]> git.ipfire.org Git - thirdparty/bash.git/blame - subst.c
bash-5.1 beta release
[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
712f80b0 7/* Copyright (C) 1987-2020 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
a0c0a00f
CR
134#define CHECK_STRING_OVERRUN(oind, ind, len, ch) \
135 if (ind >= len) \
136 { \
137 oind = len; \
138 ch = 0; \
139 break; \
140 } \
141 else \
bb70624e 142
f73dda09
JA
143/* An expansion function that takes a string and a quoted flag and returns
144 a WORD_LIST *. Used as the type of the third argument to
145 expand_string_if_necessary(). */
712f80b0 146typedef WORD_LIST *EXPFUNC PARAMS((char *, int));
f73dda09 147
726f6388
JA
148/* Process ID of the last command executed within command substitution. */
149pid_t last_command_subst_pid = NO_PID;
b72432fd 150pid_t current_command_subst_pid = NO_PID;
726f6388 151
7117c2d2
JA
152/* Variables used to keep track of the characters in IFS. */
153SHELL_VAR *ifs_var;
154char *ifs_value;
155unsigned char ifs_cmap[UCHAR_MAX + 1];
ac50fbac 156int ifs_is_set, ifs_is_null;
95732b49
JA
157
158#if defined (HANDLE_MULTIBYTE)
159unsigned char ifs_firstc[MB_LEN_MAX];
160size_t ifs_firstc_len;
161#else
7117c2d2 162unsigned char ifs_firstc;
95732b49 163#endif
7117c2d2 164
a0c0a00f
CR
165/* If non-zero, command substitution inherits the value of errexit option */
166int inherit_errexit = 0;
167
0001803f
CR
168/* Sentinel to tell when we are performing variable assignments preceding a
169 command name and putting them into the environment. Used to make sure
170 we use the temporary environment when looking up variable values. */
3185942a
JA
171int assigning_in_environment;
172
0001803f
CR
173/* Used to hold a list of variable assignments preceding a command. Global
174 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
175 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
176WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
177
a0c0a00f
CR
178/* Tell the expansion functions to not longjmp back to top_level on fatal
179 errors. Enabled when doing completion and prompt string expansion. */
180int no_longjmp_on_fatal_error = 0;
181
d233b485
CR
182/* Non-zero means to allow unmatched globbed filenames to expand to
183 a null file. */
184int allow_null_glob_expansion;
185
186/* Non-zero means to throw an error when globbing fails to match anything. */
187int fail_glob_expansion;
188
726f6388 189/* Extern functions and variables from different files. */
ccc6cda3 190extern struct fd_bitmap *current_fds_to_close;
cce855bc 191extern int wordexp_only;
726f6388 192
a0c0a00f
CR
193#if defined (JOB_CONTROL) && defined (PROCESS_SUBSTITUTION)
194extern PROCESS *last_procsub_child;
195#endif
196
0628567a 197#if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
712f80b0 198extern wchar_t *wcsdup PARAMS((const wchar_t *));
0628567a
JA
199#endif
200
f73dda09 201#if 0
ccc6cda3
JA
202/* Variables to keep track of which words in an expanded word list (the
203 output of expand_word_list_internal) are the result of globbing
f73dda09
JA
204 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
205 (CURRENTLY UNUSED). */
ccc6cda3
JA
206char *glob_argv_flags;
207static int glob_argv_flags_size;
f73dda09 208#endif
726f6388 209
a0c0a00f
CR
210static WORD_LIST *cached_quoted_dollar_at = 0;
211
d233b485 212/* Distinguished error values to return from expansion functions */
726f6388 213static WORD_LIST expand_word_error, expand_word_fatal;
95732b49 214static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
d233b485 215static char expand_param_error, expand_param_fatal, expand_param_unset;
95732b49 216static char extract_string_error, extract_string_fatal;
726f6388 217
d233b485
CR
218/* Set by expand_word_unsplit and several of the expand_string_XXX functions;
219 used to inhibit splitting and re-joining $* on $IFS, primarily when doing
220 assignment statements. The idea is that if we're in a context where this
221 is set, we're not going to be performing word splitting, so we use the same
222 rules to expand $* as we would if it appeared within double quotes. */
28ef6c31 223static int expand_no_split_dollar_star = 0;
bb70624e 224
bb70624e
JA
225/* A WORD_LIST of words to be expanded by expand_word_list_internal,
226 without any leading variable assignments. */
227static WORD_LIST *garglist = (WORD_LIST *)NULL;
b72432fd 228
712f80b0
CR
229static char *quoted_substring PARAMS((char *, int, int));
230static int quoted_strlen PARAMS((char *));
231static char *quoted_strchr PARAMS((char *, int, int));
f73dda09 232
712f80b0
CR
233static char *expand_string_if_necessary PARAMS((char *, int, EXPFUNC *));
234static inline char *expand_string_to_string_internal PARAMS((char *, int, EXPFUNC *));
235static WORD_LIST *call_expand_word_internal PARAMS((WORD_DESC *, int, int, int *, int *));
236static WORD_LIST *expand_string_internal PARAMS((char *, int));
237static WORD_LIST *expand_string_leave_quoted PARAMS((char *, int));
238static WORD_LIST *expand_string_for_rhs PARAMS((char *, int, int, int, int *, int *));
239static WORD_LIST *expand_string_for_pat PARAMS((char *, int, int *, int *));
d233b485 240
712f80b0 241static char *quote_escapes_internal PARAMS((const char *, int));
f73dda09 242
712f80b0
CR
243static WORD_LIST *list_quote_escapes PARAMS((WORD_LIST *));
244static WORD_LIST *list_dequote_escapes PARAMS((WORD_LIST *));
ac50fbac 245
712f80b0
CR
246static char *make_quoted_char PARAMS((int));
247static WORD_LIST *quote_list PARAMS((WORD_LIST *));
f73dda09 248
712f80b0
CR
249static int unquoted_substring PARAMS((char *, char *));
250static int unquoted_member PARAMS((int, char *));
f73dda09 251
95732b49 252#if defined (ARRAY_VARS)
712f80b0 253static SHELL_VAR *do_compound_assignment PARAMS((char *, char *, int));
95732b49 254#endif
712f80b0 255static int do_assignment_internal PARAMS((const WORD_DESC *, int));
f73dda09 256
712f80b0
CR
257static char *string_extract_verbatim PARAMS((char *, size_t, int *, char *, int));
258static char *string_extract PARAMS((char *, int *, char *, int));
259static char *string_extract_double_quoted PARAMS((char *, int *, int));
260static inline char *string_extract_single_quoted PARAMS((char *, int *));
261static inline int skip_single_quoted PARAMS((const char *, size_t, int, int));
262static int skip_double_quoted PARAMS((char *, size_t, int, int));
263static char *extract_delimited_string PARAMS((char *, int *, char *, char *, char *, int));
264static char *extract_dollar_brace_string PARAMS((char *, int *, int, int));
265static int skip_matched_pair PARAMS((const char *, int, int, int, int));
f73dda09 266
712f80b0 267static char *pos_params PARAMS((char *, int, int, int, int));
f73dda09 268
712f80b0 269static unsigned char *mb_getcharlens PARAMS((char *, int));
b80f6443 270
712f80b0 271static char *remove_upattern PARAMS((char *, char *, int));
0628567a 272#if defined (HANDLE_MULTIBYTE)
712f80b0 273static wchar_t *remove_wpattern PARAMS((wchar_t *, size_t, wchar_t *, int));
b80f6443 274#endif
712f80b0 275static char *remove_pattern PARAMS((char *, char *, int));
b80f6443 276
712f80b0 277static int match_upattern PARAMS((char *, char *, int, char **, char **));
b80f6443 278#if defined (HANDLE_MULTIBYTE)
712f80b0 279static int match_wpattern PARAMS((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
b80f6443 280#endif
712f80b0
CR
281static int match_pattern PARAMS((char *, char *, int, char **, char **));
282static int getpatspec PARAMS((int, char *));
283static char *getpattern PARAMS((char *, int, int));
284static char *variable_remove_pattern PARAMS((char *, char *, int, int));
285static char *list_remove_pattern PARAMS((WORD_LIST *, char *, int, int, int));
286static char *parameter_list_remove_pattern PARAMS((int, char *, int, int));
f73dda09 287#ifdef ARRAY_VARS
712f80b0 288static char *array_remove_pattern PARAMS((SHELL_VAR *, char *, int, int, int));
f73dda09 289#endif
712f80b0 290static char *parameter_brace_remove_pattern PARAMS((char *, char *, int, char *, int, int, int));
f73dda09 291
712f80b0 292static char *string_var_assignment PARAMS((SHELL_VAR *, char *));
a0c0a00f 293#if defined (ARRAY_VARS)
712f80b0 294static char *array_var_assignment PARAMS((SHELL_VAR *, int, int, int));
a0c0a00f 295#endif
712f80b0
CR
296static char *pos_params_assignment PARAMS((WORD_LIST *, int, int));
297static char *string_transform PARAMS((int, SHELL_VAR *, char *));
298static char *list_transform PARAMS((int, SHELL_VAR *, WORD_LIST *, int, int));
299static char *parameter_list_transform PARAMS((int, int, int));
a0c0a00f 300#if defined ARRAY_VARS
712f80b0 301static char *array_transform PARAMS((int, SHELL_VAR *, int, int));
a0c0a00f 302#endif
712f80b0 303static char *parameter_brace_transform PARAMS((char *, char *, int, char *, int, int, int, int));
3eb0018e 304static int valid_parameter_transform PARAMS((char *));
a0c0a00f 305
712f80b0 306static char *process_substitute PARAMS((char *, int));
f73dda09 307
712f80b0 308static char *read_comsub PARAMS((int, int, int, int *));
f73dda09
JA
309
310#ifdef ARRAY_VARS
712f80b0 311static arrayind_t array_length_reference PARAMS((char *));
f73dda09
JA
312#endif
313
712f80b0
CR
314static int valid_brace_expansion_word PARAMS((char *, int));
315static int chk_atstar PARAMS((char *, int, int, int *, int *));
316static int chk_arithsub PARAMS((const char *, int));
b80f6443 317
712f80b0
CR
318static WORD_DESC *parameter_brace_expand_word PARAMS((char *, int, int, int, arrayind_t *));
319static char *parameter_brace_find_indir PARAMS((char *, int, int, int));
320static WORD_DESC *parameter_brace_expand_indir PARAMS((char *, int, int, int, int *, int *));
321static WORD_DESC *parameter_brace_expand_rhs PARAMS((char *, char *, int, int, int, int *, int *));
322static void parameter_brace_expand_error PARAMS((char *, char *, int));
f73dda09 323
712f80b0
CR
324static int valid_length_expression PARAMS((char *));
325static intmax_t parameter_brace_expand_length PARAMS((char *));
f73dda09 326
712f80b0
CR
327static char *skiparith PARAMS((char *, int));
328static int verify_substring_values PARAMS((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
329static int get_var_and_type PARAMS((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
330static char *mb_substring PARAMS((char *, int, int));
331static char *parameter_brace_substring PARAMS((char *, char *, int, char *, int, int, int));
495aee44 332
712f80b0 333static int shouldexp_replacement PARAMS((char *));
f73dda09 334
712f80b0 335static char *pos_params_pat_subst PARAMS((char *, char *, char *, int));
f73dda09 336
712f80b0 337static char *parameter_brace_patsub PARAMS((char *, char *, int, char *, int, int, int));
f73dda09 338
712f80b0
CR
339static char *pos_params_casemod PARAMS((char *, char *, int, int));
340static char *parameter_brace_casemod PARAMS((char *, char *, int, int, char *, int, int, int));
3185942a 341
712f80b0
CR
342static WORD_DESC *parameter_brace_expand PARAMS((char *, int *, int, int, int *, int *));
343static WORD_DESC *param_expand PARAMS((char *, int *, int, int *, int *, int *, int *, int));
f73dda09 344
712f80b0 345static WORD_LIST *expand_word_internal PARAMS((WORD_DESC *, int, int, int *, int *));
f73dda09 346
712f80b0 347static WORD_LIST *word_list_split PARAMS((WORD_LIST *));
f73dda09 348
712f80b0 349static void exp_jump_to_top_level PARAMS((int));
b80f6443 350
712f80b0
CR
351static WORD_LIST *separate_out_assignments PARAMS((WORD_LIST *));
352static WORD_LIST *glob_expand_word_list PARAMS((WORD_LIST *, int));
f73dda09 353#ifdef BRACE_EXPANSION
712f80b0 354static WORD_LIST *brace_expand_word_list PARAMS((WORD_LIST *, int));
f73dda09 355#endif
3185942a 356#if defined (ARRAY_VARS)
712f80b0
CR
357static int make_internal_declare PARAMS((char *, char *, char *));
358static void expand_compound_assignment_word PARAMS((WORD_LIST *, int));
359static WORD_LIST *expand_declaration_argument PARAMS((WORD_LIST *, WORD_LIST *));
3185942a 360#endif
712f80b0
CR
361static WORD_LIST *shell_expand_word_list PARAMS((WORD_LIST *, int));
362static WORD_LIST *expand_word_list_internal PARAMS((WORD_LIST *, int));
726f6388
JA
363
364/* **************************************************************** */
365/* */
366/* Utility Functions */
367/* */
368/* **************************************************************** */
369
0001803f
CR
370#if defined (DEBUG)
371void
372dump_word_flags (flags)
373 int flags;
374{
375 int f;
376
377 f = flags;
378 fprintf (stderr, "%d -> ", f);
a0c0a00f
CR
379 if (f & W_ARRAYIND)
380 {
381 f &= ~W_ARRAYIND;
382 fprintf (stderr, "W_ARRAYIND%s", f ? "|" : "");
383 }
0001803f
CR
384 if (f & W_ASSIGNASSOC)
385 {
386 f &= ~W_ASSIGNASSOC;
387 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
388 }
ac50fbac
CR
389 if (f & W_ASSIGNARRAY)
390 {
391 f &= ~W_ASSIGNARRAY;
392 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
393 }
d233b485 394 if (f & W_SAWQUOTEDNULL)
0001803f 395 {
d233b485
CR
396 f &= ~W_SAWQUOTEDNULL;
397 fprintf (stderr, "W_SAWQUOTEDNULL%s", f ? "|" : "");
0001803f
CR
398 }
399 if (f & W_NOPROCSUB)
400 {
401 f &= ~W_NOPROCSUB;
402 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
403 }
404 if (f & W_DQUOTE)
405 {
406 f &= ~W_DQUOTE;
407 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
408 }
409 if (f & W_HASQUOTEDNULL)
410 {
411 f &= ~W_HASQUOTEDNULL;
412 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
413 }
414 if (f & W_ASSIGNARG)
415 {
416 f &= ~W_ASSIGNARG;
417 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
418 }
419 if (f & W_ASSNBLTIN)
420 {
421 f &= ~W_ASSNBLTIN;
422 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
423 }
6d41b715
CR
424 if (f & W_ASSNGLOBAL)
425 {
426 f &= ~W_ASSNGLOBAL;
427 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
428 }
0001803f
CR
429 if (f & W_COMPASSIGN)
430 {
431 f &= ~W_COMPASSIGN;
432 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
433 }
712f80b0 434 if (f & W_EXPANDRHS)
0001803f 435 {
712f80b0
CR
436 f &= ~W_EXPANDRHS;
437 fprintf (stderr, "W_EXPANDRHS%s", f ? "|" : "");
0001803f
CR
438 }
439 if (f & W_ITILDE)
440 {
441 f &= ~W_ITILDE;
442 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
443 }
444 if (f & W_NOTILDE)
445 {
446 f &= ~W_NOTILDE;
447 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
448 }
449 if (f & W_ASSIGNRHS)
450 {
451 f &= ~W_ASSIGNRHS;
452 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
453 }
d233b485
CR
454 if (f & W_NOASSNTILDE)
455 {
456 f &= ~W_NOASSNTILDE;
457 fprintf (stderr, "W_NOASSNTILDE%s", f ? "|" : "");
458 }
0001803f
CR
459 if (f & W_NOCOMSUB)
460 {
461 f &= ~W_NOCOMSUB;
462 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
463 }
464 if (f & W_DOLLARSTAR)
465 {
466 f &= ~W_DOLLARSTAR;
467 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
468 }
469 if (f & W_DOLLARAT)
470 {
471 f &= ~W_DOLLARAT;
472 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
473 }
474 if (f & W_TILDEEXP)
475 {
476 f &= ~W_TILDEEXP;
477 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
478 }
479 if (f & W_NOSPLIT2)
480 {
481 f &= ~W_NOSPLIT2;
482 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
483 }
0001803f
CR
484 if (f & W_NOSPLIT)
485 {
486 f &= ~W_NOSPLIT;
487 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
488 }
ac50fbac 489 if (f & W_NOBRACE)
0001803f 490 {
ac50fbac
CR
491 f &= ~W_NOBRACE;
492 fprintf (stderr, "W_NOBRACE%s", f ? "|" : "");
493 }
494 if (f & W_NOGLOB)
495 {
496 f &= ~W_NOGLOB;
497 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
498 }
499 if (f & W_SPLITSPACE)
500 {
501 f &= ~W_SPLITSPACE;
502 fprintf (stderr, "W_SPLITSPACE%s", f ? "|" : "");
0001803f
CR
503 }
504 if (f & W_ASSIGNMENT)
505 {
506 f &= ~W_ASSIGNMENT;
507 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
508 }
509 if (f & W_QUOTED)
510 {
511 f &= ~W_QUOTED;
512 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
513 }
514 if (f & W_HASDOLLAR)
515 {
516 f &= ~W_HASDOLLAR;
517 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
518 }
a0c0a00f
CR
519 if (f & W_COMPLETE)
520 {
521 f &= ~W_COMPLETE;
522 fprintf (stderr, "W_COMPLETE%s", f ? "|" : "");
523 }
d233b485
CR
524 if (f & W_CHKLOCAL)
525 {
526 f &= ~W_CHKLOCAL;
527 fprintf (stderr, "W_CHKLOCAL%s", f ? "|" : "");
528 }
712f80b0
CR
529 if (f & W_FORCELOCAL)
530 {
531 f &= ~W_FORCELOCAL;
532 fprintf (stderr, "W_FORCELOCAL%s", f ? "|" : "");
533 }
534
0001803f
CR
535 fprintf (stderr, "\n");
536 fflush (stderr);
537}
538#endif
539
7117c2d2 540#ifdef INCLUDE_UNUSED
ccc6cda3
JA
541static char *
542quoted_substring (string, start, end)
543 char *string;
544 int start, end;
545{
546 register int len, l;
547 register char *result, *s, *r;
548
549 len = end - start;
550
551 /* Move to string[start], skipping quoted characters. */
552 for (s = string, l = 0; *s && l < start; )
553 {
554 if (*s == CTLESC)
555 {
28ef6c31
JA
556 s++;
557 continue;
ccc6cda3
JA
558 }
559 l++;
560 if (*s == 0)
28ef6c31 561 break;
ccc6cda3
JA
562 }
563
f73dda09 564 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
ccc6cda3
JA
565
566 /* Copy LEN characters, including quote characters. */
567 s = string + l;
568 for (l = 0; l < len; s++)
569 {
570 if (*s == CTLESC)
28ef6c31 571 *r++ = *s++;
ccc6cda3
JA
572 *r++ = *s;
573 l++;
574 if (*s == 0)
28ef6c31 575 break;
ccc6cda3
JA
576 }
577 *r = '\0';
578 return result;
579}
7117c2d2
JA
580#endif
581
582#ifdef INCLUDE_UNUSED
583/* Return the length of S, skipping over quoted characters */
584static int
585quoted_strlen (s)
586 char *s;
587{
588 register char *p;
589 int i;
590
591 i = 0;
592 for (p = s; *p; p++)
593 {
594 if (*p == CTLESC)
595 {
596 p++;
597 if (*p == 0)
598 return (i + 1);
599 }
600 i++;
601 }
602
603 return i;
604}
605#endif
ccc6cda3 606
d233b485 607#ifdef INCLUDE_UNUSED
ccc6cda3
JA
608/* Find the first occurrence of character C in string S, obeying shell
609 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
610 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
611 escaped with CTLESC are skipped. */
7117c2d2 612static char *
ccc6cda3
JA
613quoted_strchr (s, c, flags)
614 char *s;
615 int c, flags;
616{
617 register char *p;
618
619 for (p = s; *p; p++)
620 {
621 if (((flags & ST_BACKSL) && *p == '\\')
622 || ((flags & ST_CTLESC) && *p == CTLESC))
623 {
624 p++;
625 if (*p == '\0')
626 return ((char *)NULL);
627 continue;
628 }
629 else if (*p == c)
630 return p;
631 }
632 return ((char *)NULL);
633}
634
cce855bc 635/* Return 1 if CHARACTER appears in an unquoted portion of
7117c2d2 636 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
cce855bc
JA
637static int
638unquoted_member (character, string)
639 int character;
726f6388
JA
640 char *string;
641{
7117c2d2 642 size_t slen;
cce855bc 643 int sindex, c;
7117c2d2 644 DECLARE_MBSTATE;
726f6388 645
7117c2d2
JA
646 slen = strlen (string);
647 sindex = 0;
648 while (c = string[sindex])
726f6388 649 {
cce855bc
JA
650 if (c == character)
651 return (1);
652
653 switch (c)
ccc6cda3 654 {
cce855bc 655 default:
7117c2d2 656 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
657 break;
658
659 case '\\':
660 sindex++;
661 if (string[sindex])
7117c2d2 662 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
663 break;
664
665 case '\'':
a0c0a00f 666 sindex = skip_single_quoted (string, slen, ++sindex, 0);
cce855bc
JA
667 break;
668
669 case '"':
a0c0a00f 670 sindex = skip_double_quoted (string, slen, ++sindex, 0);
cce855bc 671 break;
ccc6cda3 672 }
726f6388 673 }
cce855bc 674 return (0);
726f6388
JA
675}
676
cce855bc
JA
677/* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
678static int
679unquoted_substring (substr, string)
680 char *substr, *string;
726f6388 681{
7117c2d2 682 size_t slen;
cce855bc 683 int sindex, c, sublen;
7117c2d2 684 DECLARE_MBSTATE;
726f6388 685
cce855bc
JA
686 if (substr == 0 || *substr == '\0')
687 return (0);
688
7117c2d2 689 slen = strlen (string);
cce855bc
JA
690 sublen = strlen (substr);
691 for (sindex = 0; c = string[sindex]; )
726f6388 692 {
cce855bc
JA
693 if (STREQN (string + sindex, substr, sublen))
694 return (1);
695
696 switch (c)
697 {
698 case '\\':
699 sindex++;
cce855bc 700 if (string[sindex])
7117c2d2 701 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
702 break;
703
704 case '\'':
a0c0a00f 705 sindex = skip_single_quoted (string, slen, ++sindex, 0);
cce855bc
JA
706 break;
707
708 case '"':
a0c0a00f 709 sindex = skip_double_quoted (string, slen, ++sindex, 0);
cce855bc
JA
710 break;
711
712 default:
7117c2d2 713 ADVANCE_CHAR (string, slen, sindex);
cce855bc
JA
714 break;
715 }
726f6388 716 }
cce855bc 717 return (0);
ccc6cda3 718}
a0c0a00f 719#endif
726f6388 720
cce855bc
JA
721/* Most of the substitutions must be done in parallel. In order
722 to avoid using tons of unclear goto's, I have some functions
723 for manipulating malloc'ed strings. They all take INDX, a
724 pointer to an integer which is the offset into the string
725 where manipulation is taking place. They also take SIZE, a
726 pointer to an integer which is the current length of the
727 character array for this string. */
726f6388 728
cce855bc
JA
729/* Append SOURCE to TARGET at INDEX. SIZE is the current amount
730 of space allocated to TARGET. SOURCE can be NULL, in which
731 case nothing happens. Gets rid of SOURCE by freeing it.
732 Returns TARGET in case the location has changed. */
7117c2d2 733INLINE char *
cce855bc
JA
734sub_append_string (source, target, indx, size)
735 char *source, *target;
a0c0a00f
CR
736 int *indx;
737 size_t *size;
cce855bc
JA
738{
739 if (source)
726f6388 740 {
a0c0a00f
CR
741 int n;
742 size_t srclen;
cce855bc
JA
743
744 srclen = STRLEN (source);
745 if (srclen >= (int)(*size - *indx))
726f6388 746 {
cce855bc
JA
747 n = srclen + *indx;
748 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
f73dda09 749 target = (char *)xrealloc (target, (*size = n));
726f6388 750 }
cce855bc
JA
751
752 FASTCOPY (source, target + *indx, srclen);
753 *indx += srclen;
754 target[*indx] = '\0';
755
756 free (source);
726f6388 757 }
cce855bc
JA
758 return (target);
759}
760
761#if 0
762/* UNUSED */
763/* Append the textual representation of NUMBER to TARGET.
764 INDX and SIZE are as in SUB_APPEND_STRING. */
765char *
766sub_append_number (number, target, indx, size)
7117c2d2 767 intmax_t number;
cce855bc 768 char *target;
a0c0a00f
CR
769 int *indx;
770 size_t *size;
cce855bc
JA
771{
772 char *temp;
773
774 temp = itos (number);
775 return (sub_append_string (temp, target, indx, size));
726f6388 776}
d166f048 777#endif
726f6388
JA
778
779/* Extract a substring from STRING, starting at SINDEX and ending with
780 one of the characters in CHARLIST. Don't make the ending character
781 part of the string. Leave SINDEX pointing at the ending character.
3185942a 782 Understand about backslashes in the string. If (flags & SX_VARNAME)
7117c2d2
JA
783 is non-zero, and array variables have been compiled into the shell,
784 everything between a `[' and a corresponding `]' is skipped over.
3185942a
JA
785 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
786 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
95732b49 787 contain a closing character from CHARLIST. */
726f6388 788static char *
7117c2d2 789string_extract (string, sindex, charlist, flags)
f73dda09
JA
790 char *string;
791 int *sindex;
792 char *charlist;
7117c2d2 793 int flags;
726f6388 794{
ccc6cda3 795 register int c, i;
95732b49 796 int found;
7117c2d2 797 size_t slen;
726f6388 798 char *temp;
7117c2d2 799 DECLARE_MBSTATE;
726f6388 800
95732b49 801 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
7117c2d2 802 i = *sindex;
95732b49 803 found = 0;
7117c2d2 804 while (c = string[i])
726f6388
JA
805 {
806 if (c == '\\')
7117c2d2
JA
807 {
808 if (string[i + 1])
809 i++;
810 else
811 break;
812 }
ccc6cda3 813#if defined (ARRAY_VARS)
d233b485 814 else if ((flags & SX_VARNAME) && c == LBRACK)
ccc6cda3
JA
815 {
816 int ni;
817 /* If this is an array subscript, skip over it and continue. */
0001803f 818 ni = skipsubscript (string, i, 0);
d233b485 819 if (string[ni] == RBRACK)
ccc6cda3
JA
820 i = ni;
821 }
822#endif
823 else if (MEMBER (c, charlist))
95732b49
JA
824 {
825 found = 1;
726f6388 826 break;
95732b49 827 }
7117c2d2
JA
828
829 ADVANCE_CHAR (string, slen, i);
726f6388 830 }
bb70624e 831
95732b49
JA
832 /* If we had to have a matching delimiter and didn't find one, return an
833 error and let the caller deal with it. */
3185942a 834 if ((flags & SX_REQMATCH) && found == 0)
95732b49
JA
835 {
836 *sindex = i;
837 return (&extract_string_error);
838 }
839
3185942a 840 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
726f6388 841 *sindex = i;
95732b49 842
726f6388
JA
843 return (temp);
844}
845
ccc6cda3
JA
846/* Extract the contents of STRING as if it is enclosed in double quotes.
847 SINDEX, when passed in, is the offset of the character immediately
848 following the opening double quote; on exit, SINDEX is left pointing after
849 the closing double quote. If STRIPDQ is non-zero, unquoted double
850 quotes are stripped and the string is terminated by a null byte.
851 Backslashes between the embedded double quotes are processed. If STRIPDQ
852 is zero, an unquoted `"' terminates the string. */
7117c2d2 853static char *
a0c0a00f 854string_extract_double_quoted (string, sindex, flags)
726f6388 855 char *string;
a0c0a00f 856 int *sindex, flags;
726f6388 857{
7117c2d2
JA
858 size_t slen;
859 char *send;
f73dda09
JA
860 int j, i, t;
861 unsigned char c;
ccc6cda3
JA
862 char *temp, *ret; /* The new string we return. */
863 int pass_next, backquote, si; /* State variables for the machine. */
864 int dquote;
a0c0a00f 865 int stripdq;
7117c2d2
JA
866 DECLARE_MBSTATE;
867
868 slen = strlen (string + *sindex) + *sindex;
869 send = string + slen;
726f6388 870
a0c0a00f
CR
871 stripdq = (flags & SX_STRIPDQ);
872
ccc6cda3 873 pass_next = backquote = dquote = 0;
7117c2d2 874 temp = (char *)xmalloc (1 + slen - *sindex);
726f6388 875
7117c2d2
JA
876 j = 0;
877 i = *sindex;
878 while (c = string[i])
726f6388 879 {
ccc6cda3
JA
880 /* Process a character that was quoted by a backslash. */
881 if (pass_next)
726f6388 882 {
495aee44 883 /* XXX - take another look at this in light of Interp 221 */
ccc6cda3 884 /* Posix.2 sez:
726f6388 885
ccc6cda3
JA
886 ``The backslash shall retain its special meaning as an escape
887 character only when followed by one of the characters:
7117c2d2 888 $ ` " \ <newline>''.
726f6388 889
ccc6cda3
JA
890 If STRIPDQ is zero, we handle the double quotes here and let
891 expand_word_internal handle the rest. If STRIPDQ is non-zero,
892 we have already been through one round of backslash stripping,
893 and want to strip these backslashes only if DQUOTE is non-zero,
894 indicating that we are inside an embedded double-quoted string. */
895
d233b485
CR
896 /* If we are in an embedded quoted string, then don't strip
897 backslashes before characters for which the backslash
898 retains its special meaning, but remove backslashes in
899 front of other characters. If we are not in an
900 embedded quoted string, don't strip backslashes at all.
901 This mess is necessary because the string was already
902 surrounded by double quotes (and sh has some really weird
903 quoting rules).
904 The returned string will be run through expansion as if
905 it were double-quoted. */
ccc6cda3 906 if ((stripdq == 0 && c != '"') ||
28ef6c31 907 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
ccc6cda3 908 temp[j++] = '\\';
ccc6cda3 909 pass_next = 0;
7117c2d2
JA
910
911add_one_character:
912 COPY_CHAR_I (temp, j, string, send, i);
ccc6cda3
JA
913 continue;
914 }
726f6388 915
ccc6cda3
JA
916 /* A backslash protects the next character. The code just above
917 handles preserving the backslash in front of any character but
918 a double quote. */
919 if (c == '\\')
726f6388 920 {
ccc6cda3 921 pass_next++;
7117c2d2 922 i++;
726f6388
JA
923 continue;
924 }
925
ccc6cda3
JA
926 /* Inside backquotes, ``the portion of the quoted string from the
927 initial backquote and the characters up to the next backquote
928 that is not preceded by a backslash, having escape characters
929 removed, defines that command''. */
930 if (backquote)
726f6388 931 {
ccc6cda3
JA
932 if (c == '`')
933 backquote = 0;
d233b485 934 temp[j++] = c; /* COPY_CHAR_I? */
7117c2d2 935 i++;
726f6388
JA
936 continue;
937 }
938
ccc6cda3 939 if (c == '`')
726f6388 940 {
ccc6cda3
JA
941 temp[j++] = c;
942 backquote++;
7117c2d2 943 i++;
ccc6cda3 944 continue;
726f6388
JA
945 }
946
ccc6cda3
JA
947 /* Pass everything between `$(' and the matching `)' or a quoted
948 ${ ... } pair through according to the Posix.2 specification. */
cce855bc 949 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
726f6388 950 {
b80f6443
JA
951 int free_ret = 1;
952
ccc6cda3 953 si = i + 2;
cce855bc 954 if (string[i + 1] == LPAREN)
a0c0a00f 955 ret = extract_command_subst (string, &si, (flags & SX_COMPLETE));
ccc6cda3 956 else
495aee44 957 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
726f6388 958
ccc6cda3
JA
959 temp[j++] = '$';
960 temp[j++] = string[i + 1];
726f6388 961
b80f6443
JA
962 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
963 is set. */
964 if (ret == 0 && no_longjmp_on_fatal_error)
965 {
966 free_ret = 0;
967 ret = string + i + 2;
968 }
969
d233b485 970 /* XXX - CHECK_STRING_OVERRUN here? */
ccc6cda3
JA
971 for (t = 0; ret[t]; t++, j++)
972 temp[j] = ret[t];
b80f6443 973 temp[j] = string[si];
726f6388 974
712f80b0
CR
975 if (si < i + 2) /* we went back? */
976 i += 2;
977 else if (string[si])
b80f6443
JA
978 {
979 j++;
980 i = si + 1;
981 }
982 else
983 i = si;
984
985 if (free_ret)
986 free (ret);
ccc6cda3 987 continue;
726f6388
JA
988 }
989
ccc6cda3 990 /* Add any character but a double quote to the quoted string we're
28ef6c31 991 accumulating. */
ccc6cda3 992 if (c != '"')
7117c2d2 993 goto add_one_character;
ccc6cda3
JA
994
995 /* c == '"' */
996 if (stripdq)
726f6388 997 {
ccc6cda3 998 dquote ^= 1;
7117c2d2 999 i++;
ccc6cda3 1000 continue;
726f6388 1001 }
ccc6cda3
JA
1002
1003 break;
726f6388 1004 }
ccc6cda3 1005 temp[j] = '\0';
726f6388 1006
ccc6cda3
JA
1007 /* Point to after the closing quote. */
1008 if (c)
1009 i++;
726f6388
JA
1010 *sindex = i;
1011
ccc6cda3
JA
1012 return (temp);
1013}
1014
1015/* This should really be another option to string_extract_double_quoted. */
f73dda09 1016static int
a0c0a00f 1017skip_double_quoted (string, slen, sind, flags)
ccc6cda3 1018 char *string;
7117c2d2 1019 size_t slen;
ccc6cda3 1020 int sind;
a0c0a00f 1021 int flags;
ccc6cda3 1022{
f73dda09 1023 int c, i;
ccc6cda3
JA
1024 char *ret;
1025 int pass_next, backquote, si;
7117c2d2 1026 DECLARE_MBSTATE;
ccc6cda3
JA
1027
1028 pass_next = backquote = 0;
7117c2d2
JA
1029 i = sind;
1030 while (c = string[i])
726f6388 1031 {
ccc6cda3
JA
1032 if (pass_next)
1033 {
1034 pass_next = 0;
7117c2d2 1035 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1036 continue;
1037 }
1038 else if (c == '\\')
1039 {
1040 pass_next++;
7117c2d2 1041 i++;
ccc6cda3
JA
1042 continue;
1043 }
1044 else if (backquote)
1045 {
1046 if (c == '`')
1047 backquote = 0;
7117c2d2 1048 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1049 continue;
1050 }
1051 else if (c == '`')
1052 {
1053 backquote++;
7117c2d2 1054 i++;
ccc6cda3
JA
1055 continue;
1056 }
cce855bc 1057 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
ccc6cda3
JA
1058 {
1059 si = i + 2;
cce855bc 1060 if (string[i + 1] == LPAREN)
a0c0a00f 1061 ret = extract_command_subst (string, &si, SX_NOALLOC|(flags&SX_COMPLETE));
ccc6cda3 1062 else
495aee44 1063 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
ccc6cda3 1064
a0c0a00f
CR
1065 /* These can consume the entire string if they are unterminated */
1066 CHECK_STRING_OVERRUN (i, si, slen, c);
1067
7117c2d2 1068 i = si + 1;
ccc6cda3
JA
1069 continue;
1070 }
1071 else if (c != '"')
7117c2d2
JA
1072 {
1073 ADVANCE_CHAR (string, slen, i);
1074 continue;
1075 }
ccc6cda3
JA
1076 else
1077 break;
726f6388 1078 }
ccc6cda3
JA
1079
1080 if (c)
1081 i++;
1082
1083 return (i);
726f6388
JA
1084}
1085
ccc6cda3
JA
1086/* Extract the contents of STRING as if it is enclosed in single quotes.
1087 SINDEX, when passed in, is the offset of the character immediately
1088 following the opening single quote; on exit, SINDEX is left pointing after
1089 the closing single quote. */
1090static inline char *
1091string_extract_single_quoted (string, sindex)
1092 char *string;
1093 int *sindex;
1094{
f73dda09 1095 register int i;
7117c2d2 1096 size_t slen;
ccc6cda3 1097 char *t;
7117c2d2 1098 DECLARE_MBSTATE;
ccc6cda3 1099
95732b49
JA
1100 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1101 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
7117c2d2
JA
1102 i = *sindex;
1103 while (string[i] && string[i] != '\'')
1104 ADVANCE_CHAR (string, slen, i);
ccc6cda3 1105
bb70624e 1106 t = substring (string, *sindex, i);
ccc6cda3
JA
1107
1108 if (string[i])
1109 i++;
1110 *sindex = i;
1111
1112 return (t);
1113}
1114
a0c0a00f
CR
1115/* Skip over a single-quoted string. We overload the SX_COMPLETE flag to mean
1116 that we are splitting out words for completion and have encountered a $'...'
1117 string, which allows backslash-escaped single quotes. */
ccc6cda3 1118static inline int
a0c0a00f 1119skip_single_quoted (string, slen, sind, flags)
0628567a 1120 const char *string;
7117c2d2 1121 size_t slen;
ccc6cda3 1122 int sind;
a0c0a00f 1123 int flags;
ccc6cda3 1124{
28ef6c31 1125 register int c;
7117c2d2
JA
1126 DECLARE_MBSTATE;
1127
1128 c = sind;
1129 while (string[c] && string[c] != '\'')
a0c0a00f
CR
1130 {
1131 if ((flags & SX_COMPLETE) && string[c] == '\\' && string[c+1] == '\'' && string[c+2])
1132 ADVANCE_CHAR (string, slen, c);
1133 ADVANCE_CHAR (string, slen, c);
1134 }
ccc6cda3 1135
28ef6c31
JA
1136 if (string[c])
1137 c++;
1138 return c;
ccc6cda3
JA
1139}
1140
1141/* Just like string_extract, but doesn't hack backslashes or any of
bb70624e 1142 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
726f6388 1143static char *
3185942a 1144string_extract_verbatim (string, slen, sindex, charlist, flags)
f73dda09 1145 char *string;
95732b49 1146 size_t slen;
ccc6cda3 1147 int *sindex;
f73dda09 1148 char *charlist;
3185942a 1149 int flags;
ccc6cda3 1150{
0001803f 1151 register int i;
95732b49 1152#if defined (HANDLE_MULTIBYTE)
95732b49
JA
1153 wchar_t *wcharlist;
1154#endif
ccc6cda3
JA
1155 int c;
1156 char *temp;
95732b49 1157 DECLARE_MBSTATE;
ccc6cda3 1158
a0c0a00f 1159 if ((flags & SX_NOCTLESC) && charlist[0] == '\'' && charlist[1] == '\0')
ccc6cda3
JA
1160 {
1161 temp = string_extract_single_quoted (string, sindex);
1162 --*sindex; /* leave *sindex at separator character */
1163 return temp;
1164 }
1165
712f80b0
CR
1166 /* This can never be called with charlist == NULL. If *charlist == NULL,
1167 we can skip the loop and just return a copy of the string, updating
1168 *sindex */
1169 if (*charlist == 0)
1170 {
1171 temp = string + *sindex;
1172 c = (*sindex == 0) ? slen : STRLEN (temp);
1173 temp = savestring (temp);
1174 *sindex += c;
1175 return temp;
1176 }
1177
95732b49 1178 i = *sindex;
95732b49 1179#if defined (HANDLE_MULTIBYTE)
95732b49
JA
1180 wcharlist = 0;
1181#endif
1182 while (c = string[i])
ccc6cda3 1183 {
95732b49
JA
1184#if defined (HANDLE_MULTIBYTE)
1185 size_t mblength;
1186#endif
3185942a
JA
1187 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1188 {
1189 i += 2;
d233b485 1190 CHECK_STRING_OVERRUN (i, i, slen, c);
3185942a
JA
1191 continue;
1192 }
1193 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1194 through, to protect the CTLNULs from later calls to
1195 remove_quoted_nulls. */
1196 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
ccc6cda3 1197 {
95732b49 1198 i += 2;
d233b485 1199 CHECK_STRING_OVERRUN (i, i, slen, c);
ccc6cda3
JA
1200 continue;
1201 }
1202
95732b49 1203#if defined (HANDLE_MULTIBYTE)
d233b485
CR
1204 if (locale_utf8locale && slen > i && UTF8_SINGLEBYTE (string[i]))
1205 mblength = (string[i] != 0) ? 1 : 0;
1206 else
1207 mblength = MBLEN (string + i, slen - i);
95732b49
JA
1208 if (mblength > 1)
1209 {
1210 wchar_t wc;
1211 mblength = mbtowc (&wc, string + i, slen - i);
1212 if (MB_INVALIDCH (mblength))
1213 {
1214 if (MEMBER (c, charlist))
1215 break;
1216 }
1217 else
1218 {
1219 if (wcharlist == 0)
1220 {
1221 size_t len;
1222 len = mbstowcs (wcharlist, charlist, 0);
1223 if (len == -1)
1224 len = 0;
0628567a
JA
1225 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1226 mbstowcs (wcharlist, charlist, len + 1);
95732b49
JA
1227 }
1228
1229 if (wcschr (wcharlist, wc))
1230 break;
1231 }
1232 }
1233 else
1234#endif
ccc6cda3
JA
1235 if (MEMBER (c, charlist))
1236 break;
95732b49
JA
1237
1238 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1239 }
1240
95732b49
JA
1241#if defined (HANDLE_MULTIBYTE)
1242 FREE (wcharlist);
1243#endif
1244
bb70624e 1245 temp = substring (string, *sindex, i);
ccc6cda3
JA
1246 *sindex = i;
1247
1248 return (temp);
1249}
1250
1251/* Extract the $( construct in STRING, and return a new string.
1252 Start extracting at (SINDEX) as if we had just seen "$(".
3185942a 1253 Make (SINDEX) get the position of the matching ")". )
0001803f 1254 XFLAGS is additional flags to pass to other extraction functions. */
ccc6cda3 1255char *
3185942a 1256extract_command_subst (string, sindex, xflags)
726f6388
JA
1257 char *string;
1258 int *sindex;
3185942a 1259 int xflags;
726f6388 1260{
a0c0a00f
CR
1261 char *ret;
1262
1263 if (string[*sindex] == LPAREN || (xflags & SX_COMPLETE))
3185942a
JA
1264 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1265 else
1266 {
1267 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
a0c0a00f
CR
1268 ret = xparse_dolparen (string, string+*sindex, sindex, xflags);
1269 return ret;
3185942a 1270 }
ccc6cda3
JA
1271}
1272
28ef6c31 1273/* Extract the $[ construct in STRING, and return a new string. (])
ccc6cda3
JA
1274 Start extracting at (SINDEX) as if we had just seen "$[".
1275 Make (SINDEX) get the position of the matching "]". */
1276char *
1277extract_arithmetic_subst (string, sindex)
1278 char *string;
1279 int *sindex;
1280{
7117c2d2 1281 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
ccc6cda3
JA
1282}
1283
1284#if defined (PROCESS_SUBSTITUTION)
1285/* Extract the <( or >( construct in STRING, and return a new string.
1286 Start extracting at (SINDEX) as if we had just seen "<(".
cce855bc 1287 Make (SINDEX) get the position of the matching ")". */ /*))*/
ccc6cda3 1288char *
85b94814 1289extract_process_subst (string, starter, sindex, xflags)
ccc6cda3
JA
1290 char *string;
1291 char *starter;
1292 int *sindex;
85b94814 1293 int xflags;
ccc6cda3 1294{
85b94814 1295#if 0
d233b485 1296 /* XXX - check xflags&SX_COMPLETE here? */
ac50fbac 1297 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
85b94814
CR
1298#else
1299 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1300 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1301#endif
ccc6cda3
JA
1302}
1303#endif /* PROCESS_SUBSTITUTION */
1304
1305#if defined (ARRAY_VARS)
95732b49
JA
1306/* This can be fooled by unquoted right parens in the passed string. If
1307 each caller verifies that the last character in STRING is a right paren,
1308 we don't even need to call extract_delimited_string. */
ccc6cda3
JA
1309char *
1310extract_array_assignment_list (string, sindex)
1311 char *string;
1312 int *sindex;
1313{
95732b49
JA
1314 int slen;
1315 char *ret;
1316
d233b485
CR
1317 slen = strlen (string);
1318 if (string[slen - 1] == RPAREN)
95732b49
JA
1319 {
1320 ret = substring (string, *sindex, slen - 1);
1321 *sindex = slen - 1;
1322 return ret;
1323 }
1324 return 0;
ccc6cda3
JA
1325}
1326#endif
1327
1328/* Extract and create a new string from the contents of STRING, a
1329 character string delimited with OPENER and CLOSER. SINDEX is
1330 the address of an int describing the current offset in STRING;
1331 it should point to just after the first OPENER found. On exit,
1332 SINDEX gets the position of the last character of the matching CLOSER.
1333 If OPENER is more than a single character, ALT_OPENER, if non-null,
1334 contains a character string that can also match CLOSER and thus
1335 needs to be skipped. */
1336static char *
7117c2d2 1337extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
ccc6cda3
JA
1338 char *string;
1339 int *sindex;
1340 char *opener, *alt_opener, *closer;
7117c2d2 1341 int flags;
ccc6cda3
JA
1342{
1343 int i, c, si;
7117c2d2 1344 size_t slen;
ccc6cda3 1345 char *t, *result;
0628567a 1346 int pass_character, nesting_level, in_comment;
ccc6cda3 1347 int len_closer, len_opener, len_alt_opener;
7117c2d2 1348 DECLARE_MBSTATE;
ccc6cda3 1349
7117c2d2 1350 slen = strlen (string + *sindex) + *sindex;
ccc6cda3
JA
1351 len_opener = STRLEN (opener);
1352 len_alt_opener = STRLEN (alt_opener);
1353 len_closer = STRLEN (closer);
726f6388 1354
0628567a 1355 pass_character = in_comment = 0;
726f6388
JA
1356
1357 nesting_level = 1;
ccc6cda3 1358 i = *sindex;
726f6388 1359
ccc6cda3 1360 while (nesting_level)
726f6388 1361 {
ccc6cda3
JA
1362 c = string[i];
1363
a0c0a00f
CR
1364 /* If a recursive call or a call to ADVANCE_CHAR leaves the index beyond
1365 the end of the string, catch it and cut the loop. */
1366 if (i > slen)
1367 {
1368 i = slen;
1369 c = string[i = slen];
1370 break;
1371 }
1372
ccc6cda3 1373 if (c == 0)
28ef6c31 1374 break;
ccc6cda3 1375
0628567a
JA
1376 if (in_comment)
1377 {
1378 if (c == '\n')
1379 in_comment = 0;
1380 ADVANCE_CHAR (string, slen, i);
1381 continue;
1382 }
1383
ccc6cda3 1384 if (pass_character) /* previous char was backslash */
726f6388
JA
1385 {
1386 pass_character = 0;
7117c2d2 1387 ADVANCE_CHAR (string, slen, i);
726f6388
JA
1388 continue;
1389 }
1390
0628567a 1391 /* Not exactly right yet; should handle shell metacharacters and
0001803f 1392 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
3185942a 1393 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
0628567a
JA
1394 {
1395 in_comment = 1;
1396 ADVANCE_CHAR (string, slen, i);
1397 continue;
1398 }
1399
7117c2d2 1400 if (c == CTLESC || c == '\\')
726f6388 1401 {
ccc6cda3
JA
1402 pass_character++;
1403 i++;
1404 continue;
726f6388
JA
1405 }
1406
495aee44
CR
1407 /* Process a nested command substitution, but only if we're parsing an
1408 arithmetic substitution. */
0001803f
CR
1409 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1410 {
1411 si = i + 2;
495aee44 1412 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
d233b485 1413 CHECK_STRING_OVERRUN (i, si, slen, c);
0001803f
CR
1414 i = si + 1;
1415 continue;
1416 }
0001803f 1417
ccc6cda3
JA
1418 /* Process a nested OPENER. */
1419 if (STREQN (string + i, opener, len_opener))
726f6388 1420 {
ccc6cda3 1421 si = i + len_opener;
3185942a 1422 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
d233b485 1423 CHECK_STRING_OVERRUN (i, si, slen, c);
ccc6cda3 1424 i = si + 1;
ccc6cda3 1425 continue;
726f6388
JA
1426 }
1427
ccc6cda3
JA
1428 /* Process a nested ALT_OPENER */
1429 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
726f6388 1430 {
ccc6cda3 1431 si = i + len_alt_opener;
3185942a 1432 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
d233b485 1433 CHECK_STRING_OVERRUN (i, si, slen, c);
ccc6cda3 1434 i = si + 1;
726f6388
JA
1435 continue;
1436 }
ccc6cda3
JA
1437
1438 /* If the current substring terminates the delimited string, decrement
1439 the nesting level. */
1440 if (STREQN (string + i, closer, len_closer))
726f6388 1441 {
7117c2d2 1442 i += len_closer - 1; /* move to last byte of the closer */
ccc6cda3
JA
1443 nesting_level--;
1444 if (nesting_level == 0)
1445 break;
726f6388 1446 }
ccc6cda3
JA
1447
1448 /* Pass old-style command substitution through verbatim. */
1449 if (c == '`')
28ef6c31
JA
1450 {
1451 si = i + 1;
3185942a 1452 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
d233b485 1453 CHECK_STRING_OVERRUN (i, si, slen, c);
28ef6c31 1454 i = si + 1;
28ef6c31
JA
1455 continue;
1456 }
ccc6cda3 1457
7117c2d2
JA
1458 /* Pass single-quoted and double-quoted strings through verbatim. */
1459 if (c == '\'' || c == '"')
28ef6c31
JA
1460 {
1461 si = i + 1;
a0c0a00f
CR
1462 i = (c == '\'') ? skip_single_quoted (string, slen, si, 0)
1463 : skip_double_quoted (string, slen, si, 0);
28ef6c31
JA
1464 continue;
1465 }
ccc6cda3 1466
7117c2d2
JA
1467 /* move past this character, which was not special. */
1468 ADVANCE_CHAR (string, slen, i);
726f6388
JA
1469 }
1470
b80f6443 1471 if (c == 0 && nesting_level)
726f6388 1472 {
b80f6443
JA
1473 if (no_longjmp_on_fatal_error == 0)
1474 {
b80f6443 1475 last_command_exit_value = EXECUTION_FAILURE;
ac50fbac 1476 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
b80f6443
JA
1477 exp_jump_to_top_level (DISCARD);
1478 }
1479 else
1480 {
1481 *sindex = i;
1482 return (char *)NULL;
1483 }
726f6388 1484 }
ccc6cda3 1485
cce855bc 1486 si = i - *sindex - len_closer + 1;
3185942a 1487 if (flags & SX_NOALLOC)
7117c2d2
JA
1488 result = (char *)NULL;
1489 else
1490 {
1491 result = (char *)xmalloc (1 + si);
1492 strncpy (result, string + *sindex, si);
1493 result[si] = '\0';
1494 }
cce855bc
JA
1495 *sindex = i;
1496
726f6388
JA
1497 return (result);
1498}
1499
ccc6cda3
JA
1500/* Extract a parameter expansion expression within ${ and } from STRING.
1501 Obey the Posix.2 rules for finding the ending `}': count braces while
1502 skipping over enclosed quoted strings and command substitutions.
1503 SINDEX is the address of an int describing the current offset in STRING;
1504 it should point to just after the first `{' found. On exit, SINDEX
1505 gets the position of the matching `}'. QUOTED is non-zero if this
1506 occurs inside double quotes. */
1507/* XXX -- this is very similar to extract_delimited_string -- XXX */
726f6388 1508static char *
7117c2d2 1509extract_dollar_brace_string (string, sindex, quoted, flags)
726f6388 1510 char *string;
7117c2d2 1511 int *sindex, quoted, flags;
726f6388 1512{
f73dda09 1513 register int i, c;
7117c2d2 1514 size_t slen;
495aee44 1515 int pass_character, nesting_level, si, dolbrace_state;
ccc6cda3 1516 char *result, *t;
7117c2d2 1517 DECLARE_MBSTATE;
726f6388 1518
ccc6cda3 1519 pass_character = 0;
ccc6cda3 1520 nesting_level = 1;
7117c2d2 1521 slen = strlen (string + *sindex) + *sindex;
ccc6cda3 1522
495aee44 1523 /* The handling of dolbrace_state needs to agree with the code in parse.y:
49ed961b
CR
1524 parse_matched_pair(). The different initial value is to handle the
1525 case where this function is called to parse the word in
1526 ${param op word} (SX_WORD). */
1527 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1528 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1529 dolbrace_state = DOLBRACE_QUOTE;
495aee44 1530
7117c2d2
JA
1531 i = *sindex;
1532 while (c = string[i])
726f6388 1533 {
ccc6cda3 1534 if (pass_character)
726f6388 1535 {
ccc6cda3 1536 pass_character = 0;
7117c2d2 1537 ADVANCE_CHAR (string, slen, i);
ccc6cda3
JA
1538 continue;
1539 }
726f6388 1540
cce855bc
JA
1541 /* CTLESCs and backslashes quote the next character. */
1542 if (c == CTLESC || c == '\\')
726f6388 1543 {
ccc6cda3 1544 pass_character++;
7117c2d2 1545 i++;
726f6388
JA
1546 continue;
1547 }
1548
cce855bc 1549 if (string[i] == '$' && string[i+1] == LBRACE)
726f6388 1550 {
ccc6cda3 1551 nesting_level++;
7117c2d2 1552 i += 2;
726f6388
JA
1553 continue;
1554 }
1555
cce855bc 1556 if (c == RBRACE)
726f6388 1557 {
ccc6cda3
JA
1558 nesting_level--;
1559 if (nesting_level == 0)
1560 break;
7117c2d2 1561 i++;
726f6388
JA
1562 continue;
1563 }
1564
ccc6cda3
JA
1565 /* Pass the contents of old-style command substitutions through
1566 verbatim. */
1567 if (c == '`')
726f6388 1568 {
ccc6cda3 1569 si = i + 1;
3185942a 1570 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
a0c0a00f
CR
1571
1572 CHECK_STRING_OVERRUN (i, si, slen, c);
d233b485 1573
7117c2d2 1574 i = si + 1;
ccc6cda3
JA
1575 continue;
1576 }
726f6388 1577
cce855bc
JA
1578 /* Pass the contents of new-style command substitutions and
1579 arithmetic substitutions through verbatim. */
1580 if (string[i] == '$' && string[i+1] == LPAREN)
ccc6cda3 1581 {
726f6388 1582 si = i + 2;
3185942a 1583 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
d233b485
CR
1584
1585 CHECK_STRING_OVERRUN (i, si, slen, c);
1586
1587 i = si + 1;
1588 continue;
1589 }
1590
1591#if defined (PROCESS_SUBSTITUTION)
1592 /* Technically this should only work at the start of a word */
1593 if ((string[i] == '<' || string[i] == '>') && string[i+1] == LPAREN)
1594 {
1595 si = i + 2;
1596 t = extract_process_subst (string, (string[i] == '<' ? "<(" : ">)"), &si, flags|SX_NOALLOC);
1597
1598 CHECK_STRING_OVERRUN (i, si, slen, c);
1599
7117c2d2 1600 i = si + 1;
726f6388
JA
1601 continue;
1602 }
d233b485 1603#endif
726f6388 1604
495aee44
CR
1605 /* Pass the contents of double-quoted strings through verbatim. */
1606 if (c == '"')
1607 {
1608 si = i + 1;
a0c0a00f 1609 i = skip_double_quoted (string, slen, si, 0);
495aee44
CR
1610 /* skip_XXX_quoted leaves index one past close quote */
1611 continue;
1612 }
1613
1614 if (c == '\'')
1615 {
1616/*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
ac50fbac 1617 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
495aee44
CR
1618 ADVANCE_CHAR (string, slen, i);
1619 else
1620 {
1621 si = i + 1;
a0c0a00f 1622 i = skip_single_quoted (string, slen, si, 0);
495aee44
CR
1623 }
1624
1625 continue;
1626 }
7117c2d2 1627
d233b485
CR
1628#if defined (ARRAY_VARS)
1629 if (c == LBRACK && dolbrace_state == DOLBRACE_PARAM)
1630 {
1631 si = skipsubscript (string, i, 0);
1632 CHECK_STRING_OVERRUN (i, si, slen, c);
1633 if (string[si] == RBRACK)
1634 c = string[i = si];
1635 }
1636#endif
1637
7117c2d2
JA
1638 /* move past this character, which was not special. */
1639 ADVANCE_CHAR (string, slen, i);
495aee44
CR
1640
1641 /* This logic must agree with parse.y:parse_matched_pair, since they
1642 share the same defines. */
1643 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1644 dolbrace_state = DOLBRACE_QUOTE;
1645 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1646 dolbrace_state = DOLBRACE_QUOTE;
1647 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
ac50fbac 1648 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
495aee44
CR
1649 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1650 dolbrace_state = DOLBRACE_QUOTE;
1651 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1652 dolbrace_state = DOLBRACE_QUOTE;
a0c0a00f
CR
1653 /* This is intended to handle all of the [:]op expansions and the substring/
1654 length/pattern removal/pattern substitution expansions. */
495aee44
CR
1655 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1656 dolbrace_state = DOLBRACE_OP;
1657 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1658 dolbrace_state = DOLBRACE_WORD;
cce855bc 1659 }
726f6388 1660
b80f6443 1661 if (c == 0 && nesting_level)
cce855bc 1662 {
b80f6443
JA
1663 if (no_longjmp_on_fatal_error == 0)
1664 { /* { */
b80f6443 1665 last_command_exit_value = EXECUTION_FAILURE;
ac50fbac 1666 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
b80f6443
JA
1667 exp_jump_to_top_level (DISCARD);
1668 }
1669 else
1670 {
1671 *sindex = i;
1672 return ((char *)NULL);
1673 }
726f6388 1674 }
726f6388 1675
3185942a 1676 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
726f6388
JA
1677 *sindex = i;
1678
ccc6cda3 1679 return (result);
726f6388
JA
1680}
1681
ccc6cda3
JA
1682/* Remove backslashes which are quoting backquotes from STRING. Modifies
1683 STRING, and returns a pointer to it. */
1684char *
1685de_backslash (string)
726f6388 1686 char *string;
ccc6cda3 1687{
7117c2d2
JA
1688 register size_t slen;
1689 register int i, j, prev_i;
1690 DECLARE_MBSTATE;
726f6388 1691
7117c2d2
JA
1692 slen = strlen (string);
1693 i = j = 0;
1694
1695 /* Loop copying string[i] to string[j], i >= j. */
1696 while (i < slen)
1697 {
1698 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
ccc6cda3 1699 string[i + 1] == '$'))
7117c2d2
JA
1700 i++;
1701 prev_i = i;
1702 ADVANCE_CHAR (string, slen, i);
1703 if (j < prev_i)
b80f6443 1704 do string[j++] = string[prev_i++]; while (prev_i < i);
7117c2d2 1705 else
b80f6443 1706 j = i;
7117c2d2
JA
1707 }
1708 string[j] = '\0';
1709
ccc6cda3
JA
1710 return (string);
1711}
726f6388 1712
ccc6cda3 1713#if 0
cce855bc 1714/*UNUSED*/
ccc6cda3
JA
1715/* Replace instances of \! in a string with !. */
1716void
1717unquote_bang (string)
1718 char *string;
1719{
1720 register int i, j;
1721 register char *temp;
726f6388 1722
f73dda09 1723 temp = (char *)xmalloc (1 + strlen (string));
726f6388 1724
ccc6cda3
JA
1725 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1726 {
1727 if (string[i] == '\\' && string[i + 1] == '!')
1728 {
1729 temp[j] = '!';
1730 i++;
1731 }
1732 }
1733 strcpy (string, temp);
1734 free (temp);
726f6388 1735}
ccc6cda3 1736#endif
726f6388 1737
a0c0a00f 1738#define CQ_RETURN(x) do { no_longjmp_on_fatal_error = oldjmp; return (x); } while (0)
3185942a 1739
89a92869 1740/* This function assumes s[i] == open; returns with s[ret] == close; used to
0001803f
CR
1741 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1742 matched pairs of quotes or backquotes, or skip word expansions; it is
1743 intended to be used after expansion has been performed and during final
d233b485
CR
1744 assignment parsing (see arrayfunc.c:assign_compound_array_list()) or
1745 during execution by a builtin which has already undergone word expansion. */
89a92869
CR
1746static int
1747skip_matched_pair (string, start, open, close, flags)
1748 const char *string;
1749 int start, open, close, flags;
1750{
a0c0a00f 1751 int i, pass_next, backq, si, c, count, oldjmp;
89a92869
CR
1752 size_t slen;
1753 char *temp, *ss;
1754 DECLARE_MBSTATE;
1755
1756 slen = strlen (string + start) + start;
a0c0a00f 1757 oldjmp = no_longjmp_on_fatal_error;
89a92869
CR
1758 no_longjmp_on_fatal_error = 1;
1759
1760 i = start + 1; /* skip over leading bracket */
1761 count = 1;
1762 pass_next = backq = 0;
1763 ss = (char *)string;
1764 while (c = string[i])
1765 {
1766 if (pass_next)
1767 {
1768 pass_next = 0;
1769 if (c == 0)
1770 CQ_RETURN(i);
1771 ADVANCE_CHAR (string, slen, i);
1772 continue;
1773 }
d233b485 1774 else if ((flags & 1) == 0 && c == '\\')
89a92869
CR
1775 {
1776 pass_next = 1;
1777 i++;
1778 continue;
1779 }
1780 else if (backq)
1781 {
1782 if (c == '`')
1783 backq = 0;
1784 ADVANCE_CHAR (string, slen, i);
1785 continue;
1786 }
0001803f 1787 else if ((flags & 1) == 0 && c == '`')
89a92869
CR
1788 {
1789 backq = 1;
1790 i++;
1791 continue;
1792 }
0001803f 1793 else if ((flags & 1) == 0 && c == open)
89a92869
CR
1794 {
1795 count++;
1796 i++;
1797 continue;
1798 }
1799 else if (c == close)
1800 {
1801 count--;
1802 if (count == 0)
1803 break;
1804 i++;
1805 continue;
1806 }
0001803f 1807 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
89a92869 1808 {
a0c0a00f
CR
1809 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i, 0)
1810 : skip_double_quoted (ss, slen, ++i, 0);
89a92869
CR
1811 /* no increment, the skip functions increment past the closing quote. */
1812 }
0001803f 1813 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
89a92869
CR
1814 {
1815 si = i + 2;
1816 if (string[si] == '\0')
1817 CQ_RETURN(si);
1818
d233b485 1819 /* XXX - extract_command_subst here? */
89a92869
CR
1820 if (string[i+1] == LPAREN)
1821 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1822 else
1823 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
a0c0a00f
CR
1824
1825 CHECK_STRING_OVERRUN (i, si, slen, c);
1826
89a92869
CR
1827 i = si;
1828 if (string[i] == '\0') /* don't increment i past EOS in loop */
1829 break;
1830 i++;
1831 continue;
1832 }
1833 else
1834 ADVANCE_CHAR (string, slen, i);
1835 }
1836
1837 CQ_RETURN(i);
1838}
1839
1840#if defined (ARRAY_VARS)
d233b485
CR
1841/* Flags has 1 as a reserved value, since skip_matched_pair uses it for
1842 skipping over quoted strings and taking the first instance of the
1843 closing character. */
89a92869 1844int
0001803f 1845skipsubscript (string, start, flags)
89a92869 1846 const char *string;
0001803f 1847 int start, flags;
89a92869 1848{
0001803f 1849 return (skip_matched_pair (string, start, '[', ']', flags));
89a92869
CR
1850}
1851#endif
1852
3185942a
JA
1853/* Skip characters in STRING until we find a character in DELIMS, and return
1854 the index of that character. START is the index into string at which we
1855 begin. This is similar in spirit to strpbrk, but it returns an index into
1856 STRING and takes a starting index. This little piece of code knows quite
1857 a lot of shell syntax. It's very similar to skip_double_quoted and other
1858 functions of that ilk. */
1859int
1860skip_to_delim (string, start, delims, flags)
1861 char *string;
1862 int start;
1863 char *delims;
1864 int flags;
1865{
a0c0a00f
CR
1866 int i, pass_next, backq, dquote, si, c, oldjmp;
1867 int invert, skipquote, skipcmd, noprocsub, completeflag;
1868 int arithexp, skipcol;
3185942a 1869 size_t slen;
495aee44 1870 char *temp, open[3];
3185942a
JA
1871 DECLARE_MBSTATE;
1872
1873 slen = strlen (string + start) + start;
a0c0a00f 1874 oldjmp = no_longjmp_on_fatal_error;
3185942a
JA
1875 if (flags & SD_NOJMP)
1876 no_longjmp_on_fatal_error = 1;
1877 invert = (flags & SD_INVERT);
0001803f 1878 skipcmd = (flags & SD_NOSKIPCMD) == 0;
a0c0a00f
CR
1879 noprocsub = (flags & SD_NOPROCSUB);
1880 completeflag = (flags & SD_COMPLETE) ? SX_COMPLETE : 0;
1881
1882 arithexp = (flags & SD_ARITHEXP);
1883 skipcol = 0;
3185942a
JA
1884
1885 i = start;
a0c0a00f 1886 pass_next = backq = dquote = 0;
3185942a
JA
1887 while (c = string[i])
1888 {
0001803f
CR
1889 /* If this is non-zero, we should not let quote characters be delimiters
1890 and the current character is a single or double quote. We should not
1891 test whether or not it's a delimiter until after we skip single- or
1892 double-quoted strings. */
1893 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
3185942a
JA
1894 if (pass_next)
1895 {
1896 pass_next = 0;
1897 if (c == 0)
1898 CQ_RETURN(i);
1899 ADVANCE_CHAR (string, slen, i);
1900 continue;
1901 }
1902 else if (c == '\\')
1903 {
1904 pass_next = 1;
1905 i++;
1906 continue;
1907 }
1908 else if (backq)
1909 {
1910 if (c == '`')
1911 backq = 0;
1912 ADVANCE_CHAR (string, slen, i);
1913 continue;
1914 }
1915 else if (c == '`')
1916 {
1917 backq = 1;
1918 i++;
1919 continue;
1920 }
a0c0a00f 1921 else if (arithexp && skipcol && c == ':')
3185942a 1922 {
a0c0a00f
CR
1923 skipcol--;
1924 i++;
1925 continue;
3185942a 1926 }
a0c0a00f
CR
1927 else if (arithexp && c == '?')
1928 {
1929 skipcol++;
1930 i++;
1931 continue;
1932 }
1933 else if (skipquote == 0 && invert == 0 && member (c, delims))
1934 break;
1935 /* the usual case is to use skip_xxx_quoted, but we don't skip over double
1936 quoted strings when looking for the history expansion character as a
1937 delimiter. */
1938 /* special case for programmable completion which takes place before
1939 parser converts backslash-escaped single quotes between $'...' to
1940 `regular' single-quoted strings. */
1941 else if (completeflag && i > 0 && string[i-1] == '$' && c == '\'')
1942 i = skip_single_quoted (string, slen, ++i, SX_COMPLETE);
1943 else if (c == '\'')
1944 i = skip_single_quoted (string, slen, ++i, 0);
1945 else if (c == '"')
1946 i = skip_double_quoted (string, slen, ++i, completeflag);
1947 else if (c == LPAREN && arithexp)
1948 {
1949 si = i + 1;
1950 if (string[si] == '\0')
1951 CQ_RETURN(si);
1952
1953 temp = extract_delimited_string (string, &si, "(", "(", ")", SX_NOALLOC); /* ) */
1954 i = si;
1955 if (string[i] == '\0') /* don't increment i past EOS in loop */
1956 break;
1957 i++;
1958 continue;
1959 }
0001803f 1960 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
3185942a
JA
1961 {
1962 si = i + 2;
1963 if (string[si] == '\0')
1964 CQ_RETURN(si);
1965
1966 if (string[i+1] == LPAREN)
1967 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1968 else
1969 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
d233b485 1970 CHECK_STRING_OVERRUN (i, si, slen, c);
3185942a
JA
1971 i = si;
1972 if (string[i] == '\0') /* don't increment i past EOS in loop */
1973 break;
1974 i++;
1975 continue;
1976 }
0001803f 1977#if defined (PROCESS_SUBSTITUTION)
a0c0a00f 1978 else if (skipcmd && noprocsub == 0 && (c == '<' || c == '>') && string[i+1] == LPAREN)
0001803f
CR
1979 {
1980 si = i + 2;
1981 if (string[si] == '\0')
1982 CQ_RETURN(si);
d233b485 1983
a0c0a00f 1984 temp = extract_delimited_string (string, &si, (c == '<') ? "<(" : ">(", "(", ")", SX_COMMAND|SX_NOALLOC); /* )) */
d233b485 1985 CHECK_STRING_OVERRUN (i, si, slen, c);
0001803f
CR
1986 i = si;
1987 if (string[i] == '\0')
1988 break;
1989 i++;
1990 continue;
1991 }
1992#endif /* PROCESS_SUBSTITUTION */
495aee44
CR
1993#if defined (EXTENDED_GLOB)
1994 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1995 {
1996 si = i + 2;
1997 if (string[si] == '\0')
1998 CQ_RETURN(si);
1999
2000 open[0] = c;
2001 open[1] = LPAREN;
2002 open[2] = '\0';
2003 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
2004
d233b485 2005 CHECK_STRING_OVERRUN (i, si, slen, c);
495aee44
CR
2006 i = si;
2007 if (string[i] == '\0') /* don't increment i past EOS in loop */
2008 break;
2009 i++;
2010 continue;
2011 }
2012#endif
ac50fbac
CR
2013 else if ((flags & SD_GLOB) && c == LBRACK)
2014 {
2015 si = i + 1;
2016 if (string[si] == '\0')
2017 CQ_RETURN(si);
2018
2019 temp = extract_delimited_string (string, &si, "[", "[", "]", SX_NOALLOC); /* ] */
2020
2021 i = si;
2022 if (string[i] == '\0') /* don't increment i past EOS in loop */
2023 break;
2024 i++;
2025 continue;
2026 }
0001803f 2027 else if ((skipquote || invert) && (member (c, delims) == 0))
3185942a
JA
2028 break;
2029 else
2030 ADVANCE_CHAR (string, slen, i);
2031 }
2032
2033 CQ_RETURN(i);
2034}
2035
a0c0a00f
CR
2036#if defined (BANG_HISTORY)
2037/* Skip to the history expansion character (delims[0]), paying attention to
2038 quoted strings and command and process substitution. This is a stripped-
2039 down version of skip_to_delims. The essential difference is that this
2040 resets the quoting state when starting a command substitution */
2041int
2042skip_to_histexp (string, start, delims, flags)
2043 char *string;
2044 int start;
2045 char *delims;
2046 int flags;
2047{
d233b485 2048 int i, pass_next, backq, dquote, c, oldjmp;
a0c0a00f
CR
2049 int histexp_comsub, histexp_backq, old_dquote;
2050 size_t slen;
a0c0a00f
CR
2051 DECLARE_MBSTATE;
2052
2053 slen = strlen (string + start) + start;
2054 oldjmp = no_longjmp_on_fatal_error;
2055 if (flags & SD_NOJMP)
2056 no_longjmp_on_fatal_error = 1;
2057
2058 histexp_comsub = histexp_backq = old_dquote = 0;
2059
2060 i = start;
2061 pass_next = backq = dquote = 0;
2062 while (c = string[i])
2063 {
2064 if (pass_next)
2065 {
2066 pass_next = 0;
2067 if (c == 0)
2068 CQ_RETURN(i);
2069 ADVANCE_CHAR (string, slen, i);
2070 continue;
2071 }
2072 else if (c == '\\')
2073 {
2074 pass_next = 1;
2075 i++;
2076 continue;
2077 }
2078 else if (backq && c == '`')
2079 {
2080 backq = 0;
2081 histexp_backq--;
2082 dquote = old_dquote;
2083 i++;
2084 continue;
2085 }
2086 else if (c == '`')
2087 {
2088 backq = 1;
2089 histexp_backq++;
2090 old_dquote = dquote; /* simple - one level for now */
2091 dquote = 0;
2092 i++;
2093 continue;
2094 }
2095 /* When in double quotes, act as if the double quote is a member of
2096 history_no_expand_chars, like the history library does */
2097 else if (dquote && c == delims[0] && string[i+1] == '"')
2098 {
2099 i++;
2100 continue;
2101 }
2102 else if (c == delims[0])
2103 break;
2104 /* the usual case is to use skip_xxx_quoted, but we don't skip over double
2105 quoted strings when looking for the history expansion character as a
2106 delimiter. */
2107 else if (dquote && c == '\'')
2108 {
2109 i++;
2110 continue;
2111 }
2112 else if (c == '\'')
2113 i = skip_single_quoted (string, slen, ++i, 0);
2114 /* The posixly_correct test makes posix-mode shells allow double quotes
2115 to quote the history expansion character */
2116 else if (posixly_correct == 0 && c == '"')
2117 {
2118 dquote = 1 - dquote;
2119 i++;
2120 continue;
2121 }
2122 else if (c == '"')
2123 i = skip_double_quoted (string, slen, ++i, 0);
2124#if defined (PROCESS_SUBSTITUTION)
2125 else if ((c == '$' || c == '<' || c == '>') && string[i+1] == LPAREN && string[i+2] != LPAREN)
2126#else
2127 else if (c == '$' && string[i+1] == LPAREN && string[i+2] != LPAREN)
2128#endif
2129 {
2130 if (string[i+2] == '\0')
2131 CQ_RETURN(i+2);
2132 i += 2;
2133 histexp_comsub++;
2134 old_dquote = dquote;
2135 dquote = 0;
2136 }
2137 else if (histexp_comsub && c == RPAREN)
2138 {
2139 histexp_comsub--;
2140 dquote = old_dquote;
2141 i++;
2142 continue;
2143 }
2144 else if (backq) /* placeholder */
2145 {
2146 ADVANCE_CHAR (string, slen, i);
2147 continue;
2148 }
2149 else
2150 ADVANCE_CHAR (string, slen, i);
2151 }
2152
2153 CQ_RETURN(i);
2154}
2155#endif /* BANG_HISTORY */
2156
ccc6cda3 2157#if defined (READLINE)
726f6388
JA
2158/* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
2159 an unclosed quoted string), or if the character at EINDEX is quoted
28ef6c31 2160 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
b72432fd 2161 single and double-quoted string parsing functions should not return an
7117c2d2
JA
2162 error if there are unclosed quotes or braces. The characters that this
2163 recognizes need to be the same as the contents of
2164 rl_completer_quote_characters. */
b72432fd 2165
726f6388
JA
2166int
2167char_is_quoted (string, eindex)
2168 char *string;
2169 int eindex;
2170{
a0c0a00f 2171 int i, pass_next, c, oldjmp;
7117c2d2
JA
2172 size_t slen;
2173 DECLARE_MBSTATE;
726f6388 2174
7117c2d2 2175 slen = strlen (string);
a0c0a00f 2176 oldjmp = no_longjmp_on_fatal_error;
28ef6c31 2177 no_longjmp_on_fatal_error = 1;
7117c2d2
JA
2178 i = pass_next = 0;
2179 while (i <= eindex)
726f6388 2180 {
7117c2d2
JA
2181 c = string[i];
2182
726f6388
JA
2183 if (pass_next)
2184 {
2185 pass_next = 0;
2186 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
b72432fd 2187 CQ_RETURN(1);
7117c2d2 2188 ADVANCE_CHAR (string, slen, i);
726f6388
JA
2189 continue;
2190 }
7117c2d2 2191 else if (c == '\\')
ccc6cda3
JA
2192 {
2193 pass_next = 1;
7117c2d2 2194 i++;
ccc6cda3
JA
2195 continue;
2196 }
a0c0a00f
CR
2197 else if (c == '$' && string[i+1] == '\'' && string[i+2])
2198 {
2199 i += 2;
2200 i = skip_single_quoted (string, slen, i, SX_COMPLETE);
2201 if (i > eindex)
2202 CQ_RETURN (i);
2203 }
7117c2d2
JA
2204 else if (c == '\'' || c == '"')
2205 {
a0c0a00f
CR
2206 i = (c == '\'') ? skip_single_quoted (string, slen, ++i, 0)
2207 : skip_double_quoted (string, slen, ++i, SX_COMPLETE);
7117c2d2
JA
2208 if (i > eindex)
2209 CQ_RETURN(1);
2210 /* no increment, the skip_xxx functions go one past end */
2211 }
2212 else
2213 ADVANCE_CHAR (string, slen, i);
726f6388 2214 }
7117c2d2 2215
b72432fd 2216 CQ_RETURN(0);
726f6388
JA
2217}
2218
726f6388
JA
2219int
2220unclosed_pair (string, eindex, openstr)
2221 char *string;
2222 int eindex;
2223 char *openstr;
2224{
ccc6cda3 2225 int i, pass_next, openc, olen;
7117c2d2
JA
2226 size_t slen;
2227 DECLARE_MBSTATE;
726f6388 2228
7117c2d2 2229 slen = strlen (string);
726f6388 2230 olen = strlen (openstr);
7117c2d2
JA
2231 i = pass_next = openc = 0;
2232 while (i <= eindex)
726f6388
JA
2233 {
2234 if (pass_next)
2235 {
2236 pass_next = 0;
2237 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
2238 return 0;
7117c2d2
JA
2239 ADVANCE_CHAR (string, slen, i);
2240 continue;
2241 }
2242 else if (string[i] == '\\')
2243 {
2244 pass_next = 1;
2245 i++;
726f6388
JA
2246 continue;
2247 }
2248 else if (STREQN (string + i, openstr, olen))
2249 {
2250 openc = 1 - openc;
7117c2d2 2251 i += olen;
726f6388 2252 }
a0c0a00f 2253 /* XXX - may want to handle $'...' specially here */
ccc6cda3 2254 else if (string[i] == '\'' || string[i] == '"')
726f6388 2255 {
a0c0a00f
CR
2256 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i, 0)
2257 : skip_double_quoted (string, slen, i, SX_COMPLETE);
726f6388
JA
2258 if (i > eindex)
2259 return 0;
2260 }
7117c2d2
JA
2261 else
2262 ADVANCE_CHAR (string, slen, i);
726f6388
JA
2263 }
2264 return (openc);
2265}
bb70624e 2266
bb70624e
JA
2267/* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
2268 individual words. If DELIMS is NULL, the current value of $IFS is used
b80f6443
JA
2269 to split the string, and the function follows the shell field splitting
2270 rules. SENTINEL is an index to look for. NWP, if non-NULL,
bb70624e
JA
2271 gets the number of words in the returned list. CWP, if non-NULL, gets
2272 the index of the word containing SENTINEL. Non-whitespace chars in
d233b485 2273 DELIMS delimit separate fields. This is used by programmable completion. */
bb70624e 2274WORD_LIST *
0001803f 2275split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
bb70624e
JA
2276 char *string;
2277 int slen;
2278 char *delims;
0001803f 2279 int sentinel, flags;
bb70624e
JA
2280 int *nwp, *cwp;
2281{
0001803f 2282 int ts, te, i, nw, cw, ifs_split, dflags;
f73dda09 2283 char *token, *d, *d2;
bb70624e
JA
2284 WORD_LIST *ret, *tl;
2285
2286 if (string == 0 || *string == '\0')
2287 {
2288 if (nwp)
2289 *nwp = 0;
2290 if (cwp)
2291 *cwp = 0;
2292 return ((WORD_LIST *)NULL);
2293 }
2294
7117c2d2 2295 d = (delims == 0) ? ifs_value : delims;
b80f6443 2296 ifs_split = delims == 0;
bb70624e
JA
2297
2298 /* Make d2 the non-whitespace characters in delims */
2299 d2 = 0;
2300 if (delims)
2301 {
95732b49
JA
2302 size_t slength;
2303#if defined (HANDLE_MULTIBYTE)
2304 size_t mblength = 1;
2305#endif
2306 DECLARE_MBSTATE;
2307
2308 slength = strlen (delims);
2309 d2 = (char *)xmalloc (slength + 1);
2310 i = ts = 0;
2311 while (delims[i])
bb70624e 2312 {
95732b49 2313#if defined (HANDLE_MULTIBYTE)
0628567a
JA
2314 mbstate_t state_bak;
2315 state_bak = state;
95732b49
JA
2316 mblength = MBRLEN (delims + i, slength, &state);
2317 if (MB_INVALIDCH (mblength))
2318 state = state_bak;
2319 else if (mblength > 1)
2320 {
2321 memcpy (d2 + ts, delims + i, mblength);
2322 ts += mblength;
2323 i += mblength;
2324 slength -= mblength;
2325 continue;
2326 }
2327#endif
2328 if (whitespace (delims[i]) == 0)
bb70624e 2329 d2[ts++] = delims[i];
95732b49
JA
2330
2331 i++;
2332 slength--;
bb70624e
JA
2333 }
2334 d2[ts] = '\0';
2335 }
2336
2337 ret = (WORD_LIST *)NULL;
2338
0001803f 2339 /* Remove sequences of whitespace characters at the start of the string, as
b80f6443
JA
2340 long as those characters are delimiters. */
2341 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
bb70624e
JA
2342 ;
2343 if (string[i] == '\0')
a0c0a00f
CR
2344 {
2345 FREE (d2);
2346 return (ret);
2347 }
bb70624e
JA
2348
2349 ts = i;
2350 nw = 0;
2351 cw = -1;
0001803f 2352 dflags = flags|SD_NOJMP;
bb70624e
JA
2353 while (1)
2354 {
0001803f 2355 te = skip_to_delim (string, ts, d, dflags);
bb70624e
JA
2356
2357 /* If we have a non-whitespace delimiter character, use it to make a
2358 separate field. This is just about what $IFS splitting does and
2359 is closer to the behavior of the shell parser. */
28ef6c31 2360 if (ts == te && d2 && member (string[ts], d2))
bb70624e
JA
2361 {
2362 te = ts + 1;
b80f6443
JA
2363 /* If we're using IFS splitting, the non-whitespace delimiter char
2364 and any additional IFS whitespace delimits a field. */
2365 if (ifs_split)
d233b485 2366 while (member (string[te], d) && spctabnl (string[te]) && ((flags&SD_NOQUOTEDELIM) == 0 || (string[te] != '\'' && string[te] != '"')))
b80f6443
JA
2367 te++;
2368 else
d233b485 2369 while (member (string[te], d2) && ((flags&SD_NOQUOTEDELIM) == 0 || (string[te] != '\'' && string[te] != '"')))
b80f6443 2370 te++;
bb70624e
JA
2371 }
2372
2373 token = substring (string, ts, te);
2374
712f80b0 2375 ret = add_string_to_list (token, ret); /* XXX */
bb70624e
JA
2376 free (token);
2377 nw++;
2378
2379 if (sentinel >= ts && sentinel <= te)
2380 cw = nw;
2381
2382 /* If the cursor is at whitespace just before word start, set the
28ef6c31 2383 sentinel word to the current word. */
bb70624e
JA
2384 if (cwp && cw == -1 && sentinel == ts-1)
2385 cw = nw;
2386
2387 /* If the cursor is at whitespace between two words, make a new, empty
28ef6c31
JA
2388 word, add it before (well, after, since the list is in reverse order)
2389 the word we just added, and set the current word to that one. */
bb70624e 2390 if (cwp && cw == -1 && sentinel < ts)
28ef6c31 2391 {
7117c2d2 2392 tl = make_word_list (make_word (""), ret->next);
28ef6c31
JA
2393 ret->next = tl;
2394 cw = nw;
2395 nw++;
2396 }
bb70624e
JA
2397
2398 if (string[te] == 0)
2399 break;
2400
b80f6443 2401 i = te;
d233b485
CR
2402 /* XXX - honor SD_NOQUOTEDELIM here */
2403 while (member (string[i], d) && (ifs_split || spctabnl(string[i])) && ((flags&SD_NOQUOTEDELIM) == 0 || (string[te] != '\'' && string[te] != '"')))
bb70624e
JA
2404 i++;
2405
2406 if (string[i])
2407 ts = i;
2408 else
2409 break;
2410 }
2411
2412 /* Special case for SENTINEL at the end of STRING. If we haven't found
2413 the word containing SENTINEL yet, and the index we're looking for is at
0001803f
CR
2414 the end of STRING (or past the end of the previously-found token,
2415 possible if the end of the line is composed solely of IFS whitespace)
2416 add an additional null argument and set the current word pointer to that. */
2417 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
bb70624e
JA
2418 {
2419 if (whitespace (string[sentinel - 1]))
28ef6c31
JA
2420 {
2421 token = "";
2422 ret = add_string_to_list (token, ret);
2423 nw++;
2424 }
bb70624e
JA
2425 cw = nw;
2426 }
2427
2428 if (nwp)
2429 *nwp = nw;
2430 if (cwp)
2431 *cwp = cw;
2432
ac50fbac
CR
2433 FREE (d2);
2434
bb70624e
JA
2435 return (REVERSE_LIST (ret, WORD_LIST *));
2436}
726f6388
JA
2437#endif /* READLINE */
2438
ccc6cda3
JA
2439#if 0
2440/* UNUSED */
726f6388
JA
2441/* Extract the name of the variable to bind to from the assignment string. */
2442char *
2443assignment_name (string)
2444 char *string;
2445{
ccc6cda3 2446 int offset;
726f6388
JA
2447 char *temp;
2448
b80f6443 2449 offset = assignment (string, 0);
ccc6cda3 2450 if (offset == 0)
726f6388 2451 return (char *)NULL;
bb70624e 2452 temp = substring (string, 0, offset);
726f6388
JA
2453 return (temp);
2454}
ccc6cda3 2455#endif
726f6388 2456
cce855bc
JA
2457/* **************************************************************** */
2458/* */
2459/* Functions to convert strings to WORD_LISTs and vice versa */
2460/* */
2461/* **************************************************************** */
2462
726f6388
JA
2463/* Return a single string of all the words in LIST. SEP is the separator
2464 to put between individual elements of LIST in the output string. */
7117c2d2 2465char *
726f6388
JA
2466string_list_internal (list, sep)
2467 WORD_LIST *list;
2468 char *sep;
2469{
2470 register WORD_LIST *t;
2471 char *result, *r;
a0c0a00f 2472 size_t word_len, sep_len, result_size;
726f6388 2473
ccc6cda3 2474 if (list == 0)
726f6388
JA
2475 return ((char *)NULL);
2476
b80f6443
JA
2477 /* Short-circuit quickly if we don't need to separate anything. */
2478 if (list->next == 0)
2479 return (savestring (list->word->word));
2480
726f6388
JA
2481 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2482 sep_len = STRLEN (sep);
2483 result_size = 0;
2484
2485 for (t = list; t; t = t->next)
2486 {
2487 if (t != list)
2488 result_size += sep_len;
2489 result_size += strlen (t->word->word);
2490 }
2491
f73dda09 2492 r = result = (char *)xmalloc (result_size + 1);
726f6388
JA
2493
2494 for (t = list; t; t = t->next)
2495 {
2496 if (t != list && sep_len)
2497 {
ccc6cda3
JA
2498 if (sep_len > 1)
2499 {
2500 FASTCOPY (sep, r, sep_len);
2501 r += sep_len;
2502 }
2503 else
2504 *r++ = sep[0];
726f6388
JA
2505 }
2506
2507 word_len = strlen (t->word->word);
2508 FASTCOPY (t->word->word, r, word_len);
2509 r += word_len;
2510 }
2511
ccc6cda3 2512 *r = '\0';
726f6388
JA
2513 return (result);
2514}
2515
2516/* Return a single string of all the words present in LIST, separating
2517 each word with a space. */
2518char *
2519string_list (list)
2520 WORD_LIST *list;
2521{
2522 return (string_list_internal (list, " "));
2523}
2524
3185942a
JA
2525/* An external interface that can be used by the rest of the shell to
2526 obtain a string containing the first character in $IFS. Handles all
2527 the multibyte complications. If LENP is non-null, it is set to the
2528 length of the returned string. */
2529char *
2530ifs_firstchar (lenp)
2531 int *lenp;
2532{
2533 char *ret;
2534 int len;
2535
2536 ret = xmalloc (MB_LEN_MAX + 1);
2537#if defined (HANDLE_MULTIBYTE)
2538 if (ifs_firstc_len == 1)
2539 {
2540 ret[0] = ifs_firstc[0];
2541 ret[1] = '\0';
2542 len = ret[0] ? 1 : 0;
2543 }
2544 else
2545 {
2546 memcpy (ret, ifs_firstc, ifs_firstc_len);
2547 ret[len = ifs_firstc_len] = '\0';
2548 }
2549#else
2550 ret[0] = ifs_firstc;
2551 ret[1] = '\0';
2552 len = ret[0] ? 0 : 1;
2553#endif
2554
2555 if (lenp)
2556 *lenp = len;
2557
2558 return ret;
2559}
2560
726f6388
JA
2561/* Return a single string of all the words present in LIST, obeying the
2562 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2563 expansion [of $*] appears within a double quoted string, it expands
2564 to a single field with the value of each parameter separated by the
2565 first character of the IFS variable, or by a <space> if IFS is unset." */
d233b485
CR
2566/* Posix interpretation 888 changes this when IFS is null by specifying
2567 that when unquoted, this expands to separate arguments */
f73dda09 2568char *
d233b485 2569string_list_dollar_star (list, quoted, flags)
726f6388 2570 WORD_LIST *list;
d233b485 2571 int quoted, flags;
726f6388 2572{
0628567a 2573 char *ret;
95732b49 2574#if defined (HANDLE_MULTIBYTE)
0628567a 2575# if defined (__GNUC__)
95732b49 2576 char sep[MB_CUR_MAX + 1];
0628567a
JA
2577# else
2578 char *sep = 0;
2579# endif
95732b49 2580#else
7117c2d2 2581 char sep[2];
95732b49 2582#endif
726f6388 2583
95732b49 2584#if defined (HANDLE_MULTIBYTE)
0628567a
JA
2585# if !defined (__GNUC__)
2586 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2587# endif /* !__GNUC__ */
95732b49
JA
2588 if (ifs_firstc_len == 1)
2589 {
2590 sep[0] = ifs_firstc[0];
2591 sep[1] = '\0';
2592 }
2593 else
2594 {
2595 memcpy (sep, ifs_firstc, ifs_firstc_len);
2596 sep[ifs_firstc_len] = '\0';
2597 }
2598#else
7117c2d2 2599 sep[0] = ifs_firstc;
726f6388 2600 sep[1] = '\0';
95732b49 2601#endif
726f6388 2602
0628567a
JA
2603 ret = string_list_internal (list, sep);
2604#if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2605 free (sep);
2606#endif
2607 return ret;
726f6388
JA
2608}
2609
cce855bc
JA
2610/* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2611 is non-zero, the $@ appears within double quotes, and we should quote
2612 the list before converting it into a string. If IFS is unset, and the
2613 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2614 in the words in the list, because the default value of $IFS is
2615 <space><tab><newline>, IFS characters in the words in the list should
2616 also be split. If IFS is null, and the word is not quoted, we need
2617 to quote the words in the list to preserve the positional parameters
a0c0a00f
CR
2618 exactly.
2619 Valid values for the FLAGS argument are the PF_ flags in command.h,
2620 the only one we care about is PF_ASSIGNRHS. $@ is supposed to expand
2621 to the positional parameters separated by spaces no matter what IFS is
2622 set to if in a context where word splitting is not performed. The only
2623 one that we didn't handle before is assignment statement arguments to
2624 declaration builtins like `declare'. */
f73dda09 2625char *
a0c0a00f 2626string_list_dollar_at (list, quoted, flags)
cce855bc
JA
2627 WORD_LIST *list;
2628 int quoted;
a0c0a00f 2629 int flags;
cce855bc 2630{
95732b49
JA
2631 char *ifs, *ret;
2632#if defined (HANDLE_MULTIBYTE)
0628567a 2633# if defined (__GNUC__)
95732b49 2634 char sep[MB_CUR_MAX + 1];
0628567a
JA
2635# else
2636 char *sep = 0;
2637# endif /* !__GNUC__ */
95732b49
JA
2638#else
2639 char sep[2];
2640#endif
cce855bc
JA
2641 WORD_LIST *tlist;
2642
7117c2d2
JA
2643 /* XXX this could just be ifs = ifs_value; */
2644 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
cce855bc 2645
95732b49 2646#if defined (HANDLE_MULTIBYTE)
0628567a
JA
2647# if !defined (__GNUC__)
2648 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2649# endif /* !__GNUC__ */
d233b485
CR
2650 /* XXX - testing PF_ASSIGNRHS to make sure positional parameters are
2651 separated with a space even when word splitting will not occur. */
a0c0a00f
CR
2652 if (flags & PF_ASSIGNRHS)
2653 {
2654 sep[0] = ' ';
2655 sep[1] = '\0';
2656 }
2657 else if (ifs && *ifs)
95732b49
JA
2658 {
2659 if (ifs_firstc_len == 1)
2660 {
2661 sep[0] = ifs_firstc[0];
2662 sep[1] = '\0';
2663 }
2664 else
2665 {
2666 memcpy (sep, ifs_firstc, ifs_firstc_len);
2667 sep[ifs_firstc_len] = '\0';
2668 }
2669 }
2670 else
2671 {
2672 sep[0] = ' ';
2673 sep[1] = '\0';
2674 }
d233b485
CR
2675#else /* !HANDLE_MULTIBYTE */
2676 /* XXX - PF_ASSIGNRHS means no word splitting, so we want positional
2677 parameters separated by a space. */
a0c0a00f 2678 sep[0] = ((flags & PF_ASSIGNRHS) || ifs == 0 || *ifs == 0) ? ' ' : *ifs;
cce855bc 2679 sep[1] = '\0';
d233b485 2680#endif /* !HANDLE_MULTIBYTE */
cce855bc 2681
f1be666c
JA
2682 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2683 it now that quote_escapes quotes spaces */
0001803f 2684 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
cce855bc
JA
2685 ? quote_list (list)
2686 : list_quote_escapes (list);
0628567a
JA
2687
2688 ret = string_list_internal (tlist, sep);
2689#if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2690 free (sep);
2691#endif
2692 return ret;
cce855bc
JA
2693}
2694
ac50fbac 2695/* Turn the positional parameters into a string, understanding quoting and
3185942a
JA
2696 the various subtleties of using the first character of $IFS as the
2697 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2698 string_list as appropriate. */
712f80b0
CR
2699/* This needs to fully understand the additional contexts where word
2700 splitting does not occur (W_ASSIGNRHS, etc.) */
3185942a 2701char *
712f80b0 2702string_list_pos_params (pchar, list, quoted, pflags)
3185942a
JA
2703 int pchar;
2704 WORD_LIST *list;
712f80b0 2705 int quoted, pflags;
3185942a
JA
2706{
2707 char *ret;
2708 WORD_LIST *tlist;
2709
2710 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2711 {
2712 tlist = quote_list (list);
2713 word_list_remove_quoted_nulls (tlist);
d233b485 2714 ret = string_list_dollar_star (tlist, 0, 0);
3185942a
JA
2715 }
2716 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2717 {
2718 tlist = quote_list (list);
2719 word_list_remove_quoted_nulls (tlist);
2720 ret = string_list (tlist);
2721 }
d233b485
CR
2722 else if (pchar == '*' && quoted == 0 && ifs_is_null) /* XXX */
2723 ret = expand_no_split_dollar_star ? string_list_dollar_star (list, quoted, 0) : string_list_dollar_at (list, quoted, 0); /* Posix interp 888 */
712f80b0
CR
2724 else if (pchar == '*' && quoted == 0 && (pflags & PF_ASSIGNRHS)) /* XXX */
2725 ret = expand_no_split_dollar_star ? string_list_dollar_star (list, quoted, 0) : string_list_dollar_at (list, quoted, 0); /* Posix interp 888 */
3185942a
JA
2726 else if (pchar == '*')
2727 {
2728 /* Even when unquoted, string_list_dollar_star does the right thing
2729 making sure that the first character of $IFS is used as the
2730 separator. */
d233b485 2731 ret = string_list_dollar_star (list, quoted, 0);
3185942a
JA
2732 }
2733 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2734 /* We use string_list_dollar_at, but only if the string is quoted, since
2735 that quotes the escapes if it's not, which we don't want. We could
2736 use string_list (the old code did), but that doesn't do the right
2737 thing if the first character of $IFS is not a space. We use
2738 string_list_dollar_star if the string is unquoted so we make sure that
2739 the elements of $@ are separated by the first character of $IFS for
2740 later splitting. */
a0c0a00f 2741 ret = string_list_dollar_at (list, quoted, 0);
d233b485
CR
2742 else if (pchar == '@' && quoted == 0 && ifs_is_null) /* XXX */
2743 ret = string_list_dollar_at (list, quoted, 0); /* Posix interp 888 */
712f80b0
CR
2744 else if (pchar == '@' && quoted == 0 && (pflags & PF_ASSIGNRHS))
2745 ret = string_list_dollar_at (list, quoted, pflags); /* Posix interp 888 */
3185942a 2746 else if (pchar == '@')
d233b485 2747 ret = string_list_dollar_star (list, quoted, 0);
3185942a
JA
2748 else
2749 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2750
2751 return ret;
2752}
2753
726f6388
JA
2754/* Return the list of words present in STRING. Separate the string into
2755 words at any of the characters found in SEPARATORS. If QUOTED is
2756 non-zero then word in the list will have its quoted flag set, otherwise
2757 the quoted flag is left as make_word () deemed fit.
2758
2759 This obeys the P1003.2 word splitting semantics. If `separators' is
2760 exactly <space><tab><newline>, then the splitting algorithm is that of
2761 the Bourne shell, which treats any sequence of characters from `separators'
2762 as a delimiter. If IFS is unset, which results in `separators' being set
2763 to "", no splitting occurs. If separators has some other value, the
2764 following rules are applied (`IFS white space' means zero or more
2765 occurrences of <space>, <tab>, or <newline>, as long as those characters
2766 are in `separators'):
2767
2768 1) IFS white space is ignored at the start and the end of the
2769 string.
2770 2) Each occurrence of a character in `separators' that is not
2771 IFS white space, along with any adjacent occurrences of
2772 IFS white space delimits a field.
2773 3) Any nonzero-length sequence of IFS white space delimits a field.
2774 */
2775
2776/* BEWARE! list_string strips null arguments. Don't call it twice and
2777 expect to have "" preserved! */
2778
726f6388
JA
2779/* This performs word splitting and quoted null character removal on
2780 STRING. */
b80f6443
JA
2781#define issep(c) \
2782 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2783 : (c) == (separators)[0]) \
2784 : 0)
726f6388 2785
d233b485
CR
2786/* member of the space character class in the current locale */
2787#define ifs_whitespace(c) ISSPACE(c)
2788
2789/* "adjacent IFS white space" */
2790#define ifs_whitesep(c) ((sh_style_split || separators == 0) ? spctabnl (c) \
2791 : ifs_whitespace (c))
2792
726f6388
JA
2793WORD_LIST *
2794list_string (string, separators, quoted)
2795 register char *string, *separators;
2796 int quoted;
2797{
ccc6cda3
JA
2798 WORD_LIST *result;
2799 WORD_DESC *t;
2800 char *current_word, *s;
712f80b0 2801 int sindex, sh_style_split, whitesep, xflags, free_word;
95732b49 2802 size_t slen;
726f6388
JA
2803
2804 if (!string || !*string)
2805 return ((WORD_LIST *)NULL);
2806
7117c2d2
JA
2807 sh_style_split = separators && separators[0] == ' ' &&
2808 separators[1] == '\t' &&
2809 separators[2] == '\n' &&
2810 separators[3] == '\0';
3185942a
JA
2811 for (xflags = 0, s = ifs_value; s && *s; s++)
2812 {
2813 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2814 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2815 }
726f6388 2816
95732b49 2817 slen = 0;
726f6388
JA
2818 /* Remove sequences of whitespace at the beginning of STRING, as
2819 long as those characters appear in IFS. Do not do this if
d233b485
CR
2820 STRING is quoted or if there are no separator characters. We use the
2821 Posix definition of whitespace as a member of the space character
2822 class in the current locale. */
712f80b0 2823#if 0
726f6388 2824 if (!quoted || !separators || !*separators)
712f80b0
CR
2825#else
2826 /* issep() requires that separators be non-null, and always returns 0 if
2827 separator is the empty string, so don't bother if we get an empty string
2828 for separators. We already returned NULL above if STRING is empty. */
2829 if (!quoted && separators && *separators)
2830#endif
726f6388 2831 {
d233b485 2832 for (s = string; *s && issep (*s) && ifs_whitespace (*s); s++);
726f6388
JA
2833
2834 if (!*s)
2835 return ((WORD_LIST *)NULL);
2836
2837 string = s;
2838 }
2839
2840 /* OK, now STRING points to a word that does not begin with white space.
2841 The splitting algorithm is:
7117c2d2 2842 extract a word, stopping at a separator
d233b485 2843 skip sequences of whitespace characters as long as they are separators
726f6388 2844 This obeys the field splitting rules in Posix.2. */
d233b485 2845 slen = STRLEN (string);
ccc6cda3 2846 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
726f6388 2847 {
d233b485
CR
2848 /* Don't need string length in ADVANCE_CHAR unless multibyte chars are
2849 possible, but need it in string_extract_verbatim for bounds checking */
3185942a 2850 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
ccc6cda3 2851 if (current_word == 0)
726f6388
JA
2852 break;
2853
712f80b0
CR
2854 free_word = 1; /* If non-zero, we free current_word */
2855
726f6388
JA
2856 /* If we have a quoted empty string, add a quoted null argument. We
2857 want to preserve the quoted null character iff this is a quoted
2858 empty string; otherwise the quoted null characters are removed
2859 below. */
2860 if (QUOTED_NULL (current_word))
2861 {
95732b49 2862 t = alloc_word_desc ();
726f6388 2863 t->word = make_quoted_char ('\0');
95732b49 2864 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
726f6388
JA
2865 result = make_word_list (t, result);
2866 }
ccc6cda3 2867 else if (current_word[0] != '\0')
726f6388
JA
2868 {
2869 /* If we have something, then add it regardless. However,
2870 perform quoted null character removal on the current word. */
2871 remove_quoted_nulls (current_word);
712f80b0
CR
2872
2873 /* We don't want to set the word flags based on the string contents
2874 here -- that's mostly for the parser -- so we just allocate a
2875 WORD_DESC *, assign current_word (noting that we don't want to
2876 free it), and skip all of make_word. */
2877 t = alloc_word_desc ();
2878 t->word = current_word;
2879 result = make_word_list (t, result);
2880 free_word = 0;
95732b49 2881 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
ccc6cda3
JA
2882 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2883 result->word->flags |= W_QUOTED;
712f80b0
CR
2884 /* If removing quoted null characters leaves an empty word, note
2885 that we saw this for the caller to act on. */
2886 if (current_word == 0 || current_word[0] == '\0')
2887 result->word->flags |= W_SAWQUOTEDNULL;
726f6388
JA
2888 }
2889
2890 /* If we're not doing sequences of separators in the traditional
2891 Bourne shell style, then add a quoted null argument. */
d233b485 2892 else if (!sh_style_split && !ifs_whitespace (string[sindex]))
726f6388 2893 {
95732b49 2894 t = alloc_word_desc ();
ccc6cda3 2895 t->word = make_quoted_char ('\0');
95732b49 2896 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
ccc6cda3 2897 result = make_word_list (t, result);
726f6388
JA
2898 }
2899
712f80b0
CR
2900 if (free_word)
2901 free (current_word);
726f6388 2902
28ef6c31 2903 /* Note whether or not the separator is IFS whitespace, used later. */
d233b485 2904 whitesep = string[sindex] && ifs_whitesep (string[sindex]);
28ef6c31 2905
726f6388
JA
2906 /* Move past the current separator character. */
2907 if (string[sindex])
95732b49
JA
2908 {
2909 DECLARE_MBSTATE;
2910 ADVANCE_CHAR (string, slen, sindex);
2911 }
726f6388 2912
d233b485 2913 /* Now skip sequences of whitespace characters if they are
726f6388 2914 in the list of separators. */
d233b485 2915 while (string[sindex] && ifs_whitesep (string[sindex]) && issep (string[sindex]))
726f6388 2916 sindex++;
28ef6c31 2917
7117c2d2
JA
2918 /* If the first separator was IFS whitespace and the current character
2919 is a non-whitespace IFS character, it should be part of the current
2920 field delimiter, not a separate delimiter that would result in an
2921 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
d233b485 2922 if (string[sindex] && whitesep && issep (string[sindex]) && !ifs_whitesep (string[sindex]))
95732b49
JA
2923 {
2924 sindex++;
2925 /* An IFS character that is not IFS white space, along with any
2926 adjacent IFS white space, shall delimit a field. (SUSv3) */
d233b485 2927 while (string[sindex] && ifs_whitesep (string[sindex]) && isifs (string[sindex]))
95732b49
JA
2928 sindex++;
2929 }
726f6388
JA
2930 }
2931 return (REVERSE_LIST (result, WORD_LIST *));
2932}
2933
2934/* Parse a single word from STRING, using SEPARATORS to separate fields.
2935 ENDPTR is set to the first character after the word. This is used by
bc007799
CR
2936 the `read' builtin.
2937
2938 This is never called with SEPARATORS != $IFS, and takes advantage of that.
7117c2d2 2939
726f6388
JA
2940 XXX - this function is very similar to list_string; they should be
2941 combined - XXX */
bc007799 2942
d233b485 2943/* character is in $IFS */
bc007799
CR
2944#define islocalsep(c) (local_cmap[(unsigned char)(c)] != 0)
2945
726f6388
JA
2946char *
2947get_word_from_string (stringp, separators, endptr)
2948 char **stringp, *separators, **endptr;
2949{
2950 register char *s;
2951 char *current_word;
3185942a 2952 int sindex, sh_style_split, whitesep, xflags;
bc007799 2953 unsigned char local_cmap[UCHAR_MAX+1]; /* really only need single-byte chars here */
95732b49 2954 size_t slen;
726f6388
JA
2955
2956 if (!stringp || !*stringp || !**stringp)
2957 return ((char *)NULL);
ccc6cda3 2958
7117c2d2
JA
2959 sh_style_split = separators && separators[0] == ' ' &&
2960 separators[1] == '\t' &&
2961 separators[2] == '\n' &&
2962 separators[3] == '\0';
bc007799
CR
2963 memset (local_cmap, '\0', sizeof (local_cmap));
2964 for (xflags = 0, s = separators; s && *s; s++)
3185942a
JA
2965 {
2966 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2967 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
bc007799 2968 local_cmap[(unsigned char)*s] = 1; /* local charmap of separators */
3185942a 2969 }
726f6388 2970
3185942a 2971 s = *stringp;
95732b49
JA
2972 slen = 0;
2973
726f6388 2974 /* Remove sequences of whitespace at the beginning of STRING, as
bc007799
CR
2975 long as those characters appear in SEPARATORS. This happens if
2976 SEPARATORS == $' \t\n' or if IFS is unset. */
2977 if (sh_style_split || separators == 0)
d233b485
CR
2978 for (; *s && spctabnl (*s) && islocalsep (*s); s++);
2979 else
2980 for (; *s && ifs_whitespace (*s) && islocalsep (*s); s++);
726f6388 2981
d233b485
CR
2982 /* If the string is nothing but whitespace, update it and return. */
2983 if (!*s)
2984 {
2985 *stringp = s;
2986 if (endptr)
2987 *endptr = s;
2988 return ((char *)NULL);
726f6388
JA
2989 }
2990
2991 /* OK, S points to a word that does not begin with white space.
2992 Now extract a word, stopping at a separator, save a pointer to
2993 the first character after the word, then skip sequences of spc,
2994 tab, or nl as long as they are separators.
ccc6cda3 2995
726f6388
JA
2996 This obeys the field splitting rules in Posix.2. */
2997 sindex = 0;
bc007799
CR
2998 /* Don't need string length in ADVANCE_CHAR unless multibyte chars are
2999 possible, but need it in string_extract_verbatim for bounds checking */
3000 slen = STRLEN (s);
3185942a 3001 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
726f6388
JA
3002
3003 /* Set ENDPTR to the first character after the end of the word. */
3004 if (endptr)
3005 *endptr = s + sindex;
3006
28ef6c31 3007 /* Note whether or not the separator is IFS whitespace, used later. */
d233b485 3008 whitesep = s[sindex] && ifs_whitesep (s[sindex]);
28ef6c31 3009
726f6388
JA
3010 /* Move past the current separator character. */
3011 if (s[sindex])
95732b49
JA
3012 {
3013 DECLARE_MBSTATE;
3014 ADVANCE_CHAR (s, slen, sindex);
3015 }
726f6388
JA
3016
3017 /* Now skip sequences of space, tab, or newline characters if they are
3018 in the list of separators. */
bc007799 3019 while (s[sindex] && spctabnl (s[sindex]) && islocalsep (s[sindex]))
726f6388
JA
3020 sindex++;
3021
28ef6c31
JA
3022 /* If the first separator was IFS whitespace and the current character is
3023 a non-whitespace IFS character, it should be part of the current field
3024 delimiter, not a separate delimiter that would result in an empty field.
3025 Look at POSIX.2, 3.6.5, (3)(b). */
d233b485 3026 if (s[sindex] && whitesep && islocalsep (s[sindex]) && !ifs_whitesep (s[sindex]))
95732b49
JA
3027 {
3028 sindex++;
3029 /* An IFS character that is not IFS white space, along with any adjacent
3030 IFS white space, shall delimit a field. */
d233b485 3031 while (s[sindex] && ifs_whitesep (s[sindex]) && islocalsep(s[sindex]))
95732b49
JA
3032 sindex++;
3033 }
28ef6c31 3034
726f6388
JA
3035 /* Update STRING to point to the next field. */
3036 *stringp = s + sindex;
3037 return (current_word);
3038}
3039
3040/* Remove IFS white space at the end of STRING. Start at the end
3041 of the string and walk backwards until the beginning of the string
3042 or we find a character that's not IFS white space and not CTLESC.
3043 Only let CTLESC escape a white space character if SAW_ESCAPE is
3044 non-zero. */
3045char *
3046strip_trailing_ifs_whitespace (string, separators, saw_escape)
3047 char *string, *separators;
3048 int saw_escape;
3049{
3050 char *s;
ccc6cda3 3051
726f6388 3052 s = string + STRLEN (string) - 1;
7117c2d2 3053 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
726f6388
JA
3054 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
3055 s--;
3056 *++s = '\0';
3057 return string;
3058}
3059
bb70624e
JA
3060#if 0
3061/* UNUSED */
3062/* Split STRING into words at whitespace. Obeys shell-style quoting with
3063 backslashes, single and double quotes. */
ccc6cda3
JA
3064WORD_LIST *
3065list_string_with_quotes (string)
3066 char *string;
3067{
3068 WORD_LIST *list;
3069 char *token, *s;
7117c2d2 3070 size_t s_len;
ccc6cda3
JA
3071 int c, i, tokstart, len;
3072
3073 for (s = string; s && *s && spctabnl (*s); s++)
3074 ;
3075 if (s == 0 || *s == 0)
3076 return ((WORD_LIST *)NULL);
3077
7117c2d2 3078 s_len = strlen (s);
ccc6cda3
JA
3079 tokstart = i = 0;
3080 list = (WORD_LIST *)NULL;
3081 while (1)
3082 {
3083 c = s[i];
3084 if (c == '\\')
3085 {
3086 i++;
3087 if (s[i])
3088 i++;
3089 }
3090 else if (c == '\'')
a0c0a00f 3091 i = skip_single_quoted (s, s_len, ++i, 0);
ccc6cda3 3092 else if (c == '"')
a0c0a00f 3093 i = skip_double_quoted (s, s_len, ++i, 0);
ccc6cda3
JA
3094 else if (c == 0 || spctabnl (c))
3095 {
3096 /* We have found the end of a token. Make a word out of it and
3097 add it to the word list. */
bb70624e 3098 token = substring (s, tokstart, i);
cce855bc 3099 list = add_string_to_list (token, list);
ccc6cda3
JA
3100 free (token);
3101 while (spctabnl (s[i]))
3102 i++;
3103 if (s[i])
3104 tokstart = i;
3105 else
3106 break;
3107 }
3108 else
3109 i++; /* normal character */
3110 }
3111 return (REVERSE_LIST (list, WORD_LIST *));
3112}
bb70624e 3113#endif
d166f048 3114
cce855bc
JA
3115/********************************************************/
3116/* */
3117/* Functions to perform assignment statements */
3118/* */
3119/********************************************************/
d166f048 3120
95732b49
JA
3121#if defined (ARRAY_VARS)
3122static SHELL_VAR *
3123do_compound_assignment (name, value, flags)
3124 char *name, *value;
3125 int flags;
3126{
3127 SHELL_VAR *v;
d233b485 3128 int mklocal, mkassoc, mkglobal, chklocal;
0628567a 3129 WORD_LIST *list;
d233b485 3130 char *newname; /* used for local nameref references */
95732b49
JA
3131
3132 mklocal = flags & ASS_MKLOCAL;
3185942a 3133 mkassoc = flags & ASS_MKASSOC;
ac50fbac 3134 mkglobal = flags & ASS_MKGLOBAL;
d233b485 3135 chklocal = flags & ASS_CHKLOCAL;
95732b49
JA
3136
3137 if (mklocal && variable_context)
3138 {
d233b485
CR
3139 v = find_variable (name); /* follows namerefs */
3140 newname = (v == 0) ? nameref_transform_name (name, flags) : v->name;
a0c0a00f
CR
3141 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3142 {
d233b485 3143 if (readonly_p (v))
a0c0a00f
CR
3144 err_readonly (name);
3145 return (v); /* XXX */
3146 }
3185942a
JA
3147 list = expand_compound_array_assignment (v, value, flags);
3148 if (mkassoc)
d233b485 3149 v = make_local_assoc_variable (newname, 0);
3185942a 3150 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
d233b485 3151 v = make_local_array_variable (newname, 0);
ac50fbac
CR
3152 if (v)
3153 assign_compound_array_list (v, list, flags);
a0c0a00f
CR
3154 if (list)
3155 dispose_words (list);
ac50fbac 3156 }
d233b485
CR
3157 /* In a function but forcing assignment in global context. CHKLOCAL means to
3158 check for an existing local variable first. */
ac50fbac
CR
3159 else if (mkglobal && variable_context)
3160 {
d233b485
CR
3161 v = chklocal ? find_variable (name) : 0;
3162 if (v && (local_p (v) == 0 || v->context != variable_context))
3163 v = 0;
3164 if (v == 0)
3165 v = find_global_variable (name);
a0c0a00f
CR
3166 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3167 {
d233b485 3168 if (readonly_p (v))
a0c0a00f
CR
3169 err_readonly (name);
3170 return (v); /* XXX */
3171 }
d233b485
CR
3172 /* sanity check */
3173 newname = (v == 0) ? nameref_transform_name (name, flags) : name;
ac50fbac
CR
3174 list = expand_compound_array_assignment (v, value, flags);
3175 if (v == 0 && mkassoc)
d233b485 3176 v = make_new_assoc_variable (newname);
ac50fbac
CR
3177 else if (v && mkassoc && assoc_p (v) == 0)
3178 v = convert_var_to_assoc (v);
3179 else if (v == 0)
d233b485 3180 v = make_new_array_variable (newname);
ac50fbac
CR
3181 else if (v && mkassoc == 0 && array_p (v) == 0)
3182 v = convert_var_to_array (v);
3183 if (v)
3184 assign_compound_array_list (v, list, flags);
a0c0a00f
CR
3185 if (list)
3186 dispose_words (list);
95732b49
JA
3187 }
3188 else
a0c0a00f
CR
3189 {
3190 v = assign_array_from_string (name, value, flags);
3191 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3192 {
d233b485 3193 if (readonly_p (v))
a0c0a00f
CR
3194 err_readonly (name);
3195 return (v); /* XXX */
3196 }
3197 }
95732b49
JA
3198
3199 return (v);
3200}
3201#endif
3202
726f6388
JA
3203/* Given STRING, an assignment string, get the value of the right side
3204 of the `=', and bind it to the left side. If EXPAND is true, then
3205 perform parameter expansion, command substitution, and arithmetic
3206 expansion on the right-hand side. Perform tilde expansion in any
3207 case. Do not perform word splitting on the result of expansion. */
3208static int
95732b49
JA
3209do_assignment_internal (word, expand)
3210 const WORD_DESC *word;
726f6388
JA
3211 int expand;
3212{
495aee44
CR
3213 int offset, appendop, assign_list, aflags, retval;
3214 char *name, *value, *temp;
ccc6cda3
JA
3215 SHELL_VAR *entry;
3216#if defined (ARRAY_VARS)
3217 char *t;
b80f6443 3218 int ni;
ccc6cda3 3219#endif
95732b49 3220 const char *string;
ccc6cda3 3221
95732b49
JA
3222 if (word == 0 || word->word == 0)
3223 return 0;
3224
3225 appendop = assign_list = aflags = 0;
3226 string = word->word;
b80f6443 3227 offset = assignment (string, 0);
ccc6cda3
JA
3228 name = savestring (string);
3229 value = (char *)NULL;
726f6388
JA
3230
3231 if (name[offset] == '=')
3232 {
95732b49
JA
3233 if (name[offset - 1] == '+')
3234 {
3235 appendop = 1;
3236 name[offset - 1] = '\0';
3237 }
3238
3239 name[offset] = 0; /* might need this set later */
726f6388
JA
3240 temp = name + offset + 1;
3241
ccc6cda3 3242#if defined (ARRAY_VARS)
95732b49 3243 if (expand && (word->flags & W_COMPASSIGN))
726f6388 3244 {
ccc6cda3 3245 assign_list = ni = 1;
95732b49 3246 value = extract_array_assignment_list (temp, &ni);
ccc6cda3
JA
3247 }
3248 else
3249#endif
ccc6cda3 3250 if (expand && temp[0])
95732b49 3251 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
726f6388
JA
3252 else
3253 value = savestring (temp);
3254 }
3255
3256 if (value == 0)
d166f048 3257 {
f73dda09 3258 value = (char *)xmalloc (1);
d166f048
JA
3259 value[0] = '\0';
3260 }
726f6388 3261
726f6388 3262 if (echo_command_at_execute)
95732b49
JA
3263 {
3264 if (appendop)
3265 name[offset - 1] = '+';
3266 xtrace_print_assignment (name, value, assign_list, 1);
3267 if (appendop)
3268 name[offset - 1] = '\0';
3269 }
726f6388 3270
d166f048 3271#define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
ccc6cda3 3272
95732b49
JA
3273 if (appendop)
3274 aflags |= ASS_APPEND;
3275
ccc6cda3 3276#if defined (ARRAY_VARS)
d233b485 3277 if (t = mbschr (name, LBRACK))
ccc6cda3
JA
3278 {
3279 if (assign_list)
3280 {
b80f6443 3281 report_error (_("%s: cannot assign list to array member"), name);
ccc6cda3
JA
3282 ASSIGN_RETURN (0);
3283 }
95732b49 3284 entry = assign_array_element (name, value, aflags);
ccc6cda3 3285 if (entry == 0)
28ef6c31 3286 ASSIGN_RETURN (0);
ccc6cda3
JA
3287 }
3288 else if (assign_list)
95732b49 3289 {
d233b485
CR
3290 if ((word->flags & W_ASSIGNARG) && (word->flags & W_CHKLOCAL))
3291 aflags |= ASS_CHKLOCAL;
6d41b715 3292 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
95732b49 3293 aflags |= ASS_MKLOCAL;
ac50fbac
CR
3294 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
3295 aflags |= ASS_MKGLOBAL;
3185942a
JA
3296 if (word->flags & W_ASSIGNASSOC)
3297 aflags |= ASS_MKASSOC;
95732b49
JA
3298 entry = do_compound_assignment (name, value, aflags);
3299 }
ccc6cda3
JA
3300 else
3301#endif /* ARRAY_VARS */
95732b49 3302 entry = bind_variable (name, value, aflags);
ccc6cda3 3303
3eb0018e
CR
3304 if (entry)
3305 stupidly_hack_special_variables (entry->name); /* might be a nameref */
3306 else
3307 stupidly_hack_special_variables (name);
726f6388 3308
3185942a
JA
3309 /* Return 1 if the assignment seems to have been performed correctly. */
3310 if (entry == 0 || readonly_p (entry))
3311 retval = 0; /* assignment failure */
3312 else if (noassign_p (entry))
3313 {
712f80b0 3314 set_exit_status (EXECUTION_FAILURE);
3185942a
JA
3315 retval = 1; /* error status, but not assignment failure */
3316 }
3317 else
3318 retval = 1;
3319
3320 if (entry && retval != 0 && noassign_p (entry) == 0)
3321 VUNSETATTR (entry, att_invisible);
3322
3323 ASSIGN_RETURN (retval);
726f6388
JA
3324}
3325
3326/* Perform the assignment statement in STRING, and expand the
95732b49 3327 right side by doing tilde, command and parameter expansion. */
ccc6cda3 3328int
726f6388 3329do_assignment (string)
95732b49 3330 char *string;
726f6388 3331{
95732b49
JA
3332 WORD_DESC td;
3333
3334 td.flags = W_ASSIGNMENT;
3335 td.word = string;
3336
3337 return do_assignment_internal (&td, 1);
3338}
3339
3340int
495aee44 3341do_word_assignment (word, flags)
95732b49 3342 WORD_DESC *word;
495aee44 3343 int flags;
95732b49
JA
3344{
3345 return do_assignment_internal (word, 1);
726f6388
JA
3346}
3347
3348/* Given STRING, an assignment string, get the value of the right side
95732b49
JA
3349 of the `=', and bind it to the left side. Do not perform any word
3350 expansions on the right hand side. */
ccc6cda3 3351int
726f6388 3352do_assignment_no_expand (string)
95732b49 3353 char *string;
726f6388 3354{
95732b49
JA
3355 WORD_DESC td;
3356
3357 td.flags = W_ASSIGNMENT;
3358 td.word = string;
3359
3360 return (do_assignment_internal (&td, 0));
726f6388
JA
3361}
3362
cce855bc
JA
3363/***************************************************
3364 * *
3365 * Functions to manage the positional parameters *
3366 * *
3367 ***************************************************/
726f6388
JA
3368
3369/* Return the word list that corresponds to `$*'. */
3370WORD_LIST *
3371list_rest_of_args ()
3372{
ccc6cda3 3373 register WORD_LIST *list, *args;
726f6388
JA
3374 int i;
3375
3376 /* Break out of the loop as soon as one of the dollar variables is null. */
ccc6cda3
JA
3377 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
3378 list = make_word_list (make_bare_word (dollar_vars[i]), list);
3379
3380 for (args = rest_of_args; args; args = args->next)
3381 list = make_word_list (make_bare_word (args->word->word), list);
726f6388 3382
726f6388
JA
3383 return (REVERSE_LIST (list, WORD_LIST *));
3384}
3385
cce855bc
JA
3386/* Return the value of a positional parameter. This handles values > 10. */
3387char *
3388get_dollar_var_value (ind)
7117c2d2 3389 intmax_t ind;
cce855bc
JA
3390{
3391 char *temp;
3392 WORD_LIST *p;
3393
3394 if (ind < 10)
3395 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
3396 else /* We want something like ${11} */
3397 {
3398 ind -= 10;
3399 for (p = rest_of_args; p && ind--; p = p->next)
28ef6c31 3400 ;
cce855bc
JA
3401 temp = p ? savestring (p->word->word) : (char *)NULL;
3402 }
3403 return (temp);
3404}
3405
726f6388
JA
3406/* Make a single large string out of the dollar digit variables,
3407 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
3408 case of "$*" with respect to IFS. */
3409char *
3410string_rest_of_args (dollar_star)
3411 int dollar_star;
3412{
ccc6cda3 3413 register WORD_LIST *list;
726f6388
JA
3414 char *string;
3415
ccc6cda3 3416 list = list_rest_of_args ();
d233b485 3417 string = dollar_star ? string_list_dollar_star (list, 0, 0) : string_list (list);
726f6388
JA
3418 dispose_words (list);
3419 return (string);
3420}
3421
cce855bc
JA
3422/* Return a string containing the positional parameters from START to
3423 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
7117c2d2
JA
3424 which only makes a difference if QUOTED is non-zero. If QUOTED includes
3425 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3426 no quoting chars are added. */
cce855bc 3427static char *
712f80b0 3428pos_params (string, start, end, quoted, pflags)
cce855bc 3429 char *string;
712f80b0 3430 int start, end, quoted, pflags;
726f6388 3431{
cce855bc
JA
3432 WORD_LIST *save, *params, *h, *t;
3433 char *ret;
3434 int i;
726f6388 3435
bb70624e
JA
3436 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3437 if (start == end)
3438 return ((char *)NULL);
3439
cce855bc 3440 save = params = list_rest_of_args ();
a0c0a00f 3441 if (save == 0 && start > 0)
cce855bc
JA
3442 return ((char *)NULL);
3443
3185942a
JA
3444 if (start == 0) /* handle ${@:0[:x]} specially */
3445 {
3446 t = make_word_list (make_word (dollar_vars[0]), params);
3447 save = params = t;
3448 }
3449
0001803f 3450 for (i = start ? 1 : 0; params && i < start; i++)
cce855bc
JA
3451 params = params->next;
3452 if (params == 0)
a0c0a00f
CR
3453 {
3454 dispose_words (save);
3455 return ((char *)NULL);
3456 }
cce855bc 3457 for (h = t = params; params && i < end; i++)
d166f048 3458 {
cce855bc
JA
3459 t = params;
3460 params = params->next;
d166f048 3461 }
cce855bc 3462 t->next = (WORD_LIST *)NULL;
3185942a 3463
712f80b0 3464 ret = string_list_pos_params (string[0], h, quoted, pflags);
3185942a 3465
bb70624e
JA
3466 if (t != params)
3467 t->next = params;
726f6388 3468
cce855bc
JA
3469 dispose_words (save);
3470 return (ret);
3471}
3472
3473/******************************************************************/
3474/* */
3475/* Functions to expand strings to strings or WORD_LISTs */
3476/* */
3477/******************************************************************/
3478
3479#if defined (PROCESS_SUBSTITUTION)
95732b49 3480#define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
cce855bc 3481#else
95732b49 3482#define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
cce855bc
JA
3483#endif
3484
3485/* If there are any characters in STRING that require full expansion,
3486 then call FUNC to expand STRING; otherwise just perform quote
3487 removal if necessary. This returns a new string. */
3488static char *
f73dda09 3489expand_string_if_necessary (string, quoted, func)
cce855bc
JA
3490 char *string;
3491 int quoted;
f73dda09 3492 EXPFUNC *func;
cce855bc
JA
3493{
3494 WORD_LIST *list;
7117c2d2 3495 size_t slen;
cce855bc
JA
3496 int i, saw_quote;
3497 char *ret;
7117c2d2 3498 DECLARE_MBSTATE;
cce855bc 3499
95732b49
JA
3500 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3501 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
7117c2d2
JA
3502 i = saw_quote = 0;
3503 while (string[i])
cce855bc
JA
3504 {
3505 if (EXP_CHAR (string[i]))
3506 break;
3507 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3508 saw_quote = 1;
7117c2d2 3509 ADVANCE_CHAR (string, slen, i);
cce855bc
JA
3510 }
3511
3512 if (string[i])
3513 {
3514 list = (*func) (string, quoted);
3515 if (list)
3516 {
3517 ret = string_list (list);
3518 dispose_words (list);
3519 }
3520 else
3521 ret = (char *)NULL;
3522 }
3523 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3524 ret = string_quote_removal (string, quoted);
3525 else
3526 ret = savestring (string);
7117c2d2 3527
cce855bc
JA
3528 return ret;
3529}
3530
3531static inline char *
f73dda09 3532expand_string_to_string_internal (string, quoted, func)
cce855bc
JA
3533 char *string;
3534 int quoted;
f73dda09 3535 EXPFUNC *func;
cce855bc
JA
3536{
3537 WORD_LIST *list;
3538 char *ret;
3539
3540 if (string == 0 || *string == '\0')
3541 return ((char *)NULL);
3542
3543 list = (*func) (string, quoted);
3544 if (list)
3545 {
3546 ret = string_list (list);
3547 dispose_words (list);
3548 }
3549 else
3550 ret = (char *)NULL;
3551
3552 return (ret);
3553}
3554
f73dda09
JA
3555char *
3556expand_string_to_string (string, quoted)
3557 char *string;
3558 int quoted;
3559{
3560 return (expand_string_to_string_internal (string, quoted, expand_string));
3561}
3562
3563char *
3564expand_string_unsplit_to_string (string, quoted)
3565 char *string;
3566 int quoted;
3567{
3568 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3569}
3570
95732b49
JA
3571char *
3572expand_assignment_string_to_string (string, quoted)
3573 char *string;
3574 int quoted;
3575{
3576 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3577}
3578
0628567a
JA
3579char *
3580expand_arith_string (string, quoted)
3581 char *string;
3185942a 3582 int quoted;
0628567a 3583{
ac50fbac
CR
3584 WORD_DESC td;
3585 WORD_LIST *list, *tlist;
3586 size_t slen;
3587 int i, saw_quote;
3588 char *ret;
3589 DECLARE_MBSTATE;
3590
3591 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3592 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3593 i = saw_quote = 0;
3594 while (string[i])
3595 {
3596 if (EXP_CHAR (string[i]))
3597 break;
3598 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3599 saw_quote = 1;
3600 ADVANCE_CHAR (string, slen, i);
3601 }
3602
3603 if (string[i])
3604 {
3605 /* This is expanded version of expand_string_internal as it's called by
3606 expand_string_leave_quoted */
d233b485 3607 td.flags = W_NOPROCSUB|W_NOTILDE; /* don't want process substitution or tilde expansion */
712f80b0
CR
3608#if 0 /* TAG:bash-5.1 */
3609 if (quoted & Q_ARRAYSUB)
3610 td.flags |= W_NOCOMSUB;
3611#endif
ac50fbac
CR
3612 td.word = savestring (string);
3613 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3614 /* This takes care of the calls from expand_string_leave_quoted and
3615 expand_string */
3616 if (list)
3617 {
3618 tlist = word_list_split (list);
3619 dispose_words (list);
3620 list = tlist;
3621 if (list)
3622 dequote_list (list);
3623 }
3624 /* This comes from expand_string_if_necessary */
3625 if (list)
3626 {
3627 ret = string_list (list);
3628 dispose_words (list);
3629 }
3630 else
3631 ret = (char *)NULL;
3632 FREE (td.word);
3633 }
a0c0a00f
CR
3634 else if (saw_quote && (quoted & Q_ARITH))
3635 ret = string_quote_removal (string, quoted);
ac50fbac
CR
3636 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3637 ret = string_quote_removal (string, quoted);
3638 else
3639 ret = savestring (string);
3640
3641 return ret;
0628567a
JA
3642}
3643
cce855bc
JA
3644#if defined (COND_COMMAND)
3645/* Just remove backslashes in STRING. Returns a new string. */
3646char *
3647remove_backslashes (string)
3648 char *string;
3649{
3650 char *r, *ret, *s;
3651
f73dda09 3652 r = ret = (char *)xmalloc (strlen (string) + 1);
cce855bc
JA
3653 for (s = string; s && *s; )
3654 {
3655 if (*s == '\\')
28ef6c31 3656 s++;
cce855bc 3657 if (*s == 0)
28ef6c31 3658 break;
cce855bc
JA
3659 *r++ = *s++;
3660 }
3661 *r = '\0';
3662 return ret;
3663}
3664
3665/* This needs better error handling. */
3666/* Expand W for use as an argument to a unary or binary operator in a
f1be666c 3667 [[...]] expression. If SPECIAL is 1, this is the rhs argument
cce855bc 3668 to the != or == operator, and should be treated as a pattern. In
f1be666c
JA
3669 this case, we quote the string specially for the globbing code. If
3670 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3671 be quoted appropriately for regcomp/regexec. The caller is responsible
d894cfd1
CR
3672 for removing the backslashes if the unquoted word is needed later. In
3673 any case, since we don't perform word splitting, we need to do quoted
3674 null character removal. */
cce855bc
JA
3675char *
3676cond_expand_word (w, special)
3677 WORD_DESC *w;
3678 int special;
3679{
3680 char *r, *p;
3681 WORD_LIST *l;
f1be666c 3682 int qflags;
cce855bc
JA
3683
3684 if (w->word == 0 || w->word[0] == '\0')
3685 return ((char *)NULL);
3686
3b34f6e6 3687 expand_no_split_dollar_star = 1;
0001803f 3688 w->flags |= W_NOSPLIT2;
b72432fd 3689 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3b34f6e6 3690 expand_no_split_dollar_star = 0;
cce855bc
JA
3691 if (l)
3692 {
ac50fbac 3693 if (special == 0) /* LHS */
cce855bc 3694 {
d894cfd1
CR
3695 if (l->word)
3696 word_list_remove_quoted_nulls (l);
cce855bc
JA
3697 dequote_list (l);
3698 r = string_list (l);
3699 }
3700 else
28ef6c31 3701 {
ac50fbac
CR
3702 /* Need to figure out whether or not we should call dequote_escapes
3703 or a new dequote_ctlnul function here, and under what
3704 circumstances. */
d233b485 3705 qflags = QGLOB_CVTNULL|QGLOB_CTLESC;
f1be666c
JA
3706 if (special == 2)
3707 qflags |= QGLOB_REGEXP;
a0c0a00f 3708 word_list_remove_quoted_nulls (l);
28ef6c31 3709 p = string_list (l);
f1be666c 3710 r = quote_string_for_globbing (p, qflags);
28ef6c31
JA
3711 free (p);
3712 }
cce855bc
JA
3713 dispose_words (l);
3714 }
3715 else
3716 r = (char *)NULL;
3717
3718 return r;
3719}
3720#endif
3721
3722/* Call expand_word_internal to expand W and handle error returns.
3723 A convenience function for functions that don't want to handle
3724 any errors or free any memory before aborting. */
3725static WORD_LIST *
b72432fd 3726call_expand_word_internal (w, q, i, c, e)
cce855bc 3727 WORD_DESC *w;
b72432fd 3728 int q, i, *c, *e;
cce855bc
JA
3729{
3730 WORD_LIST *result;
3731
b72432fd 3732 result = expand_word_internal (w, q, i, c, e);
bb70624e 3733 if (result == &expand_word_error || result == &expand_word_fatal)
cce855bc
JA
3734 {
3735 /* By convention, each time this error is returned, w->word has
bb70624e
JA
3736 already been freed (it sometimes may not be in the fatal case,
3737 but that doesn't result in a memory leak because we're going
3738 to exit in most cases). */
cce855bc 3739 w->word = (char *)NULL;
28ef6c31 3740 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 3741 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
cce855bc 3742 /* NOTREACHED */
ac50fbac 3743 return (NULL);
cce855bc 3744 }
cce855bc
JA
3745 else
3746 return (result);
3747}
3748
3749/* Perform parameter expansion, command substitution, and arithmetic
ac50fbac
CR
3750 expansion on STRING, as if it were a word. Leave the result quoted.
3751 Since this does not perform word splitting, it leaves quoted nulls
3752 in the result. */
cce855bc
JA
3753static WORD_LIST *
3754expand_string_internal (string, quoted)
3755 char *string;
3756 int quoted;
3757{
3758 WORD_DESC td;
3759 WORD_LIST *tresult;
3760
3761 if (string == 0 || *string == 0)
3762 return ((WORD_LIST *)NULL);
3763
28ef6c31
JA
3764 td.flags = 0;
3765 td.word = savestring (string);
3766
b72432fd 3767 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31
JA
3768
3769 FREE (td.word);
cce855bc 3770 return (tresult);
726f6388
JA
3771}
3772
3773/* Expand STRING by performing parameter expansion, command substitution,
3774 and arithmetic expansion. Dequote the resulting WORD_LIST before
3775 returning it, but do not perform word splitting. The call to
3776 remove_quoted_nulls () is in here because word splitting normally
3777 takes care of quote removal. */
3778WORD_LIST *
3779expand_string_unsplit (string, quoted)
3780 char *string;
3781 int quoted;
3782{
3783 WORD_LIST *value;
3784
28ef6c31 3785 if (string == 0 || *string == '\0')
726f6388
JA
3786 return ((WORD_LIST *)NULL);
3787
28ef6c31 3788 expand_no_split_dollar_star = 1;
726f6388 3789 value = expand_string_internal (string, quoted);
28ef6c31
JA
3790 expand_no_split_dollar_star = 0;
3791
726f6388
JA
3792 if (value)
3793 {
3794 if (value->word)
95732b49 3795 {
712f80b0 3796 remove_quoted_nulls (value->word->word); /* XXX */
95732b49
JA
3797 value->word->flags &= ~W_HASQUOTEDNULL;
3798 }
3799 dequote_list (value);
3800 }
3801 return (value);
3802}
3803
3804/* Expand the rhs of an assignment statement */
3805WORD_LIST *
3806expand_string_assignment (string, quoted)
3807 char *string;
3808 int quoted;
3809{
3810 WORD_DESC td;
3811 WORD_LIST *value;
3812
3813 if (string == 0 || *string == '\0')
3814 return ((WORD_LIST *)NULL);
3815
3816 expand_no_split_dollar_star = 1;
3817
d233b485
CR
3818#if 0
3819 /* Other shells (ksh93) do it this way, which affects how $@ is expanded
3820 in constructs like bar=${@#0} (preserves the spaces resulting from the
3821 expansion of $@ in a context where you don't do word splitting); Posix
3822 interp 888 makes the expansion of $@ in contexts where word splitting
3823 is not performed unspecified. */
3824 td.flags = W_ASSIGNRHS|W_NOSPLIT2; /* Posix interp 888 */
3825#else
95732b49 3826 td.flags = W_ASSIGNRHS;
d233b485 3827#endif
95732b49
JA
3828 td.word = savestring (string);
3829 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3830 FREE (td.word);
3831
3832 expand_no_split_dollar_star = 0;
3833
3834 if (value)
3835 {
3836 if (value->word)
3837 {
712f80b0 3838 remove_quoted_nulls (value->word->word); /* XXX */
95732b49
JA
3839 value->word->flags &= ~W_HASQUOTEDNULL;
3840 }
726f6388
JA
3841 dequote_list (value);
3842 }
3843 return (value);
3844}
3845
bb70624e
JA
3846
3847/* Expand one of the PS? prompt strings. This is a sort of combination of
3848 expand_string_unsplit and expand_string_internal, but returns the
3849 passed string when an error occurs. Might want to trap other calls
3850 to jump_to_top_level here so we don't endlessly loop. */
3851WORD_LIST *
f1be666c 3852expand_prompt_string (string, quoted, wflags)
bb70624e
JA
3853 char *string;
3854 int quoted;
f1be666c 3855 int wflags;
bb70624e
JA
3856{
3857 WORD_LIST *value;
3858 WORD_DESC td;
3859
3860 if (string == 0 || *string == 0)
3861 return ((WORD_LIST *)NULL);
3862
f1be666c 3863 td.flags = wflags;
bb70624e 3864 td.word = savestring (string);
28ef6c31
JA
3865
3866 no_longjmp_on_fatal_error = 1;
bb70624e 3867 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
28ef6c31
JA
3868 no_longjmp_on_fatal_error = 0;
3869
bb70624e
JA
3870 if (value == &expand_word_error || value == &expand_word_fatal)
3871 {
3872 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3873 return value;
3874 }
3875 FREE (td.word);
3876 if (value)
3877 {
3878 if (value->word)
95732b49 3879 {
712f80b0 3880 remove_quoted_nulls (value->word->word); /* XXX */
95732b49
JA
3881 value->word->flags &= ~W_HASQUOTEDNULL;
3882 }
bb70624e
JA
3883 dequote_list (value);
3884 }
3885 return (value);
3886}
3887
726f6388
JA
3888/* Expand STRING just as if you were expanding a word, but do not dequote
3889 the resultant WORD_LIST. This is called only from within this file,
3890 and is used to correctly preserve quoted characters when expanding
3891 things like ${1+"$@"}. This does parameter expansion, command
b72432fd 3892 substitution, arithmetic expansion, and word splitting. */
726f6388
JA
3893static WORD_LIST *
3894expand_string_leave_quoted (string, quoted)
3895 char *string;
3896 int quoted;
3897{
3898 WORD_LIST *tlist;
3899 WORD_LIST *tresult;
3900
ccc6cda3 3901 if (string == 0 || *string == '\0')
726f6388
JA
3902 return ((WORD_LIST *)NULL);
3903
3904 tlist = expand_string_internal (string, quoted);
3905
3906 if (tlist)
3907 {
3908 tresult = word_list_split (tlist);
3909 dispose_words (tlist);
3910 return (tresult);
3911 }
3912 return ((WORD_LIST *)NULL);
3913}
3914
ccc6cda3
JA
3915/* This does not perform word splitting or dequote the WORD_LIST
3916 it returns. */
3917static WORD_LIST *
d233b485
CR
3918expand_string_for_rhs (string, quoted, op, pflags, dollar_at_p, expanded_p)
3919 char *string;
3920 int quoted, op, pflags;
3921 int *dollar_at_p, *expanded_p;
3922{
3923 WORD_DESC td;
3924 WORD_LIST *tresult;
3925 int old_nosplit;
3926
3927 if (string == 0 || *string == '\0')
3928 return (WORD_LIST *)NULL;
3929
3930 /* We want field splitting to be determined by what is going to be done with
3931 the entire ${parameterOPword} expansion, so we don't want to split the RHS
3932 we expand here. However, the expansion of $* is determined by whether we
3933 are going to eventually perform word splitting, so we want to set this
3934 depending on whether or not are are going to be splitting: if the expansion
3935 is quoted, if the OP is `=', or if IFS is set to the empty string, we
3936 are not going to be splitting, so we set expand_no_split_dollar_star to
712f80b0 3937 note this to callees.
d233b485
CR
3938 We pass through PF_ASSIGNRHS as W_ASSIGNRHS if this is on the RHS of an
3939 assignment statement. */
3940 /* The updated treatment of $* is the result of Posix interp 888 */
3941 /* This was further clarified on the austin-group list in March, 2017 and
3942 in Posix bug 1129 */
3943 old_nosplit = expand_no_split_dollar_star;
3944 expand_no_split_dollar_star = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || op == '=' || ifs_is_null == 0; /* XXX - was 1 */
712f80b0
CR
3945 td.flags = W_EXPANDRHS; /* expanding RHS of ${paramOPword} */
3946 td.flags |= W_NOSPLIT2; /* no splitting, remove "" and '' */
d233b485
CR
3947 if (pflags & PF_ASSIGNRHS) /* pass through */
3948 td.flags |= W_ASSIGNRHS;
3949 if (op == '=')
3950#if 0
3951 td.flags |= W_ASSIGNRHS; /* expand b in ${a=b} like assignment */
3952#else
3953 td.flags |= W_ASSIGNRHS|W_NOASSNTILDE; /* expand b in ${a=b} like assignment */
3954#endif
3955 td.word = string;
3956 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, expanded_p);
3957 expand_no_split_dollar_star = old_nosplit;
3958
3959 return (tresult);
3960}
3961
3962/* This does not perform word splitting or dequote the WORD_LIST
3963 it returns and it treats $* as if it were quoted. */
3964static WORD_LIST *
3965expand_string_for_pat (string, quoted, dollar_at_p, expanded_p)
ccc6cda3 3966 char *string;
a0c0a00f 3967 int quoted, *dollar_at_p, *expanded_p;
ccc6cda3
JA
3968{
3969 WORD_DESC td;
3970 WORD_LIST *tresult;
d233b485 3971 int oexp;
ccc6cda3
JA
3972
3973 if (string == 0 || *string == '\0')
3974 return (WORD_LIST *)NULL;
3975
d233b485 3976 oexp = expand_no_split_dollar_star;
a0c0a00f 3977 expand_no_split_dollar_star = 1;
aeb26a67 3978 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
ccc6cda3 3979 td.word = string;
a0c0a00f 3980 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, expanded_p);
d233b485 3981 expand_no_split_dollar_star = oexp;
a0c0a00f 3982
ccc6cda3
JA
3983 return (tresult);
3984}
3985
726f6388
JA
3986/* Expand STRING just as if you were expanding a word. This also returns
3987 a list of words. Note that filename globbing is *NOT* done for word
3988 or string expansion, just when the shell is expanding a command. This
3989 does parameter expansion, command substitution, arithmetic expansion,
3990 and word splitting. Dequote the resultant WORD_LIST before returning. */
3991WORD_LIST *
3992expand_string (string, quoted)
3993 char *string;
3994 int quoted;
3995{
3996 WORD_LIST *result;
3997
28ef6c31 3998 if (string == 0 || *string == '\0')
726f6388
JA
3999 return ((WORD_LIST *)NULL);
4000
4001 result = expand_string_leave_quoted (string, quoted);
ccc6cda3 4002 return (result ? dequote_list (result) : result);
726f6388
JA
4003}
4004
d233b485
CR
4005/*******************************************
4006 * *
4007 * Functions to expand WORD_DESCs *
4008 * *
4009 *******************************************/
4010
4011/* Expand WORD, performing word splitting on the result. This does
4012 parameter expansion, command substitution, arithmetic expansion,
4013 word splitting, and quote removal. */
4014
4015WORD_LIST *
4016expand_word (word, quoted)
4017 WORD_DESC *word;
4018 int quoted;
4019{
4020 WORD_LIST *result, *tresult;
4021
4022 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4023 result = word_list_split (tresult);
4024 dispose_words (tresult);
4025 return (result ? dequote_list (result) : result);
4026}
4027
4028/* Expand WORD, but do not perform word splitting on the result. This
4029 does parameter expansion, command substitution, arithmetic expansion,
4030 and quote removal. */
4031WORD_LIST *
4032expand_word_unsplit (word, quoted)
4033 WORD_DESC *word;
4034 int quoted;
4035{
4036 WORD_LIST *result;
4037
4038 result = expand_word_leave_quoted (word, quoted);
4039 return (result ? dequote_list (result) : result);
4040}
4041
4042/* Perform shell expansions on WORD, but do not perform word splitting or
4043 quote removal on the result. Virtually identical to expand_word_unsplit;
4044 could be combined if implementations don't diverge. */
4045WORD_LIST *
4046expand_word_leave_quoted (word, quoted)
4047 WORD_DESC *word;
4048 int quoted;
4049{
4050 WORD_LIST *result;
4051
4052 expand_no_split_dollar_star = 1;
4053 if (ifs_is_null)
4054 word->flags |= W_NOSPLIT;
4055 word->flags |= W_NOSPLIT2;
4056 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4057 expand_no_split_dollar_star = 0;
4058
4059 return result;
4060}
4061
726f6388
JA
4062/***************************************************
4063 * *
4064 * Functions to handle quoting chars *
4065 * *
4066 ***************************************************/
4067
cce855bc
JA
4068/* Conventions:
4069
4070 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
4071 The parser passes CTLNUL as CTLESC CTLNUL. */
4072
cce855bc
JA
4073/* Quote escape characters in string s, but no other characters. This is
4074 used to protect CTLESC and CTLNUL in variable values from the rest of
3185942a
JA
4075 the word expansion process after the variable is expanded (word splitting
4076 and filename generation). If IFS is null, we quote spaces as well, just
4077 in case we split on spaces later (in the case of unquoted $@, we will
4078 eventually attempt to split the entire word on spaces). Corresponding
4079 code exists in dequote_escapes. Even if we don't end up splitting on
4080 spaces, quoting spaces is not a problem. This should never be called on
4081 a string that is quoted with single or double quotes or part of a here
d233b485
CR
4082 document (effectively double-quoted).
4083 FLAGS says whether or not we are going to split the result. If we are not,
4084 and there is a CTLESC or CTLNUL in IFS, we need to quote CTLESC and CTLNUL,
4085 respectively, to prevent them from being removed as part of dequoting. */
4086static char *
4087quote_escapes_internal (string, flags)
4088 const char *string;
4089 int flags;
cce855bc 4090{
d233b485
CR
4091 const char *s, *send;
4092 char *t, *result;
7117c2d2 4093 size_t slen;
d233b485 4094 int quote_spaces, skip_ctlesc, skip_ctlnul, nosplit;
7117c2d2 4095 DECLARE_MBSTATE;
cce855bc 4096
7117c2d2
JA
4097 slen = strlen (string);
4098 send = string + slen;
4099
f1be666c 4100 quote_spaces = (ifs_value && *ifs_value == 0);
d233b485 4101 nosplit = (flags & PF_NOSPLIT2);
3185942a
JA
4102
4103 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
d233b485
CR
4104 {
4105 skip_ctlesc |= (nosplit == 0 && *s == CTLESC);
4106 skip_ctlnul |= (nosplit == 0 && *s == CTLNUL);
4107 }
3185942a 4108
7117c2d2
JA
4109 t = result = (char *)xmalloc ((slen * 2) + 1);
4110 s = string;
4111
4112 while (*s)
cce855bc 4113 {
3185942a 4114 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
cce855bc 4115 *t++ = CTLESC;
7117c2d2 4116 COPY_CHAR_P (t, s, send);
cce855bc
JA
4117 }
4118 *t = '\0';
ac50fbac 4119
cce855bc
JA
4120 return (result);
4121}
4122
d233b485
CR
4123char *
4124quote_escapes (string)
4125 const char *string;
4126{
4127 return (quote_escapes_internal (string, 0));
4128}
4129
4130char *
4131quote_rhs (string)
4132 const char *string;
4133{
4134 return (quote_escapes_internal (string, PF_NOSPLIT2));
4135}
4136
cce855bc
JA
4137static WORD_LIST *
4138list_quote_escapes (list)
4139 WORD_LIST *list;
4140{
4141 register WORD_LIST *w;
4142 char *t;
4143
4144 for (w = list; w; w = w->next)
4145 {
4146 t = w->word->word;
4147 w->word->word = quote_escapes (t);
4148 free (t);
4149 }
4150 return list;
4151}
4152
7117c2d2
JA
4153/* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
4154
4155 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
4156 This is necessary to make unquoted CTLESC and CTLNUL characters in the
4157 data stream pass through properly.
4158
4159 We need to remove doubled CTLESC characters inside quoted strings before
4160 quoting the entire string, so we do not double the number of CTLESC
4161 characters.
4162
4163 Also used by parts of the pattern substitution code. */
3185942a 4164char *
cce855bc 4165dequote_escapes (string)
d233b485 4166 const char *string;
cce855bc 4167{
d233b485
CR
4168 const char *s, *send;
4169 char *t, *result;
7117c2d2 4170 size_t slen;
f1be666c 4171 int quote_spaces;
7117c2d2 4172 DECLARE_MBSTATE;
cce855bc 4173
7117c2d2 4174 if (string == 0)
d233b485 4175 return (char *)0;
7117c2d2
JA
4176
4177 slen = strlen (string);
4178 send = string + slen;
4179
4180 t = result = (char *)xmalloc (slen + 1);
7117c2d2
JA
4181
4182 if (strchr (string, CTLESC) == 0)
3185942a 4183 return (strcpy (result, string));
7117c2d2 4184
f1be666c 4185 quote_spaces = (ifs_value && *ifs_value == 0);
3185942a
JA
4186
4187 s = string;
7117c2d2 4188 while (*s)
cce855bc 4189 {
f1be666c 4190 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
cce855bc
JA
4191 {
4192 s++;
4193 if (*s == '\0')
4194 break;
4195 }
7117c2d2 4196 COPY_CHAR_P (t, s, send);
cce855bc
JA
4197 }
4198 *t = '\0';
ac50fbac 4199
cce855bc
JA
4200 return result;
4201}
726f6388 4202
d233b485 4203#if defined (INCLUDE_UNUSED)
ac50fbac
CR
4204static WORD_LIST *
4205list_dequote_escapes (list)
4206 WORD_LIST *list;
4207{
4208 register WORD_LIST *w;
4209 char *t;
4210
4211 for (w = list; w; w = w->next)
4212 {
4213 t = w->word->word;
4214 w->word->word = dequote_escapes (t);
4215 free (t);
4216 }
4217 return list;
4218}
d233b485 4219#endif
ac50fbac 4220
0628567a
JA
4221/* Return a new string with the quoted representation of character C.
4222 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
4223 set in any resultant WORD_DESC where this value is the word. */
726f6388
JA
4224static char *
4225make_quoted_char (c)
4226 int c;
4227{
4228 char *temp;
4229
f73dda09 4230 temp = (char *)xmalloc (3);
726f6388
JA
4231 if (c == 0)
4232 {
4233 temp[0] = CTLNUL;
4234 temp[1] = '\0';
4235 }
4236 else
4237 {
4238 temp[0] = CTLESC;
4239 temp[1] = c;
4240 temp[2] = '\0';
4241 }
4242 return (temp);
4243}
4244
0628567a
JA
4245/* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
4246 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
4247 this value is the word. */
ccc6cda3 4248char *
726f6388
JA
4249quote_string (string)
4250 char *string;
4251{
ccc6cda3 4252 register char *t;
7117c2d2
JA
4253 size_t slen;
4254 char *result, *send;
726f6388 4255
ccc6cda3 4256 if (*string == 0)
726f6388 4257 {
f73dda09 4258 result = (char *)xmalloc (2);
726f6388
JA
4259 result[0] = CTLNUL;
4260 result[1] = '\0';
4261 }
4262 else
4263 {
7117c2d2 4264 DECLARE_MBSTATE;
726f6388 4265
7117c2d2
JA
4266 slen = strlen (string);
4267 send = string + slen;
4268
4269 result = (char *)xmalloc ((slen * 2) + 1);
4270
4271 for (t = result; string < send; )
726f6388
JA
4272 {
4273 *t++ = CTLESC;
7117c2d2 4274 COPY_CHAR_P (t, string, send);
726f6388
JA
4275 }
4276 *t = '\0';
4277 }
4278 return (result);
4279}
4280
0628567a 4281/* De-quote quoted characters in STRING. */
726f6388
JA
4282char *
4283dequote_string (string)
4284 char *string;
4285{
7117c2d2
JA
4286 register char *s, *t;
4287 size_t slen;
4288 char *result, *send;
4289 DECLARE_MBSTATE;
726f6388 4290
a0c0a00f
CR
4291#if defined (DEBUG)
4292 if (string[0] == CTLESC && string[1] == 0)
4293 internal_inform ("dequote_string: string with bare CTLESC");
4294#endif
4295
d233b485 4296 slen = STRLEN (string);
7117c2d2
JA
4297
4298 t = result = (char *)xmalloc (slen + 1);
726f6388
JA
4299
4300 if (QUOTED_NULL (string))
4301 {
4302 result[0] = '\0';
4303 return (result);
4304 }
4305
a0c0a00f
CR
4306 /* A string consisting of only a single CTLESC should pass through unchanged */
4307 if (string[0] == CTLESC && string[1] == 0)
4308 {
4309 result[0] = CTLESC;
4310 result[1] = '\0';
4311 return (result);
4312 }
4313
726f6388
JA
4314 /* If no character in the string can be quoted, don't bother examining
4315 each character. Just return a copy of the string passed to us. */
7117c2d2
JA
4316 if (strchr (string, CTLESC) == NULL)
4317 return (strcpy (result, string));
726f6388 4318
7117c2d2
JA
4319 send = string + slen;
4320 s = string;
4321 while (*s)
726f6388 4322 {
7117c2d2 4323 if (*s == CTLESC)
726f6388 4324 {
7117c2d2
JA
4325 s++;
4326 if (*s == '\0')
726f6388
JA
4327 break;
4328 }
7117c2d2 4329 COPY_CHAR_P (t, s, send);
726f6388
JA
4330 }
4331
4332 *t = '\0';
4333 return (result);
4334}
4335
4336/* Quote the entire WORD_LIST list. */
ccc6cda3 4337static WORD_LIST *
726f6388
JA
4338quote_list (list)
4339 WORD_LIST *list;
4340{
4341 register WORD_LIST *w;
ccc6cda3 4342 char *t;
726f6388
JA
4343
4344 for (w = list; w; w = w->next)
4345 {
ccc6cda3 4346 t = w->word->word;
726f6388 4347 w->word->word = quote_string (t);
3185942a
JA
4348 if (*t == 0)
4349 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
ccc6cda3 4350 w->word->flags |= W_QUOTED;
3185942a 4351 free (t);
726f6388 4352 }
ccc6cda3 4353 return list;
726f6388
JA
4354}
4355
d233b485
CR
4356WORD_DESC *
4357dequote_word (word)
4358 WORD_DESC *word;
4359{
4360 register char *s;
4361
4362 s = dequote_string (word->word);
4363 if (QUOTED_NULL (word->word))
4364 word->flags &= ~W_HASQUOTEDNULL;
4365 free (word->word);
4366 word->word = s;
4367
4368 return word;
4369}
4370
0628567a
JA
4371/* De-quote quoted characters in each word in LIST. */
4372WORD_LIST *
7117c2d2
JA
4373dequote_list (list)
4374 WORD_LIST *list;
4375{
4376 register char *s;
4377 register WORD_LIST *tlist;
4378
4379 for (tlist = list; tlist; tlist = tlist->next)
4380 {
4381 s = dequote_string (tlist->word->word);
3185942a
JA
4382 if (QUOTED_NULL (tlist->word->word))
4383 tlist->word->flags &= ~W_HASQUOTEDNULL;
7117c2d2
JA
4384 free (tlist->word->word);
4385 tlist->word->word = s;
4386 }
4387 return list;
4388}
4389
4390/* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
4391 string. */
3185942a 4392char *
7117c2d2
JA
4393remove_quoted_escapes (string)
4394 char *string;
4395{
4396 char *t;
4397
4398 if (string)
4399 {
4400 t = dequote_escapes (string);
4401 strcpy (string, t);
4402 free (t);
4403 }
4404
4405 return (string);
4406}
4407
d233b485
CR
4408/* Remove quoted $IFS characters from STRING. Quoted IFS characters are
4409 added to protect them from word splitting, but we need to remove them
4410 if no word splitting takes place. This returns newly-allocated memory,
4411 so callers can use it to replace savestring(). */
4412char *
4413remove_quoted_ifs (string)
4414 char *string;
4415{
4416 register size_t slen;
4417 register int i, j;
4418 char *ret, *send;
4419 DECLARE_MBSTATE;
4420
4421 slen = strlen (string);
4422 send = string + slen;
4423
4424 i = j = 0;
4425 ret = (char *)xmalloc (slen + 1);
4426
4427 while (i < slen)
4428 {
4429 if (string[i] == CTLESC)
4430 {
4431 i++;
4432 if (string[i] == 0 || isifs (string[i]) == 0)
4433 ret[j++] = CTLESC;
4434 if (i == slen)
4435 break;
4436 }
4437
4438 COPY_CHAR_I (ret, j, string, send, i);
4439 }
4440 ret[j] = '\0';
4441
4442 return (ret);
4443}
4444
3185942a 4445char *
cce855bc
JA
4446remove_quoted_nulls (string)
4447 char *string;
4448{
7117c2d2
JA
4449 register size_t slen;
4450 register int i, j, prev_i;
4451 DECLARE_MBSTATE;
4452
4453 if (strchr (string, CTLNUL) == 0) /* XXX */
4454 return string; /* XXX */
4455
4456 slen = strlen (string);
4457 i = j = 0;
4458
4459 while (i < slen)
4460 {
4461 if (string[i] == CTLESC)
b80f6443
JA
4462 {
4463 /* Old code had j++, but we cannot assume that i == j at this
4464 point -- what if a CTLNUL has already been removed from the
4465 string? We don't want to drop the CTLESC or recopy characters
4466 that we've already copied down. */
d233b485
CR
4467 i++;
4468 string[j++] = CTLESC;
b80f6443
JA
4469 if (i == slen)
4470 break;
4471 }
7117c2d2 4472 else if (string[i] == CTLNUL)
a48a8ac3
CR
4473 {
4474 i++;
4475 continue;
4476 }
7117c2d2
JA
4477
4478 prev_i = i;
d233b485 4479 ADVANCE_CHAR (string, slen, i); /* COPY_CHAR_I? */
7117c2d2 4480 if (j < prev_i)
cce855bc 4481 {
7117c2d2 4482 do string[j++] = string[prev_i++]; while (prev_i < i);
cce855bc 4483 }
7117c2d2
JA
4484 else
4485 j = i;
cce855bc 4486 }
7117c2d2
JA
4487 string[j] = '\0';
4488
4489 return (string);
cce855bc
JA
4490}
4491
4492/* Perform quoted null character removal on each element of LIST.
4493 This modifies LIST. */
4494void
4495word_list_remove_quoted_nulls (list)
4496 WORD_LIST *list;
4497{
4498 register WORD_LIST *t;
4499
4500 for (t = list; t; t = t->next)
95732b49
JA
4501 {
4502 remove_quoted_nulls (t->word->word);
4503 t->word->flags &= ~W_HASQUOTEDNULL;
4504 }
cce855bc
JA
4505}
4506
4507/* **************************************************************** */
4508/* */
4509/* Functions for Matching and Removing Patterns */
4510/* */
4511/* **************************************************************** */
4512
b80f6443 4513#if defined (HANDLE_MULTIBYTE)
d233b485 4514# ifdef INCLUDE_UNUSED
b80f6443
JA
4515static unsigned char *
4516mb_getcharlens (string, len)
4517 char *string;
4518 int len;
4519{
4520 int i, offset, last;
4521 unsigned char *ret;
4522 char *p;
4523 DECLARE_MBSTATE;
4524
4525 i = offset = 0;
4526 last = 0;
4527 ret = (unsigned char *)xmalloc (len);
4528 memset (ret, 0, len);
4529 while (string[last])
4530 {
4531 ADVANCE_CHAR (string, len, offset);
4532 ret[last] = offset - last;
4533 last = offset;
4534 }
4535 return ret;
4536}
d233b485 4537# endif
b80f6443
JA
4538#endif
4539
cce855bc
JA
4540/* Remove the portion of PARAM matched by PATTERN according to OP, where OP
4541 can have one of 4 values:
4542 RP_LONG_LEFT remove longest matching portion at start of PARAM
726f6388
JA
4543 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
4544 RP_LONG_RIGHT remove longest matching portion at end of PARAM
4545 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
4546*/
4547
4548#define RP_LONG_LEFT 1
4549#define RP_SHORT_LEFT 2
4550#define RP_LONG_RIGHT 3
4551#define RP_SHORT_RIGHT 4
4552
495aee44 4553/* Returns its first argument if nothing matched; new memory otherwise */
726f6388 4554static char *
b80f6443 4555remove_upattern (param, pattern, op)
726f6388
JA
4556 char *param, *pattern;
4557 int op;
4558{
a0c0a00f 4559 register size_t len;
ccc6cda3 4560 register char *end;
726f6388
JA
4561 register char *p, *ret, c;
4562
ccc6cda3
JA
4563 len = STRLEN (param);
4564 end = param + len;
726f6388
JA
4565
4566 switch (op)
4567 {
4568 case RP_LONG_LEFT: /* remove longest match at start */
4569 for (p = end; p >= param; p--)
4570 {
4571 c = *p; *p = '\0';
f73dda09 4572 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
726f6388
JA
4573 {
4574 *p = c;
4575 return (savestring (p));
4576 }
4577 *p = c;
b80f6443 4578
726f6388
JA
4579 }
4580 break;
4581
4582 case RP_SHORT_LEFT: /* remove shortest match at start */
4583 for (p = param; p <= end; p++)
4584 {
4585 c = *p; *p = '\0';
f73dda09 4586 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
726f6388
JA
4587 {
4588 *p = c;
4589 return (savestring (p));
4590 }
4591 *p = c;
4592 }
4593 break;
4594
ccc6cda3
JA
4595 case RP_LONG_RIGHT: /* remove longest match at end */
4596 for (p = param; p <= end; p++)
4597 {
f73dda09 4598 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
ccc6cda3
JA
4599 {
4600 c = *p; *p = '\0';
4601 ret = savestring (param);
4602 *p = c;
4603 return (ret);
4604 }
4605 }
4606 break;
4607
4608 case RP_SHORT_RIGHT: /* remove shortest match at end */
4609 for (p = end; p >= param; p--)
4610 {
f73dda09 4611 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
ccc6cda3
JA
4612 {
4613 c = *p; *p = '\0';
4614 ret = savestring (param);
4615 *p = c;
4616 return (ret);
4617 }
4618 }
4619 break;
4620 }
b80f6443 4621
495aee44 4622 return (param); /* no match, return original string */
ccc6cda3
JA
4623}
4624
b80f6443 4625#if defined (HANDLE_MULTIBYTE)
495aee44 4626/* Returns its first argument if nothing matched; new memory otherwise */
b80f6443
JA
4627static wchar_t *
4628remove_wpattern (wparam, wstrlen, wpattern, op)
4629 wchar_t *wparam;
4630 size_t wstrlen;
4631 wchar_t *wpattern;
4632 int op;
4633{
0628567a
JA
4634 wchar_t wc, *ret;
4635 int n;
b80f6443
JA
4636
4637 switch (op)
4638 {
4639 case RP_LONG_LEFT: /* remove longest match at start */
4640 for (n = wstrlen; n >= 0; n--)
4641 {
4642 wc = wparam[n]; wparam[n] = L'\0';
4643 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4644 {
4645 wparam[n] = wc;
4646 return (wcsdup (wparam + n));
4647 }
4648 wparam[n] = wc;
4649 }
4650 break;
4651
4652 case RP_SHORT_LEFT: /* remove shortest match at start */
4653 for (n = 0; n <= wstrlen; n++)
4654 {
4655 wc = wparam[n]; wparam[n] = L'\0';
4656 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4657 {
4658 wparam[n] = wc;
4659 return (wcsdup (wparam + n));
4660 }
4661 wparam[n] = wc;
4662 }
4663 break;
4664
4665 case RP_LONG_RIGHT: /* remove longest match at end */
4666 for (n = 0; n <= wstrlen; n++)
4667 {
4668 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4669 {
4670 wc = wparam[n]; wparam[n] = L'\0';
4671 ret = wcsdup (wparam);
4672 wparam[n] = wc;
4673 return (ret);
4674 }
4675 }
4676 break;
4677
4678 case RP_SHORT_RIGHT: /* remove shortest match at end */
4679 for (n = wstrlen; n >= 0; n--)
4680 {
4681 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4682 {
4683 wc = wparam[n]; wparam[n] = L'\0';
4684 ret = wcsdup (wparam);
4685 wparam[n] = wc;
4686 return (ret);
4687 }
4688 }
4689 break;
4690 }
4691
495aee44 4692 return (wparam); /* no match, return original string */
b80f6443
JA
4693}
4694#endif /* HANDLE_MULTIBYTE */
4695
4696static char *
4697remove_pattern (param, pattern, op)
4698 char *param, *pattern;
4699 int op;
4700{
495aee44
CR
4701 char *xret;
4702
b80f6443
JA
4703 if (param == NULL)
4704 return (param);
4705 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4706 return (savestring (param));
4707
4708#if defined (HANDLE_MULTIBYTE)
4709 if (MB_CUR_MAX > 1)
4710 {
4711 wchar_t *ret, *oret;
4712 size_t n;
4713 wchar_t *wparam, *wpattern;
4714 mbstate_t ps;
b80f6443 4715
d233b485
CR
4716 /* XXX - could optimize here by checking param and pattern for multibyte
4717 chars with mbsmbchar and calling remove_upattern. */
4718
b80f6443
JA
4719 n = xdupmbstowcs (&wpattern, NULL, pattern);
4720 if (n == (size_t)-1)
495aee44
CR
4721 {
4722 xret = remove_upattern (param, pattern, op);
4723 return ((xret == param) ? savestring (param) : xret);
4724 }
b80f6443 4725 n = xdupmbstowcs (&wparam, NULL, param);
ac50fbac 4726
b80f6443
JA
4727 if (n == (size_t)-1)
4728 {
4729 free (wpattern);
495aee44
CR
4730 xret = remove_upattern (param, pattern, op);
4731 return ((xret == param) ? savestring (param) : xret);
b80f6443
JA
4732 }
4733 oret = ret = remove_wpattern (wparam, n, wpattern, op);
495aee44
CR
4734 /* Don't bother to convert wparam back to multibyte string if nothing
4735 matched; just return copy of original string */
4736 if (ret == wparam)
4737 {
4738 free (wparam);
4739 free (wpattern);
4740 return (savestring (param));
4741 }
b80f6443
JA
4742
4743 free (wparam);
4744 free (wpattern);
4745
4746 n = strlen (param);
0628567a 4747 xret = (char *)xmalloc (n + 1);
b80f6443
JA
4748 memset (&ps, '\0', sizeof (mbstate_t));
4749 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4750 xret[n] = '\0'; /* just to make sure */
4751 free (oret);
4752 return xret;
4753 }
4754 else
4755#endif
ccc6cda3 4756 {
495aee44
CR
4757 xret = remove_upattern (param, pattern, op);
4758 return ((xret == param) ? savestring (param) : xret);
ccc6cda3
JA
4759 }
4760}
4761
4762/* Match PAT anywhere in STRING and return the match boundaries.
4763 This returns 1 in case of a successful match, 0 otherwise. SP
4764 and EP are pointers into the string where the match begins and
4765 ends, respectively. MTYPE controls what kind of match is attempted.
4766 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4767 of the string, respectively. The longest match is returned. */
4768static int
b80f6443 4769match_upattern (string, pat, mtype, sp, ep)
ccc6cda3
JA
4770 char *string, *pat;
4771 int mtype;
4772 char **sp, **ep;
4773{
a0c0a00f
CR
4774 int c, mlen;
4775 size_t len;
95732b49 4776 register char *p, *p1, *npat;
ccc6cda3
JA
4777 char *end;
4778
95732b49
JA
4779 /* If the pattern doesn't match anywhere in the string, go ahead and
4780 short-circuit right away. A minor optimization, saves a bunch of
4781 unnecessary calls to strmatch (up to N calls for a string of N
4782 characters) if the match is unsuccessful. To preserve the semantics
4783 of the substring matches below, we make sure that the pattern has
4784 `*' as first and last character, making a new pattern if necessary. */
4785 /* XXX - check this later if I ever implement `**' with special meaning,
4786 since this will potentially result in `**' at the beginning or end */
4787 len = STRLEN (pat);
0001803f 4788 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
95732b49 4789 {
a0c0a00f
CR
4790 int unescaped_backslash;
4791 char *pp;
4792
0628567a 4793 p = npat = (char *)xmalloc (len + 3);
95732b49 4794 p1 = pat;
d233b485 4795 if ((mtype != MATCH_BEG) && (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob)))
95732b49
JA
4796 *p++ = '*';
4797 while (*p1)
4798 *p++ = *p1++;
a0c0a00f
CR
4799#if 1
4800 /* Need to also handle a pattern that ends with an unescaped backslash.
4801 For right now, we ignore it because the pattern matching code will
4802 fail the match anyway */
4803 /* If the pattern ends with a `*' we leave it alone if it's preceded by
4804 an even number of backslashes, but if it's escaped by a backslash
4805 we need to add another `*'. */
d233b485 4806 if ((mtype != MATCH_END) && (p1[-1] == '*' && (unescaped_backslash = p1[-2] == '\\')))
a0c0a00f
CR
4807 {
4808 pp = p1 - 3;
4809 while (pp >= pat && *pp-- == '\\')
4810 unescaped_backslash = 1 - unescaped_backslash;
4811 if (unescaped_backslash)
4812 *p++ = '*';
4813 }
d233b485 4814 else if (mtype != MATCH_END && p1[-1] != '*')
95732b49 4815 *p++ = '*';
a0c0a00f
CR
4816#else
4817 if (p1[-1] != '*' || p1[-2] == '\\')
4818 *p++ = '*';
4819#endif
95732b49
JA
4820 *p = '\0';
4821 }
4822 else
4823 npat = pat;
a0c0a00f 4824 c = strmatch (npat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
95732b49
JA
4825 if (npat != pat)
4826 free (npat);
4827 if (c == FNM_NOMATCH)
4828 return (0);
4829
b80f6443
JA
4830 len = STRLEN (string);
4831 end = string + len;
ccc6cda3 4832
495aee44 4833 mlen = umatchlen (pat, len);
712f80b0
CR
4834 if (mlen > (int)len)
4835 return (0);
495aee44 4836
ccc6cda3
JA
4837 switch (mtype)
4838 {
4839 case MATCH_ANY:
4840 for (p = string; p <= end; p++)
4841 {
a0c0a00f 4842 if (match_pattern_char (pat, p, FNMATCH_IGNCASE))
ccc6cda3 4843 {
495aee44
CR
4844 p1 = (mlen == -1) ? end : p + mlen;
4845 /* p1 - p = length of portion of string to be considered
4846 p = current position in string
4847 mlen = number of characters consumed by match (-1 for entire string)
4848 end = end of string
4849 we want to break immediately if the potential match len
4850 is greater than the number of characters remaining in the
4851 string
4852 */
4853 if (p1 > end)
4854 break;
4855 for ( ; p1 >= p; p1--)
ccc6cda3
JA
4856 {
4857 c = *p1; *p1 = '\0';
a0c0a00f 4858 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
ccc6cda3
JA
4859 {
4860 *p1 = c;
4861 *sp = p;
4862 *ep = p1;
4863 return 1;
4864 }
4865 *p1 = c;
495aee44
CR
4866#if 1
4867 /* If MLEN != -1, we have a fixed length pattern. */
4868 if (mlen != -1)
4869 break;
4870#endif
ccc6cda3
JA
4871 }
4872 }
4873 }
b80f6443 4874
ccc6cda3
JA
4875 return (0);
4876
4877 case MATCH_BEG:
a0c0a00f 4878 if (match_pattern_char (pat, string, FNMATCH_IGNCASE) == 0)
28ef6c31 4879 return (0);
b80f6443 4880
495aee44 4881 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
ccc6cda3
JA
4882 {
4883 c = *p; *p = '\0';
a0c0a00f 4884 if (strmatch (pat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
ccc6cda3
JA
4885 {
4886 *p = c;
4887 *sp = string;
4888 *ep = p;
4889 return 1;
4890 }
4891 *p = c;
495aee44
CR
4892 /* If MLEN != -1, we have a fixed length pattern. */
4893 if (mlen != -1)
4894 break;
ccc6cda3 4895 }
b80f6443 4896
ccc6cda3 4897 return (0);
726f6388 4898
ccc6cda3 4899 case MATCH_END:
495aee44 4900 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
b80f6443 4901 {
a0c0a00f 4902 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
4903 {
4904 *sp = p;
4905 *ep = end;
4906 return 1;
4907 }
495aee44
CR
4908 /* If MLEN != -1, we have a fixed length pattern. */
4909 if (mlen != -1)
4910 break;
b80f6443
JA
4911 }
4912
4913 return (0);
4914 }
4915
4916 return (0);
4917}
4918
4919#if defined (HANDLE_MULTIBYTE)
a0c0a00f
CR
4920
4921#define WFOLD(c) (match_ignore_case && iswupper (c) ? towlower (c) : (c))
4922
b80f6443
JA
4923/* Match WPAT anywhere in WSTRING and return the match boundaries.
4924 This returns 1 in case of a successful match, 0 otherwise. Wide
4925 character version. */
4926static int
4927match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4928 wchar_t *wstring;
4929 char **indices;
4930 size_t wstrlen;
4931 wchar_t *wpat;
4932 int mtype;
4933 char **sp, **ep;
4934{
95732b49 4935 wchar_t wc, *wp, *nwpat, *wp1;
495aee44
CR
4936 size_t len;
4937 int mlen;
4938 int n, n1, n2, simple;
4939
4940 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4941#if defined (EXTENDED_GLOB)
4942 if (extended_glob)
91717ba3 4943 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
b80f6443
JA
4944#endif
4945
95732b49
JA
4946 /* If the pattern doesn't match anywhere in the string, go ahead and
4947 short-circuit right away. A minor optimization, saves a bunch of
4948 unnecessary calls to strmatch (up to N calls for a string of N
4949 characters) if the match is unsuccessful. To preserve the semantics
4950 of the substring matches below, we make sure that the pattern has
4951 `*' as first and last character, making a new pattern if necessary. */
95732b49 4952 len = wcslen (wpat);
0001803f 4953 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
95732b49 4954 {
a0c0a00f
CR
4955 int unescaped_backslash;
4956 wchar_t *wpp;
4957
0628567a 4958 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
95732b49 4959 wp1 = wpat;
0001803f 4960 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
95732b49
JA
4961 *wp++ = L'*';
4962 while (*wp1 != L'\0')
4963 *wp++ = *wp1++;
a0c0a00f
CR
4964#if 1
4965 /* See comments above in match_upattern. */
4966 if (wp1[-1] == L'*' && (unescaped_backslash = wp1[-2] == L'\\'))
4967 {
4968 wpp = wp1 - 3;
4969 while (wpp >= wpat && *wpp-- == L'\\')
4970 unescaped_backslash = 1 - unescaped_backslash;
4971 if (unescaped_backslash)
4972 *wp++ = L'*';
4973 }
4974 else if (wp1[-1] != L'*')
4975 *wp++ = L'*';
4976#else
95732b49
JA
4977 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4978 *wp++ = L'*';
a0c0a00f 4979#endif
95732b49
JA
4980 *wp = '\0';
4981 }
4982 else
4983 nwpat = wpat;
a0c0a00f 4984 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
95732b49
JA
4985 if (nwpat != wpat)
4986 free (nwpat);
4987 if (len == FNM_NOMATCH)
4988 return (0);
4989
495aee44 4990 mlen = wmatchlen (wpat, wstrlen);
712f80b0
CR
4991 if (mlen > (int)wstrlen)
4992 return (0);
495aee44
CR
4993
4994/* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
b80f6443
JA
4995 switch (mtype)
4996 {
4997 case MATCH_ANY:
4998 for (n = 0; n <= wstrlen; n++)
4999 {
a0c0a00f 5000 n2 = simple ? (WFOLD(*wpat) == WFOLD(wstring[n])) : match_pattern_wchar (wpat, wstring + n, FNMATCH_IGNCASE);
495aee44 5001 if (n2)
b80f6443 5002 {
495aee44
CR
5003 n1 = (mlen == -1) ? wstrlen : n + mlen;
5004 if (n1 > wstrlen)
5005 break;
5006
5007 for ( ; n1 >= n; n1--)
b80f6443
JA
5008 {
5009 wc = wstring[n1]; wstring[n1] = L'\0';
a0c0a00f 5010 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
5011 {
5012 wstring[n1] = wc;
5013 *sp = indices[n];
5014 *ep = indices[n1];
5015 return 1;
5016 }
5017 wstring[n1] = wc;
495aee44
CR
5018 /* If MLEN != -1, we have a fixed length pattern. */
5019 if (mlen != -1)
5020 break;
b80f6443
JA
5021 }
5022 }
5023 }
5024
5025 return (0);
5026
5027 case MATCH_BEG:
a0c0a00f 5028 if (match_pattern_wchar (wpat, wstring, FNMATCH_IGNCASE) == 0)
b80f6443
JA
5029 return (0);
5030
495aee44 5031 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
b80f6443
JA
5032 {
5033 wc = wstring[n]; wstring[n] = L'\0';
a0c0a00f 5034 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
5035 {
5036 wstring[n] = wc;
5037 *sp = indices[0];
5038 *ep = indices[n];
5039 return 1;
5040 }
5041 wstring[n] = wc;
495aee44
CR
5042 /* If MLEN != -1, we have a fixed length pattern. */
5043 if (mlen != -1)
5044 break;
b80f6443
JA
5045 }
5046
5047 return (0);
5048
5049 case MATCH_END:
495aee44 5050 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
b80f6443 5051 {
a0c0a00f 5052 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
b80f6443
JA
5053 {
5054 *sp = indices[n];
5055 *ep = indices[wstrlen];
5056 return 1;
5057 }
495aee44
CR
5058 /* If MLEN != -1, we have a fixed length pattern. */
5059 if (mlen != -1)
5060 break;
b80f6443
JA
5061 }
5062
ccc6cda3 5063 return (0);
726f6388 5064 }
ccc6cda3
JA
5065
5066 return (0);
726f6388 5067}
a0c0a00f 5068#undef WFOLD
b80f6443
JA
5069#endif /* HANDLE_MULTIBYTE */
5070
5071static int
5072match_pattern (string, pat, mtype, sp, ep)
5073 char *string, *pat;
5074 int mtype;
5075 char **sp, **ep;
5076{
5077#if defined (HANDLE_MULTIBYTE)
5078 int ret;
5079 size_t n;
5080 wchar_t *wstring, *wpat;
5081 char **indices;
5082#endif
5083
a0c0a00f 5084 if (string == 0 || pat == 0 || *pat == 0)
b80f6443
JA
5085 return (0);
5086
5087#if defined (HANDLE_MULTIBYTE)
5088 if (MB_CUR_MAX > 1)
5089 {
495aee44 5090 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
495aee44
CR
5091 return (match_upattern (string, pat, mtype, sp, ep));
5092
b80f6443
JA
5093 n = xdupmbstowcs (&wpat, NULL, pat);
5094 if (n == (size_t)-1)
5095 return (match_upattern (string, pat, mtype, sp, ep));
5096 n = xdupmbstowcs (&wstring, &indices, string);
5097 if (n == (size_t)-1)
5098 {
5099 free (wpat);
5100 return (match_upattern (string, pat, mtype, sp, ep));
5101 }
5102 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
5103
5104 free (wpat);
5105 free (wstring);
5106 free (indices);
5107
5108 return (ret);
5109 }
5110 else
5111#endif
5112 return (match_upattern (string, pat, mtype, sp, ep));
5113}
726f6388 5114
cce855bc
JA
5115static int
5116getpatspec (c, value)
5117 int c;
5118 char *value;
5119{
5120 if (c == '#')
5121 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
5122 else /* c == '%' */
5123 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
5124}
5125
5126/* Posix.2 says that the WORD should be run through tilde expansion,
5127 parameter expansion, command substitution and arithmetic expansion.
5128 This leaves the result quoted, so quote_string_for_globbing () has
f73dda09 5129 to be called to fix it up for strmatch (). If QUOTED is non-zero,
cce855bc
JA
5130 it means that the entire expression was enclosed in double quotes.
5131 This means that quoting characters in the pattern do not make any
5132 special pattern characters quoted. For example, the `*' in the
5133 following retains its special meaning: "${foo#'*'}". */
5134static char *
5135getpattern (value, quoted, expandpat)
5136 char *value;
5137 int quoted, expandpat;
5138{
5139 char *pat, *tword;
5140 WORD_LIST *l;
0628567a 5141#if 0
cce855bc 5142 int i;
0628567a 5143#endif
7117c2d2
JA
5144 /* There is a problem here: how to handle single or double quotes in the
5145 pattern string when the whole expression is between double quotes?
5146 POSIX.2 says that enclosing double quotes do not cause the pattern to
5147 be quoted, but does that leave us a problem with @ and array[@] and their
5148 expansions inside a pattern? */
5149#if 0
cce855bc
JA
5150 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
5151 {
5152 i = 0;
a0c0a00f 5153 pat = string_extract_double_quoted (tword, &i, SX_STRIPDQ);
cce855bc
JA
5154 free (tword);
5155 tword = pat;
5156 }
7117c2d2 5157#endif
cce855bc 5158
d233b485 5159 /* expand_string_for_pat () leaves WORD quoted and does not perform
7117c2d2 5160 word splitting. */
d233b485 5161 l = *value ? expand_string_for_pat (value,
7117c2d2 5162 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
cce855bc 5163 (int *)NULL, (int *)NULL)
cce855bc 5164 : (WORD_LIST *)0;
712f80b0
CR
5165 if (l)
5166 word_list_remove_quoted_nulls (l);
cce855bc
JA
5167 pat = string_list (l);
5168 dispose_words (l);
5169 if (pat)
5170 {
5171 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
5172 free (pat);
5173 pat = tword;
5174 }
5175 return (pat);
5176}
5177
7117c2d2 5178#if 0
cce855bc
JA
5179/* Handle removing a pattern from a string as a result of ${name%[%]value}
5180 or ${name#[#]value}. */
5181static char *
7117c2d2
JA
5182variable_remove_pattern (value, pattern, patspec, quoted)
5183 char *value, *pattern;
5184 int patspec, quoted;
cce855bc 5185{
7117c2d2 5186 char *tword;
cce855bc 5187
7117c2d2 5188 tword = remove_pattern (value, pattern, patspec);
cce855bc 5189
cce855bc
JA
5190 return (tword);
5191}
a0c0a00f
CR
5192#endif
5193
5194static char *
5195list_remove_pattern (list, pattern, patspec, itype, quoted)
5196 WORD_LIST *list;
5197 char *pattern;
5198 int patspec, itype, quoted;
5199{
5200 WORD_LIST *new, *l;
5201 WORD_DESC *w;
5202 char *tword;
5203
5204 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
5205 {
5206 tword = remove_pattern (l->word->word, pattern, patspec);
5207 w = alloc_word_desc ();
5208 w->word = tword ? tword : savestring ("");
5209 new = make_word_list (w, new);
5210 }
5211
5212 l = REVERSE_LIST (new, WORD_LIST *);
712f80b0 5213 tword = string_list_pos_params (itype, l, quoted, 0);
a0c0a00f
CR
5214 dispose_words (l);
5215
5216 return (tword);
5217}
5218
5219static char *
5220parameter_list_remove_pattern (itype, pattern, patspec, quoted)
5221 int itype;
5222 char *pattern;
5223 int patspec, quoted;
5224{
5225 char *ret;
5226 WORD_LIST *list;
5227
5228 list = list_rest_of_args ();
5229 if (list == 0)
5230 return ((char *)NULL);
5231 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
5232 dispose_words (list);
5233 return (ret);
5234}
5235
5236#if defined (ARRAY_VARS)
5237static char *
712f80b0 5238array_remove_pattern (var, pattern, patspec, starsub, quoted)
a0c0a00f
CR
5239 SHELL_VAR *var;
5240 char *pattern;
5241 int patspec;
712f80b0 5242 int starsub; /* so we can figure out how it's indexed */
a0c0a00f
CR
5243 int quoted;
5244{
5245 ARRAY *a;
5246 HASH_TABLE *h;
5247 int itype;
5248 char *ret;
5249 WORD_LIST *list;
5250 SHELL_VAR *v;
5251
712f80b0 5252 v = var; /* XXX - for now */
a0c0a00f 5253
712f80b0 5254 itype = starsub ? '*' : '@';
a0c0a00f
CR
5255
5256 a = (v && array_p (v)) ? array_cell (v) : 0;
5257 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
5258
5259 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
5260 if (list == 0)
5261 return ((char *)NULL);
5262 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
5263 dispose_words (list);
5264
5265 return ret;
5266}
5267#endif /* ARRAY_VARS */
5268
5269static char *
5270parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
5271 char *varname, *value;
5272 int ind;
5273 char *patstr;
5274 int rtype, quoted, flags;
5275{
5276 int vtype, patspec, starsub;
d233b485 5277 char *temp1, *val, *pattern, *oname;
a0c0a00f
CR
5278 SHELL_VAR *v;
5279
5280 if (value == 0)
5281 return ((char *)NULL);
5282
d233b485 5283 oname = this_command_name;
a0c0a00f
CR
5284 this_command_name = varname;
5285
5286 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
5287 if (vtype == -1)
d233b485
CR
5288 {
5289 this_command_name = oname;
5290 return ((char *)NULL);
5291 }
a0c0a00f
CR
5292
5293 starsub = vtype & VT_STARSUB;
5294 vtype &= ~VT_STARSUB;
5295
5296 patspec = getpatspec (rtype, patstr);
5297 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
5298 patstr++;
5299
5300 /* Need to pass getpattern newly-allocated memory in case of expansion --
5301 the expansion code will free the passed string on an error. */
5302 temp1 = savestring (patstr);
5303 pattern = getpattern (temp1, quoted, 1);
5304 free (temp1);
5305
5306 temp1 = (char *)NULL; /* shut up gcc */
5307 switch (vtype)
5308 {
5309 case VT_VARIABLE:
5310 case VT_ARRAYMEMBER:
5311 temp1 = remove_pattern (val, pattern, patspec);
5312 if (vtype == VT_VARIABLE)
5313 FREE (val);
5314 if (temp1)
5315 {
5316 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5317 ? quote_string (temp1)
5318 : quote_escapes (temp1);
5319 free (temp1);
5320 temp1 = val;
5321 }
5322 break;
5323#if defined (ARRAY_VARS)
5324 case VT_ARRAYVAR:
712f80b0 5325 temp1 = array_remove_pattern (v, pattern, patspec, starsub, quoted);
a0c0a00f
CR
5326 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5327 {
5328 val = quote_escapes (temp1);
5329 free (temp1);
5330 temp1 = val;
5331 }
5332 break;
5333#endif
5334 case VT_POSPARMS:
5335 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
d233b485
CR
5336 if (temp1 && quoted == 0 && ifs_is_null)
5337 {
5338 /* Posix interp 888 */
5339 }
5340 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
a0c0a00f
CR
5341 {
5342 val = quote_escapes (temp1);
5343 free (temp1);
5344 temp1 = val;
5345 }
5346 break;
5347 }
5348
d233b485
CR
5349 this_command_name = oname;
5350
a0c0a00f
CR
5351 FREE (pattern);
5352 return temp1;
5353}
5354
d233b485 5355#if defined (PROCESS_SUBSTITUTION)
a0c0a00f 5356
712f80b0
CR
5357static void reap_some_procsubs PARAMS((int));
5358
d233b485
CR
5359/*****************************************************************/
5360/* */
5361/* Hacking Process Substitution */
5362/* */
5363/*****************************************************************/
5364
5365#if !defined (HAVE_DEV_FD)
5366/* Named pipes must be removed explicitly with `unlink'. This keeps a list
5367 of FIFOs the shell has open. unlink_fifo_list will walk the list and
3eb0018e
CR
5368 unlink the ones that don't have a living process on the other end.
5369 unlink_all_fifos will walk the list and unconditionally unlink them, trying
5370 to open and close the FIFO first to release any child processes sleeping on
5371 the FIFO. add_fifo_list adds the name of an open FIFO to the list.
5372 NFIFO is a count of the number of FIFOs in the list. */
d233b485
CR
5373#define FIFO_INCR 20
5374
5375/* PROC value of -1 means the process has been reaped and the FIFO needs to
5376 be removed. PROC value of 0 means the slot is unused. */
5377struct temp_fifo {
5378 char *file;
5379 pid_t proc;
5380};
5381
5382static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
5383static int nfifo;
5384static int fifo_list_size;
5385
5386void
5387clear_fifo_list ()
a0c0a00f 5388{
712f80b0
CR
5389 int i;
5390
5391 for (i = 0; i < fifo_list_size; i++)
5392 {
5393 if (fifo_list[i].file)
5394 free (fifo_list[i].file);
5395 fifo_list[i].file = NULL;
5396 fifo_list[i].proc = 0;
5397 }
5398 nfifo = 0;
d233b485 5399}
a0c0a00f 5400
9e49d343 5401void *
d233b485
CR
5402copy_fifo_list (sizep)
5403 int *sizep;
5404{
5405 if (sizep)
5406 *sizep = 0;
9e49d343 5407 return (void *)NULL;
d233b485
CR
5408}
5409
5410static void
5411add_fifo_list (pathname)
5412 char *pathname;
5413{
5414 int osize, i;
5415
5416 if (nfifo >= fifo_list_size - 1)
a0c0a00f 5417 {
d233b485
CR
5418 osize = fifo_list_size;
5419 fifo_list_size += FIFO_INCR;
5420 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
5421 fifo_list_size * sizeof (struct temp_fifo));
5422 for (i = osize; i < fifo_list_size; i++)
5423 {
5424 fifo_list[i].file = (char *)NULL;
5425 fifo_list[i].proc = 0; /* unused */
5426 }
a0c0a00f 5427 }
d233b485
CR
5428
5429 fifo_list[nfifo].file = savestring (pathname);
5430 nfifo++;
a0c0a00f 5431}
a0c0a00f 5432
d233b485
CR
5433void
5434unlink_fifo (i)
5435 int i;
a0c0a00f 5436{
d233b485
CR
5437 if ((fifo_list[i].proc == (pid_t)-1) || (fifo_list[i].proc > 0 && (kill(fifo_list[i].proc, 0) == -1)))
5438 {
5439 unlink (fifo_list[i].file);
5440 free (fifo_list[i].file);
5441 fifo_list[i].file = (char *)NULL;
5442 fifo_list[i].proc = 0;
5443 }
a0c0a00f
CR
5444}
5445
d233b485
CR
5446void
5447unlink_fifo_list ()
a0c0a00f 5448{
d233b485 5449 int saved, i, j;
a0c0a00f 5450
d233b485
CR
5451 if (nfifo == 0)
5452 return;
a0c0a00f 5453
d233b485 5454 for (i = saved = 0; i < nfifo; i++)
a0c0a00f 5455 {
d233b485
CR
5456 if ((fifo_list[i].proc == (pid_t)-1) || (fifo_list[i].proc > 0 && (kill(fifo_list[i].proc, 0) == -1)))
5457 {
5458 unlink (fifo_list[i].file);
5459 free (fifo_list[i].file);
5460 fifo_list[i].file = (char *)NULL;
5461 fifo_list[i].proc = 0;
5462 }
5463 else
5464 saved++;
a0c0a00f 5465 }
d233b485
CR
5466
5467 /* If we didn't remove some of the FIFOs, compact the list. */
5468 if (saved)
5469 {
5470 for (i = j = 0; i < nfifo; i++)
5471 if (fifo_list[i].file)
5472 {
9e49d343
CR
5473 if (i != j)
5474 {
5475 fifo_list[j].file = fifo_list[i].file;
5476 fifo_list[j].proc = fifo_list[i].proc;
5477 fifo_list[i].file = (char *)NULL;
5478 fifo_list[i].proc = 0;
5479 }
d233b485
CR
5480 j++;
5481 }
5482 nfifo = j;
5483 }
5484 else
5485 nfifo = 0;
a0c0a00f 5486}
cce855bc 5487
3eb0018e
CR
5488void
5489unlink_all_fifos ()
5490{
5491 int i, fd;
5492
5493 if (nfifo == 0)
5494 return;
5495
5496 for (i = 0; i < nfifo; i++)
5497 {
5498 fifo_list[i].proc = (pid_t)-1;
5499 fd = open (fifo_list[i].file, O_RDWR|O_NONBLOCK);
5500 unlink_fifo (i);
5501 if (fd >= 0)
5502 close (fd);
5503 }
5504
5505 nfifo = 0;
5506}
5507
d233b485
CR
5508/* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
5509 from some point in the past, and close all open FIFOs in fifo_list
5510 that are not marked as active in LIST. If LIST is NULL, close
5511 everything in fifo_list. LSIZE is the number of elements in LIST, in
5512 case it's larger than fifo_list_size (size of fifo_list). */
5513void
5514close_new_fifos (list, lsize)
9e49d343 5515 void *list;
d233b485 5516 int lsize;
cce855bc 5517{
d233b485 5518 int i;
9e49d343 5519 char *plist;
cce855bc 5520
d233b485 5521 if (list == 0)
cce855bc 5522 {
d233b485
CR
5523 unlink_fifo_list ();
5524 return;
cce855bc
JA
5525 }
5526
9e49d343
CR
5527 for (plist = (char *)list, i = 0; i < lsize; i++)
5528 if (plist[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
d233b485 5529 unlink_fifo (i);
3185942a 5530
d233b485
CR
5531 for (i = lsize; i < fifo_list_size; i++)
5532 unlink_fifo (i);
cce855bc
JA
5533}
5534
d233b485
CR
5535int
5536find_procsub_child (pid)
5537 pid_t pid;
cce855bc 5538{
d233b485 5539 int i;
cce855bc 5540
d233b485
CR
5541 for (i = 0; i < nfifo; i++)
5542 if (fifo_list[i].proc == pid)
5543 return i;
5544 return -1;
cce855bc
JA
5545}
5546
d233b485
CR
5547void
5548set_procsub_status (ind, pid, status)
5549 int ind;
5550 pid_t pid;
5551 int status;
cce855bc 5552{
d233b485
CR
5553 if (ind >= 0 && ind < nfifo)
5554 fifo_list[ind].proc = (pid_t)-1; /* sentinel */
5555}
ac50fbac 5556
d233b485
CR
5557/* If we've marked the process for this procsub as dead, close the
5558 associated file descriptor and delete the FIFO. */
712f80b0
CR
5559static void
5560reap_some_procsubs (max)
5561 int max;
d233b485
CR
5562{
5563 int i;
7117c2d2 5564
712f80b0 5565 for (i = 0; i < max; i++)
d233b485
CR
5566 if (fifo_list[i].proc == (pid_t)-1) /* reaped */
5567 unlink_fifo (i);
5568}
a0c0a00f 5569
712f80b0
CR
5570void
5571reap_procsubs ()
5572{
5573 reap_some_procsubs (nfifo);
5574}
5575
5576#if 0
5577/* UNUSED */
d233b485
CR
5578void
5579wait_procsubs ()
5580{
5581 int i, r;
7117c2d2 5582
d233b485
CR
5583 for (i = 0; i < nfifo; i++)
5584 {
5585 if (fifo_list[i].proc != (pid_t)-1 && fifo_list[i].proc > 0)
5586 {
712f80b0
CR
5587 r = wait_for (fifo_list[i].proc, 0);
5588 save_proc_status (fifo_list[i].proc, r);
d233b485
CR
5589 fifo_list[i].proc = (pid_t)-1;
5590 }
5591 }
7117c2d2 5592}
712f80b0 5593#endif
7117c2d2 5594
d233b485
CR
5595int
5596fifos_pending ()
7117c2d2 5597{
d233b485
CR
5598 return nfifo;
5599}
7117c2d2 5600
d233b485
CR
5601int
5602num_fifos ()
5603{
5604 return nfifo;
5605}
cce855bc 5606
d233b485
CR
5607static char *
5608make_named_pipe ()
5609{
5610 char *tname;
cce855bc 5611
d233b485
CR
5612 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
5613 if (mkfifo (tname, 0600) < 0)
cce855bc 5614 {
d233b485
CR
5615 free (tname);
5616 return ((char *)NULL);
cce855bc
JA
5617 }
5618
d233b485
CR
5619 add_fifo_list (tname);
5620 return (tname);
a0c0a00f 5621}
cce855bc 5622
d233b485 5623#else /* HAVE_DEV_FD */
726f6388 5624
d233b485
CR
5625/* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
5626 has open to children. NFDS is a count of the number of bits currently
5627 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
5628 of open files. */
5629/* dev_fd_list[I] value of -1 means the process has been reaped and file
5630 descriptor I needs to be closed. Value of 0 means the slot is unused. */
726f6388 5631
d233b485
CR
5632static pid_t *dev_fd_list = (pid_t *)NULL;
5633static int nfds;
5634static int totfds; /* The highest possible number of open files. */
726f6388 5635
d233b485
CR
5636void
5637clear_fifo (i)
5638 int i;
726f6388 5639{
d233b485
CR
5640 if (dev_fd_list[i])
5641 {
5642 dev_fd_list[i] = 0;
5643 nfds--;
5644 }
726f6388
JA
5645}
5646
d233b485
CR
5647void
5648clear_fifo_list ()
726f6388 5649{
d233b485 5650 register int i;
726f6388 5651
d233b485
CR
5652 if (nfds == 0)
5653 return;
f73dda09 5654
d233b485
CR
5655 for (i = 0; nfds && i < totfds; i++)
5656 clear_fifo (i);
726f6388 5657
d233b485 5658 nfds = 0;
a0c0a00f
CR
5659}
5660
9e49d343 5661void *
495aee44
CR
5662copy_fifo_list (sizep)
5663 int *sizep;
5664{
9e49d343 5665 void *ret;
d233b485
CR
5666
5667 if (nfds == 0 || totfds == 0)
5668 {
5669 if (sizep)
5670 *sizep = 0;
9e49d343 5671 return (void *)NULL;
d233b485
CR
5672 }
5673
495aee44 5674 if (sizep)
d233b485 5675 *sizep = totfds;
9e49d343 5676 ret = xmalloc (totfds * sizeof (pid_t));
d233b485 5677 return (memcpy (ret, dev_fd_list, totfds * sizeof (pid_t)));
495aee44
CR
5678}
5679
726f6388 5680static void
d233b485
CR
5681add_fifo_list (fd)
5682 int fd;
726f6388 5683{
d233b485 5684 if (dev_fd_list == 0 || fd >= totfds)
726f6388 5685 {
d233b485
CR
5686 int ofds;
5687
5688 ofds = totfds;
5689 totfds = getdtablesize ();
5690 if (totfds < 0 || totfds > 256)
5691 totfds = 256;
5692 if (fd >= totfds)
5693 totfds = fd + 2;
5694
5695 dev_fd_list = (pid_t *)xrealloc (dev_fd_list, totfds * sizeof (dev_fd_list[0]));
5696 /* XXX - might need a loop for this */
5697 memset (dev_fd_list + ofds, '\0', (totfds - ofds) * sizeof (pid_t));
726f6388
JA
5698 }
5699
d233b485
CR
5700 dev_fd_list[fd] = 1; /* marker; updated later */
5701 nfds++;
5702}
5703
5704int
5705fifos_pending ()
5706{
5707 return 0; /* used for cleanup; not needed with /dev/fd */
5708}
5709
5710int
5711num_fifos ()
5712{
5713 return nfds;
726f6388
JA
5714}
5715
495aee44 5716void
d233b485
CR
5717unlink_fifo (fd)
5718 int fd;
495aee44 5719{
d233b485 5720 if (dev_fd_list[fd])
495aee44 5721 {
d233b485
CR
5722 close (fd);
5723 dev_fd_list[fd] = 0;
5724 nfds--;
495aee44
CR
5725 }
5726}
5727
726f6388
JA
5728void
5729unlink_fifo_list ()
5730{
d233b485 5731 register int i;
f73dda09 5732
d233b485 5733 if (nfds == 0)
726f6388
JA
5734 return;
5735
d233b485
CR
5736 for (i = totfds-1; nfds && i >= 0; i--)
5737 unlink_fifo (i);
f73dda09 5738
d233b485 5739 nfds = 0;
726f6388
JA
5740}
5741
3eb0018e
CR
5742void
5743unlink_all_fifos ()
5744{
5745 unlink_fifo_list ();
5746}
5747
d233b485
CR
5748/* Take LIST, which is a snapshot copy of dev_fd_list from some point in
5749 the past, and close all open fds in dev_fd_list that are not marked
5750 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
5751 LSIZE is the number of elements in LIST, in case it's larger than
5752 totfds (size of dev_fd_list). */
495aee44
CR
5753void
5754close_new_fifos (list, lsize)
9e49d343 5755 void *list;
495aee44
CR
5756 int lsize;
5757{
5758 int i;
9e49d343 5759 pid_t *plist;
495aee44
CR
5760
5761 if (list == 0)
5762 {
5763 unlink_fifo_list ();
5764 return;
5765 }
5766
9e49d343
CR
5767 for (plist = (pid_t *)list, i = 0; i < lsize; i++)
5768 if (plist[i] == 0 && i < totfds && dev_fd_list[i])
495aee44
CR
5769 unlink_fifo (i);
5770
d233b485 5771 for (i = lsize; i < totfds; i++)
495aee44
CR
5772 unlink_fifo (i);
5773}
5774
f1be666c 5775int
d233b485
CR
5776find_procsub_child (pid)
5777 pid_t pid;
a0c0a00f 5778{
d233b485 5779 int i;
a0c0a00f
CR
5780
5781 if (nfds == 0)
d233b485 5782 return -1;
726f6388 5783
d233b485
CR
5784 for (i = 0; i < totfds; i++)
5785 if (dev_fd_list[i] == pid)
5786 return i;
f1be666c 5787
d233b485 5788 return -1;
495aee44
CR
5789}
5790
5791void
d233b485
CR
5792set_procsub_status (ind, pid, status)
5793 int ind;
5794 pid_t pid;
5795 int status;
495aee44 5796{
d233b485
CR
5797 if (ind >= 0 && ind < totfds)
5798 dev_fd_list[ind] = (pid_t)-1; /* sentinel */
495aee44
CR
5799}
5800
d233b485
CR
5801/* If we've marked the process for this procsub as dead, close the
5802 associated file descriptor. */
712f80b0
CR
5803static void
5804reap_some_procsubs (max)
5805 int max;
726f6388 5806{
d233b485 5807 int i;
726f6388 5808
712f80b0 5809 for (i = 0; nfds > 0 && i < max; i++)
d233b485
CR
5810 if (dev_fd_list[i] == (pid_t)-1)
5811 unlink_fifo (i);
726f6388
JA
5812}
5813
712f80b0
CR
5814void
5815reap_procsubs ()
5816{
5817 reap_some_procsubs (totfds);
5818}
5819
5820#if 0
5821/* UNUSED */
495aee44 5822void
d233b485 5823wait_procsubs ()
495aee44 5824{
d233b485 5825 int i, r;
495aee44 5826
d233b485 5827 for (i = 0; nfds > 0 && i < totfds; i++)
495aee44 5828 {
d233b485
CR
5829 if (dev_fd_list[i] != (pid_t)-1 && dev_fd_list[i] > 0)
5830 {
712f80b0
CR
5831 r = wait_for (dev_fd_list[i], 0);
5832 save_proc_status (dev_fd_list[i], r);
d233b485
CR
5833 dev_fd_list[i] = (pid_t)-1;
5834 }
495aee44 5835 }
495aee44 5836}
712f80b0 5837#endif
495aee44 5838
726f6388
JA
5839#if defined (NOTDEF)
5840print_dev_fd_list ()
5841{
5842 register int i;
5843
f73dda09 5844 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
726f6388
JA
5845 fflush (stderr);
5846
5847 for (i = 0; i < totfds; i++)
5848 {
5849 if (dev_fd_list[i])
5850 fprintf (stderr, " %d", i);
5851 }
5852 fprintf (stderr, "\n");
5853}
5854#endif /* NOTDEF */
5855
5856static char *
5857make_dev_fd_filename (fd)
5858 int fd;
5859{
f73dda09 5860 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
726f6388 5861
17345e5a 5862 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
bb70624e
JA
5863
5864 strcpy (ret, DEV_FD_PREFIX);
5865 p = inttostr (fd, intbuf, sizeof (intbuf));
5866 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5867
726f6388
JA
5868 add_fifo_list (fd);
5869 return (ret);
5870}
5871
5872#endif /* HAVE_DEV_FD */
5873
5874/* Return a filename that will open a connection to the process defined by
5875 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5876 a filename in /dev/fd corresponding to a descriptor that is one of the
5877 ends of the pipe. If not defined, we use named pipes on systems that have
5878 them. Systems without /dev/fd and named pipes are out of luck.
5879
5880 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5881 use the read end of the pipe and dup that file descriptor to fd 0 in
5882 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5883 writing or use the write end of the pipe in the child, and dup that
5884 file descriptor to fd 1 in the child. The parent does the opposite. */
5885
5886static char *
5887process_substitute (string, open_for_read_in_child)
5888 char *string;
5889 int open_for_read_in_child;
5890{
5891 char *pathname;
d233b485 5892 int fd, result, rc, function_value;
726f6388
JA
5893 pid_t old_pid, pid;
5894#if defined (HAVE_DEV_FD)
5895 int parent_pipe_fd, child_pipe_fd;
5896 int fildes[2];
5897#endif /* HAVE_DEV_FD */
5898#if defined (JOB_CONTROL)
5899 pid_t old_pipeline_pgrp;
ccc6cda3 5900#endif
726f6388 5901
cce855bc 5902 if (!string || !*string || wordexp_only)
726f6388
JA
5903 return ((char *)NULL);
5904
5905#if !defined (HAVE_DEV_FD)
5906 pathname = make_named_pipe ();
5907#else /* HAVE_DEV_FD */
5908 if (pipe (fildes) < 0)
5909 {
a0c0a00f 5910 sys_error ("%s", _("cannot make pipe for process substitution"));
726f6388
JA
5911 return ((char *)NULL);
5912 }
5913 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5914 the pipe in the parent, otherwise the read end. */
5915 parent_pipe_fd = fildes[open_for_read_in_child];
5916 child_pipe_fd = fildes[1 - open_for_read_in_child];
d166f048
JA
5917 /* Move the parent end of the pipe to some high file descriptor, to
5918 avoid clashes with FDs used by the script. */
5919 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5920
726f6388
JA
5921 pathname = make_dev_fd_filename (parent_pipe_fd);
5922#endif /* HAVE_DEV_FD */
5923
3185942a 5924 if (pathname == 0)
726f6388 5925 {
a0c0a00f 5926 sys_error ("%s", _("cannot make pipe for process substitution"));
726f6388
JA
5927 return ((char *)NULL);
5928 }
5929
5930 old_pid = last_made_pid;
5931
5932#if defined (JOB_CONTROL)
5933 old_pipeline_pgrp = pipeline_pgrp;
a0c0a00f
CR
5934 if (pipeline_pgrp == 0 || (subshell_environment & (SUBSHELL_PIPE|SUBSHELL_FORK|SUBSHELL_ASYNC)) == 0)
5935 pipeline_pgrp = shell_pgrp;
ccc6cda3 5936 save_pipeline (1);
ccc6cda3
JA
5937#endif /* JOB_CONTROL */
5938
712f80b0 5939 pid = make_child ((char *)NULL, FORK_ASYNC);
726f6388
JA
5940 if (pid == 0)
5941 {
712f80b0
CR
5942 /* The currently-executing shell is not interactive */
5943 interactive = 0;
5944
ccc6cda3 5945 reset_terminating_signals (); /* XXX */
b80f6443 5946 free_pushed_string_input ();
726f6388 5947 /* Cancel traps, in trap.c. */
495aee44 5948 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
a0c0a00f 5949 QUIT; /* catch any interrupts we got post-fork */
726f6388 5950 setup_async_signals ();
712f80b0
CR
5951 if (open_for_read_in_child == 0)
5952 async_redirect_stdin ();
3eb0018e 5953 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB|SUBSHELL_ASYNC;
a0c0a00f 5954
d233b485
CR
5955 /* We don't inherit the verbose option for command substitutions now, so
5956 let's try it for process substitutions. */
5957 change_flag ('v', FLAG_OFF);
5958
a0c0a00f
CR
5959 /* if we're expanding a redirection, we shouldn't have access to the
5960 temporary environment, but commands in the subshell should have
5961 access to their own temporary environment. */
5962 if (expanding_redir)
5963 flush_temporary_env ();
726f6388 5964 }
ccc6cda3
JA
5965
5966#if defined (JOB_CONTROL)
726f6388
JA
5967 set_sigchld_handler ();
5968 stop_making_children ();
3185942a 5969 /* XXX - should we only do this in the parent? (as in command subst) */
726f6388 5970 pipeline_pgrp = old_pipeline_pgrp;
a0c0a00f
CR
5971#else
5972 stop_making_children ();
ccc6cda3 5973#endif /* JOB_CONTROL */
726f6388
JA
5974
5975 if (pid < 0)
5976 {
a0c0a00f 5977 sys_error ("%s", _("cannot make child for process substitution"));
726f6388
JA
5978 free (pathname);
5979#if defined (HAVE_DEV_FD)
5980 close (parent_pipe_fd);
5981 close (child_pipe_fd);
5982#endif /* HAVE_DEV_FD */
d233b485
CR
5983#if defined (JOB_CONTROL)
5984 restore_pipeline (1);
5985#endif
726f6388
JA
5986 return ((char *)NULL);
5987 }
5988
5989 if (pid > 0)
5990 {
ccc6cda3 5991#if defined (JOB_CONTROL)
a0c0a00f 5992 last_procsub_child = restore_pipeline (0);
712f80b0
CR
5993 /* We assume that last_procsub_child->next == last_procsub_child because
5994 of how jobs.c:add_process() works. */
5995 last_procsub_child->next = 0;
5996 procsub_add (last_procsub_child);
ccc6cda3
JA
5997#endif
5998
d233b485
CR
5999#if defined (HAVE_DEV_FD)
6000 dev_fd_list[parent_pipe_fd] = pid;
6001#else
f73dda09
JA
6002 fifo_list[nfifo-1].proc = pid;
6003#endif
6004
726f6388
JA
6005 last_made_pid = old_pid;
6006
6007#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6008 close_pgrp_pipe ();
6009#endif /* JOB_CONTROL && PGRP_PIPE */
6010
6011#if defined (HAVE_DEV_FD)
6012 close (child_pipe_fd);
6013#endif /* HAVE_DEV_FD */
6014
6015 return (pathname);
6016 }
6017
6018 set_sigint_handler ();
6019
6020#if defined (JOB_CONTROL)
d233b485 6021 /* make sure we don't have any job control */
726f6388 6022 set_job_control (0);
d233b485 6023
712f80b0
CR
6024 /* Clear out any existing list of process substitutions */
6025 procsub_clear ();
6026
d233b485
CR
6027 /* The idea is that we want all the jobs we start from an async process
6028 substitution to be in the same process group, but not the same pgrp
6029 as our parent shell, since we don't want to affect our parent shell's
6030 jobs if we get a SIGHUP and end up calling hangup_all_jobs, for example.
6031 If pipeline_pgrp != shell_pgrp, we assume that there is a job control
6032 shell somewhere in our parent process chain (since make_child initializes
6033 pipeline_pgrp to shell_pgrp if job_control == 0). What we do in this
6034 case is to set pipeline_pgrp to our PID, so all jobs started by this
6035 process have that same pgrp and we are basically the process group leader.
6036 This should not have negative effects on child processes surviving
6037 after we exit, since we wait for the children we create, but that is
6038 something to watch for. */
6039
6040 if (pipeline_pgrp != shell_pgrp)
6041 pipeline_pgrp = getpid ();
726f6388
JA
6042#endif /* JOB_CONTROL */
6043
6044#if !defined (HAVE_DEV_FD)
6045 /* Open the named pipe in the child. */
ac50fbac 6046 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
726f6388
JA
6047 if (fd < 0)
6048 {
b80f6443
JA
6049 /* Two separate strings for ease of translation. */
6050 if (open_for_read_in_child)
6051 sys_error (_("cannot open named pipe %s for reading"), pathname);
6052 else
6053 sys_error (_("cannot open named pipe %s for writing"), pathname);
6054
726f6388
JA
6055 exit (127);
6056 }
bb70624e
JA
6057 if (open_for_read_in_child)
6058 {
28ef6c31 6059 if (sh_unset_nodelay_mode (fd) < 0)
bb70624e 6060 {
3185942a 6061 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
bb70624e
JA
6062 exit (127);
6063 }
6064 }
726f6388
JA
6065#else /* HAVE_DEV_FD */
6066 fd = child_pipe_fd;
6067#endif /* HAVE_DEV_FD */
6068
a0c0a00f
CR
6069 /* Discard buffered stdio output before replacing the underlying file
6070 descriptor. */
6071 if (open_for_read_in_child == 0)
6072 fpurge (stdout);
6073
726f6388
JA
6074 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
6075 {
b80f6443 6076 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
ccc6cda3 6077 open_for_read_in_child ? 0 : 1);
726f6388
JA
6078 exit (127);
6079 }
6080
f73dda09
JA
6081 if (fd != (open_for_read_in_child ? 0 : 1))
6082 close (fd);
726f6388
JA
6083
6084 /* Need to close any files that this process has open to pipes inherited
6085 from its parent. */
6086 if (current_fds_to_close)
6087 {
6088 close_fd_bitmap (current_fds_to_close);
6089 current_fds_to_close = (struct fd_bitmap *)NULL;
6090 }
6091
6092#if defined (HAVE_DEV_FD)
6093 /* Make sure we close the parent's end of the pipe and clear the slot
6094 in the fd list so it is not closed later, if reallocated by, for
6095 instance, pipe(2). */
6096 close (parent_pipe_fd);
6097 dev_fd_list[parent_pipe_fd] = 0;
6098#endif /* HAVE_DEV_FD */
6099
8dea6e87 6100 /* subshells shouldn't have this flag, which controls using the temporary
a0c0a00f
CR
6101 environment for variable lookups. We have already flushed the temporary
6102 environment above in the case we're expanding a redirection, so processes
6103 executed by this command need to be able to set it independently of their
6104 parent. */
8dea6e87
CR
6105 expanding_redir = 0;
6106
9cce630e
CR
6107 remove_quoted_escapes (string);
6108
d233b485
CR
6109 /* Give process substitution a place to jump back to on failure,
6110 so we don't go back up to main (). */
6111 result = setjmp_nosigs (top_level);
6112
6113 /* If we're running a process substitution inside a shell function,
6114 trap `return' so we don't return from the function in the subshell
6115 and go off to never-never land. */
6116 if (result == 0 && return_catch_flag)
6117 function_value = setjmp_nosigs (return_catch);
6118 else
6119 function_value = 0;
6120
6121 if (result == ERREXIT)
6122 rc = last_command_exit_value;
6123 else if (result == EXITPROG)
6124 rc = last_command_exit_value;
6125 else if (result)
6126 rc = EXECUTION_FAILURE;
6127 else if (function_value)
6128 rc = return_catch_value;
6129 else
6130 {
6131 subshell_level++;
6132 rc = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
6133 /* leave subshell level intact for any exit trap */
6134 }
726f6388
JA
6135
6136#if !defined (HAVE_DEV_FD)
6137 /* Make sure we close the named pipe in the child before we exit. */
6138 close (open_for_read_in_child ? 0 : 1);
6139#endif /* !HAVE_DEV_FD */
6140
d233b485
CR
6141 last_command_exit_value = rc;
6142 rc = run_exit_trap ();
6143 exit (rc);
726f6388
JA
6144 /*NOTREACHED*/
6145}
6146#endif /* PROCESS_SUBSTITUTION */
6147
cce855bc
JA
6148/***********************************/
6149/* */
6150/* Command Substitution */
6151/* */
6152/***********************************/
6153
d166f048 6154static char *
d233b485
CR
6155read_comsub (fd, quoted, flags, rflag)
6156 int fd, quoted, flags;
3185942a 6157 int *rflag;
d166f048 6158{
712f80b0 6159 char *istring, buf[512], *bufp;
d233b485 6160 int istring_index, c, tflag, skip_ctlesc, skip_ctlnul;
712f80b0 6161 int mb_cur_max;
d233b485 6162 size_t istring_size;
f73dda09 6163 ssize_t bufn;
280bd77d 6164 int nullbyte;
712f80b0
CR
6165#if defined (HANDLE_MULTIBYTE)
6166 mbstate_t ps;
6167 wchar_t wc;
6168 size_t mblen;
6169 int i;
6170#endif
d166f048
JA
6171
6172 istring = (char *)NULL;
3185942a
JA
6173 istring_index = istring_size = bufn = tflag = 0;
6174
d233b485
CR
6175 skip_ctlesc = ifs_cmap[CTLESC];
6176 skip_ctlnul = ifs_cmap[CTLNUL];
d166f048 6177
712f80b0 6178 mb_cur_max = MB_CUR_MAX;
280bd77d
CR
6179 nullbyte = 0;
6180
712f80b0 6181 /* Read the output of the command through the pipe. */
d166f048
JA
6182 while (1)
6183 {
6184 if (fd < 0)
28ef6c31 6185 break;
d166f048
JA
6186 if (--bufn <= 0)
6187 {
bb70624e 6188 bufn = zread (fd, buf, sizeof (buf));
d166f048
JA
6189 if (bufn <= 0)
6190 break;
6191 bufp = buf;
6192 }
6193 c = *bufp++;
6194
28ef6c31
JA
6195 if (c == 0)
6196 {
a0c0a00f 6197#if 1
280bd77d
CR
6198 if (nullbyte == 0)
6199 {
6200 internal_warning ("%s", _("command substitution: ignored null byte in input"));
6201 nullbyte = 1;
6202 }
28ef6c31
JA
6203#endif
6204 continue;
6205 }
6206
d166f048 6207 /* Add the character to ISTRING, possibly after resizing it. */
712f80b0 6208 RESIZE_MALLOCED_BUFFER (istring, istring_index, mb_cur_max+1, istring_size, 512);
d166f048 6209
f1be666c
JA
6210 /* This is essentially quote_string inline */
6211 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
6212 istring[istring_index++] = CTLESC;
d233b485
CR
6213 else if ((flags & PF_ASSIGNRHS) && skip_ctlesc && c == CTLESC)
6214 istring[istring_index++] = CTLESC;
f1be666c
JA
6215 /* Escape CTLESC and CTLNUL in the output to protect those characters
6216 from the rest of the word expansions (word splitting and globbing.)
6217 This is essentially quote_escapes inline. */
3185942a 6218 else if (skip_ctlesc == 0 && c == CTLESC)
d233b485 6219 istring[istring_index++] = CTLESC;
3185942a 6220 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
d166f048
JA
6221 istring[istring_index++] = CTLESC;
6222
712f80b0
CR
6223#if defined (HANDLE_MULTIBYTE)
6224 if ((locale_utf8locale && (c & 0x80)) ||
6225 (locale_utf8locale == 0 && mb_cur_max > 1 && (unsigned char)c > 127))
6226 {
6227 /* read a multibyte character from buf */
6228 /* punt on the hard case for now */
6229 memset (&ps, '\0', sizeof (mbstate_t));
6230 mblen = mbrtowc (&wc, bufp-1, bufn+1, &ps);
6231 if (MB_INVALIDCH (mblen) || mblen == 0 || mblen == 1)
6232 istring[istring_index++] = c;
6233 else
6234 {
6235 istring[istring_index++] = c;
6236 for (i = 0; i < mblen-1; i++)
6237 istring[istring_index++] = *bufp++;
6238 bufn -= mblen - 1;
6239 }
6240 continue;
28ef6c31
JA
6241 }
6242#endif
712f80b0
CR
6243
6244 istring[istring_index++] = c;
d166f048
JA
6245 }
6246
6247 if (istring)
6248 istring[istring_index] = '\0';
6249
6250 /* If we read no output, just return now and save ourselves some
6251 trouble. */
6252 if (istring_index == 0)
6253 {
6254 FREE (istring);
3185942a
JA
6255 if (rflag)
6256 *rflag = tflag;
d166f048
JA
6257 return (char *)NULL;
6258 }
6259
6260 /* Strip trailing newlines from the output of the command. */
6261 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6262 {
6263 while (istring_index > 0)
6264 {
6265 if (istring[istring_index - 1] == '\n')
6266 {
6267 --istring_index;
6268
6269 /* If the newline was quoted, remove the quoting char. */
6270 if (istring[istring_index - 1] == CTLESC)
6271 --istring_index;
6272 }
6273 else
6274 break;
6275 }
6276 istring[istring_index] = '\0';
6277 }
6278 else
6279 strip_trailing (istring, istring_index - 1, 1);
6280
3185942a
JA
6281 if (rflag)
6282 *rflag = tflag;
d166f048
JA
6283 return istring;
6284}
6285
3185942a
JA
6286/* Perform command substitution on STRING. This returns a WORD_DESC * with the
6287 contained string possibly quoted. */
6288WORD_DESC *
d233b485 6289command_substitute (string, quoted, flags)
726f6388
JA
6290 char *string;
6291 int quoted;
d233b485 6292 int flags;
726f6388 6293{
95732b49 6294 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
a0c0a00f 6295 char *istring, *s;
712f80b0 6296 int result, fildes[2], function_value, pflags, rc, tflag, fork_flags;
3185942a 6297 WORD_DESC *ret;
712f80b0 6298 sigset_t set, oset;
726f6388 6299
ccc6cda3 6300 istring = (char *)NULL;
726f6388
JA
6301
6302 /* Don't fork () if there is no need to. In the case of no command to
6303 run, just return NULL. */
a0c0a00f
CR
6304#if 1
6305 for (s = string; s && *s && (shellblank (*s) || *s == '\n'); s++)
6306 ;
6307 if (s == 0 || *s == 0)
6308 return ((WORD_DESC *)NULL);
6309#else
726f6388 6310 if (!string || !*string || (string[0] == '\n' && !string[1]))
3185942a 6311 return ((WORD_DESC *)NULL);
a0c0a00f 6312#endif
726f6388 6313
cce855bc
JA
6314 if (wordexp_only && read_but_dont_execute)
6315 {
0001803f 6316 last_command_exit_value = EX_WEXPCOMSUB;
cce855bc
JA
6317 jump_to_top_level (EXITPROG);
6318 }
6319
bb70624e
JA
6320 /* We're making the assumption here that the command substitution will
6321 eventually run a command from the file system. Since we'll run
6322 maybe_make_export_env in this subshell before executing that command,
6323 the parent shell and any other shells it starts will have to remake
6324 the environment. If we make it before we fork, other shells won't
6325 have to. Don't bother if we have any temporary variable assignments,
6326 though, because the export environment will be remade after this
6327 command completes anyway, but do it if all the words to be expanded
6328 are variable assignments. */
6329 if (subst_assign_varlist == 0 || garglist == 0)
6330 maybe_make_export_env (); /* XXX */
6331
b80f6443 6332 /* Flags to pass to parse_and_execute() */
0001803f 6333 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
b80f6443 6334
d233b485
CR
6335 old_pid = last_made_pid;
6336
726f6388
JA
6337 /* Pipe the output of executing STRING into the current shell. */
6338 if (pipe (fildes) < 0)
6339 {
a0c0a00f 6340 sys_error ("%s", _("cannot make pipe for command substitution"));
726f6388
JA
6341 goto error_exit;
6342 }
6343
726f6388 6344#if defined (JOB_CONTROL)
ccc6cda3 6345 old_pipeline_pgrp = pipeline_pgrp;
28ef6c31
JA
6346 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
6347 if ((subshell_environment & SUBSHELL_PIPE) == 0)
6348 pipeline_pgrp = shell_pgrp;
ccc6cda3 6349 cleanup_the_pipeline ();
95732b49 6350#endif /* JOB_CONTROL */
726f6388 6351
95732b49 6352 old_async_pid = last_asynchronous_pid;
712f80b0
CR
6353 fork_flags = (subshell_environment&SUBSHELL_ASYNC) ? FORK_ASYNC : 0;
6354 pid = make_child ((char *)NULL, fork_flags|FORK_NOTERM);
95732b49
JA
6355 last_asynchronous_pid = old_async_pid;
6356
726f6388 6357 if (pid == 0)
495aee44
CR
6358 {
6359 /* Reset the signal handlers in the child, but don't free the
6360 trap strings. Set a flag noting that we have to free the
6361 trap strings if we run trap to change a signal disposition. */
6362 reset_signal_handlers ();
a0c0a00f
CR
6363 if (ISINTERRUPT)
6364 {
6365 kill (getpid (), SIGINT);
6366 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6367 }
6368 QUIT; /* catch any interrupts we got post-fork */
495aee44
CR
6369 subshell_environment |= SUBSHELL_RESETTRAP;
6370 }
ccc6cda3
JA
6371
6372#if defined (JOB_CONTROL)
3185942a 6373 /* XXX DO THIS ONLY IN PARENT ? XXX */
ccc6cda3
JA
6374 set_sigchld_handler ();
6375 stop_making_children ();
f1be666c
JA
6376 if (pid != 0)
6377 pipeline_pgrp = old_pipeline_pgrp;
f73dda09
JA
6378#else
6379 stop_making_children ();
ccc6cda3 6380#endif /* JOB_CONTROL */
726f6388
JA
6381
6382 if (pid < 0)
6383 {
b80f6443 6384 sys_error (_("cannot make child for command substitution"));
726f6388
JA
6385 error_exit:
6386
ac50fbac
CR
6387 last_made_pid = old_pid;
6388
726f6388
JA
6389 FREE (istring);
6390 close (fildes[0]);
6391 close (fildes[1]);
3185942a 6392 return ((WORD_DESC *)NULL);
726f6388
JA
6393 }
6394
6395 if (pid == 0)
6396 {
a0c0a00f
CR
6397 /* The currently executing shell is not interactive. */
6398 interactive = 0;
6399
726f6388 6400 set_sigint_handler (); /* XXX */
28ef6c31 6401
b80f6443
JA
6402 free_pushed_string_input ();
6403
a0c0a00f
CR
6404 /* Discard buffered stdio output before replacing the underlying file
6405 descriptor. */
6406 fpurge (stdout);
6407
726f6388
JA
6408 if (dup2 (fildes[1], 1) < 0)
6409 {
a0c0a00f 6410 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
726f6388
JA
6411 exit (EXECUTION_FAILURE);
6412 }
6413
6414 /* If standard output is closed in the parent shell
6415 (such as after `exec >&-'), file descriptor 1 will be
6416 the lowest available file descriptor, and end up in
6417 fildes[0]. This can happen for stdin and stderr as well,
6418 but stdout is more important -- it will cause no output
6419 to be generated from this command. */
6420 if ((fildes[1] != fileno (stdin)) &&
6421 (fildes[1] != fileno (stdout)) &&
6422 (fildes[1] != fileno (stderr)))
6423 close (fildes[1]);
6424
6425 if ((fildes[0] != fileno (stdin)) &&
6426 (fildes[0] != fileno (stdout)) &&
6427 (fildes[0] != fileno (stderr)))
6428 close (fildes[0]);
6429
495aee44
CR
6430#ifdef __CYGWIN__
6431 /* Let stdio know the fd may have changed from text to binary mode, and
6432 make sure to preserve stdout line buffering. */
6433 freopen (NULL, "w", stdout);
6434 sh_setlinebuf (stdout);
6435#endif /* __CYGWIN__ */
6436
ccc6cda3 6437 /* This is a subshell environment. */
28ef6c31 6438 subshell_environment |= SUBSHELL_COMSUB;
ccc6cda3 6439
a0c0a00f
CR
6440 /* Many shells do not appear to inherit the -v option for command
6441 substitutions. */
6442 change_flag ('v', FLAG_OFF);
6443
6444 /* When inherit_errexit option is not enabled, command substitution does
6445 not inherit the -e flag. It is enabled when Posix mode is enabled */
6446 if (inherit_errexit == 0)
ac50fbac
CR
6447 {
6448 builtin_ignoring_errexit = 0;
6449 change_flag ('e', FLAG_OFF);
ac50fbac 6450 }
a0c0a00f
CR
6451 set_shellopts ();
6452
6453 /* If we are expanding a redirection, we can dispose of any temporary
6454 environment we received, since redirections are not supposed to have
6455 access to the temporary environment. We will have to see whether this
6456 affects temporary environments supplied to `eval', but the temporary
6457 environment gets copied to builtin_env at some point. */
6458 if (expanding_redir)
6459 {
6460 flush_temporary_env ();
6461 expanding_redir = 0;
6462 }
726f6388
JA
6463
6464 remove_quoted_escapes (string);
6465
ccc6cda3 6466 startup_state = 2; /* see if we can avoid a fork */
d233b485
CR
6467 parse_and_execute_level = 0;
6468
726f6388
JA
6469 /* Give command substitution a place to jump back to on failure,
6470 so we don't go back up to main (). */
ac50fbac 6471 result = setjmp_nosigs (top_level);
726f6388 6472
bb70624e
JA
6473 /* If we're running a command substitution inside a shell function,
6474 trap `return' so we don't return from the function in the subshell
6475 and go off to never-never land. */
6476 if (result == 0 && return_catch_flag)
ac50fbac 6477 function_value = setjmp_nosigs (return_catch);
bb70624e
JA
6478 else
6479 function_value = 0;
6480
b80f6443
JA
6481 if (result == ERREXIT)
6482 rc = last_command_exit_value;
6483 else if (result == EXITPROG)
6484 rc = last_command_exit_value;
726f6388 6485 else if (result)
b80f6443 6486 rc = EXECUTION_FAILURE;
bb70624e 6487 else if (function_value)
b80f6443 6488 rc = return_catch_value;
726f6388 6489 else
b80f6443
JA
6490 {
6491 subshell_level++;
6492 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
d233b485 6493 /* leave subshell level intact for any exit trap */
b80f6443
JA
6494 }
6495
6496 last_command_exit_value = rc;
6497 rc = run_exit_trap ();
f1be666c
JA
6498#if defined (PROCESS_SUBSTITUTION)
6499 unlink_fifo_list ();
6500#endif
b80f6443 6501 exit (rc);
726f6388
JA
6502 }
6503 else
6504 {
712f80b0
CR
6505 int dummyfd;
6506
726f6388
JA
6507#if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6508 close_pgrp_pipe ();
6509#endif /* JOB_CONTROL && PGRP_PIPE */
6510
6511 close (fildes[1]);
6512
712f80b0
CR
6513 begin_unwind_frame ("read-comsub");
6514 dummyfd = fildes[0];
6515 add_unwind_protect (close, dummyfd);
6516
6517 /* Block SIGINT while we're reading from the pipe. If the child
6518 process gets a SIGINT, it will either handle it or die, and the
6519 read will return. */
6520 BLOCK_SIGNAL (SIGINT, set, oset);
3185942a 6521 tflag = 0;
d233b485 6522 istring = read_comsub (fildes[0], quoted, flags, &tflag);
ccc6cda3 6523
726f6388 6524 close (fildes[0]);
712f80b0
CR
6525 discard_unwind_frame ("read-comsub");
6526 UNBLOCK_SIGNAL (oset);
726f6388 6527
b72432fd 6528 current_command_subst_pid = pid;
712f80b0 6529 last_command_exit_value = wait_for (pid, JWAIT_NOTERM);
726f6388
JA
6530 last_command_subst_pid = pid;
6531 last_made_pid = old_pid;
6532
6533#if defined (JOB_CONTROL)
6534 /* If last_command_exit_value > 128, then the substituted command
6535 was terminated by a signal. If that signal was SIGINT, then send
6536 SIGINT to ourselves. This will break out of loops, for instance. */
b80f6443 6537 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
726f6388 6538 kill (getpid (), SIGINT);
726f6388
JA
6539#endif /* JOB_CONTROL */
6540
3185942a
JA
6541 ret = alloc_word_desc ();
6542 ret->word = istring;
6543 ret->flags = tflag;
6544
6545 return ret;
726f6388
JA
6546 }
6547}
6548
6549/********************************************************
6550 * *
6551 * Utility functions for parameter expansion *
6552 * *
6553 ********************************************************/
6554
ccc6cda3 6555#if defined (ARRAY_VARS)
ccc6cda3 6556
f73dda09 6557static arrayind_t
ccc6cda3
JA
6558array_length_reference (s)
6559 char *s;
6560{
f73dda09
JA
6561 int len;
6562 arrayind_t ind;
3185942a 6563 char *akey;
f73dda09 6564 char *t, c;
ccc6cda3 6565 ARRAY *array;
495aee44 6566 HASH_TABLE *h;
ccc6cda3
JA
6567 SHELL_VAR *var;
6568
d233b485 6569 var = array_variable_part (s, 0, &t, &len);
726f6388 6570
ccc6cda3
JA
6571 /* If unbound variables should generate an error, report one and return
6572 failure. */
ac50fbac 6573 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
726f6388 6574 {
f73dda09 6575 c = *--t;
ccc6cda3 6576 *t = '\0';
712f80b0 6577 set_exit_status (EXECUTION_FAILURE);
7117c2d2 6578 err_unboundvar (s);
f73dda09 6579 *t = c;
ccc6cda3 6580 return (-1);
726f6388 6581 }
ac50fbac 6582 else if (var == 0 || invisible_p (var))
ccc6cda3 6583 return 0;
726f6388 6584
28ef6c31
JA
6585 /* We support a couple of expansions for variables that are not arrays.
6586 We'll return the length of the value for v[0], and 1 for v[@] or
6587 v[*]. Return 0 for everything else. */
6588
6589 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
495aee44 6590 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
726f6388 6591
d233b485 6592 if (ALL_ELEMENT_SUB (t[0]) && t[1] == RBRACK)
ccc6cda3 6593 {
3185942a 6594 if (assoc_p (var))
495aee44 6595 return (h ? assoc_num_elements (h) : 0);
3185942a 6596 else if (array_p (var))
495aee44 6597 return (array ? array_num_elements (array) : 0);
3185942a 6598 else
495aee44 6599 return (var_isset (var) ? 1 : 0);
ccc6cda3 6600 }
ccc6cda3 6601
3185942a
JA
6602 if (assoc_p (var))
6603 {
6604 t[len - 1] = '\0';
6605 akey = expand_assignment_string_to_string (t, 0); /* [ */
d233b485 6606 t[len - 1] = RBRACK;
3185942a
JA
6607 if (akey == 0 || *akey == 0)
6608 {
6609 err_badarraysub (t);
ac50fbac 6610 FREE (akey);
3185942a
JA
6611 return (-1);
6612 }
6613 t = assoc_reference (assoc_cell (var), akey);
ac50fbac 6614 free (akey);
3185942a 6615 }
28ef6c31 6616 else
3185942a 6617 {
d233b485 6618 ind = array_expand_index (var, t, len, 0);
ac50fbac
CR
6619 /* negative subscripts to indexed arrays count back from end */
6620 if (var && array_p (var) && ind < 0)
6621 ind = array_max_index (array_cell (var)) + 1 + ind;
3185942a
JA
6622 if (ind < 0)
6623 {
6624 err_badarraysub (t);
6625 return (-1);
6626 }
6627 if (array_p (var))
6628 t = array_reference (array, ind);
6629 else
6630 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6631 }
28ef6c31 6632
f1be666c 6633 len = MB_STRLEN (t);
ccc6cda3 6634 return (len);
726f6388 6635}
ccc6cda3 6636#endif /* ARRAY_VARS */
726f6388
JA
6637
6638static int
6639valid_brace_expansion_word (name, var_is_special)
6640 char *name;
6641 int var_is_special;
6642{
f73dda09 6643 if (DIGIT (*name) && all_digits (name))
726f6388
JA
6644 return 1;
6645 else if (var_is_special)
6646 return 1;
ccc6cda3 6647#if defined (ARRAY_VARS)
a0c0a00f 6648 else if (valid_array_reference (name, 0))
ccc6cda3
JA
6649 return 1;
6650#endif /* ARRAY_VARS */
726f6388
JA
6651 else if (legal_identifier (name))
6652 return 1;
6653 else
6654 return 0;
6655}
ccc6cda3 6656
b80f6443 6657static int
712f80b0 6658chk_atstar (name, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
b80f6443 6659 char *name;
712f80b0 6660 int quoted, pflags;
b80f6443
JA
6661 int *quoted_dollar_atp, *contains_dollar_at;
6662{
6663 char *temp1;
6664
6665 if (name == 0)
6666 {
6667 if (quoted_dollar_atp)
6668 *quoted_dollar_atp = 0;
6669 if (contains_dollar_at)
6670 *contains_dollar_at = 0;
6671 return 0;
6672 }
6673
6674 /* check for $@ and $* */
6675 if (name[0] == '@' && name[1] == 0)
6676 {
6677 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6678 *quoted_dollar_atp = 1;
6679 if (contains_dollar_at)
6680 *contains_dollar_at = 1;
6681 return 1;
6682 }
6683 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6684 {
712f80b0
CR
6685 /* Need more checks here that parallel what string_list_pos_params and
6686 param_expand do. Check expand_no_split_dollar_star and ??? */
6687 if (contains_dollar_at && expand_no_split_dollar_star == 0)
b80f6443
JA
6688 *contains_dollar_at = 1;
6689 return 1;
6690 }
6691
6692 /* Now check for ${array[@]} and ${array[*]} */
6693#if defined (ARRAY_VARS)
a0c0a00f 6694 else if (valid_array_reference (name, 0))
b80f6443 6695 {
d233b485
CR
6696 temp1 = mbschr (name, LBRACK);
6697 if (temp1 && temp1[1] == '@' && temp1[2] == RBRACK)
b80f6443
JA
6698 {
6699 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6700 *quoted_dollar_atp = 1;
6701 if (contains_dollar_at)
6702 *contains_dollar_at = 1;
6703 return 1;
d233b485 6704 }
b80f6443
JA
6705 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6706 which should result in separate words even when IFS is unset. */
d233b485 6707 if (temp1 && temp1[1] == '*' && temp1[2] == RBRACK && quoted == 0)
b80f6443
JA
6708 {
6709 if (contains_dollar_at)
6710 *contains_dollar_at = 1;
6711 return 1;
6712 }
6713 }
6714#endif
6715 return 0;
6716}
6717
726f6388 6718/* Parameter expand NAME, and return a new string which is the expansion,
d233b485 6719 or NULL if there was no expansion. NAME is as given in ${NAMEcWORD}.
726f6388
JA
6720 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6721 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6722 NAME was found inside of a double-quoted expression. */
95732b49 6723static WORD_DESC *
495aee44 6724parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
726f6388 6725 char *name;
89a92869 6726 int var_is_special, quoted, pflags;
495aee44 6727 arrayind_t *indp;
726f6388 6728{
95732b49 6729 WORD_DESC *ret;
ccc6cda3 6730 char *temp, *tt;
7117c2d2 6731 intmax_t arg_index;
ccc6cda3 6732 SHELL_VAR *var;
f1be666c 6733 int atype, rflags;
495aee44 6734 arrayind_t ind;
726f6388 6735
95732b49
JA
6736 ret = 0;
6737 temp = 0;
f1be666c 6738 rflags = 0;
95732b49 6739
495aee44
CR
6740 if (indp)
6741 *indp = INTMAX_MIN;
6742
95732b49 6743 /* Handle multiple digit arguments, as in ${11}. */
f73dda09 6744 if (legal_number (name, &arg_index))
7117c2d2
JA
6745 {
6746 tt = get_dollar_var_value (arg_index);
b80f6443
JA
6747 if (tt)
6748 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6749 ? quote_string (tt)
6750 : quote_escapes (tt);
6751 else
6752 temp = (char *)NULL;
7117c2d2
JA
6753 FREE (tt);
6754 }
726f6388
JA
6755 else if (var_is_special) /* ${@} */
6756 {
cce855bc 6757 int sindex;
f73dda09 6758 tt = (char *)xmalloc (2 + strlen (name));
cce855bc 6759 tt[sindex = 0] = '$';
726f6388 6760 strcpy (tt + 1, name);
7117c2d2 6761
95732b49 6762 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
89a92869 6763 (int *)NULL, (int *)NULL, pflags);
cce855bc 6764 free (tt);
726f6388 6765 }
ccc6cda3 6766#if defined (ARRAY_VARS)
a0c0a00f 6767 else if (valid_array_reference (name, 0))
ccc6cda3 6768 {
ac50fbac 6769expand_arrayref:
d233b485
CR
6770 var = array_variable_part (name, 0, &tt, (int *)0);
6771 /* These are the cases where word splitting will not be performed */
ac50fbac 6772 if (pflags & PF_ASSIGNRHS)
a0c0a00f 6773 {
d233b485 6774 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == RBRACK)
a0c0a00f
CR
6775 {
6776 /* Only treat as double quoted if array variable */
6777 if (var && (array_p (var) || assoc_p (var)))
a0c0a00f
CR
6778 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6779 else
6780 temp = array_value (name, quoted, 0, &atype, &ind);
6781 }
ac50fbac
CR
6782 else
6783 temp = array_value (name, quoted, 0, &atype, &ind);
a0c0a00f 6784 }
d233b485
CR
6785 /* Posix interp 888 */
6786 else if (pflags & PF_NOSPLIT2)
6787 {
6788 /* Special cases, then general case, for each of A[@], A[*], A[n] */
6789#if defined (HANDLE_MULTIBYTE)
6790 if (tt[0] == '@' && tt[1] == RBRACK && var && quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc[0] != ' ')
6791#else
6792 if (tt[0] == '@' && tt[1] == RBRACK && var && quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc != ' ')
6793#endif
6794 temp = array_value (name, Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6795 else if (tt[0] == '@' && tt[1] == RBRACK)
6796 temp = array_value (name, quoted, 0, &atype, &ind);
6797 else if (tt[0] == '*' && tt[1] == RBRACK && expand_no_split_dollar_star && ifs_is_null)
6798 temp = array_value (name, Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT, 0, &atype, &ind);
6799 else if (tt[0] == '*' && tt[1] == RBRACK)
6800 temp = array_value (name, quoted, 0, &atype, &ind);
6801 else
6802 temp = array_value (name, quoted, 0, &atype, &ind);
6803 }
6804 else if (tt[0] == '*' && tt[1] == RBRACK && expand_no_split_dollar_star && ifs_is_null)
6805 temp = array_value (name, Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT, 0, &atype, &ind);
ac50fbac
CR
6806 else
6807 temp = array_value (name, quoted, 0, &atype, &ind);
7117c2d2 6808 if (atype == 0 && temp)
495aee44
CR
6809 {
6810 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6811 ? quote_string (temp)
6812 : quote_escapes (temp);
6813 rflags |= W_ARRAYIND;
6814 if (indp)
6815 *indp = ind;
6816 }
f1be666c
JA
6817 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6818 rflags |= W_HASQUOTEDNULL;
ccc6cda3
JA
6819 }
6820#endif
6821 else if (var = find_variable (name))
6822 {
7117c2d2 6823 if (var_isset (var) && invisible_p (var) == 0)
28ef6c31 6824 {
ccc6cda3 6825#if defined (ARRAY_VARS)
3eb0018e
CR
6826 /* We avoid a memory leak by saving TT as the memory allocated by
6827 assoc_to_string or array_to_string and leaving it 0 otherwise,
6828 then freeing TT after quoting temp. */
6829 tt = (char *)NULL;
6830 if ((pflags & PF_ALLINDS) && assoc_p (var))
6831 tt = temp = assoc_empty (assoc_cell (var)) ? (char *)NULL : assoc_to_string (assoc_cell (var), " ", quoted);
6832 else if ((pflags & PF_ALLINDS) && array_p (var))
6833 tt = temp = array_empty (array_cell (var)) ? (char *)NULL : array_to_string (array_cell (var), " ", quoted);
6834 else if (assoc_p (var))
3185942a
JA
6835 temp = assoc_reference (assoc_cell (var), "0");
6836 else if (array_p (var))
6837 temp = array_reference (array_cell (var), 0);
6838 else
6839 temp = value_cell (var);
ccc6cda3
JA
6840#else
6841 temp = value_cell (var);
6842#endif
6843
6844 if (temp)
b80f6443
JA
6845 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6846 ? quote_string (temp)
d233b485
CR
6847 : ((pflags & PF_ASSIGNRHS) ? quote_rhs (temp)
6848 : quote_escapes (temp));
3eb0018e 6849 FREE (tt);
28ef6c31 6850 }
ccc6cda3
JA
6851 else
6852 temp = (char *)NULL;
6853 }
a0c0a00f 6854 else if (var = find_variable_last_nameref (name, 0))
ac50fbac
CR
6855 {
6856 temp = nameref_cell (var);
6857#if defined (ARRAY_VARS)
6858 /* Handle expanding nameref whose value is x[n] */
a0c0a00f 6859 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
6860 {
6861 name = temp;
6862 goto expand_arrayref;
6863 }
6864 else
6865#endif
6866 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6867 if (temp && *temp && legal_identifier (temp) == 0)
6868 {
712f80b0 6869 set_exit_status (EXECUTION_FAILURE);
ac50fbac
CR
6870 report_error (_("%s: invalid variable name for name reference"), temp);
6871 temp = &expand_param_error;
6872 }
6873 else
6874 temp = (char *)NULL;
6875 }
726f6388 6876 else
ccc6cda3 6877 temp = (char *)NULL;
726f6388 6878
95732b49
JA
6879 if (ret == 0)
6880 {
6881 ret = alloc_word_desc ();
6882 ret->word = temp;
f1be666c 6883 ret->flags |= rflags;
95732b49
JA
6884 }
6885 return ret;
726f6388
JA
6886}
6887
ac50fbac
CR
6888static char *
6889parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
ccc6cda3 6890 char *name;
ac50fbac 6891 int var_is_special, quoted, find_nameref;
ccc6cda3
JA
6892{
6893 char *temp, *t;
95732b49 6894 WORD_DESC *w;
ac50fbac 6895 SHELL_VAR *v;
d233b485 6896 int pflags, oldex;
ac50fbac 6897
a0c0a00f 6898 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
ac50fbac
CR
6899 nameref_p (v) && (t = nameref_cell (v)) && *t)
6900 return (savestring (t));
ccc6cda3 6901
ac50fbac
CR
6902 /* If var_is_special == 0, and name is not an array reference, this does
6903 more expansion than necessary. It should really look up the variable's
6904 value and not try to expand it. */
d233b485
CR
6905 pflags = PF_IGNUNBOUND;
6906 /* Note that we're not going to be doing word splitting here */
6907 if (var_is_special)
6908 {
6909 pflags |= PF_ASSIGNRHS; /* suppresses word splitting */
6910 oldex = expand_no_split_dollar_star;
6911 expand_no_split_dollar_star = 1;
6912 }
6913 w = parameter_brace_expand_word (name, var_is_special, quoted, pflags, 0);
6914 if (var_is_special)
6915 expand_no_split_dollar_star = oldex;
6916
95732b49 6917 t = w->word;
b80f6443
JA
6918 /* Have to dequote here if necessary */
6919 if (t)
6920 {
d233b485 6921 temp = ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || var_is_special)
b80f6443
JA
6922 ? dequote_string (t)
6923 : dequote_escapes (t);
6924 free (t);
6925 t = temp;
6926 }
95732b49
JA
6927 dispose_word_desc (w);
6928
ac50fbac
CR
6929 return t;
6930}
6931
6932/* Expand an indirect reference to a variable: ${!NAME} expands to the
6933 value of the variable whose name is the value of NAME. */
6934static WORD_DESC *
712f80b0 6935parameter_brace_expand_indir (name, var_is_special, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
ac50fbac 6936 char *name;
712f80b0 6937 int var_is_special, quoted, pflags;
ac50fbac
CR
6938 int *quoted_dollar_atp, *contains_dollar_at;
6939{
d233b485 6940 char *t;
ac50fbac
CR
6941 WORD_DESC *w;
6942 SHELL_VAR *v;
6943
6944 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6945 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6946 bash performs an indirect lookup on foo[0] and expands the result;
6947 ksh93 expands bar[0]. We could do that here -- there are enough usable
6948 primitives to do that -- but do not at this point. */
a0c0a00f 6949 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
ac50fbac
CR
6950 {
6951 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6952 {
6953 w = alloc_word_desc ();
6954 w->word = savestring (t);
6955 w->flags = 0;
6956 return w;
6957 }
6958 }
6959
d233b485
CR
6960 /* An indirect reference to a positional parameter or a special parameter
6961 is ok. Indirect references to array references, as explained above, are
6962 ok (currently). Only references to unset variables are errors at this
6963 point. */
6964 if (legal_identifier (name) && v == 0)
6965 {
6966 report_error (_("%s: invalid indirect expansion"), name);
6967 w = alloc_word_desc ();
6968 w->word = &expand_param_error;
6969 w->flags = 0;
6970 return (w);
6971 }
6972
ac50fbac
CR
6973 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6974
712f80b0 6975 chk_atstar (t, quoted, pflags, quoted_dollar_atp, contains_dollar_at);
d233b485
CR
6976
6977#if defined (ARRAY_VARS)
6978 /* Array references to unset variables are also an error */
6979 if (t == 0 && valid_array_reference (name, 0))
6980 {
6981 v = array_variable_part (name, 0, (char **)0, (int *)0);
6982 if (v == 0)
6983 {
6984 report_error (_("%s: invalid indirect expansion"), name);
6985 w = alloc_word_desc ();
6986 w->word = &expand_param_error;
6987 w->flags = 0;
6988 return (w);
6989 }
6990 else
6991 return (WORD_DESC *)NULL;
6992 }
6993#endif
6994
ccc6cda3 6995 if (t == 0)
95732b49
JA
6996 return (WORD_DESC *)NULL;
6997
a0c0a00f
CR
6998 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
6999 {
d233b485 7000 report_error (_("%s: invalid variable name"), t);
a0c0a00f
CR
7001 free (t);
7002 w = alloc_word_desc ();
7003 w->word = &expand_param_error;
7004 w->flags = 0;
7005 return (w);
7006 }
7007
712f80b0 7008 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, pflags, 0);
ccc6cda3 7009 free (t);
95732b49
JA
7010
7011 return w;
ccc6cda3
JA
7012}
7013
726f6388
JA
7014/* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
7015 depending on the value of C, the separating character. C can be one of
ccc6cda3
JA
7016 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
7017 between double quotes. */
95732b49 7018static WORD_DESC *
d233b485 7019parameter_brace_expand_rhs (name, value, op, quoted, pflags, qdollaratp, hasdollarat)
726f6388 7020 char *name, *value;
d233b485 7021 int op, quoted, pflags, *qdollaratp, *hasdollarat;
726f6388 7022{
95732b49 7023 WORD_DESC *w;
712f80b0 7024 WORD_LIST *l, *tl;
a0c0a00f
CR
7025 char *t, *t1, *temp, *vname;
7026 int l_hasdollat, sindex;
d233b485 7027 SHELL_VAR *v;
726f6388 7028
a0c0a00f 7029/*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
ccc6cda3
JA
7030 /* If the entire expression is between double quotes, we want to treat
7031 the value as a double-quoted string, with the exception that we strip
3185942a 7032 embedded unescaped double quotes (for sh backwards compatibility). */
95732b49 7033 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
726f6388 7034 {
a0c0a00f
CR
7035 sindex = 0;
7036 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
726f6388 7037 }
95732b49
JA
7038 else
7039 temp = value;
ccc6cda3 7040
95732b49 7041 w = alloc_word_desc ();
a0c0a00f 7042 l_hasdollat = 0;
d233b485 7043 l = *temp ? expand_string_for_rhs (temp, quoted, op, pflags, &l_hasdollat, (int *)NULL)
ccc6cda3
JA
7044 : (WORD_LIST *)0;
7045 if (hasdollarat)
a0c0a00f 7046 *hasdollarat = l_hasdollat || (l && l->next);
95732b49
JA
7047 if (temp != value)
7048 free (temp);
712f80b0
CR
7049
7050 /* list_string takes multiple CTLNULs and turns them into an empty word
7051 with W_SAWQUOTEDNULL set. Turn it back into a single CTLNUL for the
7052 rest of this function and the caller. */
7053 for (tl = l; tl; tl = tl->next)
7054 {
7055 if (tl->word && (tl->word->word == 0 || tl->word->word[0] == 0) &&
7056 (tl->word->flags | W_SAWQUOTEDNULL))
7057 {
7058 t = make_quoted_char ('\0');
7059 FREE (tl->word->word);
7060 tl->word->word = t;
7061 tl->word->flags |= W_QUOTED|W_HASQUOTEDNULL;
7062 tl->word->flags &= ~W_SAWQUOTEDNULL;
7063 }
7064 }
7065
726f6388
JA
7066 if (l)
7067 {
a0c0a00f
CR
7068 /* If l->next is not null, we know that TEMP contained "$@", since that
7069 is the only expansion that creates more than one word. */
7070 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
7071 {
7072/*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
7073 *qdollaratp = 1;
7074 }
7075
ccc6cda3 7076 /* The expansion of TEMP returned something. We need to treat things
a0c0a00f
CR
7077 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
7078 the individual words have already been quoted. We need to turn them
b80f6443
JA
7079 into a string with the words separated by the first character of
7080 $IFS without any additional quoting, so string_list_dollar_at won't
a0c0a00f
CR
7081 do the right thing. If IFS is null, we want "$@" to split into
7082 separate arguments, not be concatenated, so we use string_list_internal
7083 and mark the word to be split on spaces later. We use
7084 string_list_dollar_star for "$@" otherwise. */
7085 if (l->next && ifs_is_null)
7086 {
7087 temp = string_list_internal (l, " ");
7088 w->flags |= W_SPLITSPACE;
7089 }
d233b485
CR
7090 else if (l_hasdollat || l->next)
7091 temp = string_list_dollar_star (l, quoted, 0);
a0c0a00f 7092 else
d233b485
CR
7093 {
7094 temp = string_list (l);
7095 if (temp && (QUOTED_NULL (temp) == 0) && (l->word->flags & W_SAWQUOTEDNULL))
7096 w->flags |= W_SAWQUOTEDNULL; /* XXX */
7097 }
b80f6443 7098
cd110fdf
CR
7099 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
7100 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
7101 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
7102 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7103 (which is more paranoia than anything else), we need to return the
7104 quoted null string and set the flags to indicate it. */
ac50fbac 7105 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
cd110fdf
CR
7106 {
7107 w->flags |= W_HASQUOTEDNULL;
a0c0a00f
CR
7108/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
7109 /* If we return a quoted null with L_HASDOLLARAT, we either have a
7110 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
7111 parameters or a quoted expansion of "$@" with $1 == ''. In either
7112 case, we don't want to enable special handling of $@. */
7113 if (qdollaratp && l_hasdollat)
7114 *qdollaratp = 0;
cd110fdf 7115 }
726f6388
JA
7116 dispose_words (l);
7117 }
a0c0a00f 7118 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
726f6388 7119 {
a0c0a00f
CR
7120 /* Posix interp 221 changed the rules on this. The idea is that
7121 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
7122 when foo and xxx are unset. The problem is that it's not in any
7123 way backwards compatible and few other shells do it. We're eventually
7124 going to try and split the difference (heh) a little bit here. */
7125 /* l_hasdollat == 1 means we saw a quoted dollar at. */
7126
ccc6cda3
JA
7127 /* The brace expansion occurred between double quotes and there was
7128 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
7117c2d2 7129 it does not expand to anything. In this case, we want to return
a0c0a00f 7130 a quoted empty string. Posix interp 888 */
0628567a 7131 temp = make_quoted_char ('\0');
95732b49 7132 w->flags |= W_HASQUOTEDNULL;
a0c0a00f 7133/*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
726f6388
JA
7134 }
7135 else
7136 temp = (char *)NULL;
7137
d233b485 7138 if (op == '-' || op == '+')
95732b49
JA
7139 {
7140 w->word = temp;
7141 return w;
7142 }
726f6388 7143
d233b485
CR
7144 /* op == '=' */
7145 t1 = temp ? dequote_string (temp) : savestring ("");
7146 free (temp);
a0c0a00f
CR
7147
7148 /* bash-4.4/5.0 */
7149 vname = name;
7150 if (*name == '!' &&
7151 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
7152 {
7153 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
7154 if (vname == 0 || *vname == 0)
7155 {
7156 report_error (_("%s: invalid indirect expansion"), name);
7157 free (vname);
d233b485 7158 free (t1);
a0c0a00f
CR
7159 dispose_word (w);
7160 return &expand_wdesc_error;
7161 }
7162 if (legal_identifier (vname) == 0)
7163 {
7164 report_error (_("%s: invalid variable name"), vname);
7165 free (vname);
d233b485 7166 free (t1);
a0c0a00f
CR
7167 dispose_word (w);
7168 return &expand_wdesc_error;
7169 }
7170 }
7171
b80f6443 7172#if defined (ARRAY_VARS)
a0c0a00f 7173 if (valid_array_reference (vname, 0))
d233b485 7174 v = assign_array_element (vname, t1, 0);
b80f6443
JA
7175 else
7176#endif /* ARRAY_VARS */
d233b485
CR
7177 v = bind_variable (vname, t1, 0);
7178
7179 if (v == 0 || readonly_p (v) || noassign_p (v)) /* expansion error */
7180 {
7181 if ((v == 0 || readonly_p (v)) && interactive_shell == 0 && posixly_correct)
7182 {
7183 last_command_exit_value = EXECUTION_FAILURE;
7184 exp_jump_to_top_level (FORCE_EOF);
7185 }
7186 else
7187 {
7188 if (vname != name)
7189 free (vname);
7190 last_command_exit_value = EX_BADUSAGE;
7191 exp_jump_to_top_level (DISCARD);
7192 }
7193 }
a0c0a00f
CR
7194
7195 stupidly_hack_special_variables (vname);
7196
7197 if (vname != name)
7198 free (vname);
95732b49 7199
495aee44 7200 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
495aee44 7201
d233b485
CR
7202 /* If we are double-quoted or if we are not going to be performing word
7203 splitting, we want to quote the value we return appropriately, like
7204 the other expansions this function handles. */
7205 w->word = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (t1) : quote_escapes (t1);
712f80b0
CR
7206 /* If we have something that's non-null, that's not a quoted null string,
7207 and we're not going to be performing word splitting (we know we're not
7208 because the operator is `='), we can forget we saw a quoted null. */
7209 if (w->word && w->word[0] && QUOTED_NULL (w->word) == 0)
7210 w->flags &= ~W_SAWQUOTEDNULL;
d233b485
CR
7211 free (t1);
7212
712f80b0
CR
7213 /* If we convert a null string into a quoted null, make sure the caller
7214 knows it. */
7215 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && QUOTED_NULL (w->word))
7216 w->flags |= W_HASQUOTEDNULL;
7217
95732b49 7218 return w;
726f6388
JA
7219}
7220
7221/* Deal with the right hand side of a ${name:?value} expansion in the case
7222 that NAME is null or not set. If VALUE is non-null it is expanded and
7223 used as the error message to print, otherwise a standard message is
7224 printed. */
7225static void
d233b485 7226parameter_brace_expand_error (name, value, check_null)
726f6388 7227 char *name, *value;
d233b485 7228 int check_null;
726f6388 7229{
ccc6cda3
JA
7230 WORD_LIST *l;
7231 char *temp;
7232
712f80b0 7233 set_exit_status (EXECUTION_FAILURE); /* ensure it's non-zero */
726f6388
JA
7234 if (value && *value)
7235 {
95732b49 7236 l = expand_string (value, 0);
ccc6cda3
JA
7237 temp = string_list (l);
7238 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
7239 FREE (temp);
726f6388
JA
7240 dispose_words (l);
7241 }
d233b485
CR
7242 else if (check_null == 0)
7243 report_error (_("%s: parameter not set"), name);
726f6388 7244 else
b80f6443 7245 report_error (_("%s: parameter null or not set"), name);
726f6388
JA
7246
7247 /* Free the data we have allocated during this expansion, since we
7248 are about to longjmp out. */
7249 free (name);
7250 FREE (value);
7251}
7252
7253/* Return 1 if NAME is something for which parameter_brace_expand_length is
7254 OK to do. */
7255static int
7256valid_length_expression (name)
7257 char *name;
7258{
28ef6c31 7259 return (name[1] == '\0' || /* ${#} */
f73dda09
JA
7260 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
7261 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
ccc6cda3 7262#if defined (ARRAY_VARS)
a0c0a00f 7263 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
ccc6cda3 7264#endif
726f6388
JA
7265 legal_identifier (name + 1)); /* ${#PS1} */
7266}
7267
7268/* Handle the parameter brace expansion that requires us to return the
7269 length of a parameter. */
7117c2d2 7270static intmax_t
726f6388
JA
7271parameter_brace_expand_length (name)
7272 char *name;
7273{
ccc6cda3 7274 char *t, *newname;
7117c2d2 7275 intmax_t number, arg_index;
ccc6cda3 7276 WORD_LIST *list;
ccc6cda3 7277 SHELL_VAR *var;
712f80b0
CR
7278
7279 var = (SHELL_VAR *)NULL;
ccc6cda3
JA
7280
7281 if (name[1] == '\0') /* ${#} */
7282 number = number_of_args ();
a0c0a00f 7283 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
cce855bc 7284 number = number_of_args ();
f73dda09 7285 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
cce855bc
JA
7286 {
7287 /* Take the lengths of some of the shell's special parameters. */
7288 switch (name[1])
7289 {
7290 case '-':
7291 t = which_set_flags ();
7292 break;
7293 case '?':
7294 t = itos (last_command_exit_value);
7295 break;
7296 case '$':
7297 t = itos (dollar_dollar_pid);
7298 break;
7299 case '!':
7300 if (last_asynchronous_pid == NO_PID)
495aee44 7301 t = (char *)NULL; /* XXX - error if set -u set? */
cce855bc 7302 else
f73dda09 7303 t = itos (last_asynchronous_pid);
cce855bc
JA
7304 break;
7305 case '#':
7306 t = itos (number_of_args ());
7307 break;
7308 }
7309 number = STRLEN (t);
7310 FREE (t);
7311 }
ccc6cda3 7312#if defined (ARRAY_VARS)
a0c0a00f 7313 else if (valid_array_reference (name + 1, 0))
ccc6cda3
JA
7314 number = array_length_reference (name + 1);
7315#endif /* ARRAY_VARS */
cce855bc 7316 else
ccc6cda3
JA
7317 {
7318 number = 0;
7319
f73dda09 7320 if (legal_number (name + 1, &arg_index)) /* ${#1} */
ccc6cda3 7321 {
f73dda09 7322 t = get_dollar_var_value (arg_index);
495aee44
CR
7323 if (t == 0 && unbound_vars_is_error)
7324 return INTMAX_MIN;
eb873671 7325 number = MB_STRLEN (t);
ccc6cda3
JA
7326 FREE (t);
7327 }
7328#if defined (ARRAY_VARS)
3185942a 7329 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
ccc6cda3 7330 {
3185942a
JA
7331 if (assoc_p (var))
7332 t = assoc_reference (assoc_cell (var), "0");
7333 else
7334 t = array_reference (array_cell (var), 0);
495aee44
CR
7335 if (t == 0 && unbound_vars_is_error)
7336 return INTMAX_MIN;
eb873671 7337 number = MB_STRLEN (t);
ccc6cda3
JA
7338 }
7339#endif
712f80b0
CR
7340 /* Fast path for the common case of taking the length of a non-dynamic
7341 scalar variable value. */
7342 else if ((var || (var = find_variable (name + 1))) &&
7343 invisible_p (var) == 0 &&
7344 array_p (var) == 0 && assoc_p (var) == 0 &&
7345 var->dynamic_value == 0)
7346 number = value_cell (var) ? MB_STRLEN (value_cell (var)) : 0;
7347 else if (var == 0 && unbound_vars_is_error == 0)
7348 number = 0;
ccc6cda3
JA
7349 else /* ${#PS1} */
7350 {
7351 newname = savestring (name);
7352 newname[0] = '$';
7353 list = expand_string (newname, Q_DOUBLE_QUOTES);
7354 t = list ? string_list (list) : (char *)NULL;
7355 free (newname);
7356 if (list)
7357 dispose_words (list);
7358
495aee44 7359 number = t ? MB_STRLEN (t) : 0;
ccc6cda3
JA
7360 FREE (t);
7361 }
7362 }
ccc6cda3
JA
7363
7364 return (number);
7365}
7366
28ef6c31
JA
7367/* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
7368 so we do some ad-hoc parsing of an arithmetic expression to find
7369 the first DELIM, instead of using strchr(3). Two rules:
7370 1. If the substring contains a `(', read until closing `)'.
7371 2. If the substring contains a `?', read past one `:' for each `?'.
a0c0a00f 7372 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
28ef6c31
JA
7373*/
7374
7375static char *
7376skiparith (substr, delim)
7377 char *substr;
7378 int delim;
7379{
a0c0a00f
CR
7380 int i;
7381 char delims[2];
28ef6c31 7382
a0c0a00f
CR
7383 delims[0] = delim;
7384 delims[1] = '\0';
7117c2d2 7385
a0c0a00f 7386 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
7117c2d2 7387 return (substr + i);
28ef6c31
JA
7388}
7389
ccc6cda3
JA
7390/* Verify and limit the start and end of the desired substring. If
7391 VTYPE == 0, a regular shell variable is being used; if it is 1,
cce855bc 7392 then the positional parameters are being used; if it is 2, then
e8ce775d
JA
7393 VALUE is really a pointer to an array variable that should be used.
7394 Return value is 1 if both values were OK, 0 if there was a problem
7395 with an invalid expression, or -1 if the values were out of range. */
ccc6cda3 7396static int
3185942a
JA
7397verify_substring_values (v, value, substr, vtype, e1p, e2p)
7398 SHELL_VAR *v;
ccc6cda3 7399 char *value, *substr;
f73dda09 7400 int vtype;
7117c2d2 7401 intmax_t *e1p, *e2p;
ccc6cda3 7402{
bb70624e 7403 char *t, *temp1, *temp2;
f73dda09
JA
7404 arrayind_t len;
7405 int expok;
ccc6cda3
JA
7406#if defined (ARRAY_VARS)
7407 ARRAY *a;
3185942a 7408 HASH_TABLE *h;
ccc6cda3
JA
7409#endif
7410
28ef6c31
JA
7411 /* duplicate behavior of strchr(3) */
7412 t = skiparith (substr, ':');
7413 if (*t && *t == ':')
7117c2d2 7414 *t = '\0';
28ef6c31
JA
7415 else
7416 t = (char *)0;
f73dda09 7417
0628567a 7418 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
d233b485 7419 *e1p = evalexp (temp1, 0, &expok); /* XXX - EXP_EXPANDED? */
ccc6cda3 7420 free (temp1);
d166f048
JA
7421 if (expok == 0)
7422 return (0);
ccc6cda3 7423
f73dda09 7424 len = -1; /* paranoia */
ccc6cda3
JA
7425 switch (vtype)
7426 {
7427 case VT_VARIABLE:
d166f048 7428 case VT_ARRAYMEMBER:
eb873671 7429 len = MB_STRLEN (value);
ccc6cda3
JA
7430 break;
7431 case VT_POSPARMS:
7432 len = number_of_args () + 1;
3185942a
JA
7433 if (*e1p == 0)
7434 len++; /* add one arg if counting from $0 */
ccc6cda3
JA
7435 break;
7436#if defined (ARRAY_VARS)
7437 case VT_ARRAYVAR:
eb873671 7438 /* For arrays, the first value deals with array indices. Negative
3185942a
JA
7439 offsets count from one past the array's maximum index. Associative
7440 arrays treat the number of elements as the maximum index. */
7441 if (assoc_p (v))
7442 {
7443 h = assoc_cell (v);
7444 len = assoc_num_elements (h) + (*e1p < 0);
7445 }
7446 else
7447 {
7448 a = (ARRAY *)value;
7449 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7450 }
ccc6cda3
JA
7451 break;
7452#endif
7453 }
7454
f73dda09
JA
7455 if (len == -1) /* paranoia */
7456 return -1;
7457
ccc6cda3
JA
7458 if (*e1p < 0) /* negative offsets count from end */
7459 *e1p += len;
7460
eb873671 7461 if (*e1p > len || *e1p < 0)
e8ce775d 7462 return (-1);
d166f048 7463
b80f6443
JA
7464#if defined (ARRAY_VARS)
7465 /* For arrays, the second offset deals with the number of elements. */
7466 if (vtype == VT_ARRAYVAR)
3185942a 7467 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
b80f6443
JA
7468#endif
7469
ccc6cda3
JA
7470 if (t)
7471 {
7472 t++;
bb70624e 7473 temp2 = savestring (t);
0628567a 7474 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
bb70624e 7475 free (temp2);
ccc6cda3 7476 t[-1] = ':';
d233b485 7477 *e2p = evalexp (temp1, 0, &expok); /* XXX - EXP_EXPANDED? */
ccc6cda3 7478 free (temp1);
d166f048 7479 if (expok == 0)
28ef6c31 7480 return (0);
d233b485
CR
7481
7482 /* Should we allow positional parameter length < 0 to count backwards
7483 from end of positional parameters? */
ac50fbac 7484#if 1
495aee44 7485 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
712f80b0 7486#else /* XXX - TAG: bash-5.1 */
ac50fbac
CR
7487 if (vtype == VT_ARRAYVAR && *e2p < 0)
7488#endif
28ef6c31 7489 {
b80f6443 7490 internal_error (_("%s: substring expression < 0"), t);
ccc6cda3 7491 return (0);
28ef6c31 7492 }
b80f6443
JA
7493#if defined (ARRAY_VARS)
7494 /* In order to deal with sparse arrays, push the intelligence about how
7495 to deal with the number of elements desired down to the array-
7496 specific functions. */
7497 if (vtype != VT_ARRAYVAR)
7498#endif
7499 {
495aee44
CR
7500 if (*e2p < 0)
7501 {
7502 *e2p += len;
7503 if (*e2p < 0 || *e2p < *e1p)
7504 {
7505 internal_error (_("%s: substring expression < 0"), t);
7506 return (0);
7507 }
7508 }
7509 else
7510 *e2p += *e1p; /* want E2 chars starting at E1 */
b80f6443
JA
7511 if (*e2p > len)
7512 *e2p = len;
7513 }
ccc6cda3
JA
7514 }
7515 else
7516 *e2p = len;
7517
7518 return (1);
7519}
7520
ccc6cda3 7521/* Return the type of variable specified by VARNAME (simple variable,
cce855bc 7522 positional param, or array variable). Also return the value specified
7117c2d2 7523 by VARNAME (value of a variable or a reference to an array element).
495aee44
CR
7524 QUOTED is the standard description of quoting state, using Q_* defines.
7525 FLAGS is currently a set of flags to pass to array_value. If IND is
7526 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7527 passed to array_value so the array index is not computed again.
7117c2d2
JA
7528 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7529 characters in the value are quoted with CTLESC and takes appropriate
7530 steps. For convenience, *VALP is set to the dequoted VALUE. */
ccc6cda3 7531static int
495aee44 7532get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
ccc6cda3 7533 char *varname, *value;
495aee44
CR
7534 arrayind_t ind;
7535 int quoted, flags;
ccc6cda3
JA
7536 SHELL_VAR **varp;
7537 char **valp;
7538{
ac50fbac
CR
7539 int vtype, want_indir;
7540 char *temp, *vname;
ccc6cda3 7541 SHELL_VAR *v;
495aee44 7542 arrayind_t lind;
ccc6cda3 7543
ac50fbac
CR
7544 want_indir = *varname == '!' &&
7545 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7546 || VALID_INDIR_PARAM (varname[1]));
7547 if (want_indir)
7548 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
a0c0a00f 7549 /* XXX - what if vname == 0 || *vname == 0 ? */
ac50fbac
CR
7550 else
7551 vname = varname;
a0c0a00f
CR
7552
7553 if (vname == 0)
7554 {
7555 vtype = VT_VARIABLE;
7556 *varp = (SHELL_VAR *)NULL;
7557 *valp = (char *)NULL;
7558 return (vtype);
7559 }
7560
7117c2d2 7561 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
a0c0a00f 7562 vtype = STR_DOLLAR_AT_STAR (vname);
ac50fbac 7563 if (vtype == VT_POSPARMS && vname[0] == '*')
b80f6443 7564 vtype |= VT_STARSUB;
ccc6cda3
JA
7565 *varp = (SHELL_VAR *)NULL;
7566
7567#if defined (ARRAY_VARS)
a0c0a00f 7568 if (valid_array_reference (vname, 0))
ccc6cda3 7569 {
d233b485 7570 v = array_variable_part (vname, 0, &temp, (int *)0);
495aee44
CR
7571 /* If we want to signal array_value to use an already-computed index,
7572 set LIND to that index */
7573 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
ac50fbac
CR
7574 if (v && invisible_p (v))
7575 {
7576 vtype = VT_ARRAYMEMBER;
7577 *varp = (SHELL_VAR *)NULL;
7578 *valp = (char *)NULL;
7579 }
3185942a 7580 if (v && (array_p (v) || assoc_p (v)))
d233b485
CR
7581 {
7582 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == RBRACK)
ccc6cda3 7583 {
ac50fbac 7584 /* Callers have to differentiate between indexed and associative */
ccc6cda3 7585 vtype = VT_ARRAYVAR;
b80f6443
JA
7586 if (temp[0] == '*')
7587 vtype |= VT_STARSUB;
3185942a 7588 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
ccc6cda3
JA
7589 }
7590 else
7591 {
d166f048 7592 vtype = VT_ARRAYMEMBER;
ac50fbac 7593 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
ccc6cda3
JA
7594 }
7595 *varp = v;
7596 }
d233b485
CR
7597 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == RBRACK))
7598 {
7599 vtype = VT_VARIABLE;
7600 *varp = v;
7601 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
712f80b0 7602 *valp = value ? dequote_string (value) : (char *)NULL;
d233b485
CR
7603 else
7604 *valp = value ? dequote_escapes (value) : (char *)NULL;
7605 }
7606 else
7607 {
7608 vtype = VT_ARRAYMEMBER;
7609 *varp = v;
7610 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7611 }
7612 }
7613 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
7614 {
7615 vtype = VT_ARRAYMEMBER;
7616 *varp = v;
7617 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
7618 }
7619 else
7620#endif
7621 {
7622 if (value && vtype == VT_VARIABLE)
7623 {
7624 *varp = find_variable (vname);
7625 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7626 *valp = dequote_string (value);
7627 else
7628 *valp = dequote_escapes (value);
7629 }
7630 else
7631 *valp = value;
7632 }
7633
7634 if (want_indir)
7635 free (vname);
7636
7637 return vtype;
7638}
7639
7640/***********************************************************/
7641/* */
7642/* Functions to perform transformations on variable values */
7643/* */
7644/***********************************************************/
7645
7646static char *
7647string_var_assignment (v, s)
7648 SHELL_VAR *v;
7649 char *s;
7650{
7651 char flags[MAX_ATTRIBUTES], *ret, *val;
7652 int i;
7653
712f80b0 7654 val = (v && (invisible_p (v) || var_isset (v) == 0)) ? (char *)NULL : sh_quote_reusable (s, 0);
d233b485 7655 i = var_attribute_string (v, 0, flags);
712f80b0
CR
7656 if (i == 0 && val == 0)
7657 return (char *)NULL;
7658
7659 ret = (char *)xmalloc (i + STRLEN (val) + strlen (v->name) + 16 + MAX_ATTRIBUTES);
7660 if (i > 0 && val == 0)
7661 sprintf (ret, "declare -%s %s", flags, v->name);
7662 else if (i > 0)
d233b485
CR
7663 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
7664 else
7665 sprintf (ret, "%s=%s", v->name, val);
7666 free (val);
7667 return ret;
7668}
7669
7670#if defined (ARRAY_VARS)
7671static char *
712f80b0 7672array_var_assignment (v, itype, quoted, atype)
d233b485 7673 SHELL_VAR *v;
712f80b0 7674 int itype, quoted, atype;
d233b485
CR
7675{
7676 char *ret, *val, flags[MAX_ATTRIBUTES];
7677 int i;
7678
7679 if (v == 0)
7680 return (char *)NULL;
712f80b0
CR
7681 if (atype == 2)
7682 val = array_p (v) ? array_to_kvpair (array_cell (v), 0)
7683 : assoc_to_kvpair (assoc_cell (v), 0);
7684 else
7685 val = array_p (v) ? array_to_assign (array_cell (v), 0)
7686 : assoc_to_assign (assoc_cell (v), 0);
7687
7688 if (val == 0 && (invisible_p (v) || var_isset (v) == 0))
7689 ; /* placeholder */
7690 else if (val == 0)
d233b485
CR
7691 {
7692 val = (char *)xmalloc (3);
7693 val[0] = LPAREN;
7694 val[1] = RPAREN;
7695 val[2] = 0;
7696 }
7697 else
7698 {
7699 ret = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (val) : quote_escapes (val);
7700 free (val);
7701 val = ret;
7702 }
712f80b0
CR
7703
7704 if (atype == 2)
7705 return val;
7706
d233b485 7707 i = var_attribute_string (v, 0, flags);
712f80b0
CR
7708 ret = (char *)xmalloc (i + STRLEN (val) + strlen (v->name) + 16);
7709 if (val)
7710 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
7711 else
7712 sprintf (ret, "declare -%s %s", flags, v->name);
d233b485
CR
7713 free (val);
7714 return ret;
7715}
7716#endif
7717
7718static char *
7719pos_params_assignment (list, itype, quoted)
7720 WORD_LIST *list;
7721 int itype;
7722 int quoted;
7723{
7724 char *temp, *ret;
7725
7726 /* first, we transform the list to quote each word. */
7727 temp = list_transform ('Q', (SHELL_VAR *)0, list, itype, quoted);
7728 ret = (char *)xmalloc (strlen (temp) + 8);
7729 strcpy (ret, "set -- ");
7730 strcpy (ret + 7, temp);
7731 free (temp);
7732 return ret;
7733}
7734
7735static char *
7736string_transform (xc, v, s)
7737 int xc;
7738 SHELL_VAR *v;
7739 char *s;
7740{
7741 char *ret, flags[MAX_ATTRIBUTES], *t;
7742 int i;
7743
712f80b0
CR
7744 if (((xc == 'A' || xc == 'a') && v == 0))
7745 return (char *)NULL;
7746 else if (xc != 'a' && xc != 'A' && s == 0)
d233b485
CR
7747 return (char *)NULL;
7748
7749 switch (xc)
7750 {
7751 /* Transformations that interrogate the variable */
7752 case 'a':
7753 i = var_attribute_string (v, 0, flags);
7754 ret = (i > 0) ? savestring (flags) : (char *)NULL;
7755 break;
7756 case 'A':
7757 ret = string_var_assignment (v, s);
7758 break;
712f80b0
CR
7759 case 'K':
7760 ret = sh_quote_reusable (s, 0);
7761 break;
d233b485
CR
7762 /* Transformations that modify the variable's value */
7763 case 'E':
7764 t = ansiexpand (s, 0, strlen (s), (int *)0);
7765 ret = dequote_escapes (t);
7766 free (t);
7767 break;
7768 case 'P':
7769 ret = decode_prompt_string (s);
7770 break;
7771 case 'Q':
7772 ret = sh_quote_reusable (s, 0);
7773 break;
712f80b0
CR
7774 case 'U':
7775 ret = sh_modcase (s, 0, CASE_UPPER);
7776 break;
7777 case 'u':
7778 ret = sh_modcase (s, 0, CASE_UPFIRST); /* capitalize */
7779 break;
7780 case 'L':
7781 ret = sh_modcase (s, 0, CASE_LOWER);
7782 break;
d233b485
CR
7783 default:
7784 ret = (char *)NULL;
7785 break;
7786 }
7787 return ret;
7788}
7789
7790static char *
7791list_transform (xc, v, list, itype, quoted)
7792 int xc;
7793 SHELL_VAR *v;
7794 WORD_LIST *list;
7795 int itype, quoted;
7796{
7797 WORD_LIST *new, *l;
7798 WORD_DESC *w;
7799 char *tword;
7800 int qflags;
7801
7802 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
7803 {
7804 tword = string_transform (xc, v, l->word->word);
7805 w = alloc_word_desc ();
7806 w->word = tword ? tword : savestring (""); /* XXX */
7807 new = make_word_list (w, new);
7808 }
7809 l = REVERSE_LIST (new, WORD_LIST *);
7810
7811 qflags = quoted;
7812 /* If we are expanding in a context where word splitting will not be
7813 performed, treat as quoted. This changes how $* will be expanded. */
7814 if (itype == '*' && expand_no_split_dollar_star && ifs_is_null)
7815 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
7816
712f80b0 7817 tword = string_list_pos_params (itype, l, qflags, 0);
d233b485
CR
7818 dispose_words (l);
7819
7820 return (tword);
7821}
7822
7823static char *
7824parameter_list_transform (xc, itype, quoted)
7825 int xc;
7826 int itype;
7827 int quoted;
7828{
7829 char *ret;
7830 WORD_LIST *list;
7831
7832 list = list_rest_of_args ();
7833 if (list == 0)
7834 return ((char *)NULL);
7835 if (xc == 'A')
7836 ret = pos_params_assignment (list, itype, quoted);
7837 else
7838 ret = list_transform (xc, (SHELL_VAR *)0, list, itype, quoted);
7839 dispose_words (list);
7840 return (ret);
7841}
7842
7843#if defined (ARRAY_VARS)
7844static char *
712f80b0 7845array_transform (xc, var, starsub, quoted)
d233b485
CR
7846 int xc;
7847 SHELL_VAR *var;
712f80b0 7848 int starsub; /* so we can figure out how it's indexed */
d233b485
CR
7849 int quoted;
7850{
7851 ARRAY *a;
7852 HASH_TABLE *h;
7853 int itype;
7854 char *ret;
7855 WORD_LIST *list;
7856 SHELL_VAR *v;
7857
712f80b0 7858 v = var; /* XXX - for now */
d233b485 7859
712f80b0 7860 itype = starsub ? '*' : '@';
d233b485
CR
7861
7862 if (xc == 'A')
712f80b0
CR
7863 return (array_var_assignment (v, itype, quoted, 1));
7864 else if (xc == 'K')
7865 return (array_var_assignment (v, itype, quoted, 2));
7866
7867 /* special case for unset arrays and attributes */
7868 if (xc == 'a' && (invisible_p (v) || var_isset (v) == 0))
7869 {
7870 char flags[MAX_ATTRIBUTES];
7871 int i;
7872
7873 i = var_attribute_string (v, 0, flags);
7874 return ((i > 0) ? savestring (flags) : (char *)NULL);
7875 }
d233b485
CR
7876
7877 a = (v && array_p (v)) ? array_cell (v) : 0;
7878 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
712f80b0 7879
d233b485
CR
7880 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
7881 if (list == 0)
7882 return ((char *)NULL);
7883 ret = list_transform (xc, v, list, itype, quoted);
7884 dispose_words (list);
7885
7886 return ret;
7887}
7888#endif /* ARRAY_VARS */
7889
3eb0018e
CR
7890static int
7891valid_parameter_transform (xform)
7892 char *xform;
7893{
7894 if (xform[1])
7895 return 0;
7896
7897 /* check for valid values of xform[0] */
7898 switch (xform[0])
7899 {
7900 case 'a': /* expand to a string with just attributes */
7901 case 'A': /* expand as an assignment statement with attributes */
7902 case 'K': /* expand assoc array to list of key/value pairs */
7903 case 'E': /* expand like $'...' */
7904 case 'P': /* expand like prompt string */
7905 case 'Q': /* quote reusably */
7906 case 'U': /* transform to uppercase */
7907 case 'u': /* tranform by capitalizing */
7908 case 'L': /* transform to lowercase */
7909 return 1;
7910 default:
7911 return 0;
7912 }
7913}
7914
d233b485
CR
7915static char *
7916parameter_brace_transform (varname, value, ind, xform, rtype, quoted, pflags, flags)
7917 char *varname, *value;
7918 int ind;
7919 char *xform;
7920 int rtype, quoted, pflags, flags;
7921{
712f80b0 7922 int vtype, xc, starsub;
d233b485
CR
7923 char *temp1, *val, *oname;
7924 SHELL_VAR *v;
7925
7926 xc = xform[0];
7927 if (value == 0 && xc != 'A' && xc != 'a')
7928 return ((char *)NULL);
7929
7930 oname = this_command_name;
7931 this_command_name = varname;
7932
7933 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7934 if (vtype == -1)
7935 {
7936 this_command_name = oname;
7937 return ((char *)NULL);
7938 }
7939
3eb0018e 7940 if (valid_parameter_transform (xform) == 0)
d233b485 7941 {
d233b485
CR
7942 this_command_name = oname;
7943 return &expand_param_error;
7944 }
7945
712f80b0
CR
7946 starsub = vtype & VT_STARSUB;
7947 vtype &= ~VT_STARSUB;
7948
d233b485
CR
7949 /* If we are asked to display the attributes of an unset variable, V will
7950 be NULL after the call to get_var_and_type. Double-check here. */
712f80b0 7951 if ((xc == 'a' || xc == 'A') && vtype == VT_VARIABLE && varname && v == 0)
d233b485
CR
7952 v = find_variable (varname);
7953
7954 temp1 = (char *)NULL; /* shut up gcc */
712f80b0 7955 switch (vtype)
d233b485
CR
7956 {
7957 case VT_VARIABLE:
7958 case VT_ARRAYMEMBER:
7959 temp1 = string_transform (xc, v, val);
7960 if (vtype == VT_VARIABLE)
7961 FREE (val);
7962 if (temp1)
7963 {
7964 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7965 ? quote_string (temp1)
7966 : quote_escapes (temp1);
7967 free (temp1);
7968 temp1 = val;
7969 }
7970 break;
7971#if defined (ARRAY_VARS)
7972 case VT_ARRAYVAR:
712f80b0 7973 temp1 = array_transform (xc, v, starsub, quoted);
d233b485 7974 if (temp1 && quoted == 0 && ifs_is_null)
95732b49 7975 {
d233b485 7976 /* Posix interp 888 */
95732b49 7977 }
d233b485 7978 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3185942a 7979 {
d233b485
CR
7980 val = quote_escapes (temp1);
7981 free (temp1);
7982 temp1 = val;
3185942a 7983 }
d233b485 7984 break;
ccc6cda3 7985#endif
d233b485
CR
7986 case VT_POSPARMS:
7987 temp1 = parameter_list_transform (xc, varname[0], quoted);
7988 if (temp1 && quoted == 0 && ifs_is_null)
b80f6443 7989 {
d233b485 7990 /* Posix interp 888 */
b80f6443 7991 }
d233b485
CR
7992 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
7993 {
7994 val = quote_escapes (temp1);
7995 free (temp1);
7996 temp1 = val;
7997 }
7998 break;
b80f6443 7999 }
ccc6cda3 8000
d233b485
CR
8001 this_command_name = oname;
8002 return temp1;
ccc6cda3
JA
8003}
8004
cce855bc
JA
8005/******************************************************/
8006/* */
8007/* Functions to extract substrings of variable values */
8008/* */
8009/******************************************************/
8010
b80f6443
JA
8011#if defined (HANDLE_MULTIBYTE)
8012/* Character-oriented rather than strictly byte-oriented substrings. S and
8013 E, rather being strict indices into STRING, indicate character (possibly
8014 multibyte character) positions that require calculation.
8015 Used by the ${param:offset[:length]} expansion. */
8016static char *
8017mb_substring (string, s, e)
8018 char *string;
8019 int s, e;
8020{
8021 char *tt;
a0c0a00f
CR
8022 int start, stop, i;
8023 size_t slen;
b80f6443
JA
8024 DECLARE_MBSTATE;
8025
8026 start = 0;
95732b49
JA
8027 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
8028 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
b80f6443
JA
8029
8030 i = s;
8031 while (string[start] && i--)
8032 ADVANCE_CHAR (string, slen, start);
8033 stop = start;
8034 i = e - s;
8035 while (string[stop] && i--)
8036 ADVANCE_CHAR (string, slen, stop);
8037 tt = substring (string, start, stop);
8038 return tt;
8039}
8040#endif
8041
ccc6cda3
JA
8042/* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
8043 is `@', use the positional parameters; otherwise, use the value of
8044 VARNAME. If VARNAME is an array variable, use the array elements. */
8045
8046static char *
d233b485 8047parameter_brace_substring (varname, value, ind, substr, quoted, pflags, flags)
495aee44
CR
8048 char *varname, *value;
8049 int ind;
8050 char *substr;
d233b485 8051 int quoted, pflags, flags;
ccc6cda3 8052{
7117c2d2 8053 intmax_t e1, e2;
b80f6443 8054 int vtype, r, starsub;
0628567a 8055 char *temp, *val, *tt, *oname;
ccc6cda3
JA
8056 SHELL_VAR *v;
8057
a0c0a00f 8058 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
ccc6cda3
JA
8059 return ((char *)NULL);
8060
0628567a 8061 oname = this_command_name;
ccc6cda3
JA
8062 this_command_name = varname;
8063
495aee44 8064 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3 8065 if (vtype == -1)
0628567a
JA
8066 {
8067 this_command_name = oname;
8068 return ((char *)NULL);
8069 }
ccc6cda3 8070
b80f6443
JA
8071 starsub = vtype & VT_STARSUB;
8072 vtype &= ~VT_STARSUB;
8073
3185942a 8074 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
0628567a 8075 this_command_name = oname;
e8ce775d 8076 if (r <= 0)
ac50fbac
CR
8077 {
8078 if (vtype == VT_VARIABLE)
8079 FREE (val);
8080 return ((r == 0) ? &expand_param_error : (char *)NULL);
8081 }
ccc6cda3
JA
8082
8083 switch (vtype)
8084 {
8085 case VT_VARIABLE:
d166f048 8086 case VT_ARRAYMEMBER:
b80f6443
JA
8087#if defined (HANDLE_MULTIBYTE)
8088 if (MB_CUR_MAX > 1)
8089 tt = mb_substring (val, e1, e2);
8090 else
8091#endif
7117c2d2 8092 tt = substring (val, e1, e2);
b80f6443 8093
7117c2d2
JA
8094 if (vtype == VT_VARIABLE)
8095 FREE (val);
8096 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
8097 temp = quote_string (tt);
8098 else
8099 temp = tt ? quote_escapes (tt) : (char *)NULL;
8100 FREE (tt);
ccc6cda3
JA
8101 break;
8102 case VT_POSPARMS:
d233b485
CR
8103 case VT_ARRAYVAR:
8104 if (vtype == VT_POSPARMS)
712f80b0 8105 tt = pos_params (varname, e1, e2, quoted, pflags);
d233b485
CR
8106#if defined (ARRAY_VARS)
8107 /* assoc_subrange and array_subrange both call string_list_pos_params,
8108 so we can treat this case just like VT_POSPARAMS. */
8109 else if (assoc_p (v))
8110 /* we convert to list and take first e2 elements starting at e1th
8111 element -- officially undefined for now */
712f80b0 8112 tt = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted, pflags);
d233b485
CR
8113 else
8114 /* We want E2 to be the number of elements desired (arrays can be
8115 sparse, so verify_substring_values just returns the numbers
8116 specified and we rely on array_subrange to understand how to
8117 deal with them). */
712f80b0 8118 tt = array_subrange (array_cell (v), e1, e2, starsub, quoted, pflags);
d233b485
CR
8119#endif
8120 /* We want to leave this alone in every case where pos_params/
8121 string_list_pos_params quotes the list members */
8122 if (tt && quoted == 0 && ifs_is_null)
8123 {
8124 temp = tt; /* Posix interp 888 */
8125 }
712f80b0
CR
8126 else if (tt && quoted == 0 && (pflags & PF_ASSIGNRHS))
8127 {
8128 temp = tt; /* Posix interp 888 */
8129 }
d233b485 8130 else if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
7117c2d2
JA
8131 {
8132 temp = tt ? quote_escapes (tt) : (char *)NULL;
8133 FREE (tt);
8134 }
8135 else
8136 temp = tt;
ccc6cda3 8137 break;
d233b485 8138
f73dda09
JA
8139 default:
8140 temp = (char *)NULL;
ccc6cda3
JA
8141 }
8142
8143 return temp;
8144}
8145
cce855bc
JA
8146/****************************************************************/
8147/* */
8148/* Functions to perform pattern substitution on variable values */
8149/* */
8150/****************************************************************/
8151
d233b485 8152#ifdef INCLUDE_UNUSED
495aee44
CR
8153static int
8154shouldexp_replacement (s)
8155 char *s;
8156{
8157 register char *p;
8158
8159 for (p = s; p && *p; p++)
8160 {
8161 if (*p == '\\')
8162 p++;
8163 else if (*p == '&')
8164 return 1;
8165 }
8166 return 0;
8167}
a0c0a00f 8168#endif
495aee44 8169
ccc6cda3
JA
8170char *
8171pat_subst (string, pat, rep, mflags)
8172 char *string, *pat, *rep;
8173 int mflags;
8174{
712f80b0 8175 char *ret, *s, *e, *str, *rstr, *mstr, *send;
a0c0a00f
CR
8176 int rptr, mtype, rxpand, mlen;
8177 size_t rsize, l, replen, rslen;
712f80b0 8178 DECLARE_MBSTATE;
495aee44 8179
d233b485 8180 if (string == 0)
495aee44 8181 return (savestring (""));
ccc6cda3 8182
b72432fd
JA
8183 mtype = mflags & MATCH_TYPEMASK;
8184
712f80b0 8185#if 0 /* TAG: bash-5.2? */
495aee44
CR
8186 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
8187#else
8188 rxpand = 0;
8189#endif
8190
b72432fd
JA
8191 /* Special cases:
8192 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
8193 * with REP and return the result.
8194 * 2. A null pattern with mtype == MATCH_END means to append REP to
8195 * STRING and return the result.
a0c0a00f
CR
8196 * 3. A null STRING with a matching pattern means to append REP to
8197 * STRING and return the result.
495aee44 8198 * These don't understand or process `&' in the replacement string.
b72432fd
JA
8199 */
8200 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
8201 {
8202 replen = STRLEN (rep);
495aee44 8203 l = STRLEN (string);
f73dda09 8204 ret = (char *)xmalloc (replen + l + 2);
bb70624e
JA
8205 if (replen == 0)
8206 strcpy (ret, string);
8207 else if (mtype == MATCH_BEG)
b72432fd
JA
8208 {
8209 strcpy (ret, rep);
8210 strcpy (ret + replen, string);
8211 }
8212 else
8213 {
8214 strcpy (ret, string);
8215 strcpy (ret + l, rep);
8216 }
8217 return (ret);
8218 }
a0c0a00f
CR
8219 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
8220 {
8221 replen = STRLEN (rep);
8222 ret = (char *)xmalloc (replen + 1);
8223 if (replen == 0)
8224 ret[0] = '\0';
8225 else
8226 strcpy (ret, rep);
8227 return (ret);
8228 }
b72432fd 8229
f73dda09 8230 ret = (char *)xmalloc (rsize = 64);
ccc6cda3 8231 ret[0] = '\0';
712f80b0 8232 send = string + strlen (string);
ccc6cda3 8233
a0c0a00f 8234 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
ccc6cda3
JA
8235 {
8236 if (match_pattern (str, pat, mtype, &s, &e) == 0)
8237 break;
8238 l = s - str;
495aee44 8239
a0c0a00f 8240 if (rep && rxpand)
495aee44 8241 {
a0c0a00f
CR
8242 int x;
8243 mlen = e - s;
8244 mstr = xmalloc (mlen + 1);
495aee44
CR
8245 for (x = 0; x < mlen; x++)
8246 mstr[x] = s[x];
a0c0a00f
CR
8247 mstr[mlen] = '\0';
8248 rstr = strcreplace (rep, '&', mstr, 0);
8249 free (mstr);
8250 rslen = strlen (rstr);
495aee44
CR
8251 }
8252 else
a0c0a00f
CR
8253 {
8254 rstr = rep;
8255 rslen = replen;
8256 }
495aee44
CR
8257
8258 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
ccc6cda3
JA
8259
8260 /* OK, now copy the leading unmatched portion of the string (from
8261 str to s) to ret starting at rptr (the current offset). Then copy
28ef6c31
JA
8262 the replacement string at ret + rptr + (s - str). Increment
8263 rptr (if necessary) and str and go on. */
ccc6cda3
JA
8264 if (l)
8265 {
8266 strncpy (ret + rptr, str, l);
8267 rptr += l;
8268 }
8269 if (replen)
8270 {
495aee44
CR
8271 strncpy (ret + rptr, rstr, rslen);
8272 rptr += rslen;
ccc6cda3
JA
8273 }
8274 str = e; /* e == end of match */
b80f6443 8275
495aee44
CR
8276 if (rstr != rep)
8277 free (rstr);
8278
ccc6cda3 8279 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
28ef6c31 8280 break;
b80f6443
JA
8281
8282 if (s == e)
0001803f
CR
8283 {
8284 /* On a zero-length match, make sure we copy one character, since
8285 we increment one character to avoid infinite recursion. */
712f80b0
CR
8286 char *p, *origp, *origs;
8287 size_t clen;
8288
8289 RESIZE_MALLOCED_BUFFER (ret, rptr, locale_mb_cur_max, rsize, 64);
8290#if defined (HANDLE_MULTIBYTE)
8291 p = origp = ret + rptr;
8292 origs = str;
8293 COPY_CHAR_P (p, str, send);
8294 rptr += p - origp;
8295 e += str - origs;
8296#else
0001803f
CR
8297 ret[rptr++] = *str++;
8298 e++; /* avoid infinite recursion on zero-length match */
712f80b0 8299#endif
0001803f 8300 }
ccc6cda3
JA
8301 }
8302
8303 /* Now copy the unmatched portion of the input string */
495aee44 8304 if (str && *str)
d166f048
JA
8305 {
8306 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
8307 strcpy (ret + rptr, str);
8308 }
ccc6cda3
JA
8309 else
8310 ret[rptr] = '\0';
8311
8312 return ret;
8313}
8314
8315/* Do pattern match and replacement on the positional parameters. */
8316static char *
8317pos_params_pat_subst (string, pat, rep, mflags)
8318 char *string, *pat, *rep;
8319 int mflags;
8320{
8321 WORD_LIST *save, *params;
8322 WORD_DESC *w;
0628567a 8323 char *ret;
712f80b0 8324 int pchar, qflags, pflags;
ccc6cda3
JA
8325
8326 save = params = list_rest_of_args ();
8327 if (save == 0)
8328 return ((char *)NULL);
8329
8330 for ( ; params; params = params->next)
8331 {
8332 ret = pat_subst (params->word->word, pat, rep, mflags);
95732b49
JA
8333 w = alloc_word_desc ();
8334 w->word = ret ? ret : savestring ("");
ccc6cda3
JA
8335 dispose_word (params->word);
8336 params->word = w;
ccc6cda3
JA
8337 }
8338
3185942a
JA
8339 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
8340 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
712f80b0 8341 pflags = (mflags & MATCH_ASSIGNRHS) == MATCH_ASSIGNRHS ? PF_ASSIGNRHS : 0;
3185942a 8342
d233b485
CR
8343 /* If we are expanding in a context where word splitting will not be
8344 performed, treat as quoted. This changes how $* will be expanded. */
8345 if (pchar == '*' && (mflags & MATCH_ASSIGNRHS) && expand_no_split_dollar_star && ifs_is_null)
8346 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
3185942a 8347
712f80b0 8348 ret = string_list_pos_params (pchar, save, qflags, pflags);
ccc6cda3
JA
8349 dispose_words (save);
8350
8351 return (ret);
8352}
8353
cce855bc
JA
8354/* Perform pattern substitution on VALUE, which is the expansion of
8355 VARNAME. PATSUB is an expression supplying the pattern to match
8356 and the string to substitute. QUOTED is a flags word containing
8357 the type of quoting currently in effect. */
ccc6cda3 8358static char *
a0c0a00f 8359parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
495aee44
CR
8360 char *varname, *value;
8361 int ind;
8362 char *patsub;
a0c0a00f 8363 int quoted, pflags, flags;
ccc6cda3 8364{
3185942a 8365 int vtype, mflags, starsub, delim;
d233b485 8366 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt, *oname;
ccc6cda3
JA
8367 SHELL_VAR *v;
8368
8369 if (value == 0)
8370 return ((char *)NULL);
8371
d233b485
CR
8372 oname = this_command_name;
8373 this_command_name = varname; /* error messages */
ccc6cda3 8374
495aee44 8375 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
ccc6cda3 8376 if (vtype == -1)
d233b485
CR
8377 {
8378 this_command_name = oname;
8379 return ((char *)NULL);
8380 }
ccc6cda3 8381
b80f6443
JA
8382 starsub = vtype & VT_STARSUB;
8383 vtype &= ~VT_STARSUB;
8384
ccc6cda3 8385 mflags = 0;
ac50fbac
CR
8386 /* PATSUB is never NULL when this is called. */
8387 if (*patsub == '/')
f1be666c
JA
8388 {
8389 mflags |= MATCH_GLOBREP;
8390 patsub++;
8391 }
7117c2d2
JA
8392
8393 /* Malloc this because expand_string_if_necessary or one of the expansion
8394 functions in its call chain may free it on a substitution error. */
bb70624e 8395 lpatsub = savestring (patsub);
ccc6cda3
JA
8396
8397 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8398 mflags |= MATCH_QUOTED;
8399
b80f6443
JA
8400 if (starsub)
8401 mflags |= MATCH_STARSUB;
8402
a0c0a00f
CR
8403 if (pflags & PF_ASSIGNRHS)
8404 mflags |= MATCH_ASSIGNRHS;
8405
0628567a
JA
8406 /* If the pattern starts with a `/', make sure we skip over it when looking
8407 for the replacement delimiter. */
3185942a
JA
8408 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
8409 if (lpatsub[delim] == '/')
8410 {
8411 lpatsub[delim] = 0;
8412 rep = lpatsub + delim + 1;
8413 }
8414 else
8415 rep = (char *)NULL;
ccc6cda3
JA
8416
8417 if (rep && *rep == '\0')
8418 rep = (char *)NULL;
8419
b80f6443
JA
8420 /* Perform the same expansions on the pattern as performed by the
8421 pattern removal expansions. */
8422 pat = getpattern (lpatsub, quoted, 1);
bb70624e 8423
ccc6cda3 8424 if (rep)
d166f048 8425 {
ac50fbac
CR
8426 /* We want to perform quote removal on the expanded replacement even if
8427 the entire expansion is double-quoted because the parser and string
8428 extraction functions treated quotes in the replacement string as
8429 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
8430 if (shell_compatibility_level > 42)
8431 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
8432 /* This is the bash-4.2 code. */
8433 else if ((mflags & MATCH_QUOTED) == 0)
f73dda09 8434 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
d166f048 8435 else
f73dda09 8436 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
d166f048 8437 }
ccc6cda3 8438
0628567a 8439 /* ksh93 doesn't allow the match specifier to be a part of the expanded
f1be666c
JA
8440 pattern. This is an extension. Make sure we don't anchor the pattern
8441 at the beginning or end of the string if we're doing global replacement,
8442 though. */
ccc6cda3 8443 p = pat;
f1be666c
JA
8444 if (mflags & MATCH_GLOBREP)
8445 mflags |= MATCH_ANY;
0628567a 8446 else if (pat && pat[0] == '#')
ccc6cda3
JA
8447 {
8448 mflags |= MATCH_BEG;
8449 p++;
8450 }
d166f048 8451 else if (pat && pat[0] == '%')
ccc6cda3
JA
8452 {
8453 mflags |= MATCH_END;
8454 p++;
8455 }
8456 else
8457 mflags |= MATCH_ANY;
8458
cce855bc
JA
8459 /* OK, we now want to substitute REP for PAT in VAL. If
8460 flags & MATCH_GLOBREP is non-zero, the substitution is done
8461 everywhere, otherwise only the first occurrence of PAT is
7117c2d2
JA
8462 replaced. The pattern matching code doesn't understand
8463 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
8464 values passed in (VT_VARIABLE) so the pattern substitution
8465 code works right. We need to requote special chars after
8466 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
8467 other cases if QUOTED == 0, since the posparams and arrays
8468 indexed by * or @ do special things when QUOTED != 0. */
8469
ccc6cda3
JA
8470 switch (vtype)
8471 {
8472 case VT_VARIABLE:
d166f048 8473 case VT_ARRAYMEMBER:
ccc6cda3 8474 temp = pat_subst (val, p, rep, mflags);
7117c2d2
JA
8475 if (vtype == VT_VARIABLE)
8476 FREE (val);
8477 if (temp)
8478 {
3185942a 8479 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7117c2d2
JA
8480 free (temp);
8481 temp = tt;
8482 }
ccc6cda3
JA
8483 break;
8484 case VT_POSPARMS:
d233b485
CR
8485 /* This does the right thing for the case where we are not performing
8486 word splitting. MATCH_STARSUB restricts it to ${* /foo/bar}, and
8487 pos_params_pat_subst/string_list_pos_params will do the right thing
8488 in turn for the case where ifs_is_null. Posix interp 888 */
8489 if ((pflags & PF_NOSPLIT2) && (mflags & MATCH_STARSUB))
8490 mflags |= MATCH_ASSIGNRHS;
ccc6cda3 8491 temp = pos_params_pat_subst (val, p, rep, mflags);
d233b485
CR
8492 if (temp && quoted == 0 && ifs_is_null)
8493 {
8494 /* Posix interp 888 */
8495 }
712f80b0
CR
8496 else if (temp && quoted == 0 && (pflags & PF_ASSIGNRHS))
8497 {
8498 /* Posix interp 888 */
8499 }
d233b485 8500 else if (temp && (mflags & MATCH_QUOTED) == 0)
7117c2d2
JA
8501 {
8502 tt = quote_escapes (temp);
8503 free (temp);
8504 temp = tt;
8505 }
ccc6cda3
JA
8506 break;
8507#if defined (ARRAY_VARS)
8508 case VT_ARRAYVAR:
d233b485
CR
8509 /* If we are expanding in a context where word splitting will not be
8510 performed, treat as quoted. This changes how ${A[*]} will be
8511 expanded to make it identical to $*. */
8512 if ((mflags & MATCH_STARSUB) && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8513 mflags |= MATCH_QUOTED; /* Posix interp 888 */
8514
8515 /* these eventually call string_list_pos_params */
8516 if (assoc_p (v))
8517 temp = assoc_patsub (assoc_cell (v), p, rep, mflags);
8518 else
8519 temp = array_patsub (array_cell (v), p, rep, mflags);
8520
8521 if (temp && quoted == 0 && ifs_is_null)
8522 {
8523 /* Posix interp 888 */
8524 }
8525 else if (temp && (mflags & MATCH_QUOTED) == 0)
8526 {
8527 tt = quote_escapes (temp);
8528 free (temp);
8529 temp = tt;
8530 }
3185942a
JA
8531 break;
8532#endif
8533 }
8534
8535 FREE (pat);
8536 FREE (rep);
8537 free (lpatsub);
8538
d233b485
CR
8539 this_command_name = oname;
8540
3185942a
JA
8541 return temp;
8542}
8543
8544/****************************************************************/
8545/* */
8546/* Functions to perform case modification on variable values */
8547/* */
8548/****************************************************************/
8549
8550/* Do case modification on the positional parameters. */
8551
8552static char *
8553pos_params_modcase (string, pat, modop, mflags)
8554 char *string, *pat;
8555 int modop;
8556 int mflags;
8557{
8558 WORD_LIST *save, *params;
8559 WORD_DESC *w;
8560 char *ret;
712f80b0 8561 int pchar, qflags, pflags;
3185942a
JA
8562
8563 save = params = list_rest_of_args ();
8564 if (save == 0)
8565 return ((char *)NULL);
8566
8567 for ( ; params; params = params->next)
8568 {
8569 ret = sh_modcase (params->word->word, pat, modop);
8570 w = alloc_word_desc ();
8571 w->word = ret ? ret : savestring ("");
8572 dispose_word (params->word);
8573 params->word = w;
8574 }
8575
8576 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
8577 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
712f80b0 8578 pflags = (mflags & MATCH_ASSIGNRHS) == MATCH_ASSIGNRHS ? PF_ASSIGNRHS : 0;
3185942a 8579
d233b485
CR
8580 /* If we are expanding in a context where word splitting will not be
8581 performed, treat as quoted. This changes how $* will be expanded. */
8582 if (pchar == '*' && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8583 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
8584
712f80b0 8585 ret = string_list_pos_params (pchar, save, qflags, pflags);
3185942a
JA
8586 dispose_words (save);
8587
8588 return (ret);
8589}
8590
8591/* Perform case modification on VALUE, which is the expansion of
8592 VARNAME. MODSPEC is an expression supplying the type of modification
8593 to perform. QUOTED is a flags word containing the type of quoting
8594 currently in effect. */
8595static char *
d233b485 8596parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, pflags, flags)
3185942a 8597 char *varname, *value;
495aee44 8598 int ind, modspec;
3185942a 8599 char *patspec;
d233b485 8600 int quoted, pflags, flags;
3185942a
JA
8601{
8602 int vtype, starsub, modop, mflags, x;
d233b485 8603 char *val, *temp, *pat, *p, *lpat, *tt, *oname;
3185942a
JA
8604 SHELL_VAR *v;
8605
8606 if (value == 0)
8607 return ((char *)NULL);
8608
d233b485 8609 oname = this_command_name;
3185942a
JA
8610 this_command_name = varname;
8611
495aee44 8612 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
3185942a 8613 if (vtype == -1)
d233b485
CR
8614 {
8615 this_command_name = oname;
8616 return ((char *)NULL);
8617 }
3185942a
JA
8618
8619 starsub = vtype & VT_STARSUB;
8620 vtype &= ~VT_STARSUB;
8621
8622 modop = 0;
8623 mflags = 0;
8624 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8625 mflags |= MATCH_QUOTED;
8626 if (starsub)
8627 mflags |= MATCH_STARSUB;
d233b485
CR
8628 if (pflags & PF_ASSIGNRHS)
8629 mflags |= MATCH_ASSIGNRHS;
3185942a
JA
8630
8631 p = patspec;
8632 if (modspec == '^')
8633 {
8634 x = p && p[0] == modspec;
17345e5a 8635 modop = x ? CASE_UPPER : CASE_UPFIRST;
3185942a
JA
8636 p += x;
8637 }
8638 else if (modspec == ',')
8639 {
8640 x = p && p[0] == modspec;
17345e5a 8641 modop = x ? CASE_LOWER : CASE_LOWFIRST;
3185942a
JA
8642 p += x;
8643 }
8644 else if (modspec == '~')
8645 {
8646 x = p && p[0] == modspec;
8647 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
8648 p += x;
8649 }
8650
8651 lpat = p ? savestring (p) : 0;
8652 /* Perform the same expansions on the pattern as performed by the
d233b485 8653 pattern removal expansions. */
3185942a
JA
8654 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
8655
8656 /* OK, now we do the case modification. */
8657 switch (vtype)
8658 {
8659 case VT_VARIABLE:
8660 case VT_ARRAYMEMBER:
8661 temp = sh_modcase (val, pat, modop);
8662 if (vtype == VT_VARIABLE)
8663 FREE (val);
8664 if (temp)
8665 {
8666 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
8667 free (temp);
8668 temp = tt;
8669 }
8670 break;
8671
8672 case VT_POSPARMS:
8673 temp = pos_params_modcase (val, pat, modop, mflags);
d233b485
CR
8674 if (temp && quoted == 0 && ifs_is_null)
8675 {
8676 /* Posix interp 888 */
8677 }
8678 else if (temp && (mflags & MATCH_QUOTED) == 0)
7117c2d2
JA
8679 {
8680 tt = quote_escapes (temp);
8681 free (temp);
8682 temp = tt;
8683 }
3185942a
JA
8684 break;
8685
8686#if defined (ARRAY_VARS)
8687 case VT_ARRAYVAR:
d233b485
CR
8688 /* If we are expanding in a context where word splitting will not be
8689 performed, treat as quoted. This changes how ${A[*]} will be
8690 expanded to make it identical to $*. */
8691 if ((mflags & MATCH_STARSUB) && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8692 mflags |= MATCH_QUOTED; /* Posix interp 888 */
8693
3185942a
JA
8694 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
8695 : array_modcase (array_cell (v), pat, modop, mflags);
d233b485
CR
8696
8697 if (temp && quoted == 0 && ifs_is_null)
8698 {
8699 /* Posix interp 888 */
8700 }
8701 else if (temp && (mflags & MATCH_QUOTED) == 0)
8702 {
8703 tt = quote_escapes (temp);
8704 free (temp);
8705 temp = tt;
8706 }
8707
ccc6cda3
JA
8708 break;
8709#endif
8710 }
8711
8712 FREE (pat);
3185942a 8713 free (lpat);
ccc6cda3 8714
d233b485
CR
8715 this_command_name = oname;
8716
ccc6cda3
JA
8717 return temp;
8718}
8719
0628567a
JA
8720/* Check for unbalanced parens in S, which is the contents of $(( ... )). If
8721 any occur, this must be a nested command substitution, so return 0.
8722 Otherwise, return 1. A valid arithmetic expression must always have a
8723 ( before a matching ), so any cases where there are more right parens
8724 means that this must not be an arithmetic expression, though the parser
8725 will not accept it without a balanced total number of parens. */
8726static int
8727chk_arithsub (s, len)
8728 const char *s;
8729 int len;
8730{
8731 int i, count;
8732 DECLARE_MBSTATE;
8733
8734 i = count = 0;
8735 while (i < len)
8736 {
0001803f 8737 if (s[i] == LPAREN)
0628567a 8738 count++;
0001803f 8739 else if (s[i] == RPAREN)
0628567a
JA
8740 {
8741 count--;
8742 if (count < 0)
8743 return 0;
8744 }
8745
8746 switch (s[i])
8747 {
8748 default:
8749 ADVANCE_CHAR (s, len, i);
8750 break;
8751
8752 case '\\':
8753 i++;
8754 if (s[i])
8755 ADVANCE_CHAR (s, len, i);
8756 break;
8757
8758 case '\'':
a0c0a00f 8759 i = skip_single_quoted (s, len, ++i, 0);
0628567a
JA
8760 break;
8761
8762 case '"':
a0c0a00f 8763 i = skip_double_quoted ((char *)s, len, ++i, 0);
0628567a
JA
8764 break;
8765 }
8766 }
8767
8768 return (count == 0);
8769}
8770
cce855bc
JA
8771/****************************************************************/
8772/* */
8773/* Functions to perform parameter expansion on a string */
8774/* */
8775/****************************************************************/
8776
3185942a 8777/* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
95732b49 8778static WORD_DESC *
0001803f 8779parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
ccc6cda3 8780 char *string;
a0c0a00f 8781 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
ccc6cda3
JA
8782{
8783 int check_nullness, var_is_set, var_is_null, var_is_special;
3eb0018e 8784 int want_substring, want_indir, want_patsub, want_casemod, want_attributes;
ccc6cda3 8785 char *name, *value, *temp, *temp1;
95732b49 8786 WORD_DESC *tdesc, *ret;
3eb0018e 8787 int t_index, sindex, c, tflag, modspec, local_pflags, all_element_arrayref;
7117c2d2 8788 intmax_t number;
495aee44 8789 arrayind_t ind;
ccc6cda3 8790
3185942a 8791 temp = temp1 = value = (char *)NULL;
ccc6cda3 8792 var_is_set = var_is_null = var_is_special = check_nullness = 0;
3eb0018e 8793 want_substring = want_indir = want_patsub = want_casemod = want_attributes = 0;
ccc6cda3 8794
3eb0018e 8795 local_pflags = 0;
a0c0a00f
CR
8796 all_element_arrayref = 0;
8797
cce855bc
JA
8798 sindex = *indexp;
8799 t_index = ++sindex;
0628567a
JA
8800 /* ${#var} doesn't have any of the other parameter expansions on it. */
8801 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
3185942a 8802 name = string_extract (string, &t_index, "}", SX_VARNAME);
0628567a 8803 else
3185942a
JA
8804#if defined (CASEMOD_EXPANSIONS)
8805 /* To enable case-toggling expansions using the `~' operator character
3eb0018e 8806 define CASEMOD_CAPCASE in config-top.h */
3185942a 8807# if defined (CASEMOD_CAPCASE)
a0c0a00f 8808 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
3185942a 8809# else
a0c0a00f 8810 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
3185942a
JA
8811# endif /* CASEMOD_CAPCASE */
8812#else
a0c0a00f 8813 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
3185942a 8814#endif /* CASEMOD_EXPANSIONS */
cce855bc 8815
a0c0a00f
CR
8816 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
8817 the cleanest code ever. */
8818 if (*name == 0 && sindex == t_index && string[sindex] == '@')
8819 {
8820 name = (char *)xrealloc (name, 2);
8821 name[0] = '@';
8822 name[1] = '\0';
8823 t_index++;
8824 }
d233b485 8825 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == RBRACE)
a0c0a00f
CR
8826 {
8827 name = (char *)xrealloc (name, t_index - sindex + 2);
8828 name[t_index - sindex] = '@';
8829 name[t_index - sindex + 1] = '\0';
8830 t_index++;
8831 }
8832
95732b49
JA
8833 ret = 0;
8834 tflag = 0;
8835
495aee44
CR
8836 ind = INTMAX_MIN;
8837
cce855bc
JA
8838 /* If the name really consists of a special variable, then make sure
8839 that we have the entire name. We don't allow indirect references
a0c0a00f
CR
8840 to special variables except `#', `?', `@' and `*'. This clause is
8841 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
8842 general. */
495aee44 8843 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
a0c0a00f 8844 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
495aee44 8845 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
ccc6cda3
JA
8846 {
8847 t_index++;
a0c0a00f 8848 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
ac50fbac 8849 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
ccc6cda3
JA
8850 *name = string[sindex];
8851 if (string[sindex] == '!')
8852 {
28ef6c31
JA
8853 /* indirect reference of $#, $?, $@, or $* */
8854 name[1] = string[sindex + 1];
8855 strcpy (name + 2, temp1);
ccc6cda3 8856 }
cce855bc 8857 else
ccc6cda3
JA
8858 strcpy (name + 1, temp1);
8859 free (temp1);
8860 }
8861 sindex = t_index;
8862
8863 /* Find out what character ended the variable name. Then
8864 do the appropriate thing. */
8865 if (c = string[sindex])
8866 sindex++;
8867
8868 /* If c is followed by one of the valid parameter expansion
8869 characters, move past it as normal. If not, assume that
8870 a substring specification is being given, and do not move
8871 past it. */
28ef6c31 8872 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
ccc6cda3
JA
8873 {
8874 check_nullness++;
8875 if (c = string[sindex])
8876 sindex++;
8877 }
cce855bc 8878 else if (c == ':' && string[sindex] != RBRACE)
ccc6cda3 8879 want_substring = 1;
ac50fbac 8880 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
ccc6cda3 8881 want_patsub = 1;
3185942a
JA
8882#if defined (CASEMOD_EXPANSIONS)
8883 else if (c == '^' || c == ',' || c == '~')
8884 {
8885 modspec = c;
8886 want_casemod = 1;
8887 }
8888#endif
3eb0018e
CR
8889 else if (c == '@' && (string[sindex] == 'a' || string[sindex] == 'A') && string[sindex+1] == RBRACE)
8890 {
8891 /* special case because we do not want to shortcut foo as foo[0] here */
8892 want_attributes = 1;
8893 local_pflags |= PF_ALLINDS;
8894 }
ccc6cda3 8895
cce855bc
JA
8896 /* Catch the valid and invalid brace expressions that made it through the
8897 tests above. */
8898 /* ${#-} is a valid expansion and means to take the length of $-.
8899 Similarly for ${#?} and ${##}... */
8900 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
28ef6c31 8901 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
cce855bc 8902 {
f73dda09 8903 name = (char *)xrealloc (name, 3);
cce855bc
JA
8904 name[1] = c;
8905 name[2] = '\0';
8906 c = string[sindex++];
8907 }
8908
8909 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8910 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8911 member (c, "%:=+/") && string[sindex] == RBRACE)
8912 {
8913 temp = (char *)NULL;
a0c0a00f 8914 goto bad_substitution; /* XXX - substitution error */
cce855bc
JA
8915 }
8916
8917 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8918 either a variable name, one of the positional parameters or a special
8919 variable that expands to one of the positional parameters. */
8920 want_indir = *name == '!' &&
f73dda09 8921 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7117c2d2 8922 || VALID_INDIR_PARAM (name[1]));
ccc6cda3 8923
d233b485 8924 /* Determine the value of this variable whose name is NAME. */
ccc6cda3 8925
cce855bc 8926 /* Check for special variables, directly referenced. */
bb70624e 8927 if (SPECIAL_VAR (name, want_indir))
ccc6cda3
JA
8928 var_is_special++;
8929
cce855bc
JA
8930 /* Check for special expansion things, like the length of a parameter */
8931 if (*name == '#' && name[1])
ccc6cda3 8932 {
cce855bc 8933 /* If we are not pointing at the character just after the
28ef6c31
JA
8934 closing brace, then we haven't gotten all of the name.
8935 Since it begins with a special character, this is a bad
8936 substitution. Also check NAME for validity before trying
8937 to go on. */
cce855bc 8938 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
ccc6cda3
JA
8939 {
8940 temp = (char *)NULL;
a0c0a00f 8941 goto bad_substitution; /* substitution error */
ccc6cda3
JA
8942 }
8943
8944 number = parameter_brace_expand_length (name);
495aee44
CR
8945 if (number == INTMAX_MIN && unbound_vars_is_error)
8946 {
712f80b0 8947 set_exit_status (EXECUTION_FAILURE);
495aee44
CR
8948 err_unboundvar (name+1);
8949 free (name);
8950 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8951 }
ccc6cda3
JA
8952 free (name);
8953
8954 *indexp = sindex;
95732b49
JA
8955 if (number < 0)
8956 return (&expand_wdesc_error);
8957 else
8958 {
8959 ret = alloc_word_desc ();
8960 ret->word = itos (number);
8961 return ret;
8962 }
ccc6cda3
JA
8963 }
8964
8965 /* ${@} is identical to $@. */
8966 if (name[0] == '@' && name[1] == '\0')
8967 {
8968 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8969 *quoted_dollar_atp = 1;
8970
8971 if (contains_dollar_at)
8972 *contains_dollar_at = 1;
ac50fbac
CR
8973
8974 tflag |= W_DOLLARAT;
ccc6cda3
JA
8975 }
8976
b80f6443 8977 /* Process ${!PREFIX*} expansion. */
bb70624e
JA
8978 if (want_indir && string[sindex - 1] == RBRACE &&
8979 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
f73dda09 8980 legal_variable_starter ((unsigned char) name[1]))
bb70624e
JA
8981 {
8982 char **x;
8983 WORD_LIST *xlist;
8984
8985 temp1 = savestring (name + 1);
8986 number = strlen (temp1);
8987 temp1[number - 1] = '\0';
8988 x = all_variables_matching_prefix (temp1);
7117c2d2 8989 xlist = strvec_to_word_list (x, 0, 0);
28ef6c31 8990 if (string[sindex - 2] == '*')
d233b485 8991 temp = string_list_dollar_star (xlist, quoted, 0);
28ef6c31
JA
8992 else
8993 {
a0c0a00f 8994 temp = string_list_dollar_at (xlist, quoted, 0);
28ef6c31
JA
8995 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8996 *quoted_dollar_atp = 1;
8997 if (contains_dollar_at)
8998 *contains_dollar_at = 1;
ac50fbac
CR
8999
9000 tflag |= W_DOLLARAT;
28ef6c31 9001 }
bb70624e 9002 free (x);
89a92869 9003 dispose_words (xlist);
bb70624e
JA
9004 free (temp1);
9005 *indexp = sindex;
95732b49 9006
ac50fbac
CR
9007 free (name);
9008
95732b49
JA
9009 ret = alloc_word_desc ();
9010 ret->word = temp;
ac50fbac 9011 ret->flags = tflag; /* XXX */
95732b49 9012 return ret;
bb70624e 9013 }
b80f6443
JA
9014
9015#if defined (ARRAY_VARS)
d233b485 9016 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */
b80f6443 9017 if (want_indir && string[sindex - 1] == RBRACE &&
d233b485 9018 string[sindex - 2] == RBRACK && valid_array_reference (name+1, 0))
b80f6443
JA
9019 {
9020 char *x, *x1;
9021
9022 temp1 = savestring (name + 1);
d233b485 9023 x = array_variable_name (temp1, 0, &x1, (int *)0);
b80f6443 9024 FREE (x);
d233b485 9025 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == RBRACK)
b80f6443 9026 {
712f80b0 9027 temp = array_keys (temp1, quoted, pflags); /* handles assoc vars too */
b80f6443
JA
9028 if (x1[0] == '@')
9029 {
9030 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9031 *quoted_dollar_atp = 1;
9032 if (contains_dollar_at)
9033 *contains_dollar_at = 1;
ac50fbac
CR
9034
9035 tflag |= W_DOLLARAT;
b80f6443
JA
9036 }
9037
a0c0a00f 9038 free (name);
b80f6443
JA
9039 free (temp1);
9040 *indexp = sindex;
95732b49
JA
9041
9042 ret = alloc_word_desc ();
9043 ret->word = temp;
ac50fbac 9044 ret->flags = tflag; /* XXX */
95732b49 9045 return ret;
b80f6443
JA
9046 }
9047
9048 free (temp1);
9049 }
9050#endif /* ARRAY_VARS */
bb70624e 9051
ccc6cda3
JA
9052 /* Make sure that NAME is valid before trying to go on. */
9053 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
9054 var_is_special) == 0)
9055 {
9056 temp = (char *)NULL;
a0c0a00f 9057 goto bad_substitution; /* substitution error */
ccc6cda3
JA
9058 }
9059
9060 if (want_indir)
1a1f8b54 9061 {
3eb0018e 9062 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, pflags|local_pflags, quoted_dollar_atp, contains_dollar_at);
a0c0a00f
CR
9063 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9064 {
9065 temp = (char *)NULL;
9066 goto bad_substitution;
9067 }
d233b485 9068
1a1f8b54
CR
9069 /* Turn off the W_ARRAYIND flag because there is no way for this function
9070 to return the index we're supposed to be using. */
9071 if (tdesc && tdesc->flags)
9072 tdesc->flags &= ~W_ARRAYIND;
9073 }
95732b49 9074 else
3eb0018e
CR
9075 {
9076 local_pflags |= PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS));
9077 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, local_pflags, &ind);
9078 }
95732b49 9079
d233b485
CR
9080 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9081 {
9082 tflag = 0;
9083 tdesc = 0;
9084 }
9085
95732b49
JA
9086 if (tdesc)
9087 {
9088 temp = tdesc->word;
9089 tflag = tdesc->flags;
9090 dispose_word_desc (tdesc);
9091 }
ccc6cda3 9092 else
95732b49 9093 temp = (char *)0;
ccc6cda3 9094
ac50fbac
CR
9095 if (temp == &expand_param_error || temp == &expand_param_fatal)
9096 {
9097 FREE (name);
9098 FREE (value);
9099 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9100 }
9101
ccc6cda3 9102#if defined (ARRAY_VARS)
a0c0a00f
CR
9103 if (valid_array_reference (name, 0))
9104 {
9105 int qflags;
9106 char *t;
9107
9108 qflags = quoted;
9109 /* If in a context where word splitting will not take place, treat as
9110 if double-quoted. Has effects with $* and ${array[*]} */
d233b485 9111
a0c0a00f
CR
9112 if (pflags & PF_ASSIGNRHS)
9113 qflags |= Q_DOUBLE_QUOTES;
a0c0a00f 9114 /* We duplicate a little code here */
d233b485
CR
9115 t = mbschr (name, LBRACK);
9116 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == RBRACK)
9117 {
9118 all_element_arrayref = 1;
9119 if (expand_no_split_dollar_star && t[1] == '*') /* XXX */
9120 qflags |= Q_DOUBLE_QUOTES;
9121 }
712f80b0 9122 chk_atstar (name, qflags, pflags, quoted_dollar_atp, contains_dollar_at);
a0c0a00f 9123 }
ccc6cda3
JA
9124#endif
9125
9126 var_is_set = temp != (char *)0;
9127 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
ac50fbac
CR
9128 /* XXX - this may not need to be restricted to special variables */
9129 if (check_nullness)
9130 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
d233b485
CR
9131#if defined (ARRAY_VARS)
9132 if (check_nullness)
9133 var_is_null |= var_is_set &&
9134 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) &&
9135 QUOTED_NULL (temp) &&
9136 valid_array_reference (name, 0) &&
712f80b0 9137 chk_atstar (name, 0, 0, (int *)0, (int *)0);
d233b485 9138#endif
ccc6cda3
JA
9139
9140 /* Get the rest of the stuff inside the braces. */
cce855bc 9141 if (c && c != RBRACE)
ccc6cda3
JA
9142 {
9143 /* Extract the contents of the ${ ... } expansion
28ef6c31 9144 according to the Posix.2 rules. */
49ed961b 9145 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
cce855bc 9146 if (string[sindex] == RBRACE)
28ef6c31 9147 sindex++;
ccc6cda3 9148 else
a0c0a00f 9149 goto bad_substitution; /* substitution error */
ccc6cda3
JA
9150 }
9151 else
9152 value = (char *)NULL;
726f6388 9153
ccc6cda3
JA
9154 *indexp = sindex;
9155
495aee44
CR
9156 /* All the cases where an expansion can possibly generate an unbound
9157 variable error. */
712f80b0 9158 if (want_substring || want_patsub || want_casemod || c == '@' || c == '#' || c == '%' || c == RBRACE)
495aee44 9159 {
a0c0a00f 9160 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
495aee44 9161 {
712f80b0 9162 set_exit_status (EXECUTION_FAILURE);
495aee44
CR
9163 err_unboundvar (name);
9164 FREE (value);
9165 FREE (temp);
9166 free (name);
9167 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9168 }
9169 }
9170
ccc6cda3
JA
9171 /* If this is a substring spec, process it and add the result. */
9172 if (want_substring)
726f6388 9173 {
d233b485 9174 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
9175 FREE (value);
9176 FREE (temp);
95732b49 9177
d233b485
CR
9178 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9179 {
9180 FREE (name);
9181 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9182 }
95732b49
JA
9183
9184 ret = alloc_word_desc ();
9185 ret->word = temp1;
ac50fbac
CR
9186 /* We test quoted_dollar_atp because we want variants with double-quoted
9187 "$@" to take a different code path. In fact, we make sure at the end
9188 of expand_word_internal that we're only looking at these flags if
9189 quoted_dollar_at == 0. */
712f80b0 9190 if (temp1 &&
ac50fbac
CR
9191 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9192 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
0628567a 9193 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
712f80b0
CR
9194 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 &&
9195 (pflags & PF_ASSIGNRHS))
9196 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
d233b485
CR
9197 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9198 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9199 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9200
9201 FREE (name);
95732b49 9202 return ret;
726f6388 9203 }
ccc6cda3 9204 else if (want_patsub)
726f6388 9205 {
a0c0a00f 9206 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
9207 FREE (value);
9208 FREE (temp);
95732b49 9209
d233b485
CR
9210 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9211 {
9212 FREE (name);
9213 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9214 }
95732b49
JA
9215
9216 ret = alloc_word_desc ();
9217 ret->word = temp1;
ac50fbac
CR
9218 if (temp1 &&
9219 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9220 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a 9221 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
d233b485
CR
9222 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9223 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9224 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9225
9226 FREE (name);
95732b49 9227 return ret;
ccc6cda3 9228 }
3185942a
JA
9229#if defined (CASEMOD_EXPANSIONS)
9230 else if (want_casemod)
9231 {
d233b485 9232 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
3185942a
JA
9233 FREE (value);
9234 FREE (temp);
9235
d233b485
CR
9236 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9237 {
9238 FREE (name);
9239 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9240 }
3185942a
JA
9241
9242 ret = alloc_word_desc ();
9243 ret->word = temp1;
ac50fbac
CR
9244 if (temp1 &&
9245 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9246 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
3185942a 9247 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
d233b485
CR
9248 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9249 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9250 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9251
9252 FREE (name);
3185942a
JA
9253 return ret;
9254 }
9255#endif
726f6388 9256
ccc6cda3
JA
9257 /* Do the right thing based on which character ended the variable name. */
9258 switch (c)
9259 {
9260 default:
9261 case '\0':
a0c0a00f 9262bad_substitution:
712f80b0 9263 set_exit_status (EXECUTION_FAILURE);
b80f6443 9264 report_error (_("%s: bad substitution"), string ? string : "??");
ccc6cda3
JA
9265 FREE (value);
9266 FREE (temp);
9267 free (name);
a0c0a00f
CR
9268 if (shell_compatibility_level <= 43)
9269 return &expand_wdesc_error;
9270 else
9271 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
ccc6cda3 9272
cce855bc 9273 case RBRACE:
ccc6cda3 9274 break;
726f6388 9275
a0c0a00f 9276 case '@':
d233b485 9277 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
a0c0a00f
CR
9278 free (temp);
9279 free (value);
d233b485 9280
a0c0a00f
CR
9281 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9282 {
d233b485 9283 free (name);
712f80b0 9284 set_exit_status (EXECUTION_FAILURE);
a0c0a00f
CR
9285 report_error (_("%s: bad substitution"), string ? string : "??");
9286 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9287 }
9288
9289 ret = alloc_word_desc ();
9290 ret->word = temp1;
9291 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9292 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
d233b485
CR
9293 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9294 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9295 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9296
9297 free (name);
a0c0a00f
CR
9298 return ret;
9299
ccc6cda3
JA
9300 case '#': /* ${param#[#]pattern} */
9301 case '%': /* ${param%[%]pattern} */
9302 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
28ef6c31
JA
9303 {
9304 FREE (value);
ccc6cda3 9305 break;
28ef6c31 9306 }
495aee44 9307 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
ccc6cda3
JA
9308 free (temp);
9309 free (value);
3185942a
JA
9310
9311 ret = alloc_word_desc ();
9312 ret->word = temp1;
9313 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9314 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
d233b485
CR
9315 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9316 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9317 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9318
9319 free (name);
3185942a 9320 return ret;
ccc6cda3
JA
9321
9322 case '-':
9323 case '=':
9324 case '?':
9325 case '+':
9326 if (var_is_set && var_is_null == 0)
28ef6c31
JA
9327 {
9328 /* If the operator is `+', we don't want the value of the named
9329 variable for anything, just the value of the right hand side. */
ccc6cda3
JA
9330 if (c == '+')
9331 {
28ef6c31
JA
9332 /* XXX -- if we're double-quoted and the named variable is "$@",
9333 we want to turn off any special handling of "$@" --
9334 we're not using it, so whatever is on the rhs applies. */
9335 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9336 *quoted_dollar_atp = 0;
9337 if (contains_dollar_at)
9338 *contains_dollar_at = 0;
9339
ccc6cda3
JA
9340 FREE (temp);
9341 if (value)
28ef6c31 9342 {
495aee44
CR
9343 /* From Posix discussion on austin-group list. Issue 221
9344 requires that backslashes escaping `}' inside
9345 double-quoted ${...} be removed. */
9346 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9347 quoted |= Q_DOLBRACE;
95732b49
JA
9348 ret = parameter_brace_expand_rhs (name, value, c,
9349 quoted,
a0c0a00f 9350 pflags,
95732b49
JA
9351 quoted_dollar_atp,
9352 contains_dollar_at);
9353 /* XXX - fix up later, esp. noting presence of
9354 W_HASQUOTEDNULL in ret->flags */
ccc6cda3
JA
9355 free (value);
9356 }
9357 else
28ef6c31 9358 temp = (char *)NULL;
ccc6cda3
JA
9359 }
9360 else
9361 {
9362 FREE (value);
9363 }
9364 /* Otherwise do nothing; just use the value in TEMP. */
726f6388 9365 }
ccc6cda3 9366 else /* VAR not set or VAR is NULL. */
28ef6c31 9367 {
ccc6cda3
JA
9368 FREE (temp);
9369 temp = (char *)NULL;
9370 if (c == '=' && var_is_special)
9371 {
712f80b0 9372 set_exit_status (EXECUTION_FAILURE);
b80f6443 9373 report_error (_("$%s: cannot assign in this way"), name);
ccc6cda3
JA
9374 free (name);
9375 free (value);
95732b49 9376 return &expand_wdesc_error;
ccc6cda3
JA
9377 }
9378 else if (c == '?')
9379 {
d233b485 9380 parameter_brace_expand_error (name, value, check_nullness);
95732b49 9381 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
ccc6cda3
JA
9382 }
9383 else if (c != '+')
28ef6c31
JA
9384 {
9385 /* XXX -- if we're double-quoted and the named variable is "$@",
9386 we want to turn off any special handling of "$@" --
9387 we're not using it, so whatever is on the rhs applies. */
9388 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9389 *quoted_dollar_atp = 0;
9390 if (contains_dollar_at)
9391 *contains_dollar_at = 0;
9392
495aee44
CR
9393 /* From Posix discussion on austin-group list. Issue 221 requires
9394 that backslashes escaping `}' inside double-quoted ${...} be
9395 removed. */
9396 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9397 quoted |= Q_DOLBRACE;
a0c0a00f 9398 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
95732b49
JA
9399 quoted_dollar_atp,
9400 contains_dollar_at);
9401 /* XXX - fix up later, esp. noting presence of
9402 W_HASQUOTEDNULL in tdesc->flags */
28ef6c31 9403 }
ccc6cda3 9404 free (value);
726f6388 9405 }
28ef6c31 9406
ccc6cda3 9407 break;
726f6388 9408 }
ccc6cda3 9409 free (name);
95732b49
JA
9410
9411 if (ret == 0)
9412 {
9413 ret = alloc_word_desc ();
9414 ret->flags = tflag;
9415 ret->word = temp;
9416 }
9417 return (ret);
726f6388
JA
9418}
9419
cce855bc
JA
9420/* Expand a single ${xxx} expansion. The braces are optional. When
9421 the braces are used, parameter_brace_expand() does the work,
9422 possibly calling param_expand recursively. */
95732b49 9423static WORD_DESC *
cce855bc
JA
9424param_expand (string, sindex, quoted, expanded_something,
9425 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
9426 pflags)
9427 char *string;
9428 int *sindex, quoted, *expanded_something, *contains_dollar_at;
9429 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
9430{
a0c0a00f 9431 char *temp, *temp1, uerror[3], *savecmd;
f73dda09
JA
9432 int zindex, t_index, expok;
9433 unsigned char c;
7117c2d2 9434 intmax_t number;
cce855bc 9435 SHELL_VAR *var;
712f80b0 9436 WORD_LIST *list, *l;
95732b49 9437 WORD_DESC *tdesc, *ret;
712f80b0 9438 int tflag, nullarg;
cce855bc 9439
a0c0a00f 9440/*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
cce855bc
JA
9441 zindex = *sindex;
9442 c = string[++zindex];
9443
9444 temp = (char *)NULL;
95732b49
JA
9445 ret = tdesc = (WORD_DESC *)NULL;
9446 tflag = 0;
cce855bc
JA
9447
9448 /* Do simple cases first. Switch on what follows '$'. */
9449 switch (c)
9450 {
9451 /* $0 .. $9? */
9452 case '0':
9453 case '1':
9454 case '2':
9455 case '3':
9456 case '4':
9457 case '5':
9458 case '6':
9459 case '7':
9460 case '8':
9461 case '9':
f73dda09 9462 temp1 = dollar_vars[TODIGIT (c)];
d233b485 9463 /* This doesn't get called when (pflags&PF_IGNUNBOUND) != 0 */
cce855bc
JA
9464 if (unbound_vars_is_error && temp1 == (char *)NULL)
9465 {
7117c2d2
JA
9466 uerror[0] = '$';
9467 uerror[1] = c;
9468 uerror[2] = '\0';
712f80b0 9469 set_exit_status (EXECUTION_FAILURE);
0001803f 9470 err_unboundvar (uerror);
95732b49 9471 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc 9472 }
b80f6443
JA
9473 if (temp1)
9474 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9475 ? quote_string (temp1)
9476 : quote_escapes (temp1);
9477 else
9478 temp = (char *)NULL;
95732b49 9479
cce855bc
JA
9480 break;
9481
9482 /* $$ -- pid of the invoking shell. */
9483 case '$':
9484 temp = itos (dollar_dollar_pid);
9485 break;
9486
9487 /* $# -- number of positional parameters. */
9488 case '#':
9489 temp = itos (number_of_args ());
9490 break;
9491
9492 /* $? -- return value of the last synchronous command. */
9493 case '?':
9494 temp = itos (last_command_exit_value);
9495 break;
9496
9497 /* $- -- flags supplied to the shell on invocation or by `set'. */
9498 case '-':
9499 temp = which_set_flags ();
9500 break;
9501
9502 /* $! -- Pid of the last asynchronous command. */
9503 case '!':
9504 /* If no asynchronous pids have been created, expand to nothing.
9505 If `set -u' has been executed, and no async processes have
9506 been created, this is an expansion error. */
9507 if (last_asynchronous_pid == NO_PID)
9508 {
9509 if (expanded_something)
9510 *expanded_something = 0;
9511 temp = (char *)NULL;
d233b485 9512 if (unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
cce855bc 9513 {
7117c2d2
JA
9514 uerror[0] = '$';
9515 uerror[1] = c;
9516 uerror[2] = '\0';
712f80b0 9517 set_exit_status (EXECUTION_FAILURE);
0001803f 9518 err_unboundvar (uerror);
95732b49 9519 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
cce855bc
JA
9520 }
9521 }
9522 else
f73dda09 9523 temp = itos (last_asynchronous_pid);
cce855bc
JA
9524 break;
9525
9526 /* The only difference between this and $@ is when the arg is quoted. */
9527 case '*': /* `$*' */
9528 list = list_rest_of_args ();
9529
89a92869
CR
9530#if 0
9531 /* According to austin-group posix proposal by Geoff Clare in
9532 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
9533
9534 "The shell shall write a message to standard error and
9535 immediately exit when it tries to expand an unset parameter
9536 other than the '@' and '*' special parameters."
9537 */
9538
9539 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
9540 {
9541 uerror[0] = '$';
9542 uerror[1] = '*';
9543 uerror[2] = '\0';
712f80b0 9544 set_exit_status (EXECUTION_FAILURE);
89a92869 9545 err_unboundvar (uerror);
3185942a
JA
9546 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9547 }
89a92869 9548#endif
3185942a 9549
cce855bc
JA
9550 /* If there are no command-line arguments, this should just
9551 disappear if there are other characters in the expansion,
9552 even if it's quoted. */
9553 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
9554 temp = (char *)NULL;
0001803f 9555 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
cce855bc
JA
9556 {
9557 /* If we have "$*" we want to make a string of the positional
9558 parameters, separated by the first character of $IFS, and
9559 quote the whole string, including the separators. If IFS
9560 is unset, the parameters are separated by ' '; if $IFS is
9561 null, the parameters are concatenated. */
d233b485 9562 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list, quoted, 0) : string_list (list);
495aee44
CR
9563 if (temp)
9564 {
a0c0a00f 9565 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
495aee44
CR
9566 if (*temp == 0)
9567 tflag |= W_HASQUOTEDNULL;
a0c0a00f
CR
9568 if (temp != temp1)
9569 free (temp);
495aee44
CR
9570 temp = temp1;
9571 }
cce855bc
JA
9572 }
9573 else
28ef6c31 9574 {
95732b49
JA
9575 /* We check whether or not we're eventually going to split $* here,
9576 for example when IFS is empty and we are processing the rhs of
9577 an assignment statement. In that case, we don't separate the
9578 arguments at all. Otherwise, if the $* is not quoted it is
9579 identical to $@ */
d233b485
CR
9580 if (expand_no_split_dollar_star && quoted == 0 && ifs_is_set == 0 && (pflags & PF_ASSIGNRHS))
9581 {
712f80b0
CR
9582 /* Posix interp 888: RHS of assignment, IFS unset: no splitting,
9583 separate with space */
9584 temp1 = string_list_dollar_star (list, quoted, pflags);
9585 temp = temp1 ? quote_string (temp1) : temp1;
9586 /* XXX - tentative - note that we saw a quoted null here */
9587 if (temp1 && *temp1 == 0 && QUOTED_NULL (temp))
9588 tflag |= W_SAWQUOTEDNULL;
9589 FREE (temp1);
d233b485
CR
9590 }
9591 else if (expand_no_split_dollar_star && quoted == 0 && ifs_is_null && (pflags & PF_ASSIGNRHS))
9592 {
9593 /* Posix interp 888: RHS of assignment, IFS set to '' */
9594 temp1 = string_list_dollar_star (list, quoted, pflags);
9595 temp = temp1 ? quote_escapes (temp1) : temp1;
9596 FREE (temp1);
9597 }
9598 else if (expand_no_split_dollar_star && quoted == 0 && ifs_is_set && ifs_is_null == 0 && (pflags & PF_ASSIGNRHS))
9599 {
9600 /* Posix interp 888: RHS of assignment, IFS set to non-null value */
9601 temp1 = string_list_dollar_star (list, quoted, pflags);
9602 temp = temp1 ? quote_string (temp1) : temp1;
712f80b0
CR
9603
9604 /* XXX - tentative - note that we saw a quoted null here */
9605 if (temp1 && *temp1 == 0 && QUOTED_NULL (temp))
9606 tflag |= W_SAWQUOTEDNULL;
d233b485
CR
9607 FREE (temp1);
9608 }
9609 /* XXX - should we check ifs_is_set here as well? */
95732b49 9610# if defined (HANDLE_MULTIBYTE)
d233b485 9611 else if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
95732b49 9612# else
d233b485 9613 else if (expand_no_split_dollar_star && ifs_firstc == 0)
95732b49 9614# endif
d233b485
CR
9615 /* Posix interp 888: not RHS, no splitting, IFS set to '' */
9616 temp = string_list_dollar_star (list, quoted, 0);
95732b49 9617 else
ac50fbac 9618 {
a0c0a00f 9619 temp = string_list_dollar_at (list, quoted, 0);
d233b485
CR
9620 /* Set W_SPLITSPACE to make sure the individual positional
9621 parameters are split into separate arguments */
9622#if 0
ac50fbac 9623 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
d233b485
CR
9624#else /* change with bash-5.0 */
9625 if (quoted == 0 && ifs_is_null)
9626#endif
ac50fbac 9627 tflag |= W_SPLITSPACE;
a0c0a00f
CR
9628 /* If we're not quoted but we still don't want word splitting, make
9629 we quote the IFS characters to protect them from splitting (e.g.,
9630 when $@ is in the string as well). */
9631 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
9632 {
9633 temp1 = quote_string (temp);
9634 free (temp);
9635 temp = temp1;
9636 }
ac50fbac
CR
9637 }
9638
28ef6c31
JA
9639 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
9640 *contains_dollar_at = 1;
9641 }
cce855bc
JA
9642
9643 dispose_words (list);
9644 break;
9645
9646 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
9647 means that we have to turn quoting off after we split into
9648 the individually quoted arguments so that the final split
9649 on the first character of $IFS is still done. */
9650 case '@': /* `$@' */
9651 list = list_rest_of_args ();
9652
89a92869
CR
9653#if 0
9654 /* According to austin-group posix proposal by Geoff Clare in
9655 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
9656
9657 "The shell shall write a message to standard error and
9658 immediately exit when it tries to expand an unset parameter
9659 other than the '@' and '*' special parameters."
9660 */
9661
9662 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
3185942a
JA
9663 {
9664 uerror[0] = '$';
9665 uerror[1] = '@';
9666 uerror[2] = '\0';
712f80b0 9667 set_exit_status (EXECUTION_FAILURE);
89a92869 9668 err_unboundvar (uerror);
3185942a
JA
9669 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9670 }
89a92869 9671#endif
3185942a 9672
712f80b0
CR
9673 for (nullarg = 0, l = list; l; l = l->next)
9674 {
9675 if (l->word && (l->word->word == 0 || l->word->word[0] == 0))
9676 nullarg = 1;
9677 }
9678
cce855bc
JA
9679 /* We want to flag the fact that we saw this. We can't turn
9680 off quoting entirely, because other characters in the
9681 string might need it (consider "\"$@\""), but we need some
9682 way to signal that the final split on the first character
9683 of $IFS should be done, even though QUOTED is 1. */
0001803f 9684 /* XXX - should this test include Q_PATQUOTE? */
cce855bc
JA
9685 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9686 *quoted_dollar_at_p = 1;
9687 if (contains_dollar_at)
9688 *contains_dollar_at = 1;
9689
9690 /* We want to separate the positional parameters with the first
9691 character of $IFS in case $IFS is something other than a space.
9692 We also want to make sure that splitting is done no matter what --
9693 according to POSIX.2, this expands to a list of the positional
9694 parameters no matter what IFS is set to. */
3b34f6e6
CR
9695 /* XXX - what to do when in a context where word splitting is not
9696 performed? Even when IFS is not the default, posix seems to imply
712f80b0
CR
9697 that we have to expand $@ to all the positional parameters and
9698 separate them with spaces, which are preserved because word splitting
9699 doesn't take place. See below for how we use PF_NOSPLIT2 here. */
d233b485
CR
9700
9701 /* These are the cases where word splitting will not be performed. */
9702 if (pflags & PF_ASSIGNRHS)
712f80b0
CR
9703 {
9704 temp = string_list_dollar_at (list, (quoted|Q_DOUBLE_QUOTES), pflags);
9705 if (nullarg)
9706 tflag |= W_HASQUOTEDNULL; /* we know quoting produces quoted nulls */
9707 }
9708
d233b485
CR
9709 /* This needs to match what expand_word_internal does with non-quoted $@
9710 does with separating with spaces. Passing Q_DOUBLE_QUOTES means that
9711 the characters in LIST will be quoted, and PF_ASSIGNRHS ensures that
9712 they will separated by spaces. After doing this, we need the special
9713 handling for PF_NOSPLIT2 in expand_word_internal to remove the CTLESC
9714 quotes. */
9715 else if (pflags & PF_NOSPLIT2)
9716 {
9717#if defined (HANDLE_MULTIBYTE)
9718 if (quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc[0] != ' ')
9719#else
9720 if (quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc != ' ')
9721#endif
9722 /* Posix interp 888 */
9723 temp = string_list_dollar_at (list, Q_DOUBLE_QUOTES, pflags);
9724 else
9725 temp = string_list_dollar_at (list, quoted, pflags);
9726 }
9727 else
9728 temp = string_list_dollar_at (list, quoted, pflags);
cce855bc 9729
ac50fbac 9730 tflag |= W_DOLLARAT;
cce855bc
JA
9731 dispose_words (list);
9732 break;
9733
9734 case LBRACE:
0001803f 9735 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
95732b49
JA
9736 quoted_dollar_at_p,
9737 contains_dollar_at);
9738
95732b49
JA
9739 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9740 return (tdesc);
9741 temp = tdesc ? tdesc->word : (char *)0;
cce855bc
JA
9742
9743 /* XXX */
bb70624e 9744 /* Quoted nulls should be removed if there is anything else
cce855bc
JA
9745 in the string. */
9746 /* Note that we saw the quoted null so we can add one back at
9747 the end of this function if there are no other characters
28ef6c31
JA
9748 in the string, discard TEMP, and go on. The exception to
9749 this is when we have "${@}" and $1 is '', since $@ needs
9750 special handling. */
95732b49 9751 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
cce855bc
JA
9752 {
9753 if (had_quoted_null_p)
9754 *had_quoted_null_p = 1;
28ef6c31
JA
9755 if (*quoted_dollar_at_p == 0)
9756 {
9757 free (temp);
95732b49 9758 tdesc->word = temp = (char *)NULL;
28ef6c31
JA
9759 }
9760
cce855bc
JA
9761 }
9762
95732b49 9763 ret = tdesc;
cce855bc
JA
9764 goto return0;
9765
9766 /* Do command or arithmetic substitution. */
9767 case LPAREN:
9768 /* We have to extract the contents of this paren substitution. */
9769 t_index = zindex + 1;
a0c0a00f
CR
9770 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
9771 as arithmetic substitution immediately. */
9772 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
cce855bc
JA
9773 zindex = t_index;
9774
9775 /* For Posix.2-style `$(( ))' arithmetic substitution,
28ef6c31 9776 extract the expression and pass it to the evaluator. */
cce855bc
JA
9777 if (temp && *temp == LPAREN)
9778 {
9779 char *temp2;
9780 temp1 = temp + 1;
9781 temp2 = savestring (temp1);
9782 t_index = strlen (temp2) - 1;
9783
9784 if (temp2[t_index] != RPAREN)
9785 {
9786 free (temp2);
9787 goto comsub;
9788 }
9789
9790 /* Cut off ending `)' */
9791 temp2[t_index] = '\0';
9792
0628567a
JA
9793 if (chk_arithsub (temp2, t_index) == 0)
9794 {
9795 free (temp2);
0001803f
CR
9796#if 0
9797 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
9798#endif
0628567a
JA
9799 goto comsub;
9800 }
9801
cce855bc 9802 /* Expand variables found inside the expression. */
a0c0a00f 9803 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
9804 free (temp2);
9805
9806arithsub:
9807 /* No error messages. */
a0c0a00f 9808 savecmd = this_command_name;
cce855bc 9809 this_command_name = (char *)NULL;
d233b485 9810 number = evalexp (temp1, EXP_EXPANDED, &expok);
a0c0a00f 9811 this_command_name = savecmd;
cce855bc
JA
9812 free (temp);
9813 free (temp1);
9814 if (expok == 0)
9815 {
9816 if (interactive_shell == 0 && posixly_correct)
9817 {
712f80b0 9818 set_exit_status (EXECUTION_FAILURE);
95732b49 9819 return (&expand_wdesc_fatal);
cce855bc
JA
9820 }
9821 else
95732b49 9822 return (&expand_wdesc_error);
cce855bc
JA
9823 }
9824 temp = itos (number);
9825 break;
9826 }
9827
9828comsub:
b80f6443
JA
9829 if (pflags & PF_NOCOMSUB)
9830 /* we need zindex+1 because string[zindex] == RPAREN */
9831 temp1 = substring (string, *sindex, zindex+1);
9832 else
3185942a 9833 {
d233b485 9834 tdesc = command_substitute (temp, quoted, pflags&PF_ASSIGNRHS);
3185942a
JA
9835 temp1 = tdesc ? tdesc->word : (char *)NULL;
9836 if (tdesc)
9837 dispose_word_desc (tdesc);
9838 }
cce855bc
JA
9839 FREE (temp);
9840 temp = temp1;
9841 break;
9842
9843 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
9844 away in a future bash release. */
d233b485 9845 case '[': /*]*/
bb70624e 9846 /* Extract the contents of this arithmetic substitution. */
cce855bc
JA
9847 t_index = zindex + 1;
9848 temp = extract_arithmetic_subst (string, &t_index);
9849 zindex = t_index;
3185942a
JA
9850 if (temp == 0)
9851 {
9852 temp = savestring (string);
9853 if (expanded_something)
9854 *expanded_something = 0;
9855 goto return0;
9856 }
cce855bc
JA
9857
9858 /* Do initial variable expansion. */
a0c0a00f 9859 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
cce855bc
JA
9860
9861 goto arithsub;
9862
9863 default:
9864 /* Find the variable in VARIABLE_LIST. */
9865 temp = (char *)NULL;
9866
9867 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
9868 ;
9869 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
9870
9871 /* If this isn't a variable name, then just output the `$'. */
9872 if (temp1 == 0 || *temp1 == '\0')
9873 {
9874 FREE (temp1);
f73dda09 9875 temp = (char *)xmalloc (2);
cce855bc
JA
9876 temp[0] = '$';
9877 temp[1] = '\0';
9878 if (expanded_something)
9879 *expanded_something = 0;
9880 goto return0;
9881 }
9882
9883 /* If the variable exists, return its value cell. */
9884 var = find_variable (temp1);
9885
7117c2d2 9886 if (var && invisible_p (var) == 0 && var_isset (var))
cce855bc
JA
9887 {
9888#if defined (ARRAY_VARS)
3185942a 9889 if (assoc_p (var) || array_p (var))
cce855bc 9890 {
3185942a
JA
9891 temp = array_p (var) ? array_reference (array_cell (var), 0)
9892 : assoc_reference (assoc_cell (var), "0");
cce855bc 9893 if (temp)
b80f6443
JA
9894 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9895 ? quote_string (temp)
9896 : quote_escapes (temp);
9897 else if (unbound_vars_is_error)
9898 goto unbound_variable;
cce855bc
JA
9899 }
9900 else
9901#endif
b80f6443
JA
9902 {
9903 temp = value_cell (var);
9904
9905 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9906 ? quote_string (temp)
d233b485
CR
9907 : ((pflags & PF_ASSIGNRHS) ? quote_rhs (temp)
9908 : quote_escapes (temp));
b80f6443
JA
9909 }
9910
cce855bc 9911 free (temp1);
7117c2d2 9912
cce855bc
JA
9913 goto return0;
9914 }
b64a0e1d
CR
9915 else if (var && (invisible_p (var) || var_isset (var) == 0))
9916 temp = (char *)NULL;
a0c0a00f 9917 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
ac50fbac
CR
9918 {
9919 temp = nameref_cell (var);
9920#if defined (ARRAY_VARS)
a0c0a00f 9921 if (temp && *temp && valid_array_reference (temp, 0))
ac50fbac
CR
9922 {
9923 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
9924 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9925 return (tdesc);
9926 ret = tdesc;
9927 goto return0;
9928 }
9929 else
9930#endif
9931 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
9932 if (temp && *temp && legal_identifier (temp) == 0)
9933 {
712f80b0 9934 set_exit_status (EXECUTION_FAILURE);
ac50fbac
CR
9935 report_error (_("%s: invalid variable name for name reference"), temp);
9936 return (&expand_wdesc_error); /* XXX */
9937 }
9938 else
9939 temp = (char *)NULL;
9940 }
cce855bc
JA
9941
9942 temp = (char *)NULL;
9943
b80f6443 9944unbound_variable:
cce855bc 9945 if (unbound_vars_is_error)
0001803f 9946 {
712f80b0 9947 set_exit_status (EXECUTION_FAILURE);
0001803f
CR
9948 err_unboundvar (temp1);
9949 }
cce855bc
JA
9950 else
9951 {
9952 free (temp1);
9953 goto return0;
9954 }
9955
9956 free (temp1);
712f80b0 9957 set_exit_status (EXECUTION_FAILURE);
cce855bc 9958 return ((unbound_vars_is_error && interactive_shell == 0)
95732b49
JA
9959 ? &expand_wdesc_fatal
9960 : &expand_wdesc_error);
cce855bc
JA
9961 }
9962
9963 if (string[zindex])
9964 zindex++;
9965
9966return0:
9967 *sindex = zindex;
95732b49
JA
9968
9969 if (ret == 0)
9970 {
9971 ret = alloc_word_desc ();
9972 ret->flags = tflag; /* XXX */
9973 ret->word = temp;
9974 }
9975 return ret;
cce855bc
JA
9976}
9977
a0c0a00f
CR
9978void
9979invalidate_cached_quoted_dollar_at ()
9980{
9981 dispose_words (cached_quoted_dollar_at);
9982 cached_quoted_dollar_at = 0;
9983}
9984
cce855bc
JA
9985/* Make a word list which is the result of parameter and variable
9986 expansion, command substitution, arithmetic substitution, and
9987 quote removal of WORD. Return a pointer to a WORD_LIST which is
9988 the result of the expansion. If WORD contains a null word, the
9989 word list returned is also null.
726f6388 9990
ccc6cda3
JA
9991 QUOTED contains flag values defined in shell.h.
9992
b72432fd
JA
9993 ISEXP is used to tell expand_word_internal that the word should be
9994 treated as the result of an expansion. This has implications for
9995 how IFS characters in the word are treated.
9996
726f6388
JA
9997 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
9998 they point to an integer value which receives information about expansion.
9999 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
10000 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
10001 else zero.
10002
10003 This only does word splitting in the case of $@ expansion. In that
10004 case, we split on ' '. */
10005
10006/* Values for the local variable quoted_state. */
10007#define UNQUOTED 0
10008#define PARTIALLY_QUOTED 1
10009#define WHOLLY_QUOTED 2
10010
10011static WORD_LIST *
b72432fd 10012expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
726f6388 10013 WORD_DESC *word;
b72432fd 10014 int quoted, isexp;
726f6388
JA
10015 int *contains_dollar_at;
10016 int *expanded_something;
10017{
ccc6cda3
JA
10018 WORD_LIST *list;
10019 WORD_DESC *tword;
726f6388
JA
10020
10021 /* The intermediate string that we build while expanding. */
ccc6cda3 10022 char *istring;
726f6388
JA
10023
10024 /* The current size of the above object. */
a0c0a00f 10025 size_t istring_size;
726f6388
JA
10026
10027 /* Index into ISTRING. */
ccc6cda3 10028 int istring_index;
726f6388
JA
10029
10030 /* Temporary string storage. */
ccc6cda3 10031 char *temp, *temp1;
726f6388
JA
10032
10033 /* The text of WORD. */
ccc6cda3 10034 register char *string;
726f6388 10035
7117c2d2
JA
10036 /* The size of STRING. */
10037 size_t string_size;
10038
726f6388 10039 /* The index into STRING. */
ccc6cda3 10040 int sindex;
726f6388
JA
10041
10042 /* This gets 1 if we see a $@ while quoted. */
ccc6cda3 10043 int quoted_dollar_at;
726f6388
JA
10044
10045 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
10046 whether WORD contains no quoting characters, a partially quoted
10047 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
ccc6cda3
JA
10048 int quoted_state;
10049
95732b49 10050 /* State flags */
ccc6cda3 10051 int had_quoted_null;
d233b485 10052 int has_quoted_ifs; /* did we add a quoted $IFS character here? */
0b913689 10053 int has_dollar_at, temp_has_dollar_at;
ac50fbac 10054 int split_on_spaces;
d233b485 10055 int local_expanded;
28ef6c31 10056 int tflag;
0001803f 10057 int pflags; /* flags passed to param_expand */
a0c0a00f 10058 int mb_cur_max;
726f6388 10059
95732b49
JA
10060 int assignoff; /* If assignment, offset of `=' */
10061
f73dda09 10062 register unsigned char c; /* Current character. */
726f6388 10063 int t_index; /* For calls to string_extract_xxx. */
726f6388 10064
bb70624e 10065 char twochars[2];
b72432fd 10066
7117c2d2
JA
10067 DECLARE_MBSTATE;
10068
a0c0a00f
CR
10069 /* OK, let's see if we can optimize a common idiom: "$@" */
10070 if (STREQ (word->word, "\"$@\"") &&
10071 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
10072 dollar_vars[1]) /* XXX - check IFS here as well? */
10073 {
10074 if (contains_dollar_at)
10075 *contains_dollar_at = 1;
10076 if (expanded_something)
10077 *expanded_something = 1;
10078 if (cached_quoted_dollar_at)
10079 return (copy_word_list (cached_quoted_dollar_at));
10080 list = list_rest_of_args ();
10081 list = quote_list (list);
10082 cached_quoted_dollar_at = copy_word_list (list);
10083 return (list);
10084 }
10085
f73dda09 10086 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
ccc6cda3 10087 istring[istring_index = 0] = '\0';
cce855bc 10088 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
d233b485 10089 has_quoted_ifs = 0;
ac50fbac 10090 split_on_spaces = 0;
ccc6cda3
JA
10091 quoted_state = UNQUOTED;
10092
10093 string = word->word;
10094 if (string == 0)
10095 goto finished_with_string;
a0c0a00f
CR
10096 mb_cur_max = MB_CUR_MAX;
10097
95732b49 10098 /* Don't need the string length for the SADD... and COPY_ macros unless
d233b485 10099 multibyte characters are possible, but do need it for bounds checking. */
a0c0a00f 10100 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
726f6388
JA
10101
10102 if (contains_dollar_at)
10103 *contains_dollar_at = 0;
10104
95732b49
JA
10105 assignoff = -1;
10106
726f6388
JA
10107 /* Begin the expansion. */
10108
ccc6cda3 10109 for (sindex = 0; ;)
726f6388
JA
10110 {
10111 c = string[sindex];
10112
ac50fbac 10113 /* Case on top-level character. */
726f6388
JA
10114 switch (c)
10115 {
10116 case '\0':
10117 goto finished_with_string;
10118
10119 case CTLESC:
7117c2d2
JA
10120 sindex++;
10121#if HANDLE_MULTIBYTE
a0c0a00f 10122 if (mb_cur_max > 1 && string[sindex])
7117c2d2 10123 {
b80f6443 10124 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
10125 }
10126 else
10127#endif
10128 {
10129 temp = (char *)xmalloc (3);
10130 temp[0] = CTLESC;
10131 temp[1] = c = string[sindex];
10132 temp[2] = '\0';
10133 }
726f6388 10134
cce855bc 10135dollar_add_string:
726f6388
JA
10136 if (string[sindex])
10137 sindex++;
10138
cce855bc
JA
10139add_string:
10140 if (temp)
10141 {
10142 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
10143 temp = (char *)0;
10144 }
10145
10146 break;
726f6388
JA
10147
10148#if defined (PROCESS_SUBSTITUTION)
10149 /* Process substitution. */
10150 case '<':
10151 case '>':
10152 {
d233b485 10153 /* XXX - technically this should only be expanded at the start
a0c0a00f 10154 of a word */
712f80b0 10155 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)))
726f6388 10156 {
bb70624e 10157 sindex--; /* add_character: label increments sindex */
726f6388
JA
10158 goto add_character;
10159 }
10160 else
cce855bc 10161 t_index = sindex + 1; /* skip past both '<' and LPAREN */
726f6388 10162
85b94814 10163 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
ccc6cda3 10164 sindex = t_index;
726f6388
JA
10165
10166 /* If the process substitution specification is `<()', we want to
10167 open the pipe for writing in the child and produce output; if
10168 it is `>()', we want to open the pipe for reading in the child
10169 and consume input. */
ccc6cda3 10170 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
726f6388
JA
10171
10172 FREE (temp1);
10173
10174 goto dollar_add_string;
10175 }
10176#endif /* PROCESS_SUBSTITUTION */
10177
95732b49
JA
10178 case '=':
10179 /* Posix.2 section 3.6.1 says that tildes following `=' in words
10180 which are not assignment statements are not expanded. If the
10181 shell isn't in posix mode, though, we perform tilde expansion
10182 on `likely candidate' unquoted assignment statements (flags
10183 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
10184 contains an unquoted :~ or =~. Something to think about: we
10185 now have a flag that says to perform tilde expansion on arguments
10186 to `assignment builtins' like declare and export that look like
10187 assignment statements. We now do tilde expansion on such words
10188 even in POSIX mode. */
10189 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
17345e5a 10190 {
0001803f 10191 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
10192 goto add_ifs_character;
10193 else
10194 goto add_character;
10195 }
95732b49 10196 /* If we're not in posix mode or forcing assignment-statement tilde
3eb0018e
CR
10197 expansion, note where the first `=' appears in the word and prepare
10198 to do tilde expansion following the first `='. We have to keep
10199 track of the first `=' (using assignoff) to avoid being confused
10200 by an `=' in the rhs of the assignment statement. */
95732b49
JA
10201 if ((word->flags & W_ASSIGNMENT) &&
10202 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
10203 assignoff == -1 && sindex > 0)
10204 assignoff = sindex;
10205 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
10206 word->flags |= W_ITILDE;
a0c0a00f 10207
a0c0a00f
CR
10208 if (word->flags & W_ASSIGNARG)
10209 word->flags |= W_ASSIGNRHS; /* affects $@ */
10210
0001803f 10211 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
d233b485
CR
10212 {
10213 has_quoted_ifs++;
10214 goto add_ifs_character;
10215 }
17345e5a
JA
10216 else
10217 goto add_character;
95732b49
JA
10218
10219 case ':':
d233b485 10220 if (word->flags & (W_NOTILDE|W_NOASSNTILDE))
17345e5a 10221 {
0001803f 10222 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
10223 goto add_ifs_character;
10224 else
10225 goto add_character;
10226 }
95732b49 10227
3eb0018e
CR
10228 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS)) &&
10229 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
95732b49
JA
10230 string[sindex+1] == '~')
10231 word->flags |= W_ITILDE;
17345e5a 10232
0001803f 10233 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
17345e5a
JA
10234 goto add_ifs_character;
10235 else
10236 goto add_character;
95732b49
JA
10237
10238 case '~':
10239 /* If the word isn't supposed to be tilde expanded, or we're not
10240 at the start of a word or after an unquoted : or = in an
d233b485
CR
10241 assignment statement, we don't do tilde expansion. We don't
10242 do tilde expansion if quoted or in an arithmetic context. */
10243
95732b49
JA
10244 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
10245 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
d233b485 10246 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
95732b49
JA
10247 {
10248 word->flags &= ~W_ITILDE;
0001803f 10249 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
17345e5a
JA
10250 goto add_ifs_character;
10251 else
10252 goto add_character;
95732b49
JA
10253 }
10254
10255 if (word->flags & W_ASSIGNRHS)
10256 tflag = 2;
10257 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
10258 tflag = 1;
10259 else
10260 tflag = 0;
10261
10262 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
10263
10264 word->flags &= ~W_ITILDE;
10265
10266 if (temp && *temp && t_index > 0)
10267 {
10268 temp1 = bash_tilde_expand (temp, tflag);
0628567a
JA
10269 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
10270 {
10271 FREE (temp);
10272 FREE (temp1);
10273 goto add_character; /* tilde expansion failed */
10274 }
95732b49
JA
10275 free (temp);
10276 temp = temp1;
10277 sindex += t_index;
3185942a 10278 goto add_quoted_string; /* XXX was add_string */
95732b49
JA
10279 }
10280 else
10281 {
10282 FREE (temp);
10283 goto add_character;
10284 }
10285
726f6388 10286 case '$':
726f6388
JA
10287 if (expanded_something)
10288 *expanded_something = 1;
d233b485 10289 local_expanded = 1;
726f6388 10290
0b913689 10291 temp_has_dollar_at = 0;
0001803f
CR
10292 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
10293 if (word->flags & W_NOSPLIT2)
10294 pflags |= PF_NOSPLIT2;
ac50fbac
CR
10295 if (word->flags & W_ASSIGNRHS)
10296 pflags |= PF_ASSIGNRHS;
a0c0a00f
CR
10297 if (word->flags & W_COMPLETE)
10298 pflags |= PF_COMPLETE;
712f80b0 10299
95732b49 10300 tword = param_expand (string, &sindex, quoted, expanded_something,
0b913689 10301 &temp_has_dollar_at, &quoted_dollar_at,
0001803f 10302 &had_quoted_null, pflags);
0b913689 10303 has_dollar_at += temp_has_dollar_at;
ac50fbac 10304 split_on_spaces += (tword->flags & W_SPLITSPACE);
726f6388 10305
95732b49 10306 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
726f6388 10307 {
cce855bc
JA
10308 free (string);
10309 free (istring);
95732b49
JA
10310 return ((tword == &expand_wdesc_error) ? &expand_word_error
10311 : &expand_word_fatal);
cce855bc
JA
10312 }
10313 if (contains_dollar_at && has_dollar_at)
10314 *contains_dollar_at = 1;
95732b49
JA
10315
10316 if (tword && (tword->flags & W_HASQUOTEDNULL))
d233b485
CR
10317 had_quoted_null = 1; /* note for later */
10318 if (tword && (tword->flags & W_SAWQUOTEDNULL))
10319 had_quoted_null = 1; /* XXX */
95732b49 10320
ac50fbac 10321 temp = tword ? tword->word : (char *)NULL;
95732b49
JA
10322 dispose_word_desc (tword);
10323
a601c749
CR
10324 /* Kill quoted nulls; we will add them back at the end of
10325 expand_word_internal if nothing else in the string */
10326 if (had_quoted_null && temp && QUOTED_NULL (temp))
10327 {
10328 FREE (temp);
10329 temp = (char *)NULL;
10330 }
10331
cce855bc
JA
10332 goto add_string;
10333 break;
726f6388 10334
cce855bc
JA
10335 case '`': /* Backquoted command substitution. */
10336 {
b80f6443 10337 t_index = sindex++;
726f6388 10338
3185942a 10339 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
95732b49
JA
10340 /* The test of sindex against t_index is to allow bare instances of
10341 ` to pass through, for backwards compatibility. */
10342 if (temp == &extract_string_error || temp == &extract_string_fatal)
10343 {
10344 if (sindex - 1 == t_index)
10345 {
10346 sindex = t_index;
10347 goto add_character;
10348 }
712f80b0 10349 set_exit_status (EXECUTION_FAILURE);
3185942a 10350 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
95732b49
JA
10351 free (string);
10352 free (istring);
10353 return ((temp == &extract_string_error) ? &expand_word_error
10354 : &expand_word_fatal);
10355 }
10356
cce855bc
JA
10357 if (expanded_something)
10358 *expanded_something = 1;
d233b485 10359 local_expanded = 1;
726f6388 10360
b80f6443
JA
10361 if (word->flags & W_NOCOMSUB)
10362 /* sindex + 1 because string[sindex] == '`' */
10363 temp1 = substring (string, t_index, sindex + 1);
10364 else
10365 {
10366 de_backslash (temp);
d233b485 10367 tword = command_substitute (temp, quoted, 0);
3185942a
JA
10368 temp1 = tword ? tword->word : (char *)NULL;
10369 if (tword)
10370 dispose_word_desc (tword);
b80f6443 10371 }
cce855bc
JA
10372 FREE (temp);
10373 temp = temp1;
10374 goto dollar_add_string;
10375 }
ccc6cda3 10376
cce855bc
JA
10377 case '\\':
10378 if (string[sindex + 1] == '\n')
10379 {
10380 sindex += 2;
10381 continue;
10382 }
726f6388 10383
cce855bc 10384 c = string[++sindex];
726f6388 10385
cce855bc 10386 if (quoted & Q_HERE_DOCUMENT)
28ef6c31 10387 tflag = CBSHDOC;
cce855bc 10388 else if (quoted & Q_DOUBLE_QUOTES)
28ef6c31 10389 tflag = CBSDQUOTE;
cce855bc 10390 else
28ef6c31
JA
10391 tflag = 0;
10392
495aee44
CR
10393 /* From Posix discussion on austin-group list: Backslash escaping
10394 a } in ${...} is removed. Issue 0000221 */
10395 if ((quoted & Q_DOLBRACE) && c == RBRACE)
10396 {
ac50fbac
CR
10397 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10398 }
10399 /* This is the fix for " $@\ " */
10400 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
10401 {
10402 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10403 DEFAULT_ARRAY_SIZE);
10404 istring[istring_index++] = CTLESC;
10405 istring[istring_index++] = '\\';
10406 istring[istring_index] = '\0';
10407
495aee44
CR
10408 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10409 }
d233b485
CR
10410 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && c == 0)
10411 {
10412 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10413 DEFAULT_ARRAY_SIZE);
10414 istring[istring_index++] = CTLESC;
10415 istring[istring_index++] = '\\';
10416 istring[istring_index] = '\0';
10417 break;
10418 }
495aee44 10419 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
cce855bc 10420 {
7117c2d2 10421 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
bb70624e
JA
10422 }
10423 else if (c == 0)
10424 {
10425 c = CTLNUL;
10426 sindex--; /* add_character: label increments sindex */
10427 goto add_character;
cce855bc
JA
10428 }
10429 else
bb70624e 10430 {
7117c2d2 10431 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
bb70624e 10432 }
726f6388 10433
bb70624e
JA
10434 sindex++;
10435add_twochars:
10436 /* BEFORE jumping here, we need to increment sindex if appropriate */
10437 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10438 DEFAULT_ARRAY_SIZE);
10439 istring[istring_index++] = twochars[0];
10440 istring[istring_index++] = twochars[1];
10441 istring[istring_index] = '\0';
10442
10443 break;
726f6388 10444
cce855bc 10445 case '"':
a0c0a00f 10446 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
cce855bc 10447 goto add_character;
ccc6cda3
JA
10448
10449 t_index = ++sindex;
a0c0a00f 10450 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
ccc6cda3
JA
10451
10452 /* If the quotes surrounded the entire string, then the
10453 whole word was quoted. */
10454 quoted_state = (t_index == 1 && string[sindex] == '\0')
10455 ? WHOLLY_QUOTED
7117c2d2 10456 : PARTIALLY_QUOTED;
ccc6cda3
JA
10457
10458 if (temp && *temp)
726f6388 10459 {
95732b49
JA
10460 tword = alloc_word_desc ();
10461 tword->word = temp;
10462
a0c0a00f 10463 if (word->flags & W_ASSIGNARG)
d233b485 10464 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
a0c0a00f
CR
10465 if (word->flags & W_COMPLETE)
10466 tword->flags |= W_COMPLETE; /* for command substitutions */
4f747edc
CR
10467 if (word->flags & W_NOCOMSUB)
10468 tword->flags |= W_NOCOMSUB;
10469 if (word->flags & W_NOPROCSUB)
10470 tword->flags |= W_NOPROCSUB;
a0c0a00f 10471
712f80b0
CR
10472 if (word->flags & W_ASSIGNRHS)
10473 tword->flags |= W_ASSIGNRHS;
10474
ccc6cda3
JA
10475 temp = (char *)NULL;
10476
d233b485 10477 temp_has_dollar_at = 0; /* does this quoted (sub)string include $@? */
95732b49 10478 /* Need to get W_HASQUOTEDNULL flag through this function. */
0b913689
CR
10479 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
10480 has_dollar_at += temp_has_dollar_at;
726f6388 10481
ccc6cda3
JA
10482 if (list == &expand_word_error || list == &expand_word_fatal)
10483 {
10484 free (istring);
10485 free (string);
10486 /* expand_word_internal has already freed temp_word->word
10487 for us because of the way it prints error messages. */
10488 tword->word = (char *)NULL;
10489 dispose_word (tword);
10490 return list;
10491 }
726f6388 10492
ccc6cda3 10493 dispose_word (tword);
726f6388 10494
ccc6cda3
JA
10495 /* "$@" (a double-quoted dollar-at) expands into nothing,
10496 not even a NULL word, when there are no positional
d233b485
CR
10497 parameters. Posix interp 888 says that other parts of the
10498 word that expand to quoted nulls result in quoted nulls, so
10499 we can't just throw the entire word away if we have "$@"
10500 anywhere in it. We use had_quoted_null to keep track */
a0c0a00f 10501 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
726f6388 10502 {
ccc6cda3
JA
10503 quoted_dollar_at++;
10504 break;
10505 }
10506
d233b485
CR
10507 /* If this list comes back with a quoted null from expansion,
10508 we have either "$x" or "$@" with $1 == ''. In either case,
10509 we need to make sure we add a quoted null argument and
10510 disable the special handling that "$@" gets. */
10511 if (list && list->word && list->next == 0 && (list->word->flags & W_HASQUOTEDNULL))
10512 {
d233b485 10513 if (had_quoted_null && temp_has_dollar_at)
712f80b0 10514 quoted_dollar_at++;
d233b485
CR
10515 had_quoted_null = 1; /* XXX */
10516 }
10517
ccc6cda3
JA
10518 /* If we get "$@", we know we have expanded something, so we
10519 need to remember it for the final split on $IFS. This is
10520 a special case; it's the only case where a quoted string
10521 can expand into more than one word. It's going to come back
10522 from the above call to expand_word_internal as a list with
712f80b0 10523 multiple words. */
ccc6cda3
JA
10524 if (list)
10525 dequote_list (list);
10526
a0c0a00f 10527 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
ccc6cda3
JA
10528 {
10529 quoted_dollar_at++;
10530 if (contains_dollar_at)
10531 *contains_dollar_at = 1;
10532 if (expanded_something)
10533 *expanded_something = 1;
d233b485 10534 local_expanded = 1;
ccc6cda3
JA
10535 }
10536 }
10537 else
10538 {
10539 /* What we have is "". This is a minor optimization. */
f73dda09 10540 FREE (temp);
ccc6cda3 10541 list = (WORD_LIST *)NULL;
d233b485 10542 had_quoted_null = 1; /* note for later */
ccc6cda3
JA
10543 }
10544
10545 /* The code above *might* return a list (consider the case of "$@",
10546 where it returns "$1", "$2", etc.). We can't throw away the
10547 rest of the list, and we have to make sure each word gets added
10548 as quoted. We test on tresult->next: if it is non-NULL, we
10549 quote the whole list, save it to a string with string_list, and
10550 add that string. We don't need to quote the results of this
10551 (and it would be wrong, since that would quote the separators
10552 as well), so we go directly to add_string. */
10553 if (list)
10554 {
10555 if (list->next)
10556 {
bc4cd23c
JA
10557 /* Testing quoted_dollar_at makes sure that "$@" is
10558 split correctly when $IFS does not contain a space. */
10559 temp = quoted_dollar_at
a0c0a00f 10560 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
bc4cd23c 10561 : string_list (quote_list (list));
ccc6cda3 10562 dispose_words (list);
726f6388
JA
10563 goto add_string;
10564 }
10565 else
10566 {
ccc6cda3 10567 temp = savestring (list->word->word);
95732b49 10568 tflag = list->word->flags;
ccc6cda3 10569 dispose_words (list);
95732b49 10570
cce855bc
JA
10571 /* If the string is not a quoted null string, we want
10572 to remove any embedded unquoted CTLNUL characters.
10573 We do not want to turn quoted null strings back into
10574 the empty string, though. We do this because we
10575 want to remove any quoted nulls from expansions that
10576 contain other characters. For example, if we have
10577 x"$*"y or "x$*y" and there are no positional parameters,
7117c2d2 10578 the $* should expand into nothing. */
95732b49
JA
10579 /* We use the W_HASQUOTEDNULL flag to differentiate the
10580 cases: a quoted null character as above and when
10581 CTLNUL is contained in the (non-null) expansion
10582 of some variable. We use the had_quoted_null flag to
10583 pass the value through this function to its caller. */
10584 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
cce855bc 10585 remove_quoted_nulls (temp); /* XXX */
726f6388
JA
10586 }
10587 }
ccc6cda3
JA
10588 else
10589 temp = (char *)NULL;
726f6388 10590
d233b485
CR
10591 if (temp == 0 && quoted_state == PARTIALLY_QUOTED)
10592 had_quoted_null = 1; /* note for later */
10593
ccc6cda3 10594 /* We do not want to add quoted nulls to strings that are only
ac50fbac
CR
10595 partially quoted; we can throw them away. The exception to
10596 this is when we are going to be performing word splitting,
10597 since we have to preserve a null argument if the next character
10598 will cause word splitting. */
712f80b0
CR
10599 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && quoted == 0 && (word->flags & (W_NOSPLIT|W_EXPANDRHS|W_ASSIGNRHS)) == W_EXPANDRHS)
10600 {
10601 c = CTLNUL;
10602 sindex--;
10603 had_quoted_null = 1;
10604 goto add_character;
10605 }
495aee44 10606 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
cce855bc 10607 continue;
726f6388 10608
ccc6cda3 10609 add_quoted_string:
726f6388 10610
ccc6cda3
JA
10611 if (temp)
10612 {
10613 temp1 = temp;
10614 temp = quote_string (temp);
10615 free (temp1);
bb70624e 10616 goto add_string;
ccc6cda3
JA
10617 }
10618 else
10619 {
10620 /* Add NULL arg. */
bb70624e
JA
10621 c = CTLNUL;
10622 sindex--; /* add_character: label increments sindex */
d233b485 10623 had_quoted_null = 1; /* note for later */
bb70624e 10624 goto add_character;
ccc6cda3 10625 }
bb70624e 10626
ccc6cda3 10627 /* break; */
726f6388 10628
ccc6cda3 10629 case '\'':
95732b49 10630 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
ccc6cda3 10631 goto add_character;
726f6388 10632
ccc6cda3
JA
10633 t_index = ++sindex;
10634 temp = string_extract_single_quoted (string, &sindex);
726f6388 10635
ccc6cda3
JA
10636 /* If the entire STRING was surrounded by single quotes,
10637 then the string is wholly quoted. */
10638 quoted_state = (t_index == 1 && string[sindex] == '\0')
10639 ? WHOLLY_QUOTED
7117c2d2 10640 : PARTIALLY_QUOTED;
726f6388 10641
ccc6cda3
JA
10642 /* If all we had was '', it is a null expansion. */
10643 if (*temp == '\0')
10644 {
10645 free (temp);
10646 temp = (char *)NULL;
10647 }
10648 else
7117c2d2 10649 remove_quoted_escapes (temp); /* ??? */
726f6388 10650
d233b485
CR
10651 if (temp == 0 && quoted_state == PARTIALLY_QUOTED)
10652 had_quoted_null = 1; /* note for later */
10653
ccc6cda3 10654 /* We do not want to add quoted nulls to strings that are only
a0c0a00f 10655 partially quoted; such nulls are discarded. See above for the
d233b485 10656 exception, which is when the string is going to be split.
712f80b0
CR
10657 Posix interp 888/1129 */
10658 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && quoted == 0 && (word->flags & (W_NOSPLIT|W_EXPANDRHS|W_ASSIGNRHS)) == W_EXPANDRHS)
10659 {
10660 c = CTLNUL;
10661 sindex--;
10662 goto add_character;
10663 }
10664
a0c0a00f 10665 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
ccc6cda3 10666 continue;
726f6388 10667
bb70624e
JA
10668 /* If we have a quoted null expansion, add a quoted NULL to istring. */
10669 if (temp == 0)
10670 {
10671 c = CTLNUL;
10672 sindex--; /* add_character: label increments sindex */
10673 goto add_character;
10674 }
10675 else
10676 goto add_quoted_string;
10677
ccc6cda3 10678 /* break; */
726f6388 10679
712f80b0
CR
10680 case ' ':
10681 /* If we are in a context where the word is not going to be split, but
10682 we need to account for $@ and $* producing one word for each
10683 positional parameter, add quoted spaces so the spaces in the
10684 expansion of "$@", if any, behave correctly. We still may need to
10685 split if we are expanding the rhs of a word expansion. */
10686 if (ifs_is_null || split_on_spaces || ((word->flags & (W_NOSPLIT|W_NOSPLIT2|W_ASSIGNRHS)) && (word->flags & W_EXPANDRHS) == 0))
10687 {
10688 if (string[sindex])
10689 sindex++;
10690 twochars[0] = CTLESC;
10691 twochars[1] = c;
10692 goto add_twochars;
10693 }
10694 /* FALLTHROUGH */
10695
726f6388 10696 default:
726f6388 10697 /* This is the fix for " $@ " */
712f80b0 10698add_ifs_character:
a0c0a00f 10699 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
726f6388 10700 {
d233b485
CR
10701 if ((quoted&(Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)
10702 has_quoted_ifs++;
712f80b0 10703add_quoted_character:
bb70624e
JA
10704 if (string[sindex]) /* from old goto dollar_add_string */
10705 sindex++;
10706 if (c == 0)
10707 {
10708 c = CTLNUL;
10709 goto add_character;
10710 }
10711 else
10712 {
7117c2d2 10713#if HANDLE_MULTIBYTE
d233b485
CR
10714 /* XXX - should make sure that c is actually multibyte,
10715 otherwise we can use the twochars branch */
a0c0a00f 10716 if (mb_cur_max > 1)
b80f6443
JA
10717 sindex--;
10718
a0c0a00f 10719 if (mb_cur_max > 1)
7117c2d2 10720 {
b80f6443 10721 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7117c2d2
JA
10722 }
10723 else
10724#endif
10725 {
10726 twochars[0] = CTLESC;
10727 twochars[1] = c;
10728 goto add_twochars;
10729 }
bb70624e 10730 }
726f6388
JA
10731 }
10732
7117c2d2
JA
10733 SADD_MBCHAR (temp, string, sindex, string_size);
10734
712f80b0 10735add_character:
ccc6cda3
JA
10736 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
10737 DEFAULT_ARRAY_SIZE);
726f6388
JA
10738 istring[istring_index++] = c;
10739 istring[istring_index] = '\0';
10740
10741 /* Next character. */
10742 sindex++;
10743 }
10744 }
10745
10746finished_with_string:
726f6388
JA
10747 /* OK, we're ready to return. If we have a quoted string, and
10748 quoted_dollar_at is not set, we do no splitting at all; otherwise
10749 we split on ' '. The routines that call this will handle what to
10750 do if nothing has been expanded. */
ccc6cda3
JA
10751
10752 /* Partially and wholly quoted strings which expand to the empty
10753 string are retained as an empty arguments. Unquoted strings
10754 which expand to the empty string are discarded. The single
10755 exception is the case of expanding "$@" when there are no
10756 positional parameters. In that case, we discard the expansion. */
10757
10758 /* Because of how the code that handles "" and '' in partially
10759 quoted strings works, we need to make ISTRING into a QUOTED_NULL
10760 if we saw quoting characters, but the expansion was empty.
10761 "" and '' are tossed away before we get to this point when
10762 processing partially quoted strings. This makes "" and $xxx""
10763 equivalent when xxx is unset. We also look to see whether we
10764 saw a quoted null from a ${} expansion and add one back if we
10765 need to. */
10766
10767 /* If we expand to nothing and there were no single or double quotes
10768 in the word, we throw it away. Otherwise, we return a NULL word.
10769 The single exception is for $@ surrounded by double quotes when
10770 there are no positional parameters. In that case, we also throw
10771 the word away. */
10772
10773 if (*istring == '\0')
10774 {
10775 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
726f6388 10776 {
726f6388
JA
10777 istring[0] = CTLNUL;
10778 istring[1] = '\0';
d233b485
CR
10779 tword = alloc_word_desc ();
10780 tword->word = istring;
10781 istring = 0; /* avoid later free() */
95732b49 10782 tword->flags |= W_HASQUOTEDNULL; /* XXX */
ccc6cda3
JA
10783 list = make_word_list (tword, (WORD_LIST *)NULL);
10784 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
10785 tword->flags |= W_QUOTED;
726f6388 10786 }
ccc6cda3
JA
10787 /* According to sh, ksh, and Posix.2, if a word expands into nothing
10788 and a double-quoted "$@" appears anywhere in it, then the entire
10789 word is removed. */
a0c0a00f
CR
10790 /* XXX - exception appears to be that quoted null strings result in
10791 null arguments */
ccc6cda3
JA
10792 else if (quoted_state == UNQUOTED || quoted_dollar_at)
10793 list = (WORD_LIST *)NULL;
f73dda09
JA
10794 else
10795 list = (WORD_LIST *)NULL;
ccc6cda3
JA
10796 }
10797 else if (word->flags & W_NOSPLIT)
10798 {
d233b485
CR
10799 tword = alloc_word_desc ();
10800 tword->word = istring;
10801 if (had_quoted_null && QUOTED_NULL (istring))
10802 tword->flags |= W_HASQUOTEDNULL;
10803 istring = 0; /* avoid later free() */
ccc6cda3
JA
10804 if (word->flags & W_ASSIGNMENT)
10805 tword->flags |= W_ASSIGNMENT; /* XXX */
95732b49
JA
10806 if (word->flags & W_COMPASSIGN)
10807 tword->flags |= W_COMPASSIGN; /* XXX */
b72432fd
JA
10808 if (word->flags & W_NOGLOB)
10809 tword->flags |= W_NOGLOB; /* XXX */
ac50fbac
CR
10810 if (word->flags & W_NOBRACE)
10811 tword->flags |= W_NOBRACE; /* XXX */
ccc6cda3 10812 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
28ef6c31 10813 tword->flags |= W_QUOTED;
95732b49 10814 list = make_word_list (tword, (WORD_LIST *)NULL);
ccc6cda3 10815 }
712f80b0
CR
10816 else if (word->flags & W_ASSIGNRHS)
10817 {
10818 list = list_string (istring, "", quoted);
10819 tword = list->word;
10820 if (had_quoted_null && QUOTED_NULL (istring))
10821 tword->flags |= W_HASQUOTEDNULL;
10822 free (list);
10823 free (istring);
10824 istring = 0; /* avoid later free() */
10825 goto set_word_flags;
10826 }
ccc6cda3
JA
10827 else
10828 {
10829 char *ifs_chars;
10830
7117c2d2 10831 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
726f6388 10832
cce855bc
JA
10833 /* If we have $@, we need to split the results no matter what. If
10834 IFS is unset or NULL, string_list_dollar_at has separated the
10835 positional parameters with a space, so we split on space (we have
10836 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
10837 string_list_dollar_at has separated the positional parameters
ac50fbac
CR
10838 with the first character of $IFS, so we split on $IFS. If
10839 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
10840 unset or null, and we want to make sure that we split on spaces
a0c0a00f
CR
10841 regardless of what else has happened to IFS since the expansion,
10842 or we expanded "$@" with IFS null and we need to split the positional
10843 parameters into separate words. */
ac50fbac 10844 if (split_on_spaces)
d233b485
CR
10845 {
10846 /* If IFS is not set, and the word is not quoted, we want to split
10847 the individual words on $' \t\n'. We rely on previous steps to
10848 quote the portions of the word that should not be split */
10849 if (ifs_is_set == 0)
10850 list = list_string (istring, " \t\n", 1); /* XXX quoted == 1? */
10851 else
10852 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
10853 }
10854
3b34f6e6
CR
10855 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
10856 don't want to split the result (W_NOSPLIT2), and we are not quoted,
10857 we have already separated the arguments with the first character of
10858 $IFS. In this case, we want to return a list with a single word
10859 with the separator possibly replaced with a space (it's what other
10860 shells seem to do).
10861 quoted_dollar_at is internal to this function and is set if we are
10862 passed an argument that is unquoted (quoted == 0) but we encounter a
10863 double-quoted $@ while expanding it. */
10864 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
10865 {
d233b485 10866 tword = alloc_word_desc ();
3b34f6e6
CR
10867 /* Only split and rejoin if we have to */
10868 if (*ifs_chars && *ifs_chars != ' ')
10869 {
d233b485
CR
10870 /* list_string dequotes CTLESCs in the string it's passed, so we
10871 need it to get the space separation right if space isn't the
10872 first character in IFS (but is present) and to remove the
10873 quoting we added back in param_expand(). */
3b34f6e6 10874 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
d233b485
CR
10875 /* This isn't exactly right in the case where we're expanding
10876 the RHS of an expansion like ${var-$@} where IFS=: (for
10877 example). The W_NOSPLIT2 means we do the separation with :;
10878 the list_string removes the quotes and breaks the string into
10879 a list, and the string_list rejoins it on spaces. When we
10880 return, we expect to be able to split the results, but the
10881 space separation means the right split doesn't happen. */
10882 tword->word = string_list (list);
3b34f6e6
CR
10883 }
10884 else
d233b485
CR
10885 tword->word = istring;
10886 if (had_quoted_null && QUOTED_NULL (istring))
10887 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10888 if (tword->word != istring)
10889 free (istring);
10890 istring = 0; /* avoid later free() */
a0c0a00f
CR
10891 goto set_word_flags;
10892 }
ac50fbac 10893 else if (has_dollar_at && ifs_chars)
cce855bc 10894 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
ccc6cda3
JA
10895 else
10896 {
d233b485
CR
10897 tword = alloc_word_desc ();
10898 if (expanded_something && *expanded_something == 0 && has_quoted_ifs)
10899 tword->word = remove_quoted_ifs (istring);
10900 else
10901 tword->word = istring;
712f80b0 10902 if (had_quoted_null && QUOTED_NULL (istring)) /* should check for more than one */
d233b485
CR
10903 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10904 else if (had_quoted_null)
10905 tword->flags |= W_SAWQUOTEDNULL; /* XXX */
10906 if (tword->word != istring)
10907 free (istring);
10908 istring = 0; /* avoid later free() */
3b34f6e6 10909set_word_flags:
ccc6cda3
JA
10910 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
10911 tword->flags |= W_QUOTED;
10912 if (word->flags & W_ASSIGNMENT)
10913 tword->flags |= W_ASSIGNMENT;
95732b49
JA
10914 if (word->flags & W_COMPASSIGN)
10915 tword->flags |= W_COMPASSIGN;
b72432fd
JA
10916 if (word->flags & W_NOGLOB)
10917 tword->flags |= W_NOGLOB;
ac50fbac
CR
10918 if (word->flags & W_NOBRACE)
10919 tword->flags |= W_NOBRACE;
95732b49 10920 list = make_word_list (tword, (WORD_LIST *)NULL);
726f6388 10921 }
726f6388 10922 }
726f6388 10923
ccc6cda3
JA
10924 free (istring);
10925 return (list);
726f6388
JA
10926}
10927
10928/* **************************************************************** */
10929/* */
10930/* Functions for Quote Removal */
10931/* */
10932/* **************************************************************** */
10933
10934/* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
7117c2d2 10935 backslash quoting rules for within double quotes or a here document. */
726f6388
JA
10936char *
10937string_quote_removal (string, quoted)
10938 char *string;
10939 int quoted;
10940{
7117c2d2
JA
10941 size_t slen;
10942 char *r, *result_string, *temp, *send;
f73dda09
JA
10943 int sindex, tindex, dquote;
10944 unsigned char c;
7117c2d2 10945 DECLARE_MBSTATE;
726f6388
JA
10946
10947 /* The result can be no longer than the original string. */
7117c2d2
JA
10948 slen = strlen (string);
10949 send = string + slen;
10950
10951 r = result_string = (char *)xmalloc (slen + 1);
726f6388 10952
ccc6cda3 10953 for (dquote = sindex = 0; c = string[sindex];)
726f6388
JA
10954 {
10955 switch (c)
10956 {
10957 case '\\':
10958 c = string[++sindex];
3185942a
JA
10959 if (c == 0)
10960 {
10961 *r++ = '\\';
10962 break;
10963 }
28ef6c31 10964 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
726f6388 10965 *r++ = '\\';
ccc6cda3 10966 /* FALLTHROUGH */
726f6388
JA
10967
10968 default:
7117c2d2 10969 SCOPY_CHAR_M (r, string, send, sindex);
726f6388
JA
10970 break;
10971
10972 case '\'':
ccc6cda3 10973 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
726f6388
JA
10974 {
10975 *r++ = c;
10976 sindex++;
ccc6cda3 10977 break;
726f6388 10978 }
ccc6cda3
JA
10979 tindex = sindex + 1;
10980 temp = string_extract_single_quoted (string, &tindex);
10981 if (temp)
726f6388 10982 {
ccc6cda3
JA
10983 strcpy (r, temp);
10984 r += strlen (r);
10985 free (temp);
726f6388 10986 }
ccc6cda3 10987 sindex = tindex;
726f6388
JA
10988 break;
10989
10990 case '"':
10991 dquote = 1 - dquote;
10992 sindex++;
10993 break;
10994 }
10995 }
10996 *r = '\0';
10997 return (result_string);
10998}
10999
ccc6cda3
JA
11000#if 0
11001/* UNUSED */
726f6388
JA
11002/* Perform quote removal on word WORD. This allocates and returns a new
11003 WORD_DESC *. */
11004WORD_DESC *
11005word_quote_removal (word, quoted)
11006 WORD_DESC *word;
11007 int quoted;
11008{
11009 WORD_DESC *w;
11010 char *t;
11011
11012 t = string_quote_removal (word->word, quoted);
95732b49
JA
11013 w = alloc_word_desc ();
11014 w->word = t ? t : savestring ("");
726f6388
JA
11015 return (w);
11016}
11017
11018/* Perform quote removal on all words in LIST. If QUOTED is non-zero,
11019 the members of the list are treated as if they are surrounded by
11020 double quotes. Return a new list, or NULL if LIST is NULL. */
11021WORD_LIST *
11022word_list_quote_removal (list, quoted)
11023 WORD_LIST *list;
11024 int quoted;
11025{
95732b49 11026 WORD_LIST *result, *t, *tresult, *e;
726f6388 11027
ccc6cda3 11028 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 11029 {
7117c2d2 11030 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
95732b49 11031#if 0
726f6388 11032 result = (WORD_LIST *) list_append (result, tresult);
95732b49
JA
11033#else
11034 if (result == 0)
11035 result = e = tresult;
11036 else
11037 {
11038 e->next = tresult;
11039 while (e->next)
11040 e = e->next;
11041 }
11042#endif
726f6388
JA
11043 }
11044 return (result);
11045}
ccc6cda3 11046#endif
726f6388 11047
726f6388
JA
11048/*******************************************
11049 * *
11050 * Functions to perform word splitting *
11051 * *
11052 *******************************************/
11053
7117c2d2
JA
11054void
11055setifs (v)
11056 SHELL_VAR *v;
b72432fd 11057{
7117c2d2
JA
11058 char *t;
11059 unsigned char uc;
11060
11061 ifs_var = v;
95732b49 11062 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
b72432fd 11063
ac50fbac
CR
11064 ifs_is_set = ifs_var != 0;
11065 ifs_is_null = ifs_is_set && (*ifs_value == 0);
11066
95732b49
JA
11067 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
11068 handle multibyte chars in IFS */
7117c2d2
JA
11069 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
11070 for (t = ifs_value ; t && *t; t++)
11071 {
11072 uc = *t;
11073 ifs_cmap[uc] = 1;
11074 }
11075
95732b49
JA
11076#if defined (HANDLE_MULTIBYTE)
11077 if (ifs_value == 0)
11078 {
d233b485 11079 ifs_firstc[0] = '\0'; /* XXX - ? */
95732b49
JA
11080 ifs_firstc_len = 1;
11081 }
11082 else
11083 {
d233b485
CR
11084 if (locale_utf8locale && UTF8_SINGLEBYTE (*ifs_value))
11085 ifs_firstc_len = (*ifs_value != 0) ? 1 : 0;
11086 else
11087 {
11088 size_t ifs_len;
11089 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
11090 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
11091 }
95732b49
JA
11092 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
11093 {
11094 ifs_firstc[0] = ifs_value[0];
11095 ifs_firstc[1] = '\0';
11096 ifs_firstc_len = 1;
11097 }
11098 else
11099 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
11100 }
11101#else
7117c2d2 11102 ifs_firstc = ifs_value ? *ifs_value : 0;
95732b49 11103#endif
7117c2d2
JA
11104}
11105
11106char *
11107getifs ()
11108{
11109 return ifs_value;
b72432fd
JA
11110}
11111
726f6388
JA
11112/* This splits a single word into a WORD LIST on $IFS, but only if the word
11113 is not quoted. list_string () performs quote removal for us, even if we
11114 don't do any splitting. */
11115WORD_LIST *
7117c2d2 11116word_split (w, ifs_chars)
726f6388 11117 WORD_DESC *w;
7117c2d2 11118 char *ifs_chars;
726f6388
JA
11119{
11120 WORD_LIST *result;
11121
11122 if (w)
11123 {
7117c2d2 11124 char *xifs;
726f6388 11125
7117c2d2
JA
11126 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
11127 result = list_string (w->word, xifs, w->flags & W_QUOTED);
726f6388
JA
11128 }
11129 else
11130 result = (WORD_LIST *)NULL;
ccc6cda3 11131
726f6388
JA
11132 return (result);
11133}
11134
11135/* Perform word splitting on LIST and return the RESULT. It is possible
11136 to return (WORD_LIST *)NULL. */
11137static WORD_LIST *
11138word_list_split (list)
11139 WORD_LIST *list;
11140{
95732b49 11141 WORD_LIST *result, *t, *tresult, *e;
d233b485 11142 WORD_DESC *w;
726f6388 11143
ccc6cda3 11144 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
726f6388 11145 {
7117c2d2 11146 tresult = word_split (t->word, ifs_value);
d233b485
CR
11147 /* POSIX 2.6: "If the complete expansion appropriate for a word results
11148 in an empty field, that empty field shall be deleted from the list
11149 of fields that form the completely expanded command, unless the
11150 original word contained single-quote or double-quote characters."
11151 This is where we handle these words that contain quoted null strings
11152 and other characters that expand to nothing after word splitting. */
11153 if (tresult == 0 && t->word && (t->word->flags & W_SAWQUOTEDNULL)) /* XXX */
11154 {
11155 w = alloc_word_desc ();
11156 w->word = (char *)xmalloc (1);
11157 w->word[0] = '\0';
11158 tresult = make_word_list (w, (WORD_LIST *)NULL);
11159 }
95732b49
JA
11160 if (result == 0)
11161 result = e = tresult;
11162 else
11163 {
11164 e->next = tresult;
11165 while (e->next)
11166 e = e->next;
11167 }
726f6388
JA
11168 }
11169 return (result);
11170}
11171
11172/**************************************************
11173 * *
cce855bc 11174 * Functions to expand an entire WORD_LIST *
726f6388
JA
11175 * *
11176 **************************************************/
11177
b80f6443
JA
11178/* Do any word-expansion-specific cleanup and jump to top_level */
11179static void
11180exp_jump_to_top_level (v)
11181 int v;
11182{
3185942a
JA
11183 set_pipestatus_from_exit (last_command_exit_value);
11184
b80f6443
JA
11185 /* Cleanup code goes here. */
11186 expand_no_split_dollar_star = 0; /* XXX */
d233b485
CR
11187 if (expanding_redir)
11188 undo_partial_redirects ();
b80f6443 11189 expanding_redir = 0;
3185942a 11190 assigning_in_environment = 0;
b80f6443 11191
f1be666c
JA
11192 if (parse_and_execute_level == 0)
11193 top_level_cleanup (); /* from sig.c */
11194
b80f6443
JA
11195 jump_to_top_level (v);
11196}
11197
cce855bc
JA
11198/* Put NLIST (which is a WORD_LIST * of only one element) at the front of
11199 ELIST, and set ELIST to the new list. */
11200#define PREPEND_LIST(nlist, elist) \
11201 do { nlist->next = elist; elist = nlist; } while (0)
11202
726f6388
JA
11203/* Separate out any initial variable assignments from TLIST. If set -k has
11204 been executed, remove all assignment statements from TLIST. Initial
11205 variable assignments and other environment assignments are placed
bb70624e 11206 on SUBST_ASSIGN_VARLIST. */
726f6388
JA
11207static WORD_LIST *
11208separate_out_assignments (tlist)
11209 WORD_LIST *tlist;
11210{
11211 register WORD_LIST *vp, *lp;
11212
0001803f 11213 if (tlist == 0)
726f6388
JA
11214 return ((WORD_LIST *)NULL);
11215
bb70624e
JA
11216 if (subst_assign_varlist)
11217 dispose_words (subst_assign_varlist); /* Clean up after previous error */
b72432fd 11218
bb70624e 11219 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
11220 vp = lp = tlist;
11221
11222 /* Separate out variable assignments at the start of the command.
11223 Loop invariant: vp->next == lp
11224 Loop postcondition:
7117c2d2
JA
11225 lp = list of words left after assignment statements skipped
11226 tlist = original list of words
726f6388 11227 */
ccc6cda3 11228 while (lp && (lp->word->flags & W_ASSIGNMENT))
726f6388
JA
11229 {
11230 vp = lp;
11231 lp = lp->next;
11232 }
11233
bb70624e
JA
11234 /* If lp != tlist, we have some initial assignment statements.
11235 We make SUBST_ASSIGN_VARLIST point to the list of assignment
11236 words and TLIST point to the remaining words. */
726f6388
JA
11237 if (lp != tlist)
11238 {
bb70624e 11239 subst_assign_varlist = tlist;
726f6388
JA
11240 /* ASSERT(vp->next == lp); */
11241 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
11242 tlist = lp; /* remainder of word list */
11243 }
11244
11245 /* vp == end of variable list */
11246 /* tlist == remainder of original word list without variable assignments */
11247 if (!tlist)
11248 /* All the words in tlist were assignment statements */
11249 return ((WORD_LIST *)NULL);
11250
11251 /* ASSERT(tlist != NULL); */
ccc6cda3 11252 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
726f6388
JA
11253
11254 /* If the -k option is in effect, we need to go through the remaining
bb70624e
JA
11255 words, separate out the assignment words, and place them on
11256 SUBST_ASSIGN_VARLIST. */
726f6388
JA
11257 if (place_keywords_in_env)
11258 {
11259 WORD_LIST *tp; /* tp == running pointer into tlist */
11260
11261 tp = tlist;
11262 lp = tlist->next;
11263
11264 /* Loop Invariant: tp->next == lp */
11265 /* Loop postcondition: tlist == word list without assignment statements */
11266 while (lp)
11267 {
ccc6cda3 11268 if (lp->word->flags & W_ASSIGNMENT)
726f6388
JA
11269 {
11270 /* Found an assignment statement, add this word to end of
bb70624e
JA
11271 subst_assign_varlist (vp). */
11272 if (!subst_assign_varlist)
11273 subst_assign_varlist = vp = lp;
726f6388
JA
11274 else
11275 {
11276 vp->next = lp;
11277 vp = lp;
11278 }
11279
11280 /* Remove the word pointed to by LP from TLIST. */
11281 tp->next = lp->next;
11282 /* ASSERT(vp == lp); */
11283 lp->next = (WORD_LIST *)NULL;
11284 lp = tp->next;
11285 }
11286 else
11287 {
11288 tp = lp;
11289 lp = lp->next;
11290 }
11291 }
11292 }
11293 return (tlist);
11294}
11295
cce855bc
JA
11296#define WEXP_VARASSIGN 0x001
11297#define WEXP_BRACEEXP 0x002
11298#define WEXP_TILDEEXP 0x004
11299#define WEXP_PARAMEXP 0x008
11300#define WEXP_PATHEXP 0x010
11301
11302/* All of the expansions, including variable assignments at the start of
11303 the list. */
11304#define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
11305
11306/* All of the expansions except variable assignments at the start of
11307 the list. */
11308#define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
11309
11310/* All of the `shell expansions': brace expansion, tilde expansion, parameter
11311 expansion, command substitution, arithmetic expansion, word splitting, and
11312 quote removal. */
11313#define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
11314
726f6388
JA
11315/* Take the list of words in LIST and do the various substitutions. Return
11316 a new list of words which is the expanded list, and without things like
11317 variable assignments. */
11318
11319WORD_LIST *
11320expand_words (list)
11321 WORD_LIST *list;
11322{
cce855bc 11323 return (expand_word_list_internal (list, WEXP_ALL));
726f6388
JA
11324}
11325
11326/* Same as expand_words (), but doesn't hack variable or environment
11327 variables. */
11328WORD_LIST *
11329expand_words_no_vars (list)
11330 WORD_LIST *list;
11331{
cce855bc 11332 return (expand_word_list_internal (list, WEXP_NOVARS));
726f6388
JA
11333}
11334
cce855bc
JA
11335WORD_LIST *
11336expand_words_shellexp (list)
726f6388 11337 WORD_LIST *list;
726f6388 11338{
cce855bc
JA
11339 return (expand_word_list_internal (list, WEXP_SHELLEXP));
11340}
726f6388 11341
cce855bc
JA
11342static WORD_LIST *
11343glob_expand_word_list (tlist, eflags)
11344 WORD_LIST *tlist;
11345 int eflags;
11346{
11347 char **glob_array, *temp_string;
11348 register int glob_index;
11349 WORD_LIST *glob_list, *output_list, *disposables, *next;
11350 WORD_DESC *tword;
712f80b0 11351 int x;
726f6388 11352
cce855bc
JA
11353 output_list = disposables = (WORD_LIST *)NULL;
11354 glob_array = (char **)NULL;
11355 while (tlist)
11356 {
11357 /* For each word, either globbing is attempted or the word is
11358 added to orig_list. If globbing succeeds, the results are
11359 added to orig_list and the word (tlist) is added to the list
11360 of disposable words. If globbing fails and failed glob
11361 expansions are left unchanged (the shell default), the
11362 original word is added to orig_list. If globbing fails and
11363 failed glob expansions are removed, the original word is
11364 added to the list of disposable words. orig_list ends up
7117c2d2 11365 in reverse order and requires a call to REVERSE_LIST to
cce855bc
JA
11366 be set right. After all words are examined, the disposable
11367 words are freed. */
11368 next = tlist->next;
726f6388 11369
cce855bc 11370 /* If the word isn't an assignment and contains an unquoted
28ef6c31 11371 pattern matching character, then glob it. */
b72432fd 11372 if ((tlist->word->flags & W_NOGLOB) == 0 &&
cce855bc 11373 unquoted_glob_pattern_p (tlist->word->word))
726f6388 11374 {
712f80b0 11375 glob_array = shell_glob_filename (tlist->word->word, QGLOB_CTLESC); /* XXX */
cce855bc
JA
11376
11377 /* Handle error cases.
11378 I don't think we should report errors like "No such file
11379 or directory". However, I would like to report errors
11380 like "Read failed". */
11381
b80f6443 11382 if (glob_array == 0 || GLOB_FAILED (glob_array))
726f6388 11383 {
bb70624e 11384 glob_array = (char **)xmalloc (sizeof (char *));
cce855bc
JA
11385 glob_array[0] = (char *)NULL;
11386 }
11387
11388 /* Dequote the current word in case we have to use it. */
11389 if (glob_array[0] == NULL)
11390 {
11391 temp_string = dequote_string (tlist->word->word);
11392 free (tlist->word->word);
11393 tlist->word->word = temp_string;
11394 }
11395
11396 /* Make the array into a word list. */
11397 glob_list = (WORD_LIST *)NULL;
11398 for (glob_index = 0; glob_array[glob_index]; glob_index++)
11399 {
11400 tword = make_bare_word (glob_array[glob_index]);
cce855bc
JA
11401 glob_list = make_word_list (tword, glob_list);
11402 }
11403
11404 if (glob_list)
11405 {
11406 output_list = (WORD_LIST *)list_append (glob_list, output_list);
11407 PREPEND_LIST (tlist, disposables);
11408 }
b80f6443
JA
11409 else if (fail_glob_expansion != 0)
11410 {
ac50fbac 11411 last_command_exit_value = EXECUTION_FAILURE;
b80f6443 11412 report_error (_("no match: %s"), tlist->word->word);
f1be666c 11413 exp_jump_to_top_level (DISCARD);
b80f6443 11414 }
cce855bc
JA
11415 else if (allow_null_glob_expansion == 0)
11416 {
11417 /* Failed glob expressions are left unchanged. */
11418 PREPEND_LIST (tlist, output_list);
11419 }
11420 else
11421 {
11422 /* Failed glob expressions are removed. */
11423 PREPEND_LIST (tlist, disposables);
726f6388 11424 }
726f6388 11425 }
cce855bc
JA
11426 else
11427 {
11428 /* Dequote the string. */
11429 temp_string = dequote_string (tlist->word->word);
11430 free (tlist->word->word);
11431 tlist->word->word = temp_string;
11432 PREPEND_LIST (tlist, output_list);
11433 }
11434
7117c2d2 11435 strvec_dispose (glob_array);
cce855bc
JA
11436 glob_array = (char **)NULL;
11437
11438 tlist = next;
726f6388
JA
11439 }
11440
cce855bc
JA
11441 if (disposables)
11442 dispose_words (disposables);
11443
11444 if (output_list)
11445 output_list = REVERSE_LIST (output_list, WORD_LIST *);
11446
11447 return (output_list);
11448}
726f6388
JA
11449
11450#if defined (BRACE_EXPANSION)
cce855bc
JA
11451static WORD_LIST *
11452brace_expand_word_list (tlist, eflags)
11453 WORD_LIST *tlist;
11454 int eflags;
11455{
11456 register char **expansions;
11457 char *temp_string;
11458 WORD_LIST *disposables, *output_list, *next;
11459 WORD_DESC *w;
11460 int eindex;
11461
11462 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
726f6388 11463 {
cce855bc 11464 next = tlist->next;
726f6388 11465
ac50fbac
CR
11466 if (tlist->word->flags & W_NOBRACE)
11467 {
11468/*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
11469 PREPEND_LIST (tlist, output_list);
11470 continue;
11471 }
11472
0001803f
CR
11473 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
11474 {
11475/*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
11476 PREPEND_LIST (tlist, output_list);
11477 continue;
11478 }
ac50fbac 11479
cce855bc
JA
11480 /* Only do brace expansion if the word has a brace character. If
11481 not, just add the word list element to BRACES and continue. In
11482 the common case, at least when running shell scripts, this will
0001803f 11483 degenerate to a bunch of calls to `mbschr', and then what is
cce855bc 11484 basically a reversal of TLIST into BRACES, which is corrected
7117c2d2 11485 by a call to REVERSE_LIST () on BRACES when the end of TLIST
cce855bc 11486 is reached. */
0001803f 11487 if (mbschr (tlist->word->word, LBRACE))
726f6388 11488 {
cce855bc 11489 expansions = brace_expand (tlist->word->word);
726f6388 11490
cce855bc 11491 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
726f6388 11492 {
ac50fbac
CR
11493 w = alloc_word_desc ();
11494 w->word = temp_string;
11495
cce855bc
JA
11496 /* If brace expansion didn't change the word, preserve
11497 the flags. We may want to preserve the flags
11498 unconditionally someday -- XXX */
11499 if (STREQ (temp_string, tlist->word->word))
11500 w->flags = tlist->word->flags;
ac50fbac
CR
11501 else
11502 w = make_word_flags (w, temp_string);
11503
cce855bc 11504 output_list = make_word_list (w, output_list);
726f6388 11505 }
cce855bc 11506 free (expansions);
726f6388 11507
cce855bc
JA
11508 /* Add TLIST to the list of words to be freed after brace
11509 expansion has been performed. */
11510 PREPEND_LIST (tlist, disposables);
11511 }
11512 else
11513 PREPEND_LIST (tlist, output_list);
726f6388 11514 }
cce855bc
JA
11515
11516 if (disposables)
11517 dispose_words (disposables);
11518
11519 if (output_list)
11520 output_list = REVERSE_LIST (output_list, WORD_LIST *);
11521
11522 return (output_list);
11523}
11524#endif
11525
3185942a 11526#if defined (ARRAY_VARS)
712f80b0
CR
11527/* Take WORD, a compound array assignment, and internally run (for example),
11528 'declare -A w', where W is the variable name portion of WORD. OPTION is
11529 the list of options to supply to `declare'. CMD is the declaration command
11530 we are expanding right now; it's unused currently. */
3185942a 11531static int
a0c0a00f 11532make_internal_declare (word, option, cmd)
3185942a
JA
11533 char *word;
11534 char *option;
a0c0a00f 11535 char *cmd;
3185942a 11536{
a0c0a00f 11537 int t, r;
3185942a
JA
11538 WORD_LIST *wl;
11539 WORD_DESC *w;
11540
11541 w = make_word (word);
11542
11543 t = assignment (w->word, 0);
a0c0a00f
CR
11544 if (w->word[t] == '=')
11545 {
11546 w->word[t] = '\0';
11547 if (w->word[t - 1] == '+') /* cut off any append op */
11548 w->word[t - 1] = '\0';
11549 }
3185942a
JA
11550
11551 wl = make_word_list (w, (WORD_LIST *)NULL);
11552 wl = make_word_list (make_word (option), wl);
11553
a0c0a00f
CR
11554 r = declare_builtin (wl);
11555
11556 dispose_words (wl);
11557 return r;
3185942a 11558}
712f80b0
CR
11559
11560/* Expand VALUE in NAME[+]=( VALUE ) to a list of words. FLAGS is 1 if NAME
11561 is an associative array.
11562
11563 If we are processing an indexed array, expand_compound_array_assignment
11564 will expand all the individual words and quote_compound_array_list will
11565 single-quote them. If we are processing an associative array, we use
11566 parse_string_to_word_list to split VALUE into a list of words instead of
11567 faking up a shell variable and calling expand_compound_array_assignment.
11568 expand_and_quote_assoc_word expands and single-quotes each word in VALUE
11569 together so we don't have problems finding the end of the subscript when
11570 quoting it.
11571
11572 Words in VALUE can be individual words, which are expanded and single-quoted,
11573 or words of the form [IND]=VALUE, which end up as explained below, as
11574 ['expanded-ind']='expanded-value'. */
11575
11576static WORD_LIST *
11577expand_oneword (value, flags)
11578 char *value;
11579 int flags;
11580{
11581 WORD_LIST *l, *nl;
11582 char *t;
11583
11584 if (flags == 0)
11585 {
11586 /* Indexed array */
11587 l = expand_compound_array_assignment ((SHELL_VAR *)NULL, value, flags);
11588 /* Now we quote the results of the expansion above to prevent double
11589 expansion. */
11590 quote_compound_array_list (l, flags);
11591 return l;
11592 }
11593 else
11594 {
11595 /* Associative array */
11596 l = parse_string_to_word_list (value, 1, "array assign");
11597 /* For associative arrays, with their arbitrary subscripts, we have to
11598 expand and quote in one step so we don't have to search for the
11599 closing right bracket more than once. */
11600 for (nl = l; nl; nl = nl->next)
11601 {
11602 if ((nl->word->flags & W_ASSIGNMENT) == 0)
11603 t = sh_single_quote (nl->word->word ? nl->word->word : "");
11604 else
11605 t = expand_and_quote_assoc_word (nl->word->word, flags);
11606 free (nl->word->word);
11607 nl->word->word = t;
11608 }
11609 return l;
11610 }
11611}
11612
11613/* Expand a single compound assignment argument to a declaration builtin.
11614 This word takes the form NAME[+]=( VALUE ). The NAME[+]= is passed through
11615 unchanged. The VALUE is expanded and each word in the result is single-
11616 quoted. Words of the form [key]=value end up as
11617 ['expanded-key']='expanded-value'. Associative arrays have special
11618 handling, see expand_oneword() above. The return value is
11619 NAME[+]=( expanded-and-quoted-VALUE ). */
11620static void
11621expand_compound_assignment_word (tlist, flags)
11622 WORD_LIST *tlist;
11623 int flags;
11624{
11625 WORD_LIST *l;
11626 int wlen, oind, t;
11627 char *value, *temp;
11628
11629/*itrace("expand_compound_assignment_word: original word = -%s-", tlist->word->word);*/
11630 t = assignment (tlist->word->word, 0);
11631
11632 /* value doesn't have the open and close parens */
11633 oind = 1;
11634 value = extract_array_assignment_list (tlist->word->word + t + 1, &oind);
11635 /* This performs one round of expansion on the index/key and value and
11636 single-quotes each word in the result. */
11637 l = expand_oneword (value, flags);
11638 free (value);
11639
11640 value = string_list (l);
11641 wlen = STRLEN (value);
11642
11643 /* Now, let's rebuild the string */
11644 temp = xmalloc (t + 3 + wlen + 1); /* name[+]=(value) */
11645 memcpy (temp, tlist->word->word, ++t);
11646 temp[t++] = '(';
11647 if (value)
11648 memcpy (temp + t, value, wlen);
11649 t += wlen;
11650 temp[t++] = ')';
11651 temp[t] = '\0';
11652/*itrace("expand_compound_assignment_word: reconstructed word = -%s-", temp);*/
11653
11654 free (tlist->word->word);
11655 tlist->word->word = temp;
11656
11657 free (value);
11658}
11659
11660/* Expand and process an argument to a declaration command. We have already
11661 set flags in TLIST->word->flags depending on the declaration command
11662 (declare, local, etc.) and the options supplied to it (-a, -A, etc.).
11663 TLIST->word->word is of the form NAME[+]=( VALUE ).
11664
11665 This does several things, all using pieces of other functions to get the
11666 evaluation sequence right. It's called for compound array assignments with
11667 the W_ASSIGNMENT flag set (basically, valid identifier names on the lhs).
11668 It parses out which flags need to be set for declare to create the variable
11669 correctly, then calls declare internally (make_internal_declare) to make
11670 sure the variable exists with the correct attributes. Before the variable
11671 is created, it calls expand_compound_assignment_word to expand VALUE to a
11672 list of words, appropriately quoted for further evaluation. This preserves
11673 the semantics of word-expansion-before-calling-builtins. Finally, it calls
11674 do_word_assignment to perform the expansion and assignment with the same
11675 expansion semantics as a standalone assignment statement (no word splitting,
11676 etc.) even though the word is single-quoted so all that needs to happen is
11677 quote removal. */
11678static WORD_LIST *
11679expand_declaration_argument (tlist, wcmd)
11680 WORD_LIST *tlist, *wcmd;
11681{
11682 char opts[16], omap[128];
11683 int t, opti, oind, skip, inheriting;
11684 WORD_LIST *l;
11685
11686 inheriting = localvar_inherit;
11687 opti = 0;
11688 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_CHKLOCAL|W_ASSIGNARRAY))
11689 opts[opti++] = '-';
11690
11691 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
11692 {
11693 opts[opti++] = 'g';
11694 opts[opti++] = 'A';
11695 }
11696 else if (tlist->word->flags & W_ASSIGNASSOC)
11697 {
11698 opts[opti++] = 'A';
11699 }
11700 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
11701 {
11702 opts[opti++] = 'g';
11703 opts[opti++] = 'a';
11704 }
11705 else if (tlist->word->flags & W_ASSIGNARRAY)
11706 {
11707 opts[opti++] = 'a';
11708 }
11709 else if (tlist->word->flags & W_ASSNGLOBAL)
11710 opts[opti++] = 'g';
11711
11712 if (tlist->word->flags & W_CHKLOCAL)
11713 opts[opti++] = 'G';
11714
11715 /* If we have special handling note the integer attribute and others
11716 that transform the value upon assignment. What we do is take all
11717 of the option arguments and scan through them looking for options
11718 that cause such transformations, and add them to the `opts' array. */
11719
11720 memset (omap, '\0', sizeof (omap));
11721 for (l = wcmd->next; l != tlist; l = l->next)
11722 {
11723 if (l->word->word[0] != '-')
11724 break; /* non-option argument */
11725 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
11726 break; /* -- signals end of options */
11727 for (oind = 1; l->word->word[oind]; oind++)
11728 switch (l->word->word[oind])
11729 {
11730 case 'I':
11731 inheriting = 1;
11732 case 'i':
11733 case 'l':
11734 case 'u':
11735 case 'c':
11736 omap[l->word->word[oind]] = 1;
11737 if (opti == 0)
11738 opts[opti++] = '-';
11739 break;
11740 default:
11741 break;
11742 }
11743 }
11744
11745 for (oind = 0; oind < sizeof (omap); oind++)
11746 if (omap[oind])
11747 opts[opti++] = oind;
11748
11749 /* If there are no -a/-A options, but we have a compound assignment,
11750 we have a choice: we can set opts[0]='-', opts[1]='a', since the
11751 default is to create an indexed array, and call
11752 make_internal_declare with that, or we can just skip the -a and let
11753 declare_builtin deal with it. Once we're here, we're better set
11754 up for the latter, since we don't want to deal with looking up
11755 any existing variable here -- better to let declare_builtin do it.
11756 We need the variable created, though, especially if it's local, so
11757 we get the scoping right before we call do_word_assignment.
11758 To ensure that make_local_declare gets called, we add `--' if there
11759 aren't any options. */
11760 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSIGNARRAY)) == 0)
11761 {
11762 if (opti == 0)
11763 {
11764 opts[opti++] = '-';
11765 opts[opti++] = '-';
11766 }
11767 }
11768 opts[opti] = '\0';
11769
11770 /* This isn't perfect, but it's a start. Improvements later. We expand
11771 tlist->word->word and single-quote the results to avoid multiple
11772 expansions by, say, do_assignment_internal(). We have to weigh the
11773 cost of reconstructing the compound assignment string with its single
11774 quoting and letting the declare builtin handle it. The single quotes
11775 will prevent any unwanted additional expansion or word splitting. */
11776 expand_compound_assignment_word (tlist, (tlist->word->flags & W_ASSIGNASSOC) ? 1 : 0);
11777
11778 skip = 0;
11779 if (opti > 0)
11780 {
11781 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
11782 if (t != EXECUTION_SUCCESS)
11783 {
11784 last_command_exit_value = t;
11785 if (tlist->word->flags & W_FORCELOCAL) /* non-fatal error */
11786 skip = 1;
11787 else
11788 exp_jump_to_top_level (DISCARD);
11789 }
11790 }
11791
11792 if (skip == 0)
11793 {
11794 t = do_word_assignment (tlist->word, 0);
11795 if (t == 0)
11796 {
11797 last_command_exit_value = EXECUTION_FAILURE;
11798 exp_jump_to_top_level (DISCARD);
11799 }
11800 }
11801
11802 /* Now transform the word as ksh93 appears to do and go on */
11803 t = assignment (tlist->word->word, 0);
11804 tlist->word->word[t] = '\0';
11805 if (tlist->word->word[t - 1] == '+')
11806 tlist->word->word[t - 1] = '\0'; /* cut off append op */
11807 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
11808
11809 return (tlist);
11810}
11811#endif /* ARRAY_VARS */
3185942a 11812
cce855bc
JA
11813static WORD_LIST *
11814shell_expand_word_list (tlist, eflags)
11815 WORD_LIST *tlist;
11816 int eflags;
11817{
a0c0a00f 11818 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
cce855bc 11819 int expanded_something, has_dollar_at;
726f6388 11820
726f6388 11821 /* We do tilde expansion all the time. This is what 1003.2 says. */
712f80b0 11822 wcmd = new_list = (WORD_LIST *)NULL;
a0c0a00f 11823
cce855bc 11824 for (orig_list = tlist; tlist; tlist = next)
726f6388 11825 {
712f80b0
CR
11826 if (wcmd == 0 && (tlist->word->flags & W_ASSNBLTIN))
11827 wcmd = tlist;
11828
726f6388
JA
11829 next = tlist->next;
11830
95732b49
JA
11831#if defined (ARRAY_VARS)
11832 /* If this is a compound array assignment to a builtin that accepts
11833 such assignments (e.g., `declare'), take the assignment and perform
11834 it separately, handling the semantics of declarations inside shell
11835 functions. This avoids the double-evaluation of such arguments,
11836 because `declare' does some evaluation of compound assignments on
11837 its own. */
11838 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
712f80b0 11839 expand_declaration_argument (tlist, wcmd);
95732b49 11840#endif
726f6388 11841
ccc6cda3 11842 expanded_something = 0;
726f6388 11843 expanded = expand_word_internal
b72432fd 11844 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
726f6388
JA
11845
11846 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
11847 {
11848 /* By convention, each time this error is returned,
11849 tlist->word->word has already been freed. */
11850 tlist->word->word = (char *)NULL;
ccc6cda3 11851
726f6388
JA
11852 /* Dispose our copy of the original list. */
11853 dispose_words (orig_list);
d166f048 11854 /* Dispose the new list we're building. */
726f6388
JA
11855 dispose_words (new_list);
11856
28ef6c31 11857 last_command_exit_value = EXECUTION_FAILURE;
726f6388 11858 if (expanded == &expand_word_error)
b80f6443 11859 exp_jump_to_top_level (DISCARD);
726f6388 11860 else
b80f6443 11861 exp_jump_to_top_level (FORCE_EOF);
726f6388
JA
11862 }
11863
ccc6cda3
JA
11864 /* Don't split words marked W_NOSPLIT. */
11865 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
726f6388 11866 {
ccc6cda3 11867 temp_list = word_list_split (expanded);
726f6388
JA
11868 dispose_words (expanded);
11869 }
11870 else
11871 {
11872 /* If no parameter expansion, command substitution, process
11873 substitution, or arithmetic substitution took place, then
11874 do not do word splitting. We still have to remove quoted
11875 null characters from the result. */
11876 word_list_remove_quoted_nulls (expanded);
ccc6cda3 11877 temp_list = expanded;
726f6388
JA
11878 }
11879
ccc6cda3
JA
11880 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
11881 new_list = (WORD_LIST *)list_append (expanded, new_list);
726f6388
JA
11882 }
11883
cce855bc
JA
11884 if (orig_list)
11885 dispose_words (orig_list);
726f6388 11886
726f6388 11887 if (new_list)
cce855bc 11888 new_list = REVERSE_LIST (new_list, WORD_LIST *);
726f6388 11889
cce855bc
JA
11890 return (new_list);
11891}
726f6388 11892
cce855bc
JA
11893/* The workhorse for expand_words () and expand_words_no_vars ().
11894 First arg is LIST, a WORD_LIST of words.
b72432fd
JA
11895 Second arg EFLAGS is a flags word controlling which expansions are
11896 performed.
726f6388 11897
cce855bc
JA
11898 This does all of the substitutions: brace expansion, tilde expansion,
11899 parameter expansion, command substitution, arithmetic expansion,
11900 process substitution, word splitting, and pathname expansion, according
11901 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
11902 set, or for which no expansion is done, do not undergo word splitting.
ac50fbac
CR
11903 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
11904 with W_NOBRACE set do not undergo brace expansion (see
11905 brace_expand_word_list above). */
cce855bc
JA
11906static WORD_LIST *
11907expand_word_list_internal (list, eflags)
11908 WORD_LIST *list;
11909 int eflags;
11910{
11911 WORD_LIST *new_list, *temp_list;
11912 int tint;
a0c0a00f 11913 char *savecmd;
726f6388 11914
ac50fbac 11915 tempenv_assign_error = 0;
cce855bc
JA
11916 if (list == 0)
11917 return ((WORD_LIST *)NULL);
726f6388 11918
bb70624e 11919 garglist = new_list = copy_word_list (list);
cce855bc
JA
11920 if (eflags & WEXP_VARASSIGN)
11921 {
bb70624e 11922 garglist = new_list = separate_out_assignments (new_list);
cce855bc
JA
11923 if (new_list == 0)
11924 {
bb70624e 11925 if (subst_assign_varlist)
cce855bc
JA
11926 {
11927 /* All the words were variable assignments, so they are placed
11928 into the shell's environment. */
bb70624e 11929 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
cce855bc 11930 {
a0c0a00f 11931 savecmd = this_command_name;
cce855bc 11932 this_command_name = (char *)NULL; /* no arithmetic errors */
495aee44 11933 tint = do_word_assignment (temp_list->word, 0);
a0c0a00f 11934 this_command_name = savecmd;
cce855bc 11935 /* Variable assignment errors in non-interactive shells
d233b485
CR
11936 running in Posix.2 mode cause the shell to exit, unless
11937 they are being run by the `command' builtin. */
28ef6c31 11938 if (tint == 0)
ccc6cda3 11939 {
cce855bc 11940 last_command_exit_value = EXECUTION_FAILURE;
d233b485 11941 if (interactive_shell == 0 && posixly_correct && executing_command_builtin == 0)
b80f6443 11942 exp_jump_to_top_level (FORCE_EOF);
28ef6c31 11943 else
b80f6443 11944 exp_jump_to_top_level (DISCARD);
ccc6cda3 11945 }
726f6388 11946 }
bb70624e
JA
11947 dispose_words (subst_assign_varlist);
11948 subst_assign_varlist = (WORD_LIST *)NULL;
cce855bc
JA
11949 }
11950 return ((WORD_LIST *)NULL);
11951 }
11952 }
726f6388 11953
cce855bc
JA
11954 /* Begin expanding the words that remain. The expansions take place on
11955 things that aren't really variable assignments. */
726f6388 11956
cce855bc
JA
11957#if defined (BRACE_EXPANSION)
11958 /* Do brace expansion on this word if there are any brace characters
11959 in the string. */
11960 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
11961 new_list = brace_expand_word_list (new_list, eflags);
11962#endif /* BRACE_EXPANSION */
726f6388 11963
cce855bc
JA
11964 /* Perform the `normal' shell expansions: tilde expansion, parameter and
11965 variable substitution, command substitution, arithmetic expansion,
11966 and word splitting. */
11967 new_list = shell_expand_word_list (new_list, eflags);
726f6388 11968
cce855bc
JA
11969 /* Okay, we're almost done. Now let's just do some filename
11970 globbing. */
11971 if (new_list)
11972 {
11973 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
11974 /* Glob expand the word list unless globbing has been disabled. */
11975 new_list = glob_expand_word_list (new_list, eflags);
726f6388 11976 else
cce855bc
JA
11977 /* Dequote the words, because we're not performing globbing. */
11978 new_list = dequote_list (new_list);
726f6388
JA
11979 }
11980
bb70624e 11981 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
726f6388 11982 {
95732b49 11983 sh_wassign_func_t *assign_func;
495aee44 11984 int is_special_builtin, is_builtin_or_func;
726f6388
JA
11985
11986 /* If the remainder of the words expand to nothing, Posix.2 requires
11987 that the variable and environment assignments affect the shell's
11988 environment. */
95732b49 11989 assign_func = new_list ? assign_in_env : do_word_assignment;
b80f6443 11990 tempenv_assign_error = 0;
726f6388 11991
495aee44
CR
11992 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
11993 /* Posix says that special builtins exit if a variable assignment error
11994 occurs in an assignment preceding it. */
11995 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
11996
bb70624e 11997 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
726f6388 11998 {
a0c0a00f 11999 savecmd = this_command_name;
ccc6cda3 12000 this_command_name = (char *)NULL;
3185942a 12001 assigning_in_environment = (assign_func == assign_in_env);
495aee44 12002 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
3185942a 12003 assigning_in_environment = 0;
a0c0a00f 12004 this_command_name = savecmd;
ccc6cda3
JA
12005 /* Variable assignment errors in non-interactive shells running
12006 in Posix.2 mode cause the shell to exit. */
b80f6443 12007 if (tint == 0)
ccc6cda3 12008 {
95732b49 12009 if (assign_func == do_word_assignment)
b80f6443
JA
12010 {
12011 last_command_exit_value = EXECUTION_FAILURE;
d233b485 12012 if (interactive_shell == 0 && posixly_correct)
b80f6443
JA
12013 exp_jump_to_top_level (FORCE_EOF);
12014 else
12015 exp_jump_to_top_level (DISCARD);
12016 }
d233b485
CR
12017 else if (interactive_shell == 0 && is_special_builtin)
12018 {
12019 last_command_exit_value = EXECUTION_FAILURE;
12020 exp_jump_to_top_level (FORCE_EOF);
12021 }
28ef6c31 12022 else
b80f6443 12023 tempenv_assign_error++;
ccc6cda3 12024 }
726f6388 12025 }
726f6388 12026
bb70624e
JA
12027 dispose_words (subst_assign_varlist);
12028 subst_assign_varlist = (WORD_LIST *)NULL;
726f6388
JA
12029 }
12030
cce855bc 12031 return (new_list);
ccc6cda3 12032}