]> git.ipfire.org Git - thirdparty/bash.git/blame - subst.c
Bash-4.4 patch 4
[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 5810 if (last_procsub_child)
2965eca9 5811 discard_last_procsub_child ();
a0c0a00f 5812 last_procsub_child = restore_pipeline (0);
ccc6cda3
JA
5813#endif
5814
f73dda09
JA
5815#if !defined (HAVE_DEV_FD)
5816 fifo_list[nfifo-1].proc = pid;
5817#endif
5818
726f6388
JA
5819 last_made_pid = old_pid;
5820
5821#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5822 close_pgrp_pipe ();
5823#endif /* JOB_CONTROL && PGRP_PIPE */
5824
5825#if defined (HAVE_DEV_FD)
5826 close (child_pipe_fd);
5827#endif /* HAVE_DEV_FD */
5828
5829 return (pathname);
5830 }
5831
5832 set_sigint_handler ();
5833
5834#if defined (JOB_CONTROL)
5835 set_job_control (0);
5836#endif /* JOB_CONTROL */
5837
5838#if !defined (HAVE_DEV_FD)
5839 /* Open the named pipe in the child. */
ac50fbac 5840 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
726f6388
JA
5841 if (fd < 0)
5842 {
b80f6443
JA
5843 /* Two separate strings for ease of translation. */
5844 if (open_for_read_in_child)
5845 sys_error (_("cannot open named pipe %s for reading"), pathname);
5846 else
5847 sys_error (_("cannot open named pipe %s for writing"), pathname);
5848
726f6388
JA
5849 exit (127);
5850 }
bb70624e
JA
5851 if (open_for_read_in_child)
5852 {
28ef6c31 5853 if (sh_unset_nodelay_mode (fd) < 0)
bb70624e 5854 {
3185942a 5855 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
bb70624e
JA
5856 exit (127);
5857 }
5858 }
726f6388
JA
5859#else /* HAVE_DEV_FD */
5860 fd = child_pipe_fd;
5861#endif /* HAVE_DEV_FD */
5862
a0c0a00f
CR
5863 /* Discard buffered stdio output before replacing the underlying file
5864 descriptor. */
5865 if (open_for_read_in_child == 0)
5866 fpurge (stdout);
5867
726f6388
JA
5868 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5869 {
b80f6443 5870 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
ccc6cda3 5871 open_for_read_in_child ? 0 : 1);
726f6388
JA
5872 exit (127);
5873 }
5874
f73dda09
JA
5875 if (fd != (open_for_read_in_child ? 0 : 1))
5876 close (fd);
726f6388
JA
5877
5878 /* Need to close any files that this process has open to pipes inherited
5879 from its parent. */
5880 if (current_fds_to_close)
5881 {
5882 close_fd_bitmap (current_fds_to_close);
5883 current_fds_to_close = (struct fd_bitmap *)NULL;
5884 }
5885
5886#if defined (HAVE_DEV_FD)
5887 /* Make sure we close the parent's end of the pipe and clear the slot
5888 in the fd list so it is not closed later, if reallocated by, for
5889 instance, pipe(2). */
5890 close (parent_pipe_fd);
5891 dev_fd_list[parent_pipe_fd] = 0;
5892#endif /* HAVE_DEV_FD */
5893
8dea6e87 5894 /* subshells shouldn't have this flag, which controls using the temporary
a0c0a00f
CR
5895 environment for variable lookups. We have already flushed the temporary
5896 environment above in the case we're expanding a redirection, so processes
5897 executed by this command need to be able to set it independently of their
5898 parent. */
8dea6e87
CR
5899 expanding_redir = 0;
5900
a0c0a00f 5901 subshell_level++;
d166f048 5902 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
a0c0a00f 5903 subshell_level--;
726f6388
JA
5904
5905#if !defined (HAVE_DEV_FD)
5906 /* Make sure we close the named pipe in the child before we exit. */
5907 close (open_for_read_in_child ? 0 : 1);
5908#endif /* !HAVE_DEV_FD */
5909
ac50fbac
CR
5910 last_command_exit_value = result;
5911 result = run_exit_trap ();
726f6388
JA
5912 exit (result);
5913 /*NOTREACHED*/
5914}
5915#endif /* PROCESS_SUBSTITUTION */
5916
cce855bc
JA
5917/***********************************/
5918/* */
5919/* Command Substitution */
5920/* */
5921/***********************************/
5922
d166f048 5923static char *
3185942a 5924read_comsub (fd, quoted, rflag)
d166f048 5925 int fd, quoted;
3185942a 5926 int *rflag;
d166f048 5927{
3185942a
JA
5928 char *istring, buf[128], *bufp, *s;
5929 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
f73dda09 5930 ssize_t bufn;
280bd77d 5931 int nullbyte;
d166f048
JA
5932
5933 istring = (char *)NULL;
3185942a
JA
5934 istring_index = istring_size = bufn = tflag = 0;
5935
5936 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5937 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
d166f048 5938
280bd77d
CR
5939 nullbyte = 0;
5940
3185942a
JA
5941 /* Read the output of the command through the pipe. This may need to be
5942 changed to understand multibyte characters in the future. */
d166f048
JA
5943 while (1)
5944 {
5945 if (fd < 0)
28ef6c31 5946 break;
d166f048
JA
5947 if (--bufn <= 0)
5948 {
bb70624e 5949 bufn = zread (fd, buf, sizeof (buf));
d166f048
JA
5950 if (bufn <= 0)
5951 break;
5952 bufp = buf;
5953 }
5954 c = *bufp++;
5955
28ef6c31
JA
5956 if (c == 0)
5957 {
a0c0a00f 5958#if 1
280bd77d
CR
5959 if (nullbyte == 0)
5960 {
5961 internal_warning ("%s", _("command substitution: ignored null byte in input"));
5962 nullbyte = 1;
5963 }
28ef6c31
JA
5964#endif
5965 continue;
5966 }
5967
d166f048
JA
5968 /* Add the character to ISTRING, possibly after resizing it. */
5969 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5970
f1be666c
JA
5971 /* This is essentially quote_string inline */
5972 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5973 istring[istring_index++] = CTLESC;
5974 /* Escape CTLESC and CTLNUL in the output to protect those characters
5975 from the rest of the word expansions (word splitting and globbing.)
5976 This is essentially quote_escapes inline. */
3185942a
JA
5977 else if (skip_ctlesc == 0 && c == CTLESC)
5978 {
5979 tflag |= W_HASCTLESC;
5980 istring[istring_index++] = CTLESC;
5981 }
5982 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
d166f048
JA
5983 istring[istring_index++] = CTLESC;
5984
5985 istring[istring_index++] = c;
28ef6c31
JA
5986
5987#if 0
5988#if defined (__CYGWIN__)
5989 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5990 {
5991 istring_index--;
5992 istring[istring_index - 1] = '\n';
5993 }
5994#endif
5995#endif
d166f048
JA
5996 }
5997
5998 if (istring)
5999 istring[istring_index] = '\0';
6000
6001 /* If we read no output, just return now and save ourselves some
6002 trouble. */
6003 if (istring_index == 0)
6004 {
6005 FREE (istring);
3185942a
JA
6006 if (rflag)
6007 *rflag = tflag;
d166f048
JA
6008 return (char *)NULL;
6009 }
6010
6011 /* Strip trailing newlines from the output of the command. */
6012 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6013 {
6014 while (istring_index > 0)
6015 {
6016 if (istring[istring_index - 1] == '\n')
6017 {
6018 --istring_index;
6019
6020 /* If the newline was quoted, remove the quoting char. */
6021 if (istring[istring_index - 1] == CTLESC)
6022 --istring_index;
6023 }
6024 else
6025 break;
6026 }
6027 istring[istring_index] = '\0';
6028 }
6029 else
6030 strip_trailing (istring, istring_index - 1, 1);
6031
3185942a
JA
6032 if (rflag)
6033 *rflag = tflag;
d166f048
JA
6034 return istring;
6035}
6036
3185942a
JA
6037/* Perform command substitution on STRING. This returns a WORD_DESC * with the
6038 contained string possibly quoted. */
6039WORD_DESC *
726f6388
JA
6040command_substitute (string, quoted)
6041 char *string;
6042 int quoted;
6043{
95732b49 6044 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
a0c0a00f 6045 char *istring, *s;
3185942a
JA
6046 int result, fildes[2], function_value, pflags, rc, tflag;
6047 WORD_DESC *ret;
726f6388 6048
ccc6cda3 6049 istring = (char *)NULL;
726f6388
JA
6050
6051 /* Don't fork () if there is no need to. In the case of no command to
6052 run, just return NULL. */
a0c0a00f
CR
6053#if 1
6054 for (s = string; s && *s && (shellblank (*s) || *s == '\n'); s++)
6055 ;
6056 if (s == 0 || *s == 0)
6057 return ((WORD_DESC *)NULL);
6058#else
726f6388 6059 if (!string || !*string || (string[0] == '\n' && !string[1]))
3185942a 6060 return ((WORD_DESC *)NULL);
a0c0a00f 6061#endif
726f6388 6062
cce855bc
JA
6063 if (wordexp_only && read_but_dont_execute)
6064 {
0001803f 6065 last_command_exit_value = EX_WEXPCOMSUB;
cce855bc
JA
6066 jump_to_top_level (EXITPROG);
6067 }
6068
bb70624e
JA
6069 /* We're making the assumption here that the command substitution will
6070 eventually run a command from the file system. Since we'll run
6071 maybe_make_export_env in this subshell before executing that command,
6072 the parent shell and any other shells it starts will have to remake
6073 the environment. If we make it before we fork, other shells won't
6074 have to. Don't bother if we have any temporary variable assignments,
6075 though, because the export environment will be remade after this
6076 command completes anyway, but do it if all the words to be expanded
6077 are variable assignments. */
6078 if (subst_assign_varlist == 0 || garglist == 0)
6079 maybe_make_export_env (); /* XXX */
6080
b80f6443 6081 /* Flags to pass to parse_and_execute() */
0001803f 6082 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
b80f6443 6083
726f6388
JA
6084 /* Pipe the output of executing STRING into the current shell. */
6085 if (pipe (fildes) < 0)
6086 {
a0c0a00f 6087 sys_error ("%s", _("cannot make pipe for command substitution"));
726f6388
JA
6088 goto error_exit;
6089 }
6090
6091 old_pid = last_made_pid;
6092#if defined (JOB_CONTROL)
ccc6cda3 6093 old_pipeline_pgrp = pipeline_pgrp;
28ef6c31
JA
6094 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
6095 if ((subshell_environment & SUBSHELL_PIPE) == 0)
6096 pipeline_pgrp = shell_pgrp;
ccc6cda3 6097 cleanup_the_pipeline ();
95732b49 6098#endif /* JOB_CONTROL */
726f6388 6099
95732b49 6100 old_async_pid = last_asynchronous_pid;
95732b49 6101 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
95732b49
JA
6102 last_asynchronous_pid = old_async_pid;
6103
726f6388 6104 if (pid == 0)
495aee44
CR
6105 {
6106 /* Reset the signal handlers in the child, but don't free the
6107 trap strings. Set a flag noting that we have to free the
6108 trap strings if we run trap to change a signal disposition. */
6109 reset_signal_handlers ();
a0c0a00f
CR
6110 if (ISINTERRUPT)
6111 {
6112 kill (getpid (), SIGINT);
6113 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6114 }
6115 QUIT; /* catch any interrupts we got post-fork */
495aee44
CR
6116 subshell_environment |= SUBSHELL_RESETTRAP;
6117 }
ccc6cda3
JA
6118
6119#if defined (JOB_CONTROL)
3185942a 6120 /* XXX DO THIS ONLY IN PARENT ? XXX */
ccc6cda3
JA
6121 set_sigchld_handler ();
6122 stop_making_children ();
f1be666c
JA
6123 if (pid != 0)
6124 pipeline_pgrp = old_pipeline_pgrp;
f73dda09
JA
6125#else
6126 stop_making_children ();
ccc6cda3 6127#endif /* JOB_CONTROL */
726f6388
JA
6128
6129 if (pid < 0)
6130 {
b80f6443 6131 sys_error (_("cannot make child for command substitution"));
726f6388
JA
6132 error_exit:
6133
ac50fbac
CR
6134 last_made_pid = old_pid;
6135
726f6388
JA
6136 FREE (istring);
6137 close (fildes[0]);
6138 close (fildes[1]);
3185942a 6139 return ((WORD_DESC *)NULL);
726f6388
JA
6140 }
6141
6142 if (pid == 0)
6143 {
a0c0a00f
CR
6144 /* The currently executing shell is not interactive. */
6145 interactive = 0;
6146
726f6388 6147 set_sigint_handler (); /* XXX */
28ef6c31 6148
b80f6443
JA
6149 free_pushed_string_input ();
6150
a0c0a00f
CR
6151 /* Discard buffered stdio output before replacing the underlying file
6152 descriptor. */
6153 fpurge (stdout);
6154
726f6388
JA
6155 if (dup2 (fildes[1], 1) < 0)
6156 {
a0c0a00f 6157 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
726f6388
JA
6158 exit (EXECUTION_FAILURE);
6159 }
6160
6161 /* If standard output is closed in the parent shell
6162 (such as after `exec >&-'), file descriptor 1 will be
6163 the lowest available file descriptor, and end up in
6164 fildes[0]. This can happen for stdin and stderr as well,
6165 but stdout is more important -- it will cause no output
6166 to be generated from this command. */
6167 if ((fildes[1] != fileno (stdin)) &&
6168 (fildes[1] != fileno (stdout)) &&
6169 (fildes[1] != fileno (stderr)))
6170 close (fildes[1]);
6171
6172 if ((fildes[0] != fileno (stdin)) &&
6173 (fildes[0] != fileno (stdout)) &&
6174 (fildes[0] != fileno (stderr)))
6175 close (fildes[0]);
6176
495aee44
CR
6177#ifdef __CYGWIN__
6178 /* Let stdio know the fd may have changed from text to binary mode, and
6179 make sure to preserve stdout line buffering. */
6180 freopen (NULL, "w", stdout);
6181 sh_setlinebuf (stdout);
6182#endif /* __CYGWIN__ */
6183
ccc6cda3 6184 /* This is a subshell environment. */
28ef6c31 6185 subshell_environment |= SUBSHELL_COMSUB;
ccc6cda3 6186
a0c0a00f
CR
6187 /* Many shells do not appear to inherit the -v option for command
6188 substitutions. */
6189 change_flag ('v', FLAG_OFF);
6190
6191 /* When inherit_errexit option is not enabled, command substitution does
6192 not inherit the -e flag. It is enabled when Posix mode is enabled */
6193 if (inherit_errexit == 0)
ac50fbac
CR
6194 {
6195 builtin_ignoring_errexit = 0;
6196 change_flag ('e', FLAG_OFF);
ac50fbac 6197 }
a0c0a00f
CR
6198 set_shellopts ();
6199
6200 /* If we are expanding a redirection, we can dispose of any temporary
6201 environment we received, since redirections are not supposed to have
6202 access to the temporary environment. We will have to see whether this
6203 affects temporary environments supplied to `eval', but the temporary
6204 environment gets copied to builtin_env at some point. */
6205 if (expanding_redir)
6206 {
6207 flush_temporary_env ();
6208 expanding_redir = 0;
6209 }
726f6388
JA
6210
6211 remove_quoted_escapes (string);
6212
ccc6cda3 6213 startup_state = 2; /* see if we can avoid a fork */
726f6388
JA
6214 /* Give command substitution a place to jump back to on failure,
6215 so we don't go back up to main (). */
ac50fbac 6216 result = setjmp_nosigs (top_level);
726f6388 6217
bb70624e
JA
6218 /* If we're running a command substitution inside a shell function,
6219 trap `return' so we don't return from the function in the subshell
6220 and go off to never-never land. */
6221 if (result == 0 && return_catch_flag)
ac50fbac 6222 function_value = setjmp_nosigs (return_catch);
bb70624e
JA
6223 else
6224 function_value = 0;
6225
b80f6443
JA
6226 if (result == ERREXIT)
6227 rc = last_command_exit_value;
6228 else if (result == EXITPROG)
6229 rc = last_command_exit_value;
726f6388 6230 else if (result)
b80f6443 6231 rc = EXECUTION_FAILURE;
bb70624e 6232 else if (function_value)
b80f6443 6233 rc = return_catch_value;
726f6388 6234 else
b80f6443
JA
6235 {
6236 subshell_level++;
6237 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
6238 subshell_level--;
6239 }
6240
6241 last_command_exit_value = rc;
6242 rc = run_exit_trap ();
f1be666c
JA
6243#if defined (PROCESS_SUBSTITUTION)
6244 unlink_fifo_list ();
6245#endif
b80f6443 6246 exit (rc);
726f6388
JA
6247 }
6248 else
6249 {
726f6388
JA
6250#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6251 close_pgrp_pipe ();
6252#endif /* JOB_CONTROL && PGRP_PIPE */
6253
6254 close (fildes[1]);
6255
3185942a
JA
6256 tflag = 0;
6257 istring = read_comsub (fildes[0], quoted, &tflag);
ccc6cda3 6258
726f6388
JA
6259 close (fildes[0]);
6260
b72432fd 6261 current_command_subst_pid = pid;
726f6388
JA
6262 last_command_exit_value = wait_for (pid);
6263 last_command_subst_pid = pid;
6264 last_made_pid = old_pid;
6265
6266#if defined (JOB_CONTROL)
6267 /* If last_command_exit_value > 128, then the substituted command
6268 was terminated by a signal. If that signal was SIGINT, then send
6269 SIGINT to ourselves. This will break out of loops, for instance. */
b80f6443 6270 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
726f6388
JA
6271 kill (getpid (), SIGINT);
6272
6273 /* wait_for gives the terminal back to shell_pgrp. If some other
cce855bc
JA
6274 process group should have it, give it away to that group here.
6275 pipeline_pgrp is non-zero only while we are constructing a
ac50fbac 6276 pipeline, so what we are concerned about is whether or not that
cce855bc 6277 pipeline was started in the background. A pipeline started in
a0c0a00f
CR
6278 the background should never get the tty back here. We duplicate
6279 the conditions that wait_for tests to make sure we only give
6280 the terminal back to pipeline_pgrp under the conditions that wait_for
6281 gave it to shell_pgrp. If wait_for doesn't mess with the terminal
6282 pgrp, we should not either. */
6283 if (interactive && pipeline_pgrp != (pid_t)0 && running_in_background == 0 &&
6284 (subshell_environment & (SUBSHELL_ASYNC|SUBSHELL_PIPE)) == 0)
28ef6c31 6285 give_terminal_to (pipeline_pgrp, 0);
726f6388
JA
6286#endif /* JOB_CONTROL */
6287
3185942a
JA
6288 ret = alloc_word_desc ();
6289 ret->word = istring;
6290 ret->flags = tflag;
6291
6292 return ret;
726f6388
JA
6293 }
6294}
6295
6296/********************************************************
6297 * *
6298 * Utility functions for parameter expansion *
6299 * *
6300 ********************************************************/
6301
ccc6cda3 6302#if defined (ARRAY_VARS)
ccc6cda3 6303
f73dda09 6304static arrayind_t
ccc6cda3
JA
6305array_length_reference (s)
6306 char *s;
6307{
f73dda09
JA
6308 int len;
6309 arrayind_t ind;
3185942a 6310 char *akey;
f73dda09 6311 char *t, c;
ccc6cda3 6312 ARRAY *array;
495aee44 6313 HASH_TABLE *h;
ccc6cda3
JA
6314 SHELL_VAR *var;
6315
6316 var = array_variable_part (s, &t, &len);
726f6388 6317
ccc6cda3
JA
6318 /* If unbound variables should generate an error, report one and return
6319 failure. */
ac50fbac 6320 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
726f6388 6321 {
f73dda09 6322 c = *--t;
ccc6cda3 6323 *t = '\0';
0001803f 6324 last_command_exit_value = EXECUTION_FAILURE;
7117c2d2 6325 err_unboundvar (s);
f73dda09 6326 *t = c;
ccc6cda3 6327 return (-1);
726f6388 6328 }
ac50fbac 6329 else if (var == 0 || invisible_p (var))
ccc6cda3 6330 return 0;
726f6388 6331
28ef6c31
JA
6332 /* We support a couple of expansions for variables that are not arrays.
6333 We'll return the length of the value for v[0], and 1 for v[@] or
6334 v[*]. Return 0 for everything else. */
6335
6336 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
495aee44 6337 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
726f6388 6338
ccc6cda3 6339 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
ccc6cda3 6340 {
3185942a 6341 if (assoc_p (var))
495aee44 6342 return (h ? assoc_num_elements (h) : 0);
3185942a 6343 else if (array_p (var))
495aee44 6344 return (array ? array_num_elements (array) : 0);
3185942a 6345 else
495aee44 6346 return (var_isset (var) ? 1 : 0);
ccc6cda3 6347 }
ccc6cda3 6348
3185942a
JA
6349 if (assoc_p (var))
6350 {
6351 t[len - 1] = '\0';
6352 akey = expand_assignment_string_to_string (t, 0); /* [ */
6353 t[len - 1] = ']';
6354 if (akey == 0 || *akey == 0)
6355 {
6356 err_badarraysub (t);
ac50fbac 6357 FREE (akey);
3185942a
JA
6358 return (-1);
6359 }
6360 t = assoc_reference (assoc_cell (var), akey);
ac50fbac 6361 free (akey);
3185942a 6362 }
28ef6c31 6363 else
3185942a 6364 {
ac50fbac
CR
6365 ind = array_expand_index (var, t, len);
6366 /* negative subscripts to indexed arrays count back from end */
6367 if (var && array_p (var) && ind < 0)
6368 ind = array_max_index (array_cell (var)) + 1 + ind;
3185942a
JA
6369 if (ind < 0)
6370 {
6371 err_badarraysub (t);
6372 return (-1);
6373 }
6374 if (array_p (var))
6375 t = array_reference (array, ind);
6376 else
6377 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6378 }
28ef6c31 6379
f1be666c 6380 len = MB_STRLEN (t);
ccc6cda3 6381 return (len);
726f6388 6382}
ccc6cda3 6383#endif /* ARRAY_VARS */
726f6388
JA
6384
6385static int
6386valid_brace_expansion_word (name, var_is_special)
6387 char *name;
6388 int var_is_special;
6389{
f73dda09 6390 if (DIGIT (*name) && all_digits (name))
726f6388
JA
6391 return 1;
6392 else if (var_is_special)
6393 return 1;
ccc6cda3 6394#if defined (ARRAY_VARS)
a0c0a00f 6395 else if (valid_array_reference (name, 0))
ccc6cda3
JA
6396 return 1;
6397#endif /* ARRAY_VARS */
726f6388
JA
6398 else if (legal_identifier (name))
6399 return 1;
6400 else
6401 return 0;
6402}
ccc6cda3 6403
b80f6443
JA
6404static int
6405chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
6406 char *name;
6407 int quoted;
6408 int *quoted_dollar_atp, *contains_dollar_at;
6409{
6410 char *temp1;
6411
6412 if (name == 0)
6413 {
6414 if (quoted_dollar_atp)
6415 *quoted_dollar_atp = 0;
6416 if (contains_dollar_at)
6417 *contains_dollar_at = 0;
6418 return 0;
6419 }
6420
6421 /* check for $@ and $* */
6422 if (name[0] == '@' && name[1] == 0)
6423 {
6424 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6425 *quoted_dollar_atp = 1;
6426 if (contains_dollar_at)
6427 *contains_dollar_at = 1;
6428 return 1;
6429 }
6430 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6431 {
6432 if (contains_dollar_at)
6433 *contains_dollar_at = 1;
6434 return 1;
6435 }
6436
6437 /* Now check for ${array[@]} and ${array[*]} */
6438#if defined (ARRAY_VARS)
a0c0a00f 6439 else if (valid_array_reference (name, 0))
b80f6443 6440 {
0001803f 6441 temp1 = mbschr (name, '[');
b80f6443
JA
6442 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
6443 {
6444 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6445 *quoted_dollar_atp = 1;
6446 if (contains_dollar_at)
6447 *contains_dollar_at = 1;
6448 return 1;
6449 } /* [ */
6450 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6451 which should result in separate words even when IFS is unset. */
6452 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
6453 {
6454 if (contains_dollar_at)
6455 *contains_dollar_at = 1;
6456 return 1;
6457 }
6458 }
6459#endif
6460 return 0;
6461}
6462
726f6388
JA
6463/* Parameter expand NAME, and return a new string which is the expansion,
6464 or NULL if there was no expansion.
6465 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6466 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6467 NAME was found inside of a double-quoted expression. */
95732b49 6468static WORD_DESC *
495aee44 6469parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
726f6388 6470 char *name;
89a92869 6471 int var_is_special, quoted, pflags;
495aee44 6472 arrayind_t *indp;
726f6388 6473{
95732b49 6474 WORD_DESC *ret;
ccc6cda3 6475 char *temp, *tt;
7117c2d2 6476 intmax_t arg_index;
ccc6cda3 6477 SHELL_VAR *var;
f1be666c 6478 int atype, rflags;
495aee44 6479 arrayind_t ind;
726f6388 6480
95732b49
JA
6481 ret = 0;
6482 temp = 0;
f1be666c 6483 rflags = 0;
95732b49 6484
495aee44
CR
6485 if (indp)
6486 *indp = INTMAX_MIN;
6487
95732b49 6488 /* Handle multiple digit arguments, as in ${11}. */
f73dda09 6489 if (legal_number (name, &arg_index))
7117c2d2
JA
6490 {
6491 tt = get_dollar_var_value (arg_index);
b80f6443
JA
6492 if (tt)
6493 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6494 ? quote_string (tt)
6495 : quote_escapes (tt);
6496 else
6497 temp = (char *)NULL;
7117c2d2
JA
6498 FREE (tt);
6499 }
726f6388
JA
6500 else if (var_is_special) /* ${@} */
6501 {
cce855bc 6502 int sindex;
f73dda09 6503 tt = (char *)xmalloc (2 + strlen (name));
cce855bc 6504 tt[sindex = 0] = '$';
726f6388 6505 strcpy (tt + 1, name);
7117c2d2 6506
95732b49 6507 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
89a92869 6508 (int *)NULL, (int *)NULL, pflags);
cce855bc 6509 free (tt);
726f6388 6510 }
ccc6cda3 6511#if defined (ARRAY_VARS)
a0c0a00f 6512 else if (valid_array_reference (name, 0))
ccc6cda3 6513 {
ac50fbac 6514expand_arrayref:
ac50fbac 6515 if (pflags & PF_ASSIGNRHS)
a0c0a00f
CR
6516 {
6517 var = array_variable_part (name, &tt, (int *)0);
6518 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
6519 {
6520 /* Only treat as double quoted if array variable */
6521 if (var && (array_p (var) || assoc_p (var)))
6522 /* XXX - bash-4.4/bash-5.0 pass AV_ASSIGNRHS */
6523 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6524 else
6525 temp = array_value (name, quoted, 0, &atype, &ind);
6526 }
ac50fbac
CR
6527 else
6528 temp = array_value (name, quoted, 0, &atype, &ind);
a0c0a00f 6529 }
ac50fbac
CR
6530 else
6531 temp = array_value (name, quoted, 0, &atype, &ind);
7117c2d2 6532 if (atype == 0 && temp)
495aee44
CR
6533 {
6534 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6535 ? quote_string (temp)
6536 : quote_escapes (temp);
6537 rflags |= W_ARRAYIND;
6538 if (indp)
6539 *indp = ind;
6540 }
f1be666c
JA
6541 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6542 rflags |= W_HASQUOTEDNULL;
ccc6cda3
JA
6543 }
6544#endif
6545 else if (var = find_variable (name))
6546 {
7117c2d2 6547 if (var_isset (var) && invisible_p (var) == 0)
28ef6c31 6548 {
ccc6cda3 6549#if defined (ARRAY_VARS)
3185942a
JA
6550 if (assoc_p (var))
6551 temp = assoc_reference (assoc_cell (var), "0");
6552 else if (array_p (var))
6553 temp = array_reference (array_cell (var), 0);
6554 else
6555 temp = value_cell (var);
ccc6cda3
JA
6556#else
6557 temp = value_cell (var);
6558#endif
6559
6560 if (temp)
b80f6443
JA
6561 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6562 ? quote_string (temp)
6563 : quote_escapes (temp);
28ef6c31 6564 }
ccc6cda3
JA
6565 else
6566 temp = (char *)NULL;
6567 }
a0c0a00f 6568 else if (var = find_variable_last_nameref (name, 0))
ac50fbac
CR
6569 {
6570 temp = nameref_cell (var);
6571#if defined (ARRAY_VARS)
6572 /* Handle expanding nameref whose value is x[n] */
a0c0a00f 6573 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
6574 {
6575 name = temp;
6576 goto expand_arrayref;
6577 }
6578 else
6579#endif
6580 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6581 if (temp && *temp && legal_identifier (temp) == 0)
6582 {
6583 last_command_exit_value = EXECUTION_FAILURE;
6584 report_error (_("%s: invalid variable name for name reference"), temp);
6585 temp = &expand_param_error;
6586 }
6587 else
6588 temp = (char *)NULL;
6589 }
726f6388 6590 else
ccc6cda3 6591 temp = (char *)NULL;
726f6388 6592
95732b49
JA
6593 if (ret == 0)
6594 {
6595 ret = alloc_word_desc ();
6596 ret->word = temp;
f1be666c 6597 ret->flags |= rflags;
95732b49
JA
6598 }
6599 return ret;
726f6388
JA
6600}
6601
ac50fbac
CR
6602static char *
6603parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
ccc6cda3 6604 char *name;
ac50fbac 6605 int var_is_special, quoted, find_nameref;
ccc6cda3
JA
6606{
6607 char *temp, *t;
95732b49 6608 WORD_DESC *w;
ac50fbac
CR
6609 SHELL_VAR *v;
6610
a0c0a00f 6611 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
ac50fbac
CR
6612 nameref_p (v) && (t = nameref_cell (v)) && *t)
6613 return (savestring (t));
ccc6cda3 6614
ac50fbac
CR
6615 /* If var_is_special == 0, and name is not an array reference, this does
6616 more expansion than necessary. It should really look up the variable's
6617 value and not try to expand it. */
495aee44 6618 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
95732b49 6619 t = w->word;
b80f6443
JA
6620 /* Have to dequote here if necessary */
6621 if (t)
6622 {
6623 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6624 ? dequote_string (t)
6625 : dequote_escapes (t);
6626 free (t);
6627 t = temp;
6628 }
95732b49
JA
6629 dispose_word_desc (w);
6630
ac50fbac
CR
6631 return t;
6632}
6633
6634/* Expand an indirect reference to a variable: ${!NAME} expands to the
6635 value of the variable whose name is the value of NAME. */
6636static WORD_DESC *
6637parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
6638 char *name;
6639 int var_is_special, quoted;
6640 int *quoted_dollar_atp, *contains_dollar_at;
6641{
6642 char *temp, *t;
6643 WORD_DESC *w;
6644 SHELL_VAR *v;
6645
6646 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6647 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6648 bash performs an indirect lookup on foo[0] and expands the result;
6649 ksh93 expands bar[0]. We could do that here -- there are enough usable
6650 primitives to do that -- but do not at this point. */
a0c0a00f 6651 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
ac50fbac
CR
6652 {
6653 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6654 {
6655 w = alloc_word_desc ();
6656 w->word = savestring (t);
6657 w->flags = 0;
6658 return w;
6659 }
6660 }
6661
6662 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6663
b80f6443 6664 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
ccc6cda3 6665 if (t == 0)
95732b49
JA
6666 return (WORD_DESC *)NULL;
6667
a0c0a00f
CR
6668 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
6669 {
6670 report_error (_("%s: bad substitution"), t);
6671 free (t);
6672 w = alloc_word_desc ();
6673 w->word = &expand_param_error;
6674 w->flags = 0;
6675 return (w);
6676 }
6677
495aee44 6678 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
ccc6cda3 6679 free (t);
95732b49
JA
6680
6681 return w;
ccc6cda3
JA
6682}
6683
726f6388
JA
6684/* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
6685 depending on the value of C, the separating character. C can be one of
ccc6cda3
JA
6686 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
6687 between double quotes. */
95732b49 6688static WORD_DESC *
a0c0a00f 6689parameter_brace_expand_rhs (name, value, c, quoted, pflags, qdollaratp, hasdollarat)
726f6388 6690 char *name, *value;
a0c0a00f 6691 int c, quoted, pflags, *qdollaratp, *hasdollarat;
726f6388 6692{
95732b49 6693 WORD_DESC *w;
726f6388 6694 WORD_LIST *l;
a0c0a00f
CR
6695 char *t, *t1, *temp, *vname;
6696 int l_hasdollat, sindex;
726f6388 6697
a0c0a00f 6698/*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
ccc6cda3
JA
6699 /* If the entire expression is between double quotes, we want to treat
6700 the value as a double-quoted string, with the exception that we strip
3185942a 6701 embedded unescaped double quotes (for sh backwards compatibility). */
95732b49 6702 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
726f6388 6703 {
a0c0a00f
CR
6704 sindex = 0;
6705 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
726f6388 6706 }
95732b49
JA
6707 else
6708 temp = value;
ccc6cda3 6709
95732b49 6710 w = alloc_word_desc ();
a0c0a00f 6711 l_hasdollat = 0;
ccc6cda3 6712 /* XXX was 0 not quoted */
a0c0a00f 6713 l = *temp ? expand_string_for_rhs (temp, quoted, &l_hasdollat, (int *)NULL)
ccc6cda3
JA
6714 : (WORD_LIST *)0;
6715 if (hasdollarat)
a0c0a00f 6716 *hasdollarat = l_hasdollat || (l && l->next);
95732b49
JA
6717 if (temp != value)
6718 free (temp);
726f6388
JA
6719 if (l)
6720 {
a0c0a00f
CR
6721 /* If l->next is not null, we know that TEMP contained "$@", since that
6722 is the only expansion that creates more than one word. */
6723 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
6724 {
6725/*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
6726 *qdollaratp = 1;
6727 }
6728
ccc6cda3 6729 /* The expansion of TEMP returned something. We need to treat things
a0c0a00f
CR
6730 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
6731 the individual words have already been quoted. We need to turn them
b80f6443
JA
6732 into a string with the words separated by the first character of
6733 $IFS without any additional quoting, so string_list_dollar_at won't
a0c0a00f
CR
6734 do the right thing. If IFS is null, we want "$@" to split into
6735 separate arguments, not be concatenated, so we use string_list_internal
6736 and mark the word to be split on spaces later. We use
6737 string_list_dollar_star for "$@" otherwise. */
6738 if (l->next && ifs_is_null)
6739 {
6740 temp = string_list_internal (l, " ");
6741 w->flags |= W_SPLITSPACE;
6742 }
6743 else
6744 temp = (l_hasdollat || l->next) ? string_list_dollar_star (l) : string_list (l);
b80f6443 6745
cd110fdf
CR
6746 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
6747 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
6748 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
6749 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6750 (which is more paranoia than anything else), we need to return the
6751 quoted null string and set the flags to indicate it. */
ac50fbac 6752 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
cd110fdf
CR
6753 {
6754 w->flags |= W_HASQUOTEDNULL;
a0c0a00f
CR
6755/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
6756 /* If we return a quoted null with L_HASDOLLARAT, we either have a
6757 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
6758 parameters or a quoted expansion of "$@" with $1 == ''. In either
6759 case, we don't want to enable special handling of $@. */
6760 if (qdollaratp && l_hasdollat)
6761 *qdollaratp = 0;
cd110fdf 6762 }
726f6388
JA
6763 dispose_words (l);
6764 }
a0c0a00f 6765 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
726f6388 6766 {
a0c0a00f
CR
6767 /* Posix interp 221 changed the rules on this. The idea is that
6768 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
6769 when foo and xxx are unset. The problem is that it's not in any
6770 way backwards compatible and few other shells do it. We're eventually
6771 going to try and split the difference (heh) a little bit here. */
6772 /* l_hasdollat == 1 means we saw a quoted dollar at. */
6773
ccc6cda3
JA
6774 /* The brace expansion occurred between double quotes and there was
6775 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
7117c2d2 6776 it does not expand to anything. In this case, we want to return
a0c0a00f 6777 a quoted empty string. Posix interp 888 */
0628567a 6778 temp = make_quoted_char ('\0');
95732b49 6779 w->flags |= W_HASQUOTEDNULL;
a0c0a00f 6780/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
726f6388
JA
6781 }
6782 else
6783 temp = (char *)NULL;
6784
6785 if (c == '-' || c == '+')
95732b49
JA
6786 {
6787 w->word = temp;
6788 return w;
6789 }
726f6388
JA
6790
6791 /* c == '=' */
ccc6cda3 6792 t = temp ? savestring (temp) : savestring ("");
726f6388
JA
6793 t1 = dequote_string (t);
6794 free (t);
a0c0a00f
CR
6795
6796 /* bash-4.4/5.0 */
6797 vname = name;
6798 if (*name == '!' &&
6799 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
6800 {
6801 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
6802 if (vname == 0 || *vname == 0)
6803 {
6804 report_error (_("%s: invalid indirect expansion"), name);
6805 free (vname);
6806 dispose_word (w);
6807 return &expand_wdesc_error;
6808 }
6809 if (legal_identifier (vname) == 0)
6810 {
6811 report_error (_("%s: invalid variable name"), vname);
6812 free (vname);
6813 dispose_word (w);
6814 return &expand_wdesc_error;
6815 }
6816 }
6817
b80f6443 6818#if defined (ARRAY_VARS)
a0c0a00f
CR
6819 if (valid_array_reference (vname, 0))
6820 assign_array_element (vname, t1, 0);
b80f6443
JA
6821 else
6822#endif /* ARRAY_VARS */
a0c0a00f
CR
6823 bind_variable (vname, t1, 0);
6824
6825 stupidly_hack_special_variables (vname);
6826
6827 if (vname != name)
6828 free (vname);
95732b49 6829
495aee44
CR
6830 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6831 free (temp);
6832
6833 w->word = t1;
95732b49 6834 return w;
726f6388
JA
6835}
6836
6837/* Deal with the right hand side of a ${name:?value} expansion in the case
6838 that NAME is null or not set. If VALUE is non-null it is expanded and
6839 used as the error message to print, otherwise a standard message is
6840 printed. */
6841static void
6842parameter_brace_expand_error (name, value)
6843 char *name, *value;
6844{
ccc6cda3
JA
6845 WORD_LIST *l;
6846 char *temp;
6847
ac50fbac 6848 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
726f6388
JA
6849 if (value && *value)
6850 {
95732b49 6851 l = expand_string (value, 0);
ccc6cda3
JA
6852 temp = string_list (l);
6853 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6854 FREE (temp);
726f6388
JA
6855 dispose_words (l);
6856 }
6857 else
b80f6443 6858 report_error (_("%s: parameter null or not set"), name);
726f6388
JA
6859
6860 /* Free the data we have allocated during this expansion, since we
6861 are about to longjmp out. */
6862 free (name);
6863 FREE (value);
6864}
6865
6866/* Return 1 if NAME is something for which parameter_brace_expand_length is
6867 OK to do. */
6868static int
6869valid_length_expression (name)
6870 char *name;
6871{
28ef6c31 6872 return (name[1] == '\0' || /* ${#} */
f73dda09
JA
6873 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6874 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
ccc6cda3 6875#if defined (ARRAY_VARS)
a0c0a00f 6876 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
ccc6cda3 6877#endif
726f6388
JA
6878 legal_identifier (name + 1)); /* ${#PS1} */
6879}
6880
6881/* Handle the parameter brace expansion that requires us to return the
6882 length of a parameter. */
7117c2d2 6883static intmax_t
726f6388
JA
6884parameter_brace_expand_length (name)
6885 char *name;
6886{
ccc6cda3 6887 char *t, *newname;
7117c2d2 6888 intmax_t number, arg_index;
ccc6cda3
JA
6889 WORD_LIST *list;
6890#if defined (ARRAY_VARS)
6891 SHELL_VAR *var;
6892#endif
6893
6894 if (name[1] == '\0') /* ${#} */
6895 number = number_of_args ();
a0c0a00f 6896 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
cce855bc 6897 number = number_of_args ();
f73dda09 6898 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
cce855bc
JA
6899 {
6900 /* Take the lengths of some of the shell's special parameters. */
6901 switch (name[1])
6902 {
6903 case '-':
6904 t = which_set_flags ();
6905 break;
6906 case '?':
6907 t = itos (last_command_exit_value);
6908 break;
6909 case '$':
6910 t = itos (dollar_dollar_pid);
6911 break;
6912 case '!':
6913 if (last_asynchronous_pid == NO_PID)
495aee44 6914 t = (char *)NULL; /* XXX - error if set -u set? */
cce855bc 6915 else
f73dda09 6916 t = itos (last_asynchronous_pid);
cce855bc
JA
6917 break;
6918 case '#':
6919 t = itos (number_of_args ());
6920 break;
6921 }
6922 number = STRLEN (t);
6923 FREE (t);
6924 }
ccc6cda3 6925#if defined (ARRAY_VARS)
a0c0a00f 6926 else if (valid_array_reference (name + 1, 0))
ccc6cda3
JA
6927 number = array_length_reference (name + 1);
6928#endif /* ARRAY_VARS */
cce855bc 6929 else
ccc6cda3
JA
6930 {
6931 number = 0;
6932
f73dda09 6933 if (legal_number (name + 1, &arg_index)) /* ${#1} */
ccc6cda3 6934 {
f73dda09 6935 t = get_dollar_var_value (arg_index);
495aee44
CR
6936 if (t == 0 && unbound_vars_is_error)
6937 return INTMAX_MIN;
eb873671 6938 number = MB_STRLEN (t);
ccc6cda3
JA
6939 FREE (t);
6940 }
6941#if defined (ARRAY_VARS)
3185942a 6942 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
ccc6cda3 6943 {
3185942a
JA
6944 if (assoc_p (var))
6945 t = assoc_reference (assoc_cell (var), "0");
6946 else
6947 t = array_reference (array_cell (var), 0);
495aee44
CR
6948 if (t == 0 && unbound_vars_is_error)
6949 return INTMAX_MIN;
eb873671 6950 number = MB_STRLEN (t);
ccc6cda3
JA
6951 }
6952#endif
6953 else /* ${#PS1} */
6954 {
6955 newname = savestring (name);
6956 newname[0] = '$';
6957 list = expand_string (newname, Q_DOUBLE_QUOTES);
6958 t = list ? string_list (list) : (char *)NULL;
6959 free (newname);
6960 if (list)
6961 dispose_words (list);
6962
495aee44 6963 number = t ? MB_STRLEN (t) : 0;
ccc6cda3
JA
6964 FREE (t);
6965 }
6966 }
ccc6cda3
JA
6967
6968 return (number);
6969}
6970
28ef6c31
JA
6971/* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6972 so we do some ad-hoc parsing of an arithmetic expression to find
6973 the first DELIM, instead of using strchr(3). Two rules:
6974 1. If the substring contains a `(', read until closing `)'.
6975 2. If the substring contains a `?', read past one `:' for each `?'.
a0c0a00f 6976 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
28ef6c31
JA
6977*/
6978
6979static char *
6980skiparith (substr, delim)
6981 char *substr;
6982 int delim;
6983{
a0c0a00f
CR
6984 int i;
6985 char delims[2];
28ef6c31 6986
a0c0a00f
CR
6987 delims[0] = delim;
6988 delims[1] = '\0';
7117c2d2 6989
a0c0a00f 6990 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
7117c2d2 6991 return (substr + i);
28ef6c31
JA
6992}
6993
ccc6cda3
JA
6994/* Verify and limit the start and end of the desired substring. If
6995 VTYPE == 0, a regular shell variable is being used; if it is 1,
cce855bc 6996 then the positional parameters are being used; if it is 2, then
e8ce775d
JA
6997 VALUE is really a pointer to an array variable that should be used.
6998 Return value is 1 if both values were OK, 0 if there was a problem
6999 with an invalid expression, or -1 if the values were out of range. */
ccc6cda3 7000static int
3185942a
JA
7001verify_substring_values (v, value, substr, vtype, e1p, e2p)
7002 SHELL_VAR *v;
ccc6cda3 7003 char *value, *substr;
f73dda09 7004 int vtype;
7117c2d2 7005 intmax_t *e1p, *e2p;
ccc6cda3 7006{
bb70624e 7007 char *t, *temp1, *temp2;
f73dda09
JA
7008 arrayind_t len;
7009 int expok;
ccc6cda3
JA
7010#if defined (ARRAY_VARS)
7011 ARRAY *a;
3185942a 7012 HASH_TABLE *h;
ccc6cda3
JA
7013#endif
7014
28ef6c31
JA
7015 /* duplicate behavior of strchr(3) */
7016 t = skiparith (substr, ':');
7017 if (*t && *t == ':')
7117c2d2 7018 *t = '\0';
28ef6c31
JA
7019 else
7020 t = (char *)0;
f73dda09 7021
0628567a 7022 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
d166f048 7023 *e1p = evalexp (temp1, &expok);
ccc6cda3 7024 free (temp1);
d166f048
JA
7025 if (expok == 0)
7026 return (0);
ccc6cda3 7027
f73dda09 7028 len = -1; /* paranoia */
ccc6cda3
JA
7029 switch (vtype)
7030 {
7031 case VT_VARIABLE:
d166f048 7032 case VT_ARRAYMEMBER:
eb873671 7033 len = MB_STRLEN (value);
ccc6cda3
JA
7034 break;
7035 case VT_POSPARMS:
7036 len = number_of_args () + 1;
3185942a
JA
7037 if (*e1p == 0)
7038 len++; /* add one arg if counting from $0 */
ccc6cda3
JA
7039 break;
7040#if defined (ARRAY_VARS)
7041 case VT_ARRAYVAR:
eb873671 7042 /* For arrays, the first value deals with array indices. Negative
3185942a
JA
7043 offsets count from one past the array's maximum index. Associative
7044 arrays treat the number of elements as the maximum index. */
7045 if (assoc_p (v))
7046 {
7047 h = assoc_cell (v);
7048 len = assoc_num_elements (h) + (*e1p < 0);
7049 }
7050 else
7051 {
7052 a = (ARRAY *)value;
7053 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7054 }
ccc6cda3
JA
7055 break;
7056#endif
7057 }
7058
f73dda09
JA
7059 if (len == -1) /* paranoia */
7060 return -1;
7061
ccc6cda3
JA
7062 if (*e1p < 0) /* negative offsets count from end */
7063 *e1p += len;
7064
eb873671 7065 if (*e1p > len || *e1p < 0)
e8ce775d 7066 return (-1);
d166f048 7067
b80f6443
JA
7068#if defined (ARRAY_VARS)
7069 /* For arrays, the second offset deals with the number of elements. */
7070 if (vtype == VT_ARRAYVAR)
3185942a 7071 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
b80f6443
JA
7072#endif
7073
ccc6cda3
JA
7074 if (t)
7075 {
7076 t++;
bb70624e 7077 temp2 = savestring (t);
0628567a 7078 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
bb70624e 7079 free (temp2);
ccc6cda3 7080 t[-1] = ':';
d166f048 7081 *e2p = evalexp (temp1, &expok);
ccc6cda3 7082 free (temp1);
d166f048 7083 if (expok == 0)
28ef6c31 7084 return (0);
ac50fbac 7085#if 1
495aee44 7086 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
ac50fbac
CR
7087#else
7088 /* bash-4.3: allow positional parameter length < 0 to count backwards
7089 from end of positional parameters */
7090 if (vtype == VT_ARRAYVAR && *e2p < 0)
7091#endif
28ef6c31 7092 {
b80f6443 7093 internal_error (_("%s: substring expression < 0"), t);
ccc6cda3 7094 return (0);
28ef6c31 7095 }
b80f6443
JA
7096#if defined (ARRAY_VARS)
7097 /* In order to deal with sparse arrays, push the intelligence about how
7098 to deal with the number of elements desired down to the array-
7099 specific functions. */
7100 if (vtype != VT_ARRAYVAR)
7101#endif
7102 {
495aee44
CR
7103 if (*e2p < 0)
7104 {
7105 *e2p += len;
7106 if (*e2p < 0 || *e2p < *e1p)
7107 {
7108 internal_error (_("%s: substring expression < 0"), t);
7109 return (0);
7110 }
7111 }
7112 else
7113 *e2p += *e1p; /* want E2 chars starting at E1 */
b80f6443
JA
7114 if (*e2p > len)
7115 *e2p = len;
7116 }
ccc6cda3
JA
7117 }
7118 else
7119 *e2p = len;
7120
7121 return (1);
7122}
7123
ccc6cda3 7124/* Return the type of variable specified by VARNAME (simple variable,
cce855bc 7125 positional param, or array variable). Also return the value specified
7117c2d2 7126 by VARNAME (value of a variable or a reference to an array element).
495aee44
CR
7127 QUOTED is the standard description of quoting state, using Q_* defines.
7128 FLAGS is currently a set of flags to pass to array_value. If IND is
7129 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7130 passed to array_value so the array index is not computed again.
7117c2d2
JA
7131 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7132 characters in the value are quoted with CTLESC and takes appropriate
7133 steps. For convenience, *VALP is set to the dequoted VALUE. */
ccc6cda3 7134static int
495aee44 7135get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
ccc6cda3 7136 char *varname, *value;
495aee44
CR
7137 arrayind_t ind;
7138 int quoted, flags;
ccc6cda3
JA
7139 SHELL_VAR **varp;
7140 char **valp;
7141{
ac50fbac
CR
7142 int vtype, want_indir;
7143 char *temp, *vname;
7144 WORD_DESC *wd;
ccc6cda3 7145 SHELL_VAR *v;
495aee44 7146 arrayind_t lind;
ccc6cda3 7147
ac50fbac
CR
7148 want_indir = *varname == '!' &&
7149 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7150 || VALID_INDIR_PARAM (varname[1]));
7151 if (want_indir)
7152 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
a0c0a00f 7153 /* XXX - what if vname == 0 || *vname == 0 ? */
ac50fbac
CR
7154 else
7155 vname = varname;
a0c0a00f
CR
7156
7157 if (vname == 0)
7158 {
7159 vtype = VT_VARIABLE;
7160 *varp = (SHELL_VAR *)NULL;
7161 *valp = (char *)NULL;
7162 return (vtype);
7163 }
7164
7117c2d2 7165 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
a0c0a00f 7166 vtype = STR_DOLLAR_AT_STAR (vname);
ac50fbac 7167 if (vtype == VT_POSPARMS && vname[0] == '*')
b80f6443 7168 vtype |= VT_STARSUB;
ccc6cda3
JA
7169 *varp = (SHELL_VAR *)NULL;
7170
7171#if defined (ARRAY_VARS)
a0c0a00f 7172 if (valid_array_reference (vname, 0))
ccc6cda3 7173 {
ac50fbac 7174 v = array_variable_part (vname, &temp, (int *)0);
495aee44
CR
7175 /* If we want to signal array_value to use an already-computed index,
7176 set LIND to that index */
7177 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
ac50fbac
CR
7178 if (v && invisible_p (v))
7179 {
7180 vtype = VT_ARRAYMEMBER;
7181 *varp = (SHELL_VAR *)NULL;
7182 *valp = (char *)NULL;
7183 }
3185942a 7184 if (v && (array_p (v) || assoc_p (v)))
f73dda09
JA
7185 { /* [ */
7186 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
ccc6cda3 7187 {
ac50fbac 7188 /* Callers have to differentiate between indexed and associative */
ccc6cda3 7189 vtype = VT_ARRAYVAR;
b80f6443
JA
7190 if (temp[0] == '*')
7191 vtype |= VT_STARSUB;
3185942a 7192 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
ccc6cda3
JA
7193 }
7194 else
7195 {
d166f048 7196 vtype = VT_ARRAYMEMBER;
ac50fbac 7197 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
ccc6cda3
JA
7198 }
7199 *varp = v;
7200 }
95732b49
JA
7201 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
7202 {
7203 vtype = VT_VARIABLE;
7204 *varp = v;
7205 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7206 *valp = dequote_string (value);
7207 else
7208 *valp = dequote_escapes (value);
7209 }
ccc6cda3 7210 else
3185942a
JA
7211 {
7212 vtype = VT_ARRAYMEMBER;
7213 *varp = v;
ac50fbac 7214 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
3185942a 7215 }
ccc6cda3 7216 }
ac50fbac 7217 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
ccc6cda3 7218 {
7117c2d2 7219 vtype = VT_ARRAYMEMBER;
ccc6cda3 7220 *varp = v;
3185942a 7221 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
ccc6cda3
JA
7222 }
7223 else
7224#endif
b80f6443
JA
7225 {
7226 if (value && vtype == VT_VARIABLE)
7227 {
a0c0a00f 7228 *varp = find_variable (vname);
b80f6443
JA
7229 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7230 *valp = dequote_string (value);
7231 else
7232 *valp = dequote_escapes (value);
7233 }
7234 else
7235 *valp = value;
7236 }
ccc6cda3 7237
ac50fbac
CR
7238 if (want_indir)
7239 free (vname);
7240
ccc6cda3
JA
7241 return vtype;
7242}
7243
cce855bc
JA
7244/******************************************************/
7245/* */
7246/* Functions to extract substrings of variable values */
7247/* */
7248/******************************************************/
7249
b80f6443
JA
7250#if defined (HANDLE_MULTIBYTE)
7251/* Character-oriented rather than strictly byte-oriented substrings. S and
7252 E, rather being strict indices into STRING, indicate character (possibly
7253 multibyte character) positions that require calculation.
7254 Used by the ${param:offset[:length]} expansion. */
7255static char *
7256mb_substring (string, s, e)
7257 char *string;
7258 int s, e;
7259{
7260 char *tt;
a0c0a00f
CR
7261 int start, stop, i;
7262 size_t slen;
b80f6443
JA
7263 DECLARE_MBSTATE;
7264
7265 start = 0;
95732b49
JA
7266 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
7267 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
b80f6443
JA
7268
7269 i = s;
7270 while (string[start] && i--)
7271 ADVANCE_CHAR (string, slen, start);
7272 stop = start;
7273 i = e - s;
7274 while (string[stop] && i--)
7275 ADVANCE_CHAR (string, slen, stop);
7276 tt = substring (string, start, stop);
7277 return tt;
7278}
7279#endif
7280
ccc6cda3
JA
7281/* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
7282 is `@', use the positional parameters; otherwise, use the value of
7283 VARNAME. If VARNAME is an array variable, use the array elements. */
7284
7285static char *
495aee44
CR
7286parameter_brace_substring (varname, value, ind, substr, quoted, flags)
7287 char *varname, *value;
7288 int ind;
7289 char *substr;
7290 int quoted, flags;
ccc6cda3 7291{
7117c2d2 7292 intmax_t e1, e2;
b80f6443 7293 int vtype, r, starsub;
0628567a 7294 char *temp, *val, *tt, *oname;
ccc6cda3
JA
7295 SHELL_VAR *v;
7296
a0c0a00f 7297 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
ccc6cda3
JA
7298 return ((char *)NULL);
7299
0628567a 7300 oname = this_command_name;
ccc6cda3
JA
7301 this_command_name = varname;
7302
495aee44 7303 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3 7304 if (vtype == -1)
0628567a
JA
7305 {
7306 this_command_name = oname;
7307 return ((char *)NULL);
7308 }
ccc6cda3 7309
b80f6443
JA
7310 starsub = vtype & VT_STARSUB;
7311 vtype &= ~VT_STARSUB;
7312
3185942a 7313 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
0628567a 7314 this_command_name = oname;
e8ce775d 7315 if (r <= 0)
ac50fbac
CR
7316 {
7317 if (vtype == VT_VARIABLE)
7318 FREE (val);
7319 return ((r == 0) ? &expand_param_error : (char *)NULL);
7320 }
ccc6cda3
JA
7321
7322 switch (vtype)
7323 {
7324 case VT_VARIABLE:
d166f048 7325 case VT_ARRAYMEMBER:
b80f6443
JA
7326#if defined (HANDLE_MULTIBYTE)
7327 if (MB_CUR_MAX > 1)
7328 tt = mb_substring (val, e1, e2);
7329 else
7330#endif
7117c2d2 7331 tt = substring (val, e1, e2);
b80f6443 7332
7117c2d2
JA
7333 if (vtype == VT_VARIABLE)
7334 FREE (val);
7335 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7336 temp = quote_string (tt);
7337 else
7338 temp = tt ? quote_escapes (tt) : (char *)NULL;
7339 FREE (tt);
ccc6cda3
JA
7340 break;
7341 case VT_POSPARMS:
7117c2d2
JA
7342 tt = pos_params (varname, e1, e2, quoted);
7343 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
7344 {
7345 temp = tt ? quote_escapes (tt) : (char *)NULL;
7346 FREE (tt);
7347 }
7348 else
7349 temp = tt;
ccc6cda3
JA
7350 break;
7351#if defined (ARRAY_VARS)
7352 case VT_ARRAYVAR:
3185942a
JA
7353 if (assoc_p (v))
7354 /* we convert to list and take first e2 elements starting at e1th
7355 element -- officially undefined for now */
7356 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
7357 else
b80f6443
JA
7358 /* We want E2 to be the number of elements desired (arrays can be sparse,
7359 so verify_substring_values just returns the numbers specified and we
7360 rely on array_subrange to understand how to deal with them). */
3185942a 7361 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
f1be666c
JA
7362 /* array_subrange now calls array_quote_escapes as appropriate, so the
7363 caller no longer needs to. */
ccc6cda3
JA
7364 break;
7365#endif
f73dda09
JA
7366 default:
7367 temp = (char *)NULL;
ccc6cda3
JA
7368 }
7369
7370 return temp;
7371}
7372
cce855bc
JA
7373/****************************************************************/
7374/* */
7375/* Functions to perform pattern substitution on variable values */
7376/* */
7377/****************************************************************/
7378
a0c0a00f 7379#if 0 /* Unused */
495aee44
CR
7380static int
7381shouldexp_replacement (s)
7382 char *s;
7383{
7384 register char *p;
7385
7386 for (p = s; p && *p; p++)
7387 {
7388 if (*p == '\\')
7389 p++;
7390 else if (*p == '&')
7391 return 1;
7392 }
7393 return 0;
7394}
a0c0a00f 7395#endif
495aee44 7396
ccc6cda3
JA
7397char *
7398pat_subst (string, pat, rep, mflags)
7399 char *string, *pat, *rep;
7400 int mflags;
7401{
495aee44 7402 char *ret, *s, *e, *str, *rstr, *mstr;
a0c0a00f
CR
7403 int rptr, mtype, rxpand, mlen;
7404 size_t rsize, l, replen, rslen;
495aee44
CR
7405
7406 if (string == 0)
7407 return (savestring (""));
ccc6cda3 7408
b72432fd
JA
7409 mtype = mflags & MATCH_TYPEMASK;
7410
495aee44
CR
7411#if 0 /* bash-4.2 ? */
7412 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
7413#else
7414 rxpand = 0;
7415#endif
7416
b72432fd
JA
7417 /* Special cases:
7418 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
7419 * with REP and return the result.
7420 * 2. A null pattern with mtype == MATCH_END means to append REP to
7421 * STRING and return the result.
a0c0a00f
CR
7422 * 3. A null STRING with a matching pattern means to append REP to
7423 * STRING and return the result.
495aee44 7424 * These don't understand or process `&' in the replacement string.
b72432fd
JA
7425 */
7426 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
7427 {
7428 replen = STRLEN (rep);
495aee44 7429 l = STRLEN (string);
f73dda09 7430 ret = (char *)xmalloc (replen + l + 2);
bb70624e
JA
7431 if (replen == 0)
7432 strcpy (ret, string);
7433 else if (mtype == MATCH_BEG)
b72432fd
JA
7434 {
7435 strcpy (ret, rep);
7436 strcpy (ret + replen, string);
7437 }
7438 else
7439 {
7440 strcpy (ret, string);
7441 strcpy (ret + l, rep);
7442 }
7443 return (ret);
7444 }
a0c0a00f
CR
7445 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
7446 {
7447 replen = STRLEN (rep);
7448 ret = (char *)xmalloc (replen + 1);
7449 if (replen == 0)
7450 ret[0] = '\0';
7451 else
7452 strcpy (ret, rep);
7453 return (ret);
7454 }
b72432fd 7455
f73dda09 7456 ret = (char *)xmalloc (rsize = 64);
ccc6cda3
JA
7457 ret[0] = '\0';
7458
a0c0a00f 7459 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
ccc6cda3
JA
7460 {
7461 if (match_pattern (str, pat, mtype, &s, &e) == 0)
7462 break;
7463 l = s - str;
495aee44 7464
a0c0a00f 7465 if (rep && rxpand)
495aee44 7466 {
a0c0a00f
CR
7467 int x;
7468 mlen = e - s;
7469 mstr = xmalloc (mlen + 1);
495aee44
CR
7470 for (x = 0; x < mlen; x++)
7471 mstr[x] = s[x];
a0c0a00f
CR
7472 mstr[mlen] = '\0';
7473 rstr = strcreplace (rep, '&', mstr, 0);
7474 free (mstr);
7475 rslen = strlen (rstr);
495aee44
CR
7476 }
7477 else
a0c0a00f
CR
7478 {
7479 rstr = rep;
7480 rslen = replen;
7481 }
495aee44
CR
7482
7483 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
ccc6cda3
JA
7484
7485 /* OK, now copy the leading unmatched portion of the string (from
7486 str to s) to ret starting at rptr (the current offset). Then copy
28ef6c31
JA
7487 the replacement string at ret + rptr + (s - str). Increment
7488 rptr (if necessary) and str and go on. */
ccc6cda3
JA
7489 if (l)
7490 {
7491 strncpy (ret + rptr, str, l);
7492 rptr += l;
7493 }
7494 if (replen)
7495 {
495aee44
CR
7496 strncpy (ret + rptr, rstr, rslen);
7497 rptr += rslen;
ccc6cda3
JA
7498 }
7499 str = e; /* e == end of match */
b80f6443 7500
495aee44
CR
7501 if (rstr != rep)
7502 free (rstr);
7503
ccc6cda3 7504 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
28ef6c31 7505 break;
b80f6443
JA
7506
7507 if (s == e)
0001803f
CR
7508 {
7509 /* On a zero-length match, make sure we copy one character, since
7510 we increment one character to avoid infinite recursion. */
7511 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
7512 ret[rptr++] = *str++;
7513 e++; /* avoid infinite recursion on zero-length match */
7514 }
ccc6cda3
JA
7515 }
7516
7517 /* Now copy the unmatched portion of the input string */
495aee44 7518 if (str && *str)
d166f048
JA
7519 {
7520 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
7521 strcpy (ret + rptr, str);
7522 }
ccc6cda3
JA
7523 else
7524 ret[rptr] = '\0';
7525
7526 return ret;
7527}
7528
7529/* Do pattern match and replacement on the positional parameters. */
7530static char *
7531pos_params_pat_subst (string, pat, rep, mflags)
7532 char *string, *pat, *rep;
7533 int mflags;
7534{
7535 WORD_LIST *save, *params;
7536 WORD_DESC *w;
0628567a 7537 char *ret;
3185942a 7538 int pchar, qflags;
ccc6cda3
JA
7539
7540 save = params = list_rest_of_args ();
7541 if (save == 0)
7542 return ((char *)NULL);
7543
7544 for ( ; params; params = params->next)
7545 {
7546 ret = pat_subst (params->word->word, pat, rep, mflags);
95732b49
JA
7547 w = alloc_word_desc ();
7548 w->word = ret ? ret : savestring ("");
ccc6cda3
JA
7549 dispose_word (params->word);
7550 params->word = w;
ccc6cda3
JA
7551 }
7552
3185942a
JA
7553 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7554 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7555
3185942a 7556 ret = string_list_pos_params (pchar, save, qflags);
3185942a 7557
ccc6cda3
JA
7558 dispose_words (save);
7559
7560 return (ret);
7561}
7562
cce855bc
JA
7563/* Perform pattern substitution on VALUE, which is the expansion of
7564 VARNAME. PATSUB is an expression supplying the pattern to match
7565 and the string to substitute. QUOTED is a flags word containing
7566 the type of quoting currently in effect. */
ccc6cda3 7567static char *
a0c0a00f 7568parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
495aee44
CR
7569 char *varname, *value;
7570 int ind;
7571 char *patsub;
a0c0a00f 7572 int quoted, pflags, flags;
ccc6cda3 7573{
3185942a 7574 int vtype, mflags, starsub, delim;
7117c2d2 7575 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
ccc6cda3
JA
7576 SHELL_VAR *v;
7577
7578 if (value == 0)
7579 return ((char *)NULL);
7580
7581 this_command_name = varname;
7582
495aee44 7583 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3
JA
7584 if (vtype == -1)
7585 return ((char *)NULL);
7586
b80f6443
JA
7587 starsub = vtype & VT_STARSUB;
7588 vtype &= ~VT_STARSUB;
7589
ccc6cda3 7590 mflags = 0;
ac50fbac
CR
7591 /* PATSUB is never NULL when this is called. */
7592 if (*patsub == '/')
f1be666c
JA
7593 {
7594 mflags |= MATCH_GLOBREP;
7595 patsub++;
7596 }
7117c2d2
JA
7597
7598 /* Malloc this because expand_string_if_necessary or one of the expansion
7599 functions in its call chain may free it on a substitution error. */
bb70624e 7600 lpatsub = savestring (patsub);
ccc6cda3
JA
7601
7602 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7603 mflags |= MATCH_QUOTED;
7604
b80f6443
JA
7605 if (starsub)
7606 mflags |= MATCH_STARSUB;
7607
a0c0a00f
CR
7608 if (pflags & PF_ASSIGNRHS)
7609 mflags |= MATCH_ASSIGNRHS;
7610
0628567a
JA
7611 /* If the pattern starts with a `/', make sure we skip over it when looking
7612 for the replacement delimiter. */
3185942a
JA
7613 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
7614 if (lpatsub[delim] == '/')
7615 {
7616 lpatsub[delim] = 0;
7617 rep = lpatsub + delim + 1;
7618 }
7619 else
7620 rep = (char *)NULL;
ccc6cda3
JA
7621
7622 if (rep && *rep == '\0')
7623 rep = (char *)NULL;
7624
b80f6443
JA
7625 /* Perform the same expansions on the pattern as performed by the
7626 pattern removal expansions. */
7627 pat = getpattern (lpatsub, quoted, 1);
bb70624e 7628
ccc6cda3 7629 if (rep)
d166f048 7630 {
ac50fbac
CR
7631 /* We want to perform quote removal on the expanded replacement even if
7632 the entire expansion is double-quoted because the parser and string
7633 extraction functions treated quotes in the replacement string as
7634 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
7635 if (shell_compatibility_level > 42)
7636 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
7637 /* This is the bash-4.2 code. */
7638 else if ((mflags & MATCH_QUOTED) == 0)
f73dda09 7639 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
d166f048 7640 else
f73dda09 7641 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
d166f048 7642 }
ccc6cda3 7643
0628567a 7644 /* ksh93 doesn't allow the match specifier to be a part of the expanded
f1be666c
JA
7645 pattern. This is an extension. Make sure we don't anchor the pattern
7646 at the beginning or end of the string if we're doing global replacement,
7647 though. */
ccc6cda3 7648 p = pat;
f1be666c
JA
7649 if (mflags & MATCH_GLOBREP)
7650 mflags |= MATCH_ANY;
0628567a 7651 else if (pat && pat[0] == '#')
ccc6cda3
JA
7652 {
7653 mflags |= MATCH_BEG;
7654 p++;
7655 }
d166f048 7656 else if (pat && pat[0] == '%')
ccc6cda3
JA
7657 {
7658 mflags |= MATCH_END;
7659 p++;
7660 }
7661 else
7662 mflags |= MATCH_ANY;
7663
cce855bc
JA
7664 /* OK, we now want to substitute REP for PAT in VAL. If
7665 flags & MATCH_GLOBREP is non-zero, the substitution is done
7666 everywhere, otherwise only the first occurrence of PAT is
7117c2d2
JA
7667 replaced. The pattern matching code doesn't understand
7668 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
7669 values passed in (VT_VARIABLE) so the pattern substitution
7670 code works right. We need to requote special chars after
7671 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
7672 other cases if QUOTED == 0, since the posparams and arrays
7673 indexed by * or @ do special things when QUOTED != 0. */
7674
ccc6cda3
JA
7675 switch (vtype)
7676 {
7677 case VT_VARIABLE:
d166f048 7678 case VT_ARRAYMEMBER:
ccc6cda3 7679 temp = pat_subst (val, p, rep, mflags);
7117c2d2
JA
7680 if (vtype == VT_VARIABLE)
7681 FREE (val);
7682 if (temp)
7683 {
3185942a 7684 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7117c2d2
JA
7685 free (temp);
7686 temp = tt;
7687 }
ccc6cda3
JA
7688 break;
7689 case VT_POSPARMS:
7690 temp = pos_params_pat_subst (val, p, rep, mflags);
7117c2d2
JA
7691 if (temp && (mflags & MATCH_QUOTED) == 0)
7692 {
7693 tt = quote_escapes (temp);
7694 free (temp);
7695 temp = tt;
7696 }
ccc6cda3
JA
7697 break;
7698#if defined (ARRAY_VARS)
7699 case VT_ARRAYVAR:
3185942a
JA
7700 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
7701 : array_patsub (array_cell (v), p, rep, mflags);
7702 /* Don't call quote_escapes anymore; array_patsub calls
7703 array_quote_escapes as appropriate before adding the
7704 space separators; ditto for assoc_patsub. */
7705 break;
7706#endif
7707 }
7708
7709 FREE (pat);
7710 FREE (rep);
7711 free (lpatsub);
7712
7713 return temp;
7714}
7715
7716/****************************************************************/
7717/* */
7718/* Functions to perform case modification on variable values */
7719/* */
7720/****************************************************************/
7721
7722/* Do case modification on the positional parameters. */
7723
7724static char *
7725pos_params_modcase (string, pat, modop, mflags)
7726 char *string, *pat;
7727 int modop;
7728 int mflags;
7729{
7730 WORD_LIST *save, *params;
7731 WORD_DESC *w;
7732 char *ret;
7733 int pchar, qflags;
7734
7735 save = params = list_rest_of_args ();
7736 if (save == 0)
7737 return ((char *)NULL);
7738
7739 for ( ; params; params = params->next)
7740 {
7741 ret = sh_modcase (params->word->word, pat, modop);
7742 w = alloc_word_desc ();
7743 w->word = ret ? ret : savestring ("");
7744 dispose_word (params->word);
7745 params->word = w;
7746 }
7747
7748 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7749 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7750
7751 ret = string_list_pos_params (pchar, save, qflags);
7752 dispose_words (save);
7753
7754 return (ret);
7755}
7756
7757/* Perform case modification on VALUE, which is the expansion of
7758 VARNAME. MODSPEC is an expression supplying the type of modification
7759 to perform. QUOTED is a flags word containing the type of quoting
7760 currently in effect. */
7761static char *
495aee44 7762parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
3185942a 7763 char *varname, *value;
495aee44 7764 int ind, modspec;
3185942a 7765 char *patspec;
495aee44 7766 int quoted, flags;
3185942a
JA
7767{
7768 int vtype, starsub, modop, mflags, x;
7769 char *val, *temp, *pat, *p, *lpat, *tt;
7770 SHELL_VAR *v;
7771
7772 if (value == 0)
7773 return ((char *)NULL);
7774
7775 this_command_name = varname;
7776
495aee44 7777 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
3185942a
JA
7778 if (vtype == -1)
7779 return ((char *)NULL);
7780
7781 starsub = vtype & VT_STARSUB;
7782 vtype &= ~VT_STARSUB;
7783
7784 modop = 0;
7785 mflags = 0;
7786 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7787 mflags |= MATCH_QUOTED;
7788 if (starsub)
7789 mflags |= MATCH_STARSUB;
7790
7791 p = patspec;
7792 if (modspec == '^')
7793 {
7794 x = p && p[0] == modspec;
17345e5a 7795 modop = x ? CASE_UPPER : CASE_UPFIRST;
3185942a
JA
7796 p += x;
7797 }
7798 else if (modspec == ',')
7799 {
7800 x = p && p[0] == modspec;
17345e5a 7801 modop = x ? CASE_LOWER : CASE_LOWFIRST;
3185942a
JA
7802 p += x;
7803 }
7804 else if (modspec == '~')
7805 {
7806 x = p && p[0] == modspec;
7807 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7808 p += x;
7809 }
7810
7811 lpat = p ? savestring (p) : 0;
7812 /* Perform the same expansions on the pattern as performed by the
7813 pattern removal expansions. FOR LATER */
7814 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7815
7816 /* OK, now we do the case modification. */
7817 switch (vtype)
7818 {
7819 case VT_VARIABLE:
7820 case VT_ARRAYMEMBER:
7821 temp = sh_modcase (val, pat, modop);
7822 if (vtype == VT_VARIABLE)
7823 FREE (val);
7824 if (temp)
7825 {
7826 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7827 free (temp);
7828 temp = tt;
7829 }
7830 break;
7831
7832 case VT_POSPARMS:
7833 temp = pos_params_modcase (val, pat, modop, mflags);
7834 if (temp && (mflags & MATCH_QUOTED) == 0)
7117c2d2
JA
7835 {
7836 tt = quote_escapes (temp);
7837 free (temp);
7838 temp = tt;
7839 }
3185942a
JA
7840 break;
7841
7842#if defined (ARRAY_VARS)
7843 case VT_ARRAYVAR:
7844 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7845 : array_modcase (array_cell (v), pat, modop, mflags);
7846 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7847 as appropriate before adding the space separators; ditto for
7848 assoc_modcase. */
ccc6cda3
JA
7849 break;
7850#endif
7851 }
7852
7853 FREE (pat);
3185942a 7854 free (lpat);
ccc6cda3
JA
7855
7856 return temp;
7857}
7858
0628567a
JA
7859/* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7860 any occur, this must be a nested command substitution, so return 0.
7861 Otherwise, return 1. A valid arithmetic expression must always have a
7862 ( before a matching ), so any cases where there are more right parens
7863 means that this must not be an arithmetic expression, though the parser
7864 will not accept it without a balanced total number of parens. */
7865static int
7866chk_arithsub (s, len)
7867 const char *s;
7868 int len;
7869{
7870 int i, count;
7871 DECLARE_MBSTATE;
7872
7873 i = count = 0;
7874 while (i < len)
7875 {
0001803f 7876 if (s[i] == LPAREN)
0628567a 7877 count++;
0001803f 7878 else if (s[i] == RPAREN)
0628567a
JA
7879 {
7880 count--;
7881 if (count < 0)
7882 return 0;
7883 }
7884
7885 switch (s[i])
7886 {
7887 default:
7888 ADVANCE_CHAR (s, len, i);
7889 break;
7890
7891 case '\\':
7892 i++;
7893 if (s[i])
7894 ADVANCE_CHAR (s, len, i);
7895 break;
7896
7897 case '\'':
a0c0a00f 7898 i = skip_single_quoted (s, len, ++i, 0);
0628567a
JA
7899 break;
7900
7901 case '"':
a0c0a00f 7902 i = skip_double_quoted ((char *)s, len, ++i, 0);
0628567a
JA
7903 break;
7904 }
7905 }
7906
7907 return (count == 0);
7908}
7909
cce855bc
JA
7910/****************************************************************/
7911/* */
7912/* Functions to perform parameter expansion on a string */
7913/* */
7914/****************************************************************/
7915
3185942a 7916/* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
95732b49 7917static WORD_DESC *
0001803f 7918parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
ccc6cda3 7919 char *string;
a0c0a00f 7920 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
ccc6cda3
JA
7921{
7922 int check_nullness, var_is_set, var_is_null, var_is_special;
3185942a 7923 int want_substring, want_indir, want_patsub, want_casemod;
ccc6cda3 7924 char *name, *value, *temp, *temp1;
95732b49 7925 WORD_DESC *tdesc, *ret;
a0c0a00f 7926 int t_index, sindex, c, tflag, modspec, all_element_arrayref;
7117c2d2 7927 intmax_t number;
495aee44 7928 arrayind_t ind;
ccc6cda3 7929
3185942a 7930 temp = temp1 = value = (char *)NULL;
ccc6cda3 7931 var_is_set = var_is_null = var_is_special = check_nullness = 0;
3185942a 7932 want_substring = want_indir = want_patsub = want_casemod = 0;
ccc6cda3 7933
a0c0a00f
CR
7934 all_element_arrayref = 0;
7935
cce855bc
JA
7936 sindex = *indexp;
7937 t_index = ++sindex;
0628567a
JA
7938 /* ${#var} doesn't have any of the other parameter expansions on it. */
7939 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
3185942a 7940 name = string_extract (string, &t_index, "}", SX_VARNAME);
0628567a 7941 else
3185942a
JA
7942#if defined (CASEMOD_EXPANSIONS)
7943 /* To enable case-toggling expansions using the `~' operator character
7944 change the 1 to 0. */
7945# if defined (CASEMOD_CAPCASE)
a0c0a00f 7946 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
3185942a 7947# else
a0c0a00f 7948 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
3185942a
JA
7949# endif /* CASEMOD_CAPCASE */
7950#else
a0c0a00f 7951 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
3185942a 7952#endif /* CASEMOD_EXPANSIONS */
cce855bc 7953
a0c0a00f
CR
7954 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
7955 the cleanest code ever. */
7956 if (*name == 0 && sindex == t_index && string[sindex] == '@')
7957 {
7958 name = (char *)xrealloc (name, 2);
7959 name[0] = '@';
7960 name[1] = '\0';
7961 t_index++;
7962 }
7963 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == '}')
7964 {
7965 name = (char *)xrealloc (name, t_index - sindex + 2);
7966 name[t_index - sindex] = '@';
7967 name[t_index - sindex + 1] = '\0';
7968 t_index++;
7969 }
7970
95732b49
JA
7971 ret = 0;
7972 tflag = 0;
7973
495aee44
CR
7974 ind = INTMAX_MIN;
7975
cce855bc
JA
7976 /* If the name really consists of a special variable, then make sure
7977 that we have the entire name. We don't allow indirect references
a0c0a00f
CR
7978 to special variables except `#', `?', `@' and `*'. This clause is
7979 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
7980 general. */
495aee44 7981 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
a0c0a00f 7982 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
495aee44 7983 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
ccc6cda3
JA
7984 {
7985 t_index++;
a0c0a00f 7986 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
ac50fbac 7987 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
ccc6cda3
JA
7988 *name = string[sindex];
7989 if (string[sindex] == '!')
7990 {
28ef6c31
JA
7991 /* indirect reference of $#, $?, $@, or $* */
7992 name[1] = string[sindex + 1];
7993 strcpy (name + 2, temp1);
ccc6cda3 7994 }
cce855bc 7995 else
ccc6cda3
JA
7996 strcpy (name + 1, temp1);
7997 free (temp1);
7998 }
7999 sindex = t_index;
8000
8001 /* Find out what character ended the variable name. Then
8002 do the appropriate thing. */
8003 if (c = string[sindex])
8004 sindex++;
8005
8006 /* If c is followed by one of the valid parameter expansion
8007 characters, move past it as normal. If not, assume that
8008 a substring specification is being given, and do not move
8009 past it. */
28ef6c31 8010 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
ccc6cda3
JA
8011 {
8012 check_nullness++;
8013 if (c = string[sindex])
8014 sindex++;
8015 }
cce855bc 8016 else if (c == ':' && string[sindex] != RBRACE)
ccc6cda3 8017 want_substring = 1;
ac50fbac 8018 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
ccc6cda3 8019 want_patsub = 1;
3185942a
JA
8020#if defined (CASEMOD_EXPANSIONS)
8021 else if (c == '^' || c == ',' || c == '~')
8022 {
8023 modspec = c;
8024 want_casemod = 1;
8025 }
8026#endif
ccc6cda3 8027
cce855bc
JA
8028 /* Catch the valid and invalid brace expressions that made it through the
8029 tests above. */
8030 /* ${#-} is a valid expansion and means to take the length of $-.
8031 Similarly for ${#?} and ${##}... */
8032 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
28ef6c31 8033 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
cce855bc 8034 {
f73dda09 8035 name = (char *)xrealloc (name, 3);
cce855bc
JA
8036 name[1] = c;
8037 name[2] = '\0';
8038 c = string[sindex++];
8039 }
8040
8041 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8042 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8043 member (c, "%:=+/") && string[sindex] == RBRACE)
8044 {
8045 temp = (char *)NULL;
a0c0a00f 8046 goto bad_substitution; /* XXX - substitution error */
cce855bc
JA
8047 }
8048
8049 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8050 either a variable name, one of the positional parameters or a special
8051 variable that expands to one of the positional parameters. */
8052 want_indir = *name == '!' &&
f73dda09 8053 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7117c2d2 8054 || VALID_INDIR_PARAM (name[1]));
ccc6cda3
JA
8055
8056 /* Determine the value of this variable. */
8057
cce855bc 8058 /* Check for special variables, directly referenced. */
bb70624e 8059 if (SPECIAL_VAR (name, want_indir))
ccc6cda3
JA
8060 var_is_special++;
8061
cce855bc
JA
8062 /* Check for special expansion things, like the length of a parameter */
8063 if (*name == '#' && name[1])
ccc6cda3 8064 {
cce855bc 8065 /* If we are not pointing at the character just after the
28ef6c31
JA
8066 closing brace, then we haven't gotten all of the name.
8067 Since it begins with a special character, this is a bad
8068 substitution. Also check NAME for validity before trying
8069 to go on. */
cce855bc 8070 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
ccc6cda3
JA
8071 {
8072 temp = (char *)NULL;
a0c0a00f 8073 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8074 }
8075
8076 number = parameter_brace_expand_length (name);
495aee44
CR
8077 if (number == INTMAX_MIN && unbound_vars_is_error)
8078 {
8079 last_command_exit_value = EXECUTION_FAILURE;
8080 err_unboundvar (name+1);
8081 free (name);
8082 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8083 }
ccc6cda3
JA
8084 free (name);
8085
8086 *indexp = sindex;
95732b49
JA
8087 if (number < 0)
8088 return (&expand_wdesc_error);
8089 else
8090 {
8091 ret = alloc_word_desc ();
8092 ret->word = itos (number);
8093 return ret;
8094 }
ccc6cda3
JA
8095 }
8096
8097 /* ${@} is identical to $@. */
8098 if (name[0] == '@' && name[1] == '\0')
8099 {
8100 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8101 *quoted_dollar_atp = 1;
8102
8103 if (contains_dollar_at)
8104 *contains_dollar_at = 1;
ac50fbac
CR
8105
8106 tflag |= W_DOLLARAT;
ccc6cda3
JA
8107 }
8108
b80f6443 8109 /* Process ${!PREFIX*} expansion. */
bb70624e
JA
8110 if (want_indir && string[sindex - 1] == RBRACE &&
8111 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
f73dda09 8112 legal_variable_starter ((unsigned char) name[1]))
bb70624e
JA
8113 {
8114 char **x;
8115 WORD_LIST *xlist;
8116
8117 temp1 = savestring (name + 1);
8118 number = strlen (temp1);
8119 temp1[number - 1] = '\0';
8120 x = all_variables_matching_prefix (temp1);
7117c2d2 8121 xlist = strvec_to_word_list (x, 0, 0);
28ef6c31
JA
8122 if (string[sindex - 2] == '*')
8123 temp = string_list_dollar_star (xlist);
8124 else
8125 {
a0c0a00f 8126 temp = string_list_dollar_at (xlist, quoted, 0);
28ef6c31
JA
8127 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8128 *quoted_dollar_atp = 1;
8129 if (contains_dollar_at)
8130 *contains_dollar_at = 1;
ac50fbac
CR
8131
8132 tflag |= W_DOLLARAT;
28ef6c31 8133 }
bb70624e 8134 free (x);
89a92869 8135 dispose_words (xlist);
bb70624e
JA
8136 free (temp1);
8137 *indexp = sindex;
95732b49 8138
ac50fbac
CR
8139 free (name);
8140
95732b49
JA
8141 ret = alloc_word_desc ();
8142 ret->word = temp;
ac50fbac 8143 ret->flags = tflag; /* XXX */
95732b49 8144 return ret;
bb70624e 8145 }
b80f6443
JA
8146
8147#if defined (ARRAY_VARS)
8148 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
8149 if (want_indir && string[sindex - 1] == RBRACE &&
a0c0a00f 8150 string[sindex - 2] == ']' && valid_array_reference (name+1, 0))
b80f6443
JA
8151 {
8152 char *x, *x1;
8153
8154 temp1 = savestring (name + 1);
8155 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
8156 FREE (x);
8157 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
8158 {
3185942a 8159 temp = array_keys (temp1, quoted); /* handles assoc vars too */
b80f6443
JA
8160 if (x1[0] == '@')
8161 {
8162 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8163 *quoted_dollar_atp = 1;
8164 if (contains_dollar_at)
8165 *contains_dollar_at = 1;
ac50fbac
CR
8166
8167 tflag |= W_DOLLARAT;
b80f6443
JA
8168 }
8169
a0c0a00f 8170 free (name);
b80f6443
JA
8171 free (temp1);
8172 *indexp = sindex;
95732b49
JA
8173
8174 ret = alloc_word_desc ();
8175 ret->word = temp;
ac50fbac 8176 ret->flags = tflag; /* XXX */
95732b49 8177 return ret;
b80f6443
JA
8178 }
8179
8180 free (temp1);
8181 }
8182#endif /* ARRAY_VARS */
bb70624e 8183
ccc6cda3
JA
8184 /* Make sure that NAME is valid before trying to go on. */
8185 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
8186 var_is_special) == 0)
8187 {
8188 temp = (char *)NULL;
a0c0a00f 8189 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8190 }
8191
8192 if (want_indir)
1a1f8b54
CR
8193 {
8194 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
a0c0a00f
CR
8195 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8196 {
8197 temp = (char *)NULL;
8198 goto bad_substitution;
8199 }
1a1f8b54
CR
8200 /* Turn off the W_ARRAYIND flag because there is no way for this function
8201 to return the index we're supposed to be using. */
8202 if (tdesc && tdesc->flags)
8203 tdesc->flags &= ~W_ARRAYIND;
8204 }
95732b49 8205 else
ac50fbac 8206 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
95732b49
JA
8207
8208 if (tdesc)
8209 {
8210 temp = tdesc->word;
8211 tflag = tdesc->flags;
8212 dispose_word_desc (tdesc);
8213 }
ccc6cda3 8214 else
95732b49 8215 temp = (char *)0;
ccc6cda3 8216
ac50fbac
CR
8217 if (temp == &expand_param_error || temp == &expand_param_fatal)
8218 {
8219 FREE (name);
8220 FREE (value);
8221 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8222 }
8223
ccc6cda3 8224#if defined (ARRAY_VARS)
a0c0a00f
CR
8225 if (valid_array_reference (name, 0))
8226 {
8227 int qflags;
8228 char *t;
8229
8230 qflags = quoted;
8231 /* If in a context where word splitting will not take place, treat as
8232 if double-quoted. Has effects with $* and ${array[*]} */
8233 if (pflags & PF_ASSIGNRHS)
8234 qflags |= Q_DOUBLE_QUOTES;
8235 chk_atstar (name, qflags, quoted_dollar_atp, contains_dollar_at);
8236 /* We duplicate a little code here */
8237 t = mbschr (name, '[');
8238 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == ']')
8239 all_element_arrayref = 1;
8240 }
ccc6cda3
JA
8241#endif
8242
8243 var_is_set = temp != (char *)0;
8244 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
ac50fbac
CR
8245 /* XXX - this may not need to be restricted to special variables */
8246 if (check_nullness)
8247 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
ccc6cda3
JA
8248
8249 /* Get the rest of the stuff inside the braces. */
cce855bc 8250 if (c && c != RBRACE)
ccc6cda3
JA
8251 {
8252 /* Extract the contents of the ${ ... } expansion
28ef6c31 8253 according to the Posix.2 rules. */
49ed961b 8254 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
cce855bc 8255 if (string[sindex] == RBRACE)
28ef6c31 8256 sindex++;
ccc6cda3 8257 else
a0c0a00f 8258 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8259 }
8260 else
8261 value = (char *)NULL;
726f6388 8262
ccc6cda3
JA
8263 *indexp = sindex;
8264
495aee44
CR
8265 /* All the cases where an expansion can possibly generate an unbound
8266 variable error. */
8267 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
8268 {
a0c0a00f 8269 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
495aee44
CR
8270 {
8271 last_command_exit_value = EXECUTION_FAILURE;
8272 err_unboundvar (name);
8273 FREE (value);
8274 FREE (temp);
8275 free (name);
8276 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8277 }
8278 }
8279
ccc6cda3
JA
8280 /* If this is a substring spec, process it and add the result. */
8281 if (want_substring)
726f6388 8282 {
495aee44 8283 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8284 FREE (name);
8285 FREE (value);
8286 FREE (temp);
95732b49
JA
8287
8288 if (temp1 == &expand_param_error)
8289 return (&expand_wdesc_error);
8290 else if (temp1 == &expand_param_fatal)
8291 return (&expand_wdesc_fatal);
8292
8293 ret = alloc_word_desc ();
8294 ret->word = temp1;
ac50fbac
CR
8295 /* We test quoted_dollar_atp because we want variants with double-quoted
8296 "$@" to take a different code path. In fact, we make sure at the end
8297 of expand_word_internal that we're only looking at these flags if
8298 quoted_dollar_at == 0. */
8299 if (temp1 &&
8300 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8301 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
0628567a 8302 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
95732b49 8303 return ret;
726f6388 8304 }
ccc6cda3 8305 else if (want_patsub)
726f6388 8306 {
a0c0a00f 8307 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8308 FREE (name);
8309 FREE (value);
8310 FREE (temp);
95732b49
JA
8311
8312 if (temp1 == &expand_param_error)
8313 return (&expand_wdesc_error);
8314 else if (temp1 == &expand_param_fatal)
8315 return (&expand_wdesc_fatal);
8316
8317 ret = alloc_word_desc ();
8318 ret->word = temp1;
ac50fbac
CR
8319 if (temp1 &&
8320 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8321 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a 8322 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
95732b49 8323 return ret;
ccc6cda3 8324 }
3185942a
JA
8325#if defined (CASEMOD_EXPANSIONS)
8326 else if (want_casemod)
8327 {
495aee44 8328 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
3185942a
JA
8329 FREE (name);
8330 FREE (value);
8331 FREE (temp);
8332
8333 if (temp1 == &expand_param_error)
8334 return (&expand_wdesc_error);
8335 else if (temp1 == &expand_param_fatal)
8336 return (&expand_wdesc_fatal);
8337
8338 ret = alloc_word_desc ();
8339 ret->word = temp1;
ac50fbac
CR
8340 if (temp1 &&
8341 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8342 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a
JA
8343 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8344 return ret;
8345 }
8346#endif
726f6388 8347
ccc6cda3
JA
8348 /* Do the right thing based on which character ended the variable name. */
8349 switch (c)
8350 {
8351 default:
8352 case '\0':
a0c0a00f 8353bad_substitution:
98043138 8354 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 8355 report_error (_("%s: bad substitution"), string ? string : "??");
ccc6cda3
JA
8356 FREE (value);
8357 FREE (temp);
8358 free (name);
a0c0a00f
CR
8359 if (shell_compatibility_level <= 43)
8360 return &expand_wdesc_error;
8361 else
8362 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
ccc6cda3 8363
cce855bc 8364 case RBRACE:
ccc6cda3 8365 break;
726f6388 8366
a0c0a00f
CR
8367 case '@':
8368 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8369 free (temp);
8370 free (value);
8371 free (name);
8372 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
8373 {
8374 last_command_exit_value = EXECUTION_FAILURE;
8375 report_error (_("%s: bad substitution"), string ? string : "??");
8376 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8377 }
8378
8379 ret = alloc_word_desc ();
8380 ret->word = temp1;
8381 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8382 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8383 return ret;
8384
ccc6cda3
JA
8385 case '#': /* ${param#[#]pattern} */
8386 case '%': /* ${param%[%]pattern} */
8387 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
28ef6c31
JA
8388 {
8389 FREE (value);
ccc6cda3 8390 break;
28ef6c31 8391 }
495aee44 8392 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
8393 free (temp);
8394 free (value);
495aee44 8395 free (name);
3185942a
JA
8396
8397 ret = alloc_word_desc ();
8398 ret->word = temp1;
8399 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8400 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8401 return ret;
ccc6cda3
JA
8402
8403 case '-':
8404 case '=':
8405 case '?':
8406 case '+':
8407 if (var_is_set && var_is_null == 0)
28ef6c31
JA
8408 {
8409 /* If the operator is `+', we don't want the value of the named
8410 variable for anything, just the value of the right hand side. */
ccc6cda3
JA
8411 if (c == '+')
8412 {
28ef6c31
JA
8413 /* XXX -- if we're double-quoted and the named variable is "$@",
8414 we want to turn off any special handling of "$@" --
8415 we're not using it, so whatever is on the rhs applies. */
8416 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8417 *quoted_dollar_atp = 0;
8418 if (contains_dollar_at)
8419 *contains_dollar_at = 0;
8420
ccc6cda3
JA
8421 FREE (temp);
8422 if (value)
28ef6c31 8423 {
495aee44
CR
8424 /* From Posix discussion on austin-group list. Issue 221
8425 requires that backslashes escaping `}' inside
8426 double-quoted ${...} be removed. */
8427 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8428 quoted |= Q_DOLBRACE;
95732b49
JA
8429 ret = parameter_brace_expand_rhs (name, value, c,
8430 quoted,
a0c0a00f 8431 pflags,
95732b49
JA
8432 quoted_dollar_atp,
8433 contains_dollar_at);
8434 /* XXX - fix up later, esp. noting presence of
8435 W_HASQUOTEDNULL in ret->flags */
ccc6cda3
JA
8436 free (value);
8437 }
8438 else
28ef6c31 8439 temp = (char *)NULL;
ccc6cda3
JA
8440 }
8441 else
8442 {
8443 FREE (value);
8444 }
8445 /* Otherwise do nothing; just use the value in TEMP. */
726f6388 8446 }
ccc6cda3 8447 else /* VAR not set or VAR is NULL. */
28ef6c31 8448 {
ccc6cda3
JA
8449 FREE (temp);
8450 temp = (char *)NULL;
8451 if (c == '=' && var_is_special)
8452 {
ac50fbac 8453 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 8454 report_error (_("$%s: cannot assign in this way"), name);
ccc6cda3
JA
8455 free (name);
8456 free (value);
95732b49 8457 return &expand_wdesc_error;
ccc6cda3
JA
8458 }
8459 else if (c == '?')
8460 {
8461 parameter_brace_expand_error (name, value);
95732b49 8462 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
ccc6cda3
JA
8463 }
8464 else if (c != '+')
28ef6c31
JA
8465 {
8466 /* XXX -- if we're double-quoted and the named variable is "$@",
8467 we want to turn off any special handling of "$@" --
8468 we're not using it, so whatever is on the rhs applies. */
8469 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8470 *quoted_dollar_atp = 0;
8471 if (contains_dollar_at)
8472 *contains_dollar_at = 0;
8473
495aee44
CR
8474 /* From Posix discussion on austin-group list. Issue 221 requires
8475 that backslashes escaping `}' inside double-quoted ${...} be
8476 removed. */
8477 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8478 quoted |= Q_DOLBRACE;
a0c0a00f 8479 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
95732b49
JA
8480 quoted_dollar_atp,
8481 contains_dollar_at);
8482 /* XXX - fix up later, esp. noting presence of
8483 W_HASQUOTEDNULL in tdesc->flags */
28ef6c31 8484 }
ccc6cda3 8485 free (value);
726f6388 8486 }
28ef6c31 8487
ccc6cda3 8488 break;
726f6388 8489 }
ccc6cda3 8490 free (name);
95732b49
JA
8491
8492 if (ret == 0)
8493 {
8494 ret = alloc_word_desc ();
8495 ret->flags = tflag;
8496 ret->word = temp;
8497 }
8498 return (ret);
726f6388
JA
8499}
8500
cce855bc
JA
8501/* Expand a single ${xxx} expansion. The braces are optional. When
8502 the braces are used, parameter_brace_expand() does the work,
8503 possibly calling param_expand recursively. */
95732b49 8504static WORD_DESC *
cce855bc
JA
8505param_expand (string, sindex, quoted, expanded_something,
8506 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
8507 pflags)
8508 char *string;
8509 int *sindex, quoted, *expanded_something, *contains_dollar_at;
8510 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
8511{
a0c0a00f 8512 char *temp, *temp1, uerror[3], *savecmd;
f73dda09
JA
8513 int zindex, t_index, expok;
8514 unsigned char c;
7117c2d2 8515 intmax_t number;
cce855bc 8516 SHELL_VAR *var;
f73dda09 8517 WORD_LIST *list;
95732b49
JA
8518 WORD_DESC *tdesc, *ret;
8519 int tflag;
cce855bc 8520
a0c0a00f 8521/*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
cce855bc
JA
8522 zindex = *sindex;
8523 c = string[++zindex];
8524
8525 temp = (char *)NULL;
95732b49
JA
8526 ret = tdesc = (WORD_DESC *)NULL;
8527 tflag = 0;
cce855bc
JA
8528
8529 /* Do simple cases first. Switch on what follows '$'. */
8530 switch (c)
8531 {
8532 /* $0 .. $9? */
8533 case '0':
8534 case '1':
8535 case '2':
8536 case '3':
8537 case '4':
8538 case '5':
8539 case '6':
8540 case '7':
8541 case '8':
8542 case '9':
f73dda09 8543 temp1 = dollar_vars[TODIGIT (c)];
cce855bc
JA
8544 if (unbound_vars_is_error && temp1 == (char *)NULL)
8545 {
7117c2d2
JA
8546 uerror[0] = '$';
8547 uerror[1] = c;
8548 uerror[2] = '\0';
cce855bc 8549 last_command_exit_value = EXECUTION_FAILURE;
0001803f 8550 err_unboundvar (uerror);
95732b49 8551 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc 8552 }
b80f6443
JA
8553 if (temp1)
8554 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8555 ? quote_string (temp1)
8556 : quote_escapes (temp1);
8557 else
8558 temp = (char *)NULL;
95732b49 8559
cce855bc
JA
8560 break;
8561
8562 /* $$ -- pid of the invoking shell. */
8563 case '$':
8564 temp = itos (dollar_dollar_pid);
8565 break;
8566
8567 /* $# -- number of positional parameters. */
8568 case '#':
8569 temp = itos (number_of_args ());
8570 break;
8571
8572 /* $? -- return value of the last synchronous command. */
8573 case '?':
8574 temp = itos (last_command_exit_value);
8575 break;
8576
8577 /* $- -- flags supplied to the shell on invocation or by `set'. */
8578 case '-':
8579 temp = which_set_flags ();
8580 break;
8581
8582 /* $! -- Pid of the last asynchronous command. */
8583 case '!':
8584 /* If no asynchronous pids have been created, expand to nothing.
8585 If `set -u' has been executed, and no async processes have
8586 been created, this is an expansion error. */
8587 if (last_asynchronous_pid == NO_PID)
8588 {
8589 if (expanded_something)
8590 *expanded_something = 0;
8591 temp = (char *)NULL;
8592 if (unbound_vars_is_error)
8593 {
7117c2d2
JA
8594 uerror[0] = '$';
8595 uerror[1] = c;
8596 uerror[2] = '\0';
cce855bc 8597 last_command_exit_value = EXECUTION_FAILURE;
0001803f 8598 err_unboundvar (uerror);
95732b49 8599 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc
JA
8600 }
8601 }
8602 else
f73dda09 8603 temp = itos (last_asynchronous_pid);
cce855bc
JA
8604 break;
8605
8606 /* The only difference between this and $@ is when the arg is quoted. */
8607 case '*': /* `$*' */
8608 list = list_rest_of_args ();
8609
89a92869
CR
8610#if 0
8611 /* According to austin-group posix proposal by Geoff Clare in
8612 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8613
8614 "The shell shall write a message to standard error and
8615 immediately exit when it tries to expand an unset parameter
8616 other than the '@' and '*' special parameters."
8617 */
8618
8619 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
8620 {
8621 uerror[0] = '$';
8622 uerror[1] = '*';
8623 uerror[2] = '\0';
3185942a 8624 last_command_exit_value = EXECUTION_FAILURE;
89a92869 8625 err_unboundvar (uerror);
3185942a
JA
8626 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8627 }
89a92869 8628#endif
3185942a 8629
cce855bc
JA
8630 /* If there are no command-line arguments, this should just
8631 disappear if there are other characters in the expansion,
8632 even if it's quoted. */
8633 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
8634 temp = (char *)NULL;
0001803f 8635 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
cce855bc
JA
8636 {
8637 /* If we have "$*" we want to make a string of the positional
8638 parameters, separated by the first character of $IFS, and
8639 quote the whole string, including the separators. If IFS
8640 is unset, the parameters are separated by ' '; if $IFS is
8641 null, the parameters are concatenated. */
0001803f 8642 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
495aee44
CR
8643 if (temp)
8644 {
a0c0a00f 8645 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
495aee44
CR
8646 if (*temp == 0)
8647 tflag |= W_HASQUOTEDNULL;
a0c0a00f
CR
8648 if (temp != temp1)
8649 free (temp);
495aee44
CR
8650 temp = temp1;
8651 }
cce855bc
JA
8652 }
8653 else
28ef6c31 8654 {
95732b49
JA
8655 /* We check whether or not we're eventually going to split $* here,
8656 for example when IFS is empty and we are processing the rhs of
8657 an assignment statement. In that case, we don't separate the
8658 arguments at all. Otherwise, if the $* is not quoted it is
8659 identical to $@ */
95732b49
JA
8660# if defined (HANDLE_MULTIBYTE)
8661 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
8662# else
8663 if (expand_no_split_dollar_star && ifs_firstc == 0)
8664# endif
8665 temp = string_list_dollar_star (list);
8666 else
ac50fbac 8667 {
a0c0a00f 8668 temp = string_list_dollar_at (list, quoted, 0);
ac50fbac
CR
8669 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
8670 tflag |= W_SPLITSPACE;
a0c0a00f
CR
8671 /* If we're not quoted but we still don't want word splitting, make
8672 we quote the IFS characters to protect them from splitting (e.g.,
8673 when $@ is in the string as well). */
8674 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
8675 {
8676 temp1 = quote_string (temp);
8677 free (temp);
8678 temp = temp1;
8679 }
ac50fbac
CR
8680 }
8681
28ef6c31
JA
8682 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
8683 *contains_dollar_at = 1;
8684 }
cce855bc
JA
8685
8686 dispose_words (list);
8687 break;
8688
8689 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
8690 means that we have to turn quoting off after we split into
8691 the individually quoted arguments so that the final split
8692 on the first character of $IFS is still done. */
8693 case '@': /* `$@' */
8694 list = list_rest_of_args ();
8695
89a92869
CR
8696#if 0
8697 /* According to austin-group posix proposal by Geoff Clare in
8698 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8699
8700 "The shell shall write a message to standard error and
8701 immediately exit when it tries to expand an unset parameter
8702 other than the '@' and '*' special parameters."
8703 */
8704
8705 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
8706 {
8707 uerror[0] = '$';
8708 uerror[1] = '@';
8709 uerror[2] = '\0';
3185942a 8710 last_command_exit_value = EXECUTION_FAILURE;
89a92869 8711 err_unboundvar (uerror);
3185942a
JA
8712 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8713 }
89a92869 8714#endif
3185942a 8715
cce855bc
JA
8716 /* We want to flag the fact that we saw this. We can't turn
8717 off quoting entirely, because other characters in the
8718 string might need it (consider "\"$@\""), but we need some
8719 way to signal that the final split on the first character
8720 of $IFS should be done, even though QUOTED is 1. */
0001803f 8721 /* XXX - should this test include Q_PATQUOTE? */
cce855bc
JA
8722 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8723 *quoted_dollar_at_p = 1;
8724 if (contains_dollar_at)
8725 *contains_dollar_at = 1;
8726
8727 /* We want to separate the positional parameters with the first
8728 character of $IFS in case $IFS is something other than a space.
8729 We also want to make sure that splitting is done no matter what --
8730 according to POSIX.2, this expands to a list of the positional
8731 parameters no matter what IFS is set to. */
3b34f6e6
CR
8732 /* XXX - what to do when in a context where word splitting is not
8733 performed? Even when IFS is not the default, posix seems to imply
8734 that we behave like unquoted $* ? Maybe we should use PF_NOSPLIT2
8735 here. */
a0c0a00f
CR
8736 /* XXX - bash-4.4/bash-5.0 passing PFLAGS */
8737 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted, pflags);
cce855bc 8738
ac50fbac 8739 tflag |= W_DOLLARAT;
cce855bc
JA
8740 dispose_words (list);
8741 break;
8742
8743 case LBRACE:
0001803f 8744 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
95732b49
JA
8745 quoted_dollar_at_p,
8746 contains_dollar_at);
8747
95732b49
JA
8748 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8749 return (tdesc);
8750 temp = tdesc ? tdesc->word : (char *)0;
cce855bc
JA
8751
8752 /* XXX */
bb70624e 8753 /* Quoted nulls should be removed if there is anything else
cce855bc
JA
8754 in the string. */
8755 /* Note that we saw the quoted null so we can add one back at
8756 the end of this function if there are no other characters
28ef6c31
JA
8757 in the string, discard TEMP, and go on. The exception to
8758 this is when we have "${@}" and $1 is '', since $@ needs
8759 special handling. */
95732b49 8760 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
cce855bc
JA
8761 {
8762 if (had_quoted_null_p)
8763 *had_quoted_null_p = 1;
28ef6c31
JA
8764 if (*quoted_dollar_at_p == 0)
8765 {
8766 free (temp);
95732b49 8767 tdesc->word = temp = (char *)NULL;
28ef6c31
JA
8768 }
8769
cce855bc
JA
8770 }
8771
95732b49 8772 ret = tdesc;
cce855bc
JA
8773 goto return0;
8774
8775 /* Do command or arithmetic substitution. */
8776 case LPAREN:
8777 /* We have to extract the contents of this paren substitution. */
8778 t_index = zindex + 1;
a0c0a00f
CR
8779 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
8780 as arithmetic substitution immediately. */
8781 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
cce855bc
JA
8782 zindex = t_index;
8783
8784 /* For Posix.2-style `$(( ))' arithmetic substitution,
28ef6c31 8785 extract the expression and pass it to the evaluator. */
cce855bc
JA
8786 if (temp && *temp == LPAREN)
8787 {
8788 char *temp2;
8789 temp1 = temp + 1;
8790 temp2 = savestring (temp1);
8791 t_index = strlen (temp2) - 1;
8792
8793 if (temp2[t_index] != RPAREN)
8794 {
8795 free (temp2);
8796 goto comsub;
8797 }
8798
8799 /* Cut off ending `)' */
8800 temp2[t_index] = '\0';
8801
0628567a
JA
8802 if (chk_arithsub (temp2, t_index) == 0)
8803 {
8804 free (temp2);
0001803f
CR
8805#if 0
8806 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
8807#endif
0628567a
JA
8808 goto comsub;
8809 }
8810
cce855bc 8811 /* Expand variables found inside the expression. */
a0c0a00f 8812 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
8813 free (temp2);
8814
8815arithsub:
8816 /* No error messages. */
a0c0a00f 8817 savecmd = this_command_name;
cce855bc
JA
8818 this_command_name = (char *)NULL;
8819 number = evalexp (temp1, &expok);
a0c0a00f 8820 this_command_name = savecmd;
cce855bc
JA
8821 free (temp);
8822 free (temp1);
8823 if (expok == 0)
8824 {
8825 if (interactive_shell == 0 && posixly_correct)
8826 {
8827 last_command_exit_value = EXECUTION_FAILURE;
95732b49 8828 return (&expand_wdesc_fatal);
cce855bc
JA
8829 }
8830 else
95732b49 8831 return (&expand_wdesc_error);
cce855bc
JA
8832 }
8833 temp = itos (number);
8834 break;
8835 }
8836
8837comsub:
b80f6443
JA
8838 if (pflags & PF_NOCOMSUB)
8839 /* we need zindex+1 because string[zindex] == RPAREN */
8840 temp1 = substring (string, *sindex, zindex+1);
8841 else
3185942a
JA
8842 {
8843 tdesc = command_substitute (temp, quoted);
8844 temp1 = tdesc ? tdesc->word : (char *)NULL;
8845 if (tdesc)
8846 dispose_word_desc (tdesc);
8847 }
cce855bc
JA
8848 FREE (temp);
8849 temp = temp1;
8850 break;
8851
8852 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
8853 away in a future bash release. */
8854 case '[':
bb70624e 8855 /* Extract the contents of this arithmetic substitution. */
cce855bc
JA
8856 t_index = zindex + 1;
8857 temp = extract_arithmetic_subst (string, &t_index);
8858 zindex = t_index;
3185942a
JA
8859 if (temp == 0)
8860 {
8861 temp = savestring (string);
8862 if (expanded_something)
8863 *expanded_something = 0;
8864 goto return0;
8865 }
cce855bc
JA
8866
8867 /* Do initial variable expansion. */
a0c0a00f 8868 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
8869
8870 goto arithsub;
8871
8872 default:
8873 /* Find the variable in VARIABLE_LIST. */
8874 temp = (char *)NULL;
8875
8876 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
8877 ;
8878 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
8879
8880 /* If this isn't a variable name, then just output the `$'. */
8881 if (temp1 == 0 || *temp1 == '\0')
8882 {
8883 FREE (temp1);
f73dda09 8884 temp = (char *)xmalloc (2);
cce855bc
JA
8885 temp[0] = '$';
8886 temp[1] = '\0';
8887 if (expanded_something)
8888 *expanded_something = 0;
8889 goto return0;
8890 }
8891
8892 /* If the variable exists, return its value cell. */
8893 var = find_variable (temp1);
8894
7117c2d2 8895 if (var && invisible_p (var) == 0 && var_isset (var))
cce855bc
JA
8896 {
8897#if defined (ARRAY_VARS)
3185942a 8898 if (assoc_p (var) || array_p (var))
cce855bc 8899 {
3185942a
JA
8900 temp = array_p (var) ? array_reference (array_cell (var), 0)
8901 : assoc_reference (assoc_cell (var), "0");
cce855bc 8902 if (temp)
b80f6443
JA
8903 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8904 ? quote_string (temp)
8905 : quote_escapes (temp);
8906 else if (unbound_vars_is_error)
8907 goto unbound_variable;
cce855bc
JA
8908 }
8909 else
8910#endif
b80f6443
JA
8911 {
8912 temp = value_cell (var);
8913
8914 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8915 ? quote_string (temp)
8916 : quote_escapes (temp);
8917 }
8918
cce855bc 8919 free (temp1);
7117c2d2 8920
cce855bc
JA
8921 goto return0;
8922 }
b64a0e1d
CR
8923 else if (var && (invisible_p (var) || var_isset (var) == 0))
8924 temp = (char *)NULL;
a0c0a00f 8925 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
ac50fbac
CR
8926 {
8927 temp = nameref_cell (var);
8928#if defined (ARRAY_VARS)
a0c0a00f 8929 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
8930 {
8931 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8932 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8933 return (tdesc);
8934 ret = tdesc;
8935 goto return0;
8936 }
8937 else
8938#endif
8939 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8940 if (temp && *temp && legal_identifier (temp) == 0)
8941 {
8942 last_command_exit_value = EXECUTION_FAILURE;
8943 report_error (_("%s: invalid variable name for name reference"), temp);
8944 return (&expand_wdesc_error); /* XXX */
8945 }
8946 else
8947 temp = (char *)NULL;
8948 }
cce855bc
JA
8949
8950 temp = (char *)NULL;
8951
b80f6443 8952unbound_variable:
cce855bc 8953 if (unbound_vars_is_error)
0001803f
CR
8954 {
8955 last_command_exit_value = EXECUTION_FAILURE;
8956 err_unboundvar (temp1);
8957 }
cce855bc
JA
8958 else
8959 {
8960 free (temp1);
8961 goto return0;
8962 }
8963
8964 free (temp1);
8965 last_command_exit_value = EXECUTION_FAILURE;
8966 return ((unbound_vars_is_error && interactive_shell == 0)
95732b49
JA
8967 ? &expand_wdesc_fatal
8968 : &expand_wdesc_error);
cce855bc
JA
8969 }
8970
8971 if (string[zindex])
8972 zindex++;
8973
8974return0:
8975 *sindex = zindex;
95732b49
JA
8976
8977 if (ret == 0)
8978 {
8979 ret = alloc_word_desc ();
8980 ret->flags = tflag; /* XXX */
8981 ret->word = temp;
8982 }
8983 return ret;
cce855bc
JA
8984}
8985
a0c0a00f
CR
8986void
8987invalidate_cached_quoted_dollar_at ()
8988{
8989 dispose_words (cached_quoted_dollar_at);
8990 cached_quoted_dollar_at = 0;
8991}
8992
cce855bc
JA
8993/* Make a word list which is the result of parameter and variable
8994 expansion, command substitution, arithmetic substitution, and
8995 quote removal of WORD. Return a pointer to a WORD_LIST which is
8996 the result of the expansion. If WORD contains a null word, the
8997 word list returned is also null.
726f6388 8998
ccc6cda3
JA
8999 QUOTED contains flag values defined in shell.h.
9000
b72432fd
JA
9001 ISEXP is used to tell expand_word_internal that the word should be
9002 treated as the result of an expansion. This has implications for
9003 how IFS characters in the word are treated.
9004
726f6388
JA
9005 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
9006 they point to an integer value which receives information about expansion.
9007 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
9008 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
9009 else zero.
9010
9011 This only does word splitting in the case of $@ expansion. In that
9012 case, we split on ' '. */
9013
9014/* Values for the local variable quoted_state. */
9015#define UNQUOTED 0
9016#define PARTIALLY_QUOTED 1
9017#define WHOLLY_QUOTED 2
9018
9019static WORD_LIST *
b72432fd 9020expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
726f6388 9021 WORD_DESC *word;
b72432fd 9022 int quoted, isexp;
726f6388
JA
9023 int *contains_dollar_at;
9024 int *expanded_something;
9025{
ccc6cda3
JA
9026 WORD_LIST *list;
9027 WORD_DESC *tword;
726f6388
JA
9028
9029 /* The intermediate string that we build while expanding. */
ccc6cda3 9030 char *istring;
726f6388
JA
9031
9032 /* The current size of the above object. */
a0c0a00f 9033 size_t istring_size;
726f6388
JA
9034
9035 /* Index into ISTRING. */
ccc6cda3 9036 int istring_index;
726f6388
JA
9037
9038 /* Temporary string storage. */
ccc6cda3 9039 char *temp, *temp1;
726f6388
JA
9040
9041 /* The text of WORD. */
ccc6cda3 9042 register char *string;
726f6388 9043
7117c2d2
JA
9044 /* The size of STRING. */
9045 size_t string_size;
9046
726f6388 9047 /* The index into STRING. */
ccc6cda3 9048 int sindex;
726f6388
JA
9049
9050 /* This gets 1 if we see a $@ while quoted. */
ccc6cda3 9051 int quoted_dollar_at;
726f6388
JA
9052
9053 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
9054 whether WORD contains no quoting characters, a partially quoted
9055 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
ccc6cda3
JA
9056 int quoted_state;
9057
95732b49 9058 /* State flags */
ccc6cda3 9059 int had_quoted_null;
0b913689 9060 int has_dollar_at, temp_has_dollar_at;
ac50fbac 9061 int split_on_spaces;
28ef6c31 9062 int tflag;
0001803f 9063 int pflags; /* flags passed to param_expand */
a0c0a00f 9064 int mb_cur_max;
726f6388 9065
95732b49
JA
9066 int assignoff; /* If assignment, offset of `=' */
9067
f73dda09 9068 register unsigned char c; /* Current character. */
726f6388 9069 int t_index; /* For calls to string_extract_xxx. */
726f6388 9070
bb70624e 9071 char twochars[2];
b72432fd 9072
7117c2d2
JA
9073 DECLARE_MBSTATE;
9074
a0c0a00f
CR
9075 /* OK, let's see if we can optimize a common idiom: "$@" */
9076 if (STREQ (word->word, "\"$@\"") &&
9077 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
9078 dollar_vars[1]) /* XXX - check IFS here as well? */
9079 {
9080 if (contains_dollar_at)
9081 *contains_dollar_at = 1;
9082 if (expanded_something)
9083 *expanded_something = 1;
9084 if (cached_quoted_dollar_at)
9085 return (copy_word_list (cached_quoted_dollar_at));
9086 list = list_rest_of_args ();
9087 list = quote_list (list);
9088 cached_quoted_dollar_at = copy_word_list (list);
9089 return (list);
9090 }
9091
f73dda09 9092 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
ccc6cda3 9093 istring[istring_index = 0] = '\0';
cce855bc 9094 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
ac50fbac 9095 split_on_spaces = 0;
ccc6cda3
JA
9096 quoted_state = UNQUOTED;
9097
9098 string = word->word;
9099 if (string == 0)
9100 goto finished_with_string;
a0c0a00f
CR
9101 mb_cur_max = MB_CUR_MAX;
9102
95732b49
JA
9103 /* Don't need the string length for the SADD... and COPY_ macros unless
9104 multibyte characters are possible. */
a0c0a00f 9105 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
726f6388
JA
9106
9107 if (contains_dollar_at)
9108 *contains_dollar_at = 0;
9109
95732b49
JA
9110 assignoff = -1;
9111
726f6388
JA
9112 /* Begin the expansion. */
9113
ccc6cda3 9114 for (sindex = 0; ;)
726f6388
JA
9115 {
9116 c = string[sindex];
9117
ac50fbac 9118 /* Case on top-level character. */
726f6388
JA
9119 switch (c)
9120 {
9121 case '\0':
9122 goto finished_with_string;
9123
9124 case CTLESC:
7117c2d2
JA
9125 sindex++;
9126#if HANDLE_MULTIBYTE
a0c0a00f 9127 if (mb_cur_max > 1 && string[sindex])
7117c2d2 9128 {
b80f6443 9129 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
9130 }
9131 else
9132#endif
9133 {
9134 temp = (char *)xmalloc (3);
9135 temp[0] = CTLESC;
9136 temp[1] = c = string[sindex];
9137 temp[2] = '\0';
9138 }
726f6388 9139
cce855bc 9140dollar_add_string:
726f6388
JA
9141 if (string[sindex])
9142 sindex++;
9143
cce855bc
JA
9144add_string:
9145 if (temp)
9146 {
9147 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
9148 temp = (char *)0;
9149 }
9150
9151 break;
726f6388
JA
9152
9153#if defined (PROCESS_SUBSTITUTION)
9154 /* Process substitution. */
9155 case '<':
9156 case '>':
9157 {
a0c0a00f
CR
9158 /* bash-4.4/bash-5.0
9159 XXX - technically this should only be expanded at the start
9160 of a word */
0628567a 9161 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
726f6388 9162 {
bb70624e 9163 sindex--; /* add_character: label increments sindex */
726f6388
JA
9164 goto add_character;
9165 }
9166 else
cce855bc 9167 t_index = sindex + 1; /* skip past both '<' and LPAREN */
726f6388 9168
85b94814 9169 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
ccc6cda3 9170 sindex = t_index;
726f6388
JA
9171
9172 /* If the process substitution specification is `<()', we want to
9173 open the pipe for writing in the child and produce output; if
9174 it is `>()', we want to open the pipe for reading in the child
9175 and consume input. */
ccc6cda3 9176 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
726f6388
JA
9177
9178 FREE (temp1);
9179
9180 goto dollar_add_string;
9181 }
9182#endif /* PROCESS_SUBSTITUTION */
9183
95732b49
JA
9184 case '=':
9185 /* Posix.2 section 3.6.1 says that tildes following `=' in words
9186 which are not assignment statements are not expanded. If the
9187 shell isn't in posix mode, though, we perform tilde expansion
9188 on `likely candidate' unquoted assignment statements (flags
9189 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
9190 contains an unquoted :~ or =~. Something to think about: we
9191 now have a flag that says to perform tilde expansion on arguments
9192 to `assignment builtins' like declare and export that look like
9193 assignment statements. We now do tilde expansion on such words
9194 even in POSIX mode. */
9195 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
17345e5a 9196 {
0001803f 9197 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9198 goto add_ifs_character;
9199 else
9200 goto add_character;
9201 }
95732b49
JA
9202 /* If we're not in posix mode or forcing assignment-statement tilde
9203 expansion, note where the `=' appears in the word and prepare to
9204 do tilde expansion following the first `='. */
9205 if ((word->flags & W_ASSIGNMENT) &&
9206 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9207 assignoff == -1 && sindex > 0)
9208 assignoff = sindex;
9209 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
9210 word->flags |= W_ITILDE;
9211#if 0
9212 else if ((word->flags & W_ASSIGNMENT) &&
9213 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9214 string[sindex+1] == '~')
9215 word->flags |= W_ITILDE;
9216#endif
a0c0a00f
CR
9217
9218 /* XXX - bash-4.4/bash-5.0 */
9219 if (word->flags & W_ASSIGNARG)
9220 word->flags |= W_ASSIGNRHS; /* affects $@ */
9221
0001803f 9222 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9223 goto add_ifs_character;
9224 else
9225 goto add_character;
95732b49
JA
9226
9227 case ':':
9228 if (word->flags & W_NOTILDE)
17345e5a 9229 {
0001803f 9230 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9231 goto add_ifs_character;
9232 else
9233 goto add_character;
9234 }
95732b49
JA
9235
9236 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
9237 string[sindex+1] == '~')
9238 word->flags |= W_ITILDE;
17345e5a 9239
0001803f 9240 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
9241 goto add_ifs_character;
9242 else
9243 goto add_character;
95732b49
JA
9244
9245 case '~':
9246 /* If the word isn't supposed to be tilde expanded, or we're not
9247 at the start of a word or after an unquoted : or = in an
a0c0a00f
CR
9248 assignment statement, we don't do tilde expansion. If we don't want
9249 tilde expansion when expanding words to be passed to the arithmetic
9250 evaluator, remove the check for Q_ARITH. */
95732b49
JA
9251 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
9252 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
a0c0a00f 9253 ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0)))
95732b49
JA
9254 {
9255 word->flags &= ~W_ITILDE;
0001803f 9256 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
17345e5a
JA
9257 goto add_ifs_character;
9258 else
9259 goto add_character;
95732b49
JA
9260 }
9261
9262 if (word->flags & W_ASSIGNRHS)
9263 tflag = 2;
9264 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
9265 tflag = 1;
9266 else
9267 tflag = 0;
9268
9269 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
9270
9271 word->flags &= ~W_ITILDE;
9272
9273 if (temp && *temp && t_index > 0)
9274 {
9275 temp1 = bash_tilde_expand (temp, tflag);
0628567a
JA
9276 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
9277 {
9278 FREE (temp);
9279 FREE (temp1);
9280 goto add_character; /* tilde expansion failed */
9281 }
95732b49
JA
9282 free (temp);
9283 temp = temp1;
9284 sindex += t_index;
3185942a 9285 goto add_quoted_string; /* XXX was add_string */
95732b49
JA
9286 }
9287 else
9288 {
9289 FREE (temp);
9290 goto add_character;
9291 }
9292
726f6388 9293 case '$':
726f6388
JA
9294 if (expanded_something)
9295 *expanded_something = 1;
9296
0b913689 9297 temp_has_dollar_at = 0;
0001803f
CR
9298 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
9299 if (word->flags & W_NOSPLIT2)
9300 pflags |= PF_NOSPLIT2;
ac50fbac
CR
9301 if (word->flags & W_ASSIGNRHS)
9302 pflags |= PF_ASSIGNRHS;
a0c0a00f
CR
9303 if (word->flags & W_COMPLETE)
9304 pflags |= PF_COMPLETE;
95732b49 9305 tword = param_expand (string, &sindex, quoted, expanded_something,
0b913689 9306 &temp_has_dollar_at, &quoted_dollar_at,
0001803f 9307 &had_quoted_null, pflags);
0b913689 9308 has_dollar_at += temp_has_dollar_at;
ac50fbac 9309 split_on_spaces += (tword->flags & W_SPLITSPACE);
726f6388 9310
95732b49 9311 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
726f6388 9312 {
cce855bc
JA
9313 free (string);
9314 free (istring);
95732b49
JA
9315 return ((tword == &expand_wdesc_error) ? &expand_word_error
9316 : &expand_word_fatal);
cce855bc
JA
9317 }
9318 if (contains_dollar_at && has_dollar_at)
9319 *contains_dollar_at = 1;
95732b49
JA
9320
9321 if (tword && (tword->flags & W_HASQUOTEDNULL))
9322 had_quoted_null = 1;
9323
ac50fbac 9324 temp = tword ? tword->word : (char *)NULL;
95732b49
JA
9325 dispose_word_desc (tword);
9326
a601c749
CR
9327 /* Kill quoted nulls; we will add them back at the end of
9328 expand_word_internal if nothing else in the string */
9329 if (had_quoted_null && temp && QUOTED_NULL (temp))
9330 {
9331 FREE (temp);
9332 temp = (char *)NULL;
9333 }
9334
cce855bc
JA
9335 goto add_string;
9336 break;
726f6388 9337
cce855bc
JA
9338 case '`': /* Backquoted command substitution. */
9339 {
b80f6443 9340 t_index = sindex++;
726f6388 9341
3185942a 9342 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
95732b49
JA
9343 /* The test of sindex against t_index is to allow bare instances of
9344 ` to pass through, for backwards compatibility. */
9345 if (temp == &extract_string_error || temp == &extract_string_fatal)
9346 {
9347 if (sindex - 1 == t_index)
9348 {
9349 sindex = t_index;
9350 goto add_character;
9351 }
ac50fbac 9352 last_command_exit_value = EXECUTION_FAILURE;
3185942a 9353 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
95732b49
JA
9354 free (string);
9355 free (istring);
9356 return ((temp == &extract_string_error) ? &expand_word_error
9357 : &expand_word_fatal);
9358 }
9359
cce855bc
JA
9360 if (expanded_something)
9361 *expanded_something = 1;
726f6388 9362
b80f6443
JA
9363 if (word->flags & W_NOCOMSUB)
9364 /* sindex + 1 because string[sindex] == '`' */
9365 temp1 = substring (string, t_index, sindex + 1);
9366 else
9367 {
9368 de_backslash (temp);
3185942a
JA
9369 tword = command_substitute (temp, quoted);
9370 temp1 = tword ? tword->word : (char *)NULL;
9371 if (tword)
9372 dispose_word_desc (tword);
b80f6443 9373 }
cce855bc
JA
9374 FREE (temp);
9375 temp = temp1;
9376 goto dollar_add_string;
9377 }
ccc6cda3 9378
cce855bc
JA
9379 case '\\':
9380 if (string[sindex + 1] == '\n')
9381 {
9382 sindex += 2;
9383 continue;
9384 }
726f6388 9385
cce855bc 9386 c = string[++sindex];
726f6388 9387
cce855bc 9388 if (quoted & Q_HERE_DOCUMENT)
28ef6c31 9389 tflag = CBSHDOC;
cce855bc 9390 else if (quoted & Q_DOUBLE_QUOTES)
28ef6c31 9391 tflag = CBSDQUOTE;
cce855bc 9392 else
28ef6c31
JA
9393 tflag = 0;
9394
495aee44
CR
9395 /* From Posix discussion on austin-group list: Backslash escaping
9396 a } in ${...} is removed. Issue 0000221 */
9397 if ((quoted & Q_DOLBRACE) && c == RBRACE)
9398 {
ac50fbac
CR
9399 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9400 }
9401 /* This is the fix for " $@\ " */
9402 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
9403 {
9404 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9405 DEFAULT_ARRAY_SIZE);
9406 istring[istring_index++] = CTLESC;
9407 istring[istring_index++] = '\\';
9408 istring[istring_index] = '\0';
9409
495aee44
CR
9410 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9411 }
9412 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
cce855bc 9413 {
7117c2d2 9414 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
bb70624e
JA
9415 }
9416 else if (c == 0)
9417 {
9418 c = CTLNUL;
9419 sindex--; /* add_character: label increments sindex */
9420 goto add_character;
cce855bc
JA
9421 }
9422 else
bb70624e 9423 {
7117c2d2 9424 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
bb70624e 9425 }
726f6388 9426
bb70624e
JA
9427 sindex++;
9428add_twochars:
9429 /* BEFORE jumping here, we need to increment sindex if appropriate */
9430 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9431 DEFAULT_ARRAY_SIZE);
9432 istring[istring_index++] = twochars[0];
9433 istring[istring_index++] = twochars[1];
9434 istring[istring_index] = '\0';
9435
9436 break;
726f6388 9437
cce855bc 9438 case '"':
a0c0a00f 9439 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
cce855bc 9440 goto add_character;
ccc6cda3
JA
9441
9442 t_index = ++sindex;
a0c0a00f 9443 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
ccc6cda3
JA
9444
9445 /* If the quotes surrounded the entire string, then the
9446 whole word was quoted. */
9447 quoted_state = (t_index == 1 && string[sindex] == '\0')
9448 ? WHOLLY_QUOTED
7117c2d2 9449 : PARTIALLY_QUOTED;
ccc6cda3
JA
9450
9451 if (temp && *temp)
726f6388 9452 {
95732b49
JA
9453 tword = alloc_word_desc ();
9454 tword->word = temp;
9455
a0c0a00f
CR
9456 /* XXX - bash-4.4/bash-5.0 */
9457 if (word->flags & W_ASSIGNARG)
9458 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
9459 if (word->flags & W_COMPLETE)
9460 tword->flags |= W_COMPLETE; /* for command substitutions */
9461
ccc6cda3
JA
9462 temp = (char *)NULL;
9463
0b913689 9464 temp_has_dollar_at = 0; /* XXX */
95732b49 9465 /* Need to get W_HASQUOTEDNULL flag through this function. */
0b913689
CR
9466 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
9467 has_dollar_at += temp_has_dollar_at;
726f6388 9468
ccc6cda3
JA
9469 if (list == &expand_word_error || list == &expand_word_fatal)
9470 {
9471 free (istring);
9472 free (string);
9473 /* expand_word_internal has already freed temp_word->word
9474 for us because of the way it prints error messages. */
9475 tword->word = (char *)NULL;
9476 dispose_word (tword);
9477 return list;
9478 }
726f6388 9479
ccc6cda3 9480 dispose_word (tword);
726f6388 9481
ccc6cda3
JA
9482 /* "$@" (a double-quoted dollar-at) expands into nothing,
9483 not even a NULL word, when there are no positional
9484 parameters. */
a0c0a00f 9485 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
726f6388 9486 {
ccc6cda3
JA
9487 quoted_dollar_at++;
9488 break;
9489 }
9490
9491 /* If we get "$@", we know we have expanded something, so we
9492 need to remember it for the final split on $IFS. This is
9493 a special case; it's the only case where a quoted string
9494 can expand into more than one word. It's going to come back
9495 from the above call to expand_word_internal as a list with
9496 a single word, in which all characters are quoted and
9497 separated by blanks. What we want to do is to turn it back
9498 into a list for the next piece of code. */
9499 if (list)
9500 dequote_list (list);
9501
95732b49 9502 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
ac50fbac 9503 had_quoted_null = 1; /* XXX */
95732b49 9504
a0c0a00f 9505 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
ccc6cda3
JA
9506 {
9507 quoted_dollar_at++;
9508 if (contains_dollar_at)
9509 *contains_dollar_at = 1;
9510 if (expanded_something)
9511 *expanded_something = 1;
9512 }
9513 }
9514 else
9515 {
9516 /* What we have is "". This is a minor optimization. */
f73dda09 9517 FREE (temp);
ccc6cda3
JA
9518 list = (WORD_LIST *)NULL;
9519 }
9520
9521 /* The code above *might* return a list (consider the case of "$@",
9522 where it returns "$1", "$2", etc.). We can't throw away the
9523 rest of the list, and we have to make sure each word gets added
9524 as quoted. We test on tresult->next: if it is non-NULL, we
9525 quote the whole list, save it to a string with string_list, and
9526 add that string. We don't need to quote the results of this
9527 (and it would be wrong, since that would quote the separators
9528 as well), so we go directly to add_string. */
9529 if (list)
9530 {
9531 if (list->next)
9532 {
bc4cd23c
JA
9533 /* Testing quoted_dollar_at makes sure that "$@" is
9534 split correctly when $IFS does not contain a space. */
9535 temp = quoted_dollar_at
a0c0a00f 9536 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
bc4cd23c 9537 : string_list (quote_list (list));
ccc6cda3 9538 dispose_words (list);
726f6388
JA
9539 goto add_string;
9540 }
9541 else
9542 {
ccc6cda3 9543 temp = savestring (list->word->word);
95732b49 9544 tflag = list->word->flags;
ccc6cda3 9545 dispose_words (list);
95732b49 9546
cce855bc
JA
9547 /* If the string is not a quoted null string, we want
9548 to remove any embedded unquoted CTLNUL characters.
9549 We do not want to turn quoted null strings back into
9550 the empty string, though. We do this because we
9551 want to remove any quoted nulls from expansions that
9552 contain other characters. For example, if we have
9553 x"$*"y or "x$*y" and there are no positional parameters,
7117c2d2 9554 the $* should expand into nothing. */
95732b49
JA
9555 /* We use the W_HASQUOTEDNULL flag to differentiate the
9556 cases: a quoted null character as above and when
9557 CTLNUL is contained in the (non-null) expansion
9558 of some variable. We use the had_quoted_null flag to
9559 pass the value through this function to its caller. */
9560 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
cce855bc 9561 remove_quoted_nulls (temp); /* XXX */
726f6388
JA
9562 }
9563 }
ccc6cda3
JA
9564 else
9565 temp = (char *)NULL;
726f6388 9566
ccc6cda3 9567 /* We do not want to add quoted nulls to strings that are only
ac50fbac
CR
9568 partially quoted; we can throw them away. The exception to
9569 this is when we are going to be performing word splitting,
9570 since we have to preserve a null argument if the next character
9571 will cause word splitting. */
495aee44 9572 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
cce855bc 9573 continue;
726f6388 9574
ccc6cda3 9575 add_quoted_string:
726f6388 9576
ccc6cda3
JA
9577 if (temp)
9578 {
9579 temp1 = temp;
9580 temp = quote_string (temp);
9581 free (temp1);
bb70624e 9582 goto add_string;
ccc6cda3
JA
9583 }
9584 else
9585 {
9586 /* Add NULL arg. */
bb70624e
JA
9587 c = CTLNUL;
9588 sindex--; /* add_character: label increments sindex */
9589 goto add_character;
ccc6cda3 9590 }
bb70624e 9591
ccc6cda3 9592 /* break; */
726f6388 9593
ccc6cda3 9594 case '\'':
95732b49 9595 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
ccc6cda3 9596 goto add_character;
726f6388 9597
ccc6cda3
JA
9598 t_index = ++sindex;
9599 temp = string_extract_single_quoted (string, &sindex);
726f6388 9600
ccc6cda3
JA
9601 /* If the entire STRING was surrounded by single quotes,
9602 then the string is wholly quoted. */
9603 quoted_state = (t_index == 1 && string[sindex] == '\0')
9604 ? WHOLLY_QUOTED
7117c2d2 9605 : PARTIALLY_QUOTED;
726f6388 9606
ccc6cda3
JA
9607 /* If all we had was '', it is a null expansion. */
9608 if (*temp == '\0')
9609 {
9610 free (temp);
9611 temp = (char *)NULL;
9612 }
9613 else
7117c2d2 9614 remove_quoted_escapes (temp); /* ??? */
726f6388 9615
ccc6cda3 9616 /* We do not want to add quoted nulls to strings that are only
a0c0a00f
CR
9617 partially quoted; such nulls are discarded. See above for the
9618 exception, which is when the string is going to be split. */
9619 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
ccc6cda3 9620 continue;
726f6388 9621
bb70624e
JA
9622 /* If we have a quoted null expansion, add a quoted NULL to istring. */
9623 if (temp == 0)
9624 {
9625 c = CTLNUL;
9626 sindex--; /* add_character: label increments sindex */
9627 goto add_character;
9628 }
9629 else
9630 goto add_quoted_string;
9631
ccc6cda3 9632 /* break; */
726f6388
JA
9633
9634 default:
726f6388 9635 /* This is the fix for " $@ " */
17345e5a 9636 add_ifs_character:
a0c0a00f 9637 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
726f6388 9638 {
bb70624e
JA
9639 if (string[sindex]) /* from old goto dollar_add_string */
9640 sindex++;
9641 if (c == 0)
9642 {
9643 c = CTLNUL;
9644 goto add_character;
9645 }
9646 else
9647 {
7117c2d2 9648#if HANDLE_MULTIBYTE
a0c0a00f 9649 if (mb_cur_max > 1)
b80f6443
JA
9650 sindex--;
9651
a0c0a00f 9652 if (mb_cur_max > 1)
7117c2d2 9653 {
b80f6443 9654 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
9655 }
9656 else
9657#endif
9658 {
9659 twochars[0] = CTLESC;
9660 twochars[1] = c;
9661 goto add_twochars;
9662 }
bb70624e 9663 }
726f6388
JA
9664 }
9665
7117c2d2
JA
9666 SADD_MBCHAR (temp, string, sindex, string_size);
9667
726f6388 9668 add_character:
ccc6cda3
JA
9669 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
9670 DEFAULT_ARRAY_SIZE);
726f6388
JA
9671 istring[istring_index++] = c;
9672 istring[istring_index] = '\0';
9673
9674 /* Next character. */
9675 sindex++;
9676 }
9677 }
9678
9679finished_with_string:
726f6388
JA
9680 /* OK, we're ready to return. If we have a quoted string, and
9681 quoted_dollar_at is not set, we do no splitting at all; otherwise
9682 we split on ' '. The routines that call this will handle what to
9683 do if nothing has been expanded. */
ccc6cda3
JA
9684
9685 /* Partially and wholly quoted strings which expand to the empty
9686 string are retained as an empty arguments. Unquoted strings
9687 which expand to the empty string are discarded. The single
9688 exception is the case of expanding "$@" when there are no
9689 positional parameters. In that case, we discard the expansion. */
9690
9691 /* Because of how the code that handles "" and '' in partially
9692 quoted strings works, we need to make ISTRING into a QUOTED_NULL
9693 if we saw quoting characters, but the expansion was empty.
9694 "" and '' are tossed away before we get to this point when
9695 processing partially quoted strings. This makes "" and $xxx""
9696 equivalent when xxx is unset. We also look to see whether we
9697 saw a quoted null from a ${} expansion and add one back if we
9698 need to. */
9699
9700 /* If we expand to nothing and there were no single or double quotes
9701 in the word, we throw it away. Otherwise, we return a NULL word.
9702 The single exception is for $@ surrounded by double quotes when
9703 there are no positional parameters. In that case, we also throw
9704 the word away. */
9705
9706 if (*istring == '\0')
9707 {
9708 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
726f6388 9709 {
726f6388
JA
9710 istring[0] = CTLNUL;
9711 istring[1] = '\0';
ccc6cda3 9712 tword = make_bare_word (istring);
95732b49 9713 tword->flags |= W_HASQUOTEDNULL; /* XXX */
ccc6cda3
JA
9714 list = make_word_list (tword, (WORD_LIST *)NULL);
9715 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9716 tword->flags |= W_QUOTED;
726f6388 9717 }
ccc6cda3
JA
9718 /* According to sh, ksh, and Posix.2, if a word expands into nothing
9719 and a double-quoted "$@" appears anywhere in it, then the entire
9720 word is removed. */
a0c0a00f
CR
9721 /* XXX - exception appears to be that quoted null strings result in
9722 null arguments */
ccc6cda3
JA
9723 else if (quoted_state == UNQUOTED || quoted_dollar_at)
9724 list = (WORD_LIST *)NULL;
9725#if 0
9726 else
726f6388 9727 {
ccc6cda3 9728 tword = make_bare_word (istring);
ccc6cda3
JA
9729 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9730 tword->flags |= W_QUOTED;
95732b49 9731 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 9732 }
f73dda09
JA
9733#else
9734 else
9735 list = (WORD_LIST *)NULL;
ccc6cda3
JA
9736#endif
9737 }
9738 else if (word->flags & W_NOSPLIT)
9739 {
9740 tword = make_bare_word (istring);
ccc6cda3
JA
9741 if (word->flags & W_ASSIGNMENT)
9742 tword->flags |= W_ASSIGNMENT; /* XXX */
95732b49
JA
9743 if (word->flags & W_COMPASSIGN)
9744 tword->flags |= W_COMPASSIGN; /* XXX */
b72432fd
JA
9745 if (word->flags & W_NOGLOB)
9746 tword->flags |= W_NOGLOB; /* XXX */
ac50fbac
CR
9747 if (word->flags & W_NOBRACE)
9748 tword->flags |= W_NOBRACE; /* XXX */
95732b49
JA
9749 if (word->flags & W_NOEXPAND)
9750 tword->flags |= W_NOEXPAND; /* XXX */
ccc6cda3 9751 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
28ef6c31 9752 tword->flags |= W_QUOTED;
a601c749 9753 if (had_quoted_null && QUOTED_NULL (istring))
95732b49
JA
9754 tword->flags |= W_HASQUOTEDNULL;
9755 list = make_word_list (tword, (WORD_LIST *)NULL);
ccc6cda3
JA
9756 }
9757 else
9758 {
9759 char *ifs_chars;
3b34f6e6 9760 char *tstring;
ccc6cda3 9761
7117c2d2 9762 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
726f6388 9763
cce855bc
JA
9764 /* If we have $@, we need to split the results no matter what. If
9765 IFS is unset or NULL, string_list_dollar_at has separated the
9766 positional parameters with a space, so we split on space (we have
9767 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
9768 string_list_dollar_at has separated the positional parameters
ac50fbac
CR
9769 with the first character of $IFS, so we split on $IFS. If
9770 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
9771 unset or null, and we want to make sure that we split on spaces
a0c0a00f
CR
9772 regardless of what else has happened to IFS since the expansion,
9773 or we expanded "$@" with IFS null and we need to split the positional
9774 parameters into separate words. */
ac50fbac
CR
9775 if (split_on_spaces)
9776 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
3b34f6e6
CR
9777 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
9778 don't want to split the result (W_NOSPLIT2), and we are not quoted,
9779 we have already separated the arguments with the first character of
9780 $IFS. In this case, we want to return a list with a single word
9781 with the separator possibly replaced with a space (it's what other
9782 shells seem to do).
9783 quoted_dollar_at is internal to this function and is set if we are
9784 passed an argument that is unquoted (quoted == 0) but we encounter a
9785 double-quoted $@ while expanding it. */
9786 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
9787 {
9788 /* Only split and rejoin if we have to */
9789 if (*ifs_chars && *ifs_chars != ' ')
9790 {
9791 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
9792 tstring = string_list (list);
9793 }
9794 else
9795 tstring = istring;
9796 tword = make_bare_word (tstring);
9797 if (tstring != istring)
9798 free (tstring);
9799 goto set_word_flags;
9800 }
a0c0a00f
CR
9801 /* This is the attempt to make $* in an assignment context (a=$*) and
9802 array variables subscripted with * in an assignment context (a=${foo[*]})
9803 behave similarly. It has side effects that, though they increase
9804 compatibility with other shells, are not backwards compatible. */
9805#if 0
9806 else if (has_dollar_at && quoted == 0 && ifs_chars && (word->flags & W_ASSIGNRHS))
9807 {
9808 tword = make_bare_word (istring);
9809 goto set_word_flags;
9810 }
9811#endif
ac50fbac 9812 else if (has_dollar_at && ifs_chars)
cce855bc 9813 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
ccc6cda3
JA
9814 else
9815 {
9816 tword = make_bare_word (istring);
3b34f6e6 9817set_word_flags:
ccc6cda3
JA
9818 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
9819 tword->flags |= W_QUOTED;
9820 if (word->flags & W_ASSIGNMENT)
9821 tword->flags |= W_ASSIGNMENT;
95732b49
JA
9822 if (word->flags & W_COMPASSIGN)
9823 tword->flags |= W_COMPASSIGN;
b72432fd
JA
9824 if (word->flags & W_NOGLOB)
9825 tword->flags |= W_NOGLOB;
ac50fbac
CR
9826 if (word->flags & W_NOBRACE)
9827 tword->flags |= W_NOBRACE;
95732b49
JA
9828 if (word->flags & W_NOEXPAND)
9829 tword->flags |= W_NOEXPAND;
a601c749 9830 if (had_quoted_null && QUOTED_NULL (istring))
95732b49
JA
9831 tword->flags |= W_HASQUOTEDNULL; /* XXX */
9832 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 9833 }
726f6388 9834 }
726f6388 9835
ccc6cda3
JA
9836 free (istring);
9837 return (list);
726f6388
JA
9838}
9839
9840/* **************************************************************** */
9841/* */
9842/* Functions for Quote Removal */
9843/* */
9844/* **************************************************************** */
9845
9846/* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
7117c2d2 9847 backslash quoting rules for within double quotes or a here document. */
726f6388
JA
9848char *
9849string_quote_removal (string, quoted)
9850 char *string;
9851 int quoted;
9852{
7117c2d2
JA
9853 size_t slen;
9854 char *r, *result_string, *temp, *send;
f73dda09
JA
9855 int sindex, tindex, dquote;
9856 unsigned char c;
7117c2d2 9857 DECLARE_MBSTATE;
726f6388
JA
9858
9859 /* The result can be no longer than the original string. */
7117c2d2
JA
9860 slen = strlen (string);
9861 send = string + slen;
9862
9863 r = result_string = (char *)xmalloc (slen + 1);
726f6388 9864
ccc6cda3 9865 for (dquote = sindex = 0; c = string[sindex];)
726f6388
JA
9866 {
9867 switch (c)
9868 {
9869 case '\\':
9870 c = string[++sindex];
3185942a
JA
9871 if (c == 0)
9872 {
9873 *r++ = '\\';
9874 break;
9875 }
28ef6c31 9876 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
726f6388 9877 *r++ = '\\';
ccc6cda3 9878 /* FALLTHROUGH */
726f6388
JA
9879
9880 default:
7117c2d2 9881 SCOPY_CHAR_M (r, string, send, sindex);
726f6388
JA
9882 break;
9883
9884 case '\'':
ccc6cda3 9885 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
726f6388
JA
9886 {
9887 *r++ = c;
9888 sindex++;
ccc6cda3 9889 break;
726f6388 9890 }
ccc6cda3
JA
9891 tindex = sindex + 1;
9892 temp = string_extract_single_quoted (string, &tindex);
9893 if (temp)
726f6388 9894 {
ccc6cda3
JA
9895 strcpy (r, temp);
9896 r += strlen (r);
9897 free (temp);
726f6388 9898 }
ccc6cda3 9899 sindex = tindex;
726f6388
JA
9900 break;
9901
9902 case '"':
9903 dquote = 1 - dquote;
9904 sindex++;
9905 break;
9906 }
9907 }
9908 *r = '\0';
9909 return (result_string);
9910}
9911
ccc6cda3
JA
9912#if 0
9913/* UNUSED */
726f6388
JA
9914/* Perform quote removal on word WORD. This allocates and returns a new
9915 WORD_DESC *. */
9916WORD_DESC *
9917word_quote_removal (word, quoted)
9918 WORD_DESC *word;
9919 int quoted;
9920{
9921 WORD_DESC *w;
9922 char *t;
9923
9924 t = string_quote_removal (word->word, quoted);
95732b49
JA
9925 w = alloc_word_desc ();
9926 w->word = t ? t : savestring ("");
726f6388
JA
9927 return (w);
9928}
9929
9930/* Perform quote removal on all words in LIST. If QUOTED is non-zero,
9931 the members of the list are treated as if they are surrounded by
9932 double quotes. Return a new list, or NULL if LIST is NULL. */
9933WORD_LIST *
9934word_list_quote_removal (list, quoted)
9935 WORD_LIST *list;
9936 int quoted;
9937{
95732b49 9938 WORD_LIST *result, *t, *tresult, *e;
726f6388 9939
ccc6cda3 9940 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 9941 {
7117c2d2 9942 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
95732b49 9943#if 0
726f6388 9944 result = (WORD_LIST *) list_append (result, tresult);
95732b49
JA
9945#else
9946 if (result == 0)
9947 result = e = tresult;
9948 else
9949 {
9950 e->next = tresult;
9951 while (e->next)
9952 e = e->next;
9953 }
9954#endif
726f6388
JA
9955 }
9956 return (result);
9957}
ccc6cda3 9958#endif
726f6388 9959
726f6388
JA
9960/*******************************************
9961 * *
9962 * Functions to perform word splitting *
9963 * *
9964 *******************************************/
9965
7117c2d2
JA
9966void
9967setifs (v)
9968 SHELL_VAR *v;
b72432fd 9969{
7117c2d2
JA
9970 char *t;
9971 unsigned char uc;
9972
9973 ifs_var = v;
95732b49 9974 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
b72432fd 9975
ac50fbac
CR
9976 ifs_is_set = ifs_var != 0;
9977 ifs_is_null = ifs_is_set && (*ifs_value == 0);
9978
95732b49
JA
9979 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9980 handle multibyte chars in IFS */
7117c2d2
JA
9981 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9982 for (t = ifs_value ; t && *t; t++)
9983 {
9984 uc = *t;
9985 ifs_cmap[uc] = 1;
9986 }
9987
95732b49
JA
9988#if defined (HANDLE_MULTIBYTE)
9989 if (ifs_value == 0)
9990 {
9991 ifs_firstc[0] = '\0';
9992 ifs_firstc_len = 1;
9993 }
9994 else
9995 {
9996 size_t ifs_len;
9997 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
9998 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
9999 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
10000 {
10001 ifs_firstc[0] = ifs_value[0];
10002 ifs_firstc[1] = '\0';
10003 ifs_firstc_len = 1;
10004 }
10005 else
10006 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
10007 }
10008#else
7117c2d2 10009 ifs_firstc = ifs_value ? *ifs_value : 0;
95732b49 10010#endif
7117c2d2
JA
10011}
10012
10013char *
10014getifs ()
10015{
10016 return ifs_value;
b72432fd
JA
10017}
10018
726f6388
JA
10019/* This splits a single word into a WORD LIST on $IFS, but only if the word
10020 is not quoted. list_string () performs quote removal for us, even if we
10021 don't do any splitting. */
10022WORD_LIST *
7117c2d2 10023word_split (w, ifs_chars)
726f6388 10024 WORD_DESC *w;
7117c2d2 10025 char *ifs_chars;
726f6388
JA
10026{
10027 WORD_LIST *result;
10028
10029 if (w)
10030 {
7117c2d2 10031 char *xifs;
726f6388 10032
7117c2d2
JA
10033 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
10034 result = list_string (w->word, xifs, w->flags & W_QUOTED);
726f6388
JA
10035 }
10036 else
10037 result = (WORD_LIST *)NULL;
ccc6cda3 10038
726f6388
JA
10039 return (result);
10040}
10041
10042/* Perform word splitting on LIST and return the RESULT. It is possible
10043 to return (WORD_LIST *)NULL. */
10044static WORD_LIST *
10045word_list_split (list)
10046 WORD_LIST *list;
10047{
95732b49 10048 WORD_LIST *result, *t, *tresult, *e;
726f6388 10049
ccc6cda3 10050 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 10051 {
7117c2d2 10052 tresult = word_split (t->word, ifs_value);
95732b49
JA
10053 if (result == 0)
10054 result = e = tresult;
10055 else
10056 {
10057 e->next = tresult;
10058 while (e->next)
10059 e = e->next;
10060 }
726f6388
JA
10061 }
10062 return (result);
10063}
10064
10065/**************************************************
10066 * *
cce855bc 10067 * Functions to expand an entire WORD_LIST *
726f6388
JA
10068 * *
10069 **************************************************/
10070
b80f6443
JA
10071/* Do any word-expansion-specific cleanup and jump to top_level */
10072static void
10073exp_jump_to_top_level (v)
10074 int v;
10075{
3185942a
JA
10076 set_pipestatus_from_exit (last_command_exit_value);
10077
b80f6443
JA
10078 /* Cleanup code goes here. */
10079 expand_no_split_dollar_star = 0; /* XXX */
10080 expanding_redir = 0;
3185942a 10081 assigning_in_environment = 0;
b80f6443 10082
f1be666c
JA
10083 if (parse_and_execute_level == 0)
10084 top_level_cleanup (); /* from sig.c */
10085
b80f6443
JA
10086 jump_to_top_level (v);
10087}
10088
cce855bc
JA
10089/* Put NLIST (which is a WORD_LIST * of only one element) at the front of
10090 ELIST, and set ELIST to the new list. */
10091#define PREPEND_LIST(nlist, elist) \
10092 do { nlist->next = elist; elist = nlist; } while (0)
10093
726f6388
JA
10094/* Separate out any initial variable assignments from TLIST. If set -k has
10095 been executed, remove all assignment statements from TLIST. Initial
10096 variable assignments and other environment assignments are placed
bb70624e 10097 on SUBST_ASSIGN_VARLIST. */
726f6388
JA
10098static WORD_LIST *
10099separate_out_assignments (tlist)
10100 WORD_LIST *tlist;
10101{
10102 register WORD_LIST *vp, *lp;
10103
0001803f 10104 if (tlist == 0)
726f6388
JA
10105 return ((WORD_LIST *)NULL);
10106
bb70624e
JA
10107 if (subst_assign_varlist)
10108 dispose_words (subst_assign_varlist); /* Clean up after previous error */
b72432fd 10109
bb70624e 10110 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
10111 vp = lp = tlist;
10112
10113 /* Separate out variable assignments at the start of the command.
10114 Loop invariant: vp->next == lp
10115 Loop postcondition:
7117c2d2
JA
10116 lp = list of words left after assignment statements skipped
10117 tlist = original list of words
726f6388 10118 */
ccc6cda3 10119 while (lp && (lp->word->flags & W_ASSIGNMENT))
726f6388
JA
10120 {
10121 vp = lp;
10122 lp = lp->next;
10123 }
10124
bb70624e
JA
10125 /* If lp != tlist, we have some initial assignment statements.
10126 We make SUBST_ASSIGN_VARLIST point to the list of assignment
10127 words and TLIST point to the remaining words. */
726f6388
JA
10128 if (lp != tlist)
10129 {
bb70624e 10130 subst_assign_varlist = tlist;
726f6388
JA
10131 /* ASSERT(vp->next == lp); */
10132 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
10133 tlist = lp; /* remainder of word list */
10134 }
10135
10136 /* vp == end of variable list */
10137 /* tlist == remainder of original word list without variable assignments */
10138 if (!tlist)
10139 /* All the words in tlist were assignment statements */
10140 return ((WORD_LIST *)NULL);
10141
10142 /* ASSERT(tlist != NULL); */
ccc6cda3 10143 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
726f6388
JA
10144
10145 /* If the -k option is in effect, we need to go through the remaining
bb70624e
JA
10146 words, separate out the assignment words, and place them on
10147 SUBST_ASSIGN_VARLIST. */
726f6388
JA
10148 if (place_keywords_in_env)
10149 {
10150 WORD_LIST *tp; /* tp == running pointer into tlist */
10151
10152 tp = tlist;
10153 lp = tlist->next;
10154
10155 /* Loop Invariant: tp->next == lp */
10156 /* Loop postcondition: tlist == word list without assignment statements */
10157 while (lp)
10158 {
ccc6cda3 10159 if (lp->word->flags & W_ASSIGNMENT)
726f6388
JA
10160 {
10161 /* Found an assignment statement, add this word to end of
bb70624e
JA
10162 subst_assign_varlist (vp). */
10163 if (!subst_assign_varlist)
10164 subst_assign_varlist = vp = lp;
726f6388
JA
10165 else
10166 {
10167 vp->next = lp;
10168 vp = lp;
10169 }
10170
10171 /* Remove the word pointed to by LP from TLIST. */
10172 tp->next = lp->next;
10173 /* ASSERT(vp == lp); */
10174 lp->next = (WORD_LIST *)NULL;
10175 lp = tp->next;
10176 }
10177 else
10178 {
10179 tp = lp;
10180 lp = lp->next;
10181 }
10182 }
10183 }
10184 return (tlist);
10185}
10186
cce855bc
JA
10187#define WEXP_VARASSIGN 0x001
10188#define WEXP_BRACEEXP 0x002
10189#define WEXP_TILDEEXP 0x004
10190#define WEXP_PARAMEXP 0x008
10191#define WEXP_PATHEXP 0x010
10192
10193/* All of the expansions, including variable assignments at the start of
10194 the list. */
10195#define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10196
10197/* All of the expansions except variable assignments at the start of
10198 the list. */
10199#define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10200
10201/* All of the `shell expansions': brace expansion, tilde expansion, parameter
10202 expansion, command substitution, arithmetic expansion, word splitting, and
10203 quote removal. */
10204#define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
10205
726f6388
JA
10206/* Take the list of words in LIST and do the various substitutions. Return
10207 a new list of words which is the expanded list, and without things like
10208 variable assignments. */
10209
10210WORD_LIST *
10211expand_words (list)
10212 WORD_LIST *list;
10213{
cce855bc 10214 return (expand_word_list_internal (list, WEXP_ALL));
726f6388
JA
10215}
10216
10217/* Same as expand_words (), but doesn't hack variable or environment
10218 variables. */
10219WORD_LIST *
10220expand_words_no_vars (list)
10221 WORD_LIST *list;
10222{
cce855bc 10223 return (expand_word_list_internal (list, WEXP_NOVARS));
726f6388
JA
10224}
10225
cce855bc
JA
10226WORD_LIST *
10227expand_words_shellexp (list)
726f6388 10228 WORD_LIST *list;
726f6388 10229{
cce855bc
JA
10230 return (expand_word_list_internal (list, WEXP_SHELLEXP));
10231}
726f6388 10232
cce855bc
JA
10233static WORD_LIST *
10234glob_expand_word_list (tlist, eflags)
10235 WORD_LIST *tlist;
10236 int eflags;
10237{
10238 char **glob_array, *temp_string;
10239 register int glob_index;
10240 WORD_LIST *glob_list, *output_list, *disposables, *next;
10241 WORD_DESC *tword;
726f6388 10242
cce855bc
JA
10243 output_list = disposables = (WORD_LIST *)NULL;
10244 glob_array = (char **)NULL;
10245 while (tlist)
10246 {
10247 /* For each word, either globbing is attempted or the word is
10248 added to orig_list. If globbing succeeds, the results are
10249 added to orig_list and the word (tlist) is added to the list
10250 of disposable words. If globbing fails and failed glob
10251 expansions are left unchanged (the shell default), the
10252 original word is added to orig_list. If globbing fails and
10253 failed glob expansions are removed, the original word is
10254 added to the list of disposable words. orig_list ends up
7117c2d2 10255 in reverse order and requires a call to REVERSE_LIST to
cce855bc
JA
10256 be set right. After all words are examined, the disposable
10257 words are freed. */
10258 next = tlist->next;
726f6388 10259
cce855bc 10260 /* If the word isn't an assignment and contains an unquoted
28ef6c31 10261 pattern matching character, then glob it. */
b72432fd 10262 if ((tlist->word->flags & W_NOGLOB) == 0 &&
cce855bc 10263 unquoted_glob_pattern_p (tlist->word->word))
726f6388 10264 {
cce855bc
JA
10265 glob_array = shell_glob_filename (tlist->word->word);
10266
10267 /* Handle error cases.
10268 I don't think we should report errors like "No such file
10269 or directory". However, I would like to report errors
10270 like "Read failed". */
10271
b80f6443 10272 if (glob_array == 0 || GLOB_FAILED (glob_array))
726f6388 10273 {
bb70624e 10274 glob_array = (char **)xmalloc (sizeof (char *));
cce855bc
JA
10275 glob_array[0] = (char *)NULL;
10276 }
10277
10278 /* Dequote the current word in case we have to use it. */
10279 if (glob_array[0] == NULL)
10280 {
10281 temp_string = dequote_string (tlist->word->word);
10282 free (tlist->word->word);
10283 tlist->word->word = temp_string;
10284 }
10285
10286 /* Make the array into a word list. */
10287 glob_list = (WORD_LIST *)NULL;
10288 for (glob_index = 0; glob_array[glob_index]; glob_index++)
10289 {
10290 tword = make_bare_word (glob_array[glob_index]);
cce855bc
JA
10291 glob_list = make_word_list (tword, glob_list);
10292 }
10293
10294 if (glob_list)
10295 {
10296 output_list = (WORD_LIST *)list_append (glob_list, output_list);
10297 PREPEND_LIST (tlist, disposables);
10298 }
b80f6443
JA
10299 else if (fail_glob_expansion != 0)
10300 {
ac50fbac 10301 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 10302 report_error (_("no match: %s"), tlist->word->word);
f1be666c 10303 exp_jump_to_top_level (DISCARD);
b80f6443 10304 }
cce855bc
JA
10305 else if (allow_null_glob_expansion == 0)
10306 {
10307 /* Failed glob expressions are left unchanged. */
10308 PREPEND_LIST (tlist, output_list);
10309 }
10310 else
10311 {
10312 /* Failed glob expressions are removed. */
10313 PREPEND_LIST (tlist, disposables);
726f6388 10314 }
726f6388 10315 }
cce855bc
JA
10316 else
10317 {
10318 /* Dequote the string. */
10319 temp_string = dequote_string (tlist->word->word);
10320 free (tlist->word->word);
10321 tlist->word->word = temp_string;
10322 PREPEND_LIST (tlist, output_list);
10323 }
10324
7117c2d2 10325 strvec_dispose (glob_array);
cce855bc
JA
10326 glob_array = (char **)NULL;
10327
10328 tlist = next;
726f6388
JA
10329 }
10330
cce855bc
JA
10331 if (disposables)
10332 dispose_words (disposables);
10333
10334 if (output_list)
10335 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10336
10337 return (output_list);
10338}
726f6388
JA
10339
10340#if defined (BRACE_EXPANSION)
cce855bc
JA
10341static WORD_LIST *
10342brace_expand_word_list (tlist, eflags)
10343 WORD_LIST *tlist;
10344 int eflags;
10345{
10346 register char **expansions;
10347 char *temp_string;
10348 WORD_LIST *disposables, *output_list, *next;
10349 WORD_DESC *w;
10350 int eindex;
10351
10352 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
726f6388 10353 {
cce855bc 10354 next = tlist->next;
726f6388 10355
ac50fbac
CR
10356 if (tlist->word->flags & W_NOBRACE)
10357 {
10358/*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
10359 PREPEND_LIST (tlist, output_list);
10360 continue;
10361 }
10362
0001803f
CR
10363 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10364 {
10365/*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
10366 PREPEND_LIST (tlist, output_list);
10367 continue;
10368 }
ac50fbac 10369
cce855bc
JA
10370 /* Only do brace expansion if the word has a brace character. If
10371 not, just add the word list element to BRACES and continue. In
10372 the common case, at least when running shell scripts, this will
0001803f 10373 degenerate to a bunch of calls to `mbschr', and then what is
cce855bc 10374 basically a reversal of TLIST into BRACES, which is corrected
7117c2d2 10375 by a call to REVERSE_LIST () on BRACES when the end of TLIST
cce855bc 10376 is reached. */
0001803f 10377 if (mbschr (tlist->word->word, LBRACE))
726f6388 10378 {
cce855bc 10379 expansions = brace_expand (tlist->word->word);
726f6388 10380
cce855bc 10381 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
726f6388 10382 {
ac50fbac
CR
10383 w = alloc_word_desc ();
10384 w->word = temp_string;
10385
cce855bc
JA
10386 /* If brace expansion didn't change the word, preserve
10387 the flags. We may want to preserve the flags
10388 unconditionally someday -- XXX */
10389 if (STREQ (temp_string, tlist->word->word))
10390 w->flags = tlist->word->flags;
ac50fbac
CR
10391 else
10392 w = make_word_flags (w, temp_string);
10393
cce855bc 10394 output_list = make_word_list (w, output_list);
726f6388 10395 }
cce855bc 10396 free (expansions);
726f6388 10397
cce855bc
JA
10398 /* Add TLIST to the list of words to be freed after brace
10399 expansion has been performed. */
10400 PREPEND_LIST (tlist, disposables);
10401 }
10402 else
10403 PREPEND_LIST (tlist, output_list);
726f6388 10404 }
cce855bc
JA
10405
10406 if (disposables)
10407 dispose_words (disposables);
10408
10409 if (output_list)
10410 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10411
10412 return (output_list);
10413}
10414#endif
10415
3185942a
JA
10416#if defined (ARRAY_VARS)
10417/* Take WORD, a compound associative array assignment, and internally run
10418 'declare -A w', where W is the variable name portion of WORD. */
10419static int
a0c0a00f 10420make_internal_declare (word, option, cmd)
3185942a
JA
10421 char *word;
10422 char *option;
a0c0a00f 10423 char *cmd;
3185942a 10424{
a0c0a00f 10425 int t, r;
3185942a
JA
10426 WORD_LIST *wl;
10427 WORD_DESC *w;
10428
10429 w = make_word (word);
10430
10431 t = assignment (w->word, 0);
a0c0a00f
CR
10432 if (w->word[t] == '=')
10433 {
10434 w->word[t] = '\0';
10435 if (w->word[t - 1] == '+') /* cut off any append op */
10436 w->word[t - 1] = '\0';
10437 }
3185942a
JA
10438
10439 wl = make_word_list (w, (WORD_LIST *)NULL);
10440 wl = make_word_list (make_word (option), wl);
10441
a0c0a00f
CR
10442 r = declare_builtin (wl);
10443
10444 dispose_words (wl);
10445 return r;
3185942a
JA
10446}
10447#endif
10448
cce855bc
JA
10449static WORD_LIST *
10450shell_expand_word_list (tlist, eflags)
10451 WORD_LIST *tlist;
10452 int eflags;
10453{
a0c0a00f 10454 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
cce855bc
JA
10455 int expanded_something, has_dollar_at;
10456 char *temp_string;
726f6388 10457
726f6388 10458 /* We do tilde expansion all the time. This is what 1003.2 says. */
cce855bc 10459 new_list = (WORD_LIST *)NULL;
a0c0a00f
CR
10460 for (wcmd = tlist; wcmd; wcmd = wcmd->next)
10461 if (wcmd->word->flags & W_ASSNBLTIN)
10462 break;
10463
cce855bc 10464 for (orig_list = tlist; tlist; tlist = next)
726f6388 10465 {
ccc6cda3 10466 temp_string = tlist->word->word;
726f6388
JA
10467
10468 next = tlist->next;
10469
95732b49
JA
10470#if defined (ARRAY_VARS)
10471 /* If this is a compound array assignment to a builtin that accepts
10472 such assignments (e.g., `declare'), take the assignment and perform
10473 it separately, handling the semantics of declarations inside shell
10474 functions. This avoids the double-evaluation of such arguments,
10475 because `declare' does some evaluation of compound assignments on
10476 its own. */
10477 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10478 {
10479 int t;
a0c0a00f 10480 char opts[16], opti;
ac50fbac
CR
10481
10482 opti = 0;
10483 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
10484 opts[opti++] = '-';
10485
10486 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
10487 {
10488 opts[opti++] = 'g';
10489 opts[opti++] = 'A';
10490 }
10491 else if (tlist->word->flags & W_ASSIGNASSOC)
10492 opts[opti++] = 'A';
10493 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
10494 {
10495 opts[opti++] = 'g';
10496 opts[opti++] = 'a';
10497 }
10498 else if (tlist->word->flags & W_ASSIGNARRAY)
10499 opts[opti++] = 'a';
10500 else if (tlist->word->flags & W_ASSNGLOBAL)
10501 opts[opti++] = 'g';
95732b49 10502
a0c0a00f
CR
10503 /* If we have special handling note the integer attribute and others
10504 that transform the value upon assignment. What we do is take all
10505 of the option arguments and scan through them looking for options
10506 that cause such transformations, and add them to the `opts' array. */
10507/* if (opti > 0) */
10508 {
10509 char omap[128];
10510 int oind;
10511 WORD_LIST *l;
10512
10513 memset (omap, '\0', sizeof (omap));
10514 for (l = orig_list->next; l != tlist; l = l->next)
10515 {
10516 if (l->word->word[0] != '-')
10517 break; /* non-option argument */
10518 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
10519 break; /* -- signals end of options */
10520 for (oind = 1; l->word->word[oind]; oind++)
10521 switch (l->word->word[oind])
10522 {
10523 case 'i':
10524 case 'l':
10525 case 'u':
10526 case 'c':
10527 omap[l->word->word[oind]] = 1;
10528 if (opti == 0)
10529 opts[opti++] = '-';
10530 break;
10531 default:
10532 break;
10533 }
10534 }
10535
10536 for (oind = 0; oind < sizeof (omap); oind++)
10537 if (omap[oind])
10538 opts[opti++] = oind;
10539 }
ac50fbac
CR
10540
10541 opts[opti] = '\0';
10542 if (opti > 0)
30a978b7 10543 {
a0c0a00f 10544 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
30a978b7
CR
10545 if (t != EXECUTION_SUCCESS)
10546 {
10547 last_command_exit_value = t;
10548 exp_jump_to_top_level (DISCARD);
10549 }
10550 }
3185942a 10551
495aee44 10552 t = do_word_assignment (tlist->word, 0);
95732b49
JA
10553 if (t == 0)
10554 {
10555 last_command_exit_value = EXECUTION_FAILURE;
10556 exp_jump_to_top_level (DISCARD);
10557 }
10558
10559 /* Now transform the word as ksh93 appears to do and go on */
10560 t = assignment (tlist->word->word, 0);
10561 tlist->word->word[t] = '\0';
a0c0a00f
CR
10562 if (tlist->word->word[t - 1] == '+')
10563 tlist->word->word[t - 1] = '\0'; /* cut off append op */
ac50fbac 10564 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
726f6388 10565 }
95732b49 10566#endif
726f6388 10567
ccc6cda3 10568 expanded_something = 0;
726f6388 10569 expanded = expand_word_internal
b72432fd 10570 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
726f6388
JA
10571
10572 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
10573 {
10574 /* By convention, each time this error is returned,
10575 tlist->word->word has already been freed. */
10576 tlist->word->word = (char *)NULL;
ccc6cda3 10577
726f6388
JA
10578 /* Dispose our copy of the original list. */
10579 dispose_words (orig_list);
d166f048 10580 /* Dispose the new list we're building. */
726f6388
JA
10581 dispose_words (new_list);
10582
28ef6c31 10583 last_command_exit_value = EXECUTION_FAILURE;
726f6388 10584 if (expanded == &expand_word_error)
b80f6443 10585 exp_jump_to_top_level (DISCARD);
726f6388 10586 else
b80f6443 10587 exp_jump_to_top_level (FORCE_EOF);
726f6388
JA
10588 }
10589
ccc6cda3
JA
10590 /* Don't split words marked W_NOSPLIT. */
10591 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
726f6388 10592 {
ccc6cda3 10593 temp_list = word_list_split (expanded);
726f6388
JA
10594 dispose_words (expanded);
10595 }
10596 else
10597 {
10598 /* If no parameter expansion, command substitution, process
10599 substitution, or arithmetic substitution took place, then
10600 do not do word splitting. We still have to remove quoted
10601 null characters from the result. */
10602 word_list_remove_quoted_nulls (expanded);
ccc6cda3 10603 temp_list = expanded;
726f6388
JA
10604 }
10605
ccc6cda3
JA
10606 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
10607 new_list = (WORD_LIST *)list_append (expanded, new_list);
726f6388
JA
10608 }
10609
cce855bc
JA
10610 if (orig_list)
10611 dispose_words (orig_list);
726f6388 10612
726f6388 10613 if (new_list)
cce855bc 10614 new_list = REVERSE_LIST (new_list, WORD_LIST *);
726f6388 10615
cce855bc
JA
10616 return (new_list);
10617}
726f6388 10618
cce855bc
JA
10619/* The workhorse for expand_words () and expand_words_no_vars ().
10620 First arg is LIST, a WORD_LIST of words.
b72432fd
JA
10621 Second arg EFLAGS is a flags word controlling which expansions are
10622 performed.
726f6388 10623
cce855bc
JA
10624 This does all of the substitutions: brace expansion, tilde expansion,
10625 parameter expansion, command substitution, arithmetic expansion,
10626 process substitution, word splitting, and pathname expansion, according
10627 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
10628 set, or for which no expansion is done, do not undergo word splitting.
ac50fbac
CR
10629 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
10630 with W_NOBRACE set do not undergo brace expansion (see
10631 brace_expand_word_list above). */
cce855bc
JA
10632static WORD_LIST *
10633expand_word_list_internal (list, eflags)
10634 WORD_LIST *list;
10635 int eflags;
10636{
10637 WORD_LIST *new_list, *temp_list;
10638 int tint;
a0c0a00f 10639 char *savecmd;
726f6388 10640
ac50fbac 10641 tempenv_assign_error = 0;
cce855bc
JA
10642 if (list == 0)
10643 return ((WORD_LIST *)NULL);
726f6388 10644
bb70624e 10645 garglist = new_list = copy_word_list (list);
cce855bc
JA
10646 if (eflags & WEXP_VARASSIGN)
10647 {
bb70624e 10648 garglist = new_list = separate_out_assignments (new_list);
cce855bc
JA
10649 if (new_list == 0)
10650 {
bb70624e 10651 if (subst_assign_varlist)
cce855bc
JA
10652 {
10653 /* All the words were variable assignments, so they are placed
10654 into the shell's environment. */
bb70624e 10655 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
cce855bc 10656 {
a0c0a00f 10657 savecmd = this_command_name;
cce855bc 10658 this_command_name = (char *)NULL; /* no arithmetic errors */
495aee44 10659 tint = do_word_assignment (temp_list->word, 0);
a0c0a00f 10660 this_command_name = savecmd;
cce855bc
JA
10661 /* Variable assignment errors in non-interactive shells
10662 running in Posix.2 mode cause the shell to exit. */
28ef6c31 10663 if (tint == 0)
ccc6cda3 10664 {
cce855bc 10665 last_command_exit_value = EXECUTION_FAILURE;
28ef6c31 10666 if (interactive_shell == 0 && posixly_correct)
b80f6443 10667 exp_jump_to_top_level (FORCE_EOF);
28ef6c31 10668 else
b80f6443 10669 exp_jump_to_top_level (DISCARD);
ccc6cda3 10670 }
726f6388 10671 }
bb70624e
JA
10672 dispose_words (subst_assign_varlist);
10673 subst_assign_varlist = (WORD_LIST *)NULL;
cce855bc
JA
10674 }
10675 return ((WORD_LIST *)NULL);
10676 }
10677 }
726f6388 10678
cce855bc
JA
10679 /* Begin expanding the words that remain. The expansions take place on
10680 things that aren't really variable assignments. */
726f6388 10681
cce855bc
JA
10682#if defined (BRACE_EXPANSION)
10683 /* Do brace expansion on this word if there are any brace characters
10684 in the string. */
10685 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
10686 new_list = brace_expand_word_list (new_list, eflags);
10687#endif /* BRACE_EXPANSION */
726f6388 10688
cce855bc
JA
10689 /* Perform the `normal' shell expansions: tilde expansion, parameter and
10690 variable substitution, command substitution, arithmetic expansion,
10691 and word splitting. */
10692 new_list = shell_expand_word_list (new_list, eflags);
726f6388 10693
cce855bc
JA
10694 /* Okay, we're almost done. Now let's just do some filename
10695 globbing. */
10696 if (new_list)
10697 {
10698 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
10699 /* Glob expand the word list unless globbing has been disabled. */
10700 new_list = glob_expand_word_list (new_list, eflags);
726f6388 10701 else
cce855bc
JA
10702 /* Dequote the words, because we're not performing globbing. */
10703 new_list = dequote_list (new_list);
726f6388
JA
10704 }
10705
bb70624e 10706 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
726f6388 10707 {
95732b49 10708 sh_wassign_func_t *assign_func;
495aee44 10709 int is_special_builtin, is_builtin_or_func;
726f6388
JA
10710
10711 /* If the remainder of the words expand to nothing, Posix.2 requires
10712 that the variable and environment assignments affect the shell's
10713 environment. */
95732b49 10714 assign_func = new_list ? assign_in_env : do_word_assignment;
b80f6443 10715 tempenv_assign_error = 0;
726f6388 10716
495aee44
CR
10717 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
10718 /* Posix says that special builtins exit if a variable assignment error
10719 occurs in an assignment preceding it. */
10720 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
10721
bb70624e 10722 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
726f6388 10723 {
a0c0a00f 10724 savecmd = this_command_name;
ccc6cda3 10725 this_command_name = (char *)NULL;
3185942a 10726 assigning_in_environment = (assign_func == assign_in_env);
495aee44 10727 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
3185942a 10728 assigning_in_environment = 0;
a0c0a00f 10729 this_command_name = savecmd;
ccc6cda3
JA
10730 /* Variable assignment errors in non-interactive shells running
10731 in Posix.2 mode cause the shell to exit. */
b80f6443 10732 if (tint == 0)
ccc6cda3 10733 {
95732b49 10734 if (assign_func == do_word_assignment)
b80f6443
JA
10735 {
10736 last_command_exit_value = EXECUTION_FAILURE;
495aee44 10737 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
b80f6443
JA
10738 exp_jump_to_top_level (FORCE_EOF);
10739 else
10740 exp_jump_to_top_level (DISCARD);
10741 }
28ef6c31 10742 else
b80f6443 10743 tempenv_assign_error++;
ccc6cda3 10744 }
726f6388 10745 }
726f6388 10746
bb70624e
JA
10747 dispose_words (subst_assign_varlist);
10748 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
10749 }
10750
cce855bc 10751 return (new_list);
ccc6cda3 10752}