jump_to_top_level (DISCARD);
726f6388 10589 else
b80f6443 10590 exp_jump_to_top_level (FORCE_EOF);
726f6388
JA
10591 }
10592
ccc6cda3
JA
10593 /* Don't split words marked W_NOSPLIT. */
10594 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
726f6388 10595 {
ccc6cda3 10596 temp_list = word_list_split (expanded);
726f6388
JA
10597 dispose_words (expanded);
10598 }
10599 else
10600 {
10601 /* If no parameter expansion, command substitution, process
10602 substitution, or arithmetic substitution took place, then
10603 do not do word splitting. We still have to remove quoted
10604 null characters from the result. */
10605 word_list_remove_quoted_nulls (expanded);
ccc6cda3 10606 temp_list = expanded;
726f6388
JA
10607 }
10608
ccc6cda3
JA
10609 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
10610 new_list = (WORD_LIST *)list_append (expanded, new_list);
726f6388
JA
10611 }
10612
cce855bc
JA
10613 if (orig_list)
10614 dispose_words (orig_list);
726f6388 10615
726f6388 10616 if (new_list)
cce855bc 10617 new_list = REVERSE_LIST (new_list, WORD_LIST *);
726f6388 10618
cce855bc
JA
10619 return (new_list);
10620}
726f6388 10621
cce855bc
JA
10622/* The workhorse for expand_words () and expand_words_no_vars ().
10623 First arg is LIST, a WORD_LIST of words.
b72432fd
JA
10624 Second arg EFLAGS is a flags word controlling which expansions are
10625 performed.
726f6388 10626
cce855bc
JA
10627 This does all of the substitutions: brace expansion, tilde expansion,
10628 parameter expansion, command substitution, arithmetic expansion,
10629 process substitution, word splitting, and pathname expansion, according
10630 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
10631 set, or for which no expansion is done, do not undergo word splitting.
ac50fbac
CR
10632 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
10633 with W_NOBRACE set do not undergo brace expansion (see
10634 brace_expand_word_list above). */
cce855bc
JA
10635static WORD_LIST *
10636expand_word_list_internal (list, eflags)
10637 WORD_LIST *list;
10638 int eflags;
10639{
10640 WORD_LIST *new_list, *temp_list;
10641 int tint;
a0c0a00f 10642 char *savecmd;
726f6388 10643
ac50fbac 10644 tempenv_assign_error = 0;
cce855bc
JA
10645 if (list == 0)
10646 return ((WORD_LIST *)NULL);
726f6388 10647
bb70624e 10648 garglist = new_list = copy_word_list (list);
cce855bc
JA
10649 if (eflags & WEXP_VARASSIGN)
10650 {
bb70624e 10651 garglist = new_list = separate_out_assignments (new_list);
cce855bc
JA
10652 if (new_list == 0)
10653 {
bb70624e 10654 if (subst_assign_varlist)
cce855bc
JA
10655 {
10656 /* All the words were variable assignments, so they are placed
10657 into the shell's environment. */
bb70624e 10658 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
cce855bc 10659 {
a0c0a00f 10660 savecmd = this_command_name;
cce855bc 10661 this_command_name = (char *)NULL; /* no arithmetic errors */
495aee44 10662 tint = do_word_assignment (temp_list->word, 0);
a0c0a00f 10663 this_command_name = savecmd;
cce855bc
JA
10664 /* Variable assignment errors in non-interactive shells
10665 running in Posix.2 mode cause the shell to exit. */
28ef6c31 10666 if (tint == 0)
ccc6cda3 10667 {
cce855bc 10668 last_command_exit_value = EXECUTION_FAILURE;
28ef6c31 10669 if (interactive_shell == 0 && posixly_correct)
b80f6443 10670 exp_jump_to_top_level (FORCE_EOF);
28ef6c31 10671 else
b80f6443 10672 exp_jump_to_top_level (DISCARD);
ccc6cda3 10673 }
726f6388 10674 }
bb70624e
JA
10675 dispose_words (subst_assign_varlist);
10676 subst_assign_varlist = (WORD_LIST *)NULL;
cce855bc
JA
10677 }
10678 return ((WORD_LIST *)NULL);
10679 }
10680 }
726f6388 10681
cce855bc
JA
10682 /* Begin expanding the words that remain. The expansions take place on
10683 things that aren't really variable assignments. */
726f6388 10684
cce855bc
JA
10685#if defined (BRACE_EXPANSION)
10686 /* Do brace expansion on this word if there are any brace characters
10687 in the string. */
10688 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
10689 new_list = brace_expand_word_list (new_list, eflags);
10690#endif /* BRACE_EXPANSION */
726f6388 10691
cce855bc
JA
10692 /* Perform the `normal' shell expansions: tilde expansion, parameter and
10693 variable substitution, command substitution, arithmetic expansion,
10694 and word splitting. */
10695 new_list = shell_expand_word_list (new_list, eflags);
726f6388 10696
cce855bc
JA
10697 /* Okay, we're almost done. Now let's just do some filename
10698 globbing. */
10699 if (new_list)
10700 {
10701 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
10702 /* Glob expand the word list unless globbing has been disabled. */
10703 new_list = glob_expand_word_list (new_list, eflags);
726f6388 10704 else
cce855bc
JA
10705 /* Dequote the words, because we're not performing globbing. */
10706 new_list = dequote_list (new_list);
726f6388
JA
10707 }
10708
bb70624e 10709 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
726f6388 10710 {
95732b49 10711 sh_wassign_func_t *assign_func;
495aee44 10712 int is_special_builtin, is_builtin_or_func;
726f6388
JA
10713
10714 /* If the remainder of the words expand to nothing, Posix.2 requires
10715 that the variable and environment assignments affect the shell's
10716 environment. */
95732b49 10717 assign_func = new_list ? assign_in_env : do_word_assignment;
b80f6443 10718 tempenv_assign_error = 0;
726f6388 10719
495aee44
CR
10720 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
10721 /* Posix says that special builtins exit if a variable assignment error
10722 occurs in an assignment preceding it. */
10723 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
10724
bb70624e 10725 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
726f6388 10726 {
a0c0a00f 10727 savecmd = this_command_name;
ccc6cda3 10728 this_command_name = (char *)NULL;
3185942a 10729 assigning_in_environment = (assign_func == assign_in_env);
495aee44 10730 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
3185942a 10731 assigning_in_environment = 0;
a0c0a00f 10732 this_command_name = savecmd;
ccc6cda3
JA
10733 /* Variable assignment errors in non-interactive shells running
10734 in Posix.2 mode cause the shell to exit. */
b80f6443 10735 if (tint == 0)
ccc6cda3 10736 {
95732b49 10737 if (assign_func == do_word_assignment)
b80f6443
JA
10738 {
10739 last_command_exit_value = EXECUTION_FAILURE;
495aee44 10740 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
b80f6443
JA
10741 exp_jump_to_top_level (FORCE_EOF);
10742 else
10743 exp_jump_to_top_level (DISCARD);
10744 }
28ef6c31 10745 else
b80f6443 10746 tempenv_assign_error++;
ccc6cda3 10747 }
726f6388 10748 }
726f6388 10749
bb70624e
JA
10750 dispose_words (subst_assign_varlist);
10751 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
10752 }
10753
cce855bc 10754 return (new_list);
ccc6cda3 10755}