]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
Bash-5.1 patch 3: fix command substitution in here-document from child process in...
[thirdparty/bash.git] / subst.c
1 /* subst.c -- The part of the shell that does parameter, command, arithmetic,
2 and globbing substitutions. */
3
4 /* ``Have a little faith, there's magic in the night. You ain't a
5 beauty, but, hey, you're alright.'' */
6
7 /* Copyright (C) 1987-2020 Free Software Foundation, Inc.
8
9 This file is part of GNU Bash, the Bourne Again SHell.
10
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.
15
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.
20
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 */
24
25 #include "config.h"
26
27 #include "bashtypes.h"
28 #include <stdio.h>
29 #include "chartypes.h"
30 #if defined (HAVE_PWD_H)
31 # include <pwd.h>
32 #endif
33 #include <signal.h>
34 #include <errno.h>
35
36 #if defined (HAVE_UNISTD_H)
37 # include <unistd.h>
38 #endif
39
40 #define NEED_FPURGE_DECL
41
42 #include "bashansi.h"
43 #include "posixstat.h"
44 #include "bashintl.h"
45
46 #include "shell.h"
47 #include "parser.h"
48 #include "flags.h"
49 #include "jobs.h"
50 #include "execute_cmd.h"
51 #include "filecntl.h"
52 #include "trap.h"
53 #include "pathexp.h"
54 #include "mailcheck.h"
55
56 #include "shmbutil.h"
57 #if defined (HAVE_MBSTR_H) && defined (HAVE_MBSCHR)
58 # include <mbstr.h> /* mbschr */
59 #endif
60 #include "typemax.h"
61
62 #include "builtins/getopt.h"
63 #include "builtins/common.h"
64
65 #include "builtins/builtext.h"
66
67 #include <tilde/tilde.h>
68 #include <glob/strmatch.h>
69
70 #if !defined (errno)
71 extern int errno;
72 #endif /* !errno */
73
74 /* The size that strings change by. */
75 #define DEFAULT_INITIAL_ARRAY_SIZE 112
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
82 #define VT_ARRAYMEMBER 3
83 #define VT_ASSOCVAR 4
84
85 #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */
86
87 /* Flags for quoted_strchr */
88 #define ST_BACKSL 0x01
89 #define ST_CTLESC 0x02
90 #define ST_SQUOTE 0x04 /* unused yet */
91 #define ST_DQUOTE 0x08 /* unused yet */
92
93 /* These defs make it easier to use the editor. */
94 #define LBRACE '{'
95 #define RBRACE '}'
96 #define LPAREN '('
97 #define RPAREN ')'
98 #define LBRACK '['
99 #define RBRACK ']'
100
101 #if defined (HANDLE_MULTIBYTE)
102 #define WLPAREN L'('
103 #define WRPAREN L')'
104 #endif
105
106 #define DOLLAR_AT_STAR(c) ((c) == '@' || (c) == '*')
107 #define STR_DOLLAR_AT_STAR(s) (DOLLAR_AT_STAR ((s)[0]) && (s)[1] == '\0')
108
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) \
112 ((c) == '-' || (c) == '?' || (c) == '#' || (c) == '@')
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) \
117 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
118
119 /* Evaluates to 1 if C is one of the OP characters that follows the parameter
120 in ${parameter[:]OPword}. */
121 #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
122
123 /* Evaluates to 1 if this is one of the shell's special variables. */
124 #define SPECIAL_VAR(name, wi) \
125 (*name && ((DIGIT (*name) && all_digits (name)) || \
126 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
127 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1]))))
128
129 /* This can be used by all of the *_extract_* functions that have a similar
130 structure. It can't just be wrapped in a do...while(0) loop because of
131 the embedded `break'. The dangling else accommodates a trailing semicolon;
132 we could also put in a do ; while (0) */
133
134 #define CHECK_STRING_OVERRUN(oind, ind, len, ch) \
135 if (ind >= len) \
136 { \
137 oind = len; \
138 ch = 0; \
139 break; \
140 } \
141 else \
142
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(). */
146 typedef WORD_LIST *EXPFUNC PARAMS((char *, int));
147
148 /* Process ID of the last command executed within command substitution. */
149 pid_t last_command_subst_pid = NO_PID;
150 pid_t current_command_subst_pid = NO_PID;
151
152 /* Variables used to keep track of the characters in IFS. */
153 SHELL_VAR *ifs_var;
154 char *ifs_value;
155 unsigned char ifs_cmap[UCHAR_MAX + 1];
156 int ifs_is_set, ifs_is_null;
157
158 #if defined (HANDLE_MULTIBYTE)
159 unsigned char ifs_firstc[MB_LEN_MAX];
160 size_t ifs_firstc_len;
161 #else
162 unsigned char ifs_firstc;
163 #endif
164
165 /* If non-zero, command substitution inherits the value of errexit option */
166 int inherit_errexit = 0;
167
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. */
171 int assigning_in_environment;
172
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. */
176 WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
177
178 /* Tell the expansion functions to not longjmp back to top_level on fatal
179 errors. Enabled when doing completion and prompt string expansion. */
180 int no_longjmp_on_fatal_error = 0;
181
182 /* Non-zero means to allow unmatched globbed filenames to expand to
183 a null file. */
184 int allow_null_glob_expansion;
185
186 /* Non-zero means to throw an error when globbing fails to match anything. */
187 int fail_glob_expansion;
188
189 /* Extern functions and variables from different files. */
190 extern struct fd_bitmap *current_fds_to_close;
191 extern int wordexp_only;
192
193 #if defined (JOB_CONTROL) && defined (PROCESS_SUBSTITUTION)
194 extern PROCESS *last_procsub_child;
195 #endif
196
197 #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
198 extern wchar_t *wcsdup PARAMS((const wchar_t *));
199 #endif
200
201 #if 0
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
204 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
205 (CURRENTLY UNUSED). */
206 char *glob_argv_flags;
207 static int glob_argv_flags_size;
208 #endif
209
210 static WORD_LIST *cached_quoted_dollar_at = 0;
211
212 /* Distinguished error values to return from expansion functions */
213 static WORD_LIST expand_word_error, expand_word_fatal;
214 static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
215 static char expand_param_error, expand_param_fatal, expand_param_unset;
216 static char extract_string_error, extract_string_fatal;
217
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. */
223 static int expand_no_split_dollar_star = 0;
224
225 /* A WORD_LIST of words to be expanded by expand_word_list_internal,
226 without any leading variable assignments. */
227 static WORD_LIST *garglist = (WORD_LIST *)NULL;
228
229 static char *quoted_substring PARAMS((char *, int, int));
230 static int quoted_strlen PARAMS((char *));
231 static char *quoted_strchr PARAMS((char *, int, int));
232
233 static char *expand_string_if_necessary PARAMS((char *, int, EXPFUNC *));
234 static inline char *expand_string_to_string_internal PARAMS((char *, int, EXPFUNC *));
235 static WORD_LIST *call_expand_word_internal PARAMS((WORD_DESC *, int, int, int *, int *));
236 static WORD_LIST *expand_string_internal PARAMS((char *, int));
237 static WORD_LIST *expand_string_leave_quoted PARAMS((char *, int));
238 static WORD_LIST *expand_string_for_rhs PARAMS((char *, int, int, int, int *, int *));
239 static WORD_LIST *expand_string_for_pat PARAMS((char *, int, int *, int *));
240
241 static char *quote_escapes_internal PARAMS((const char *, int));
242
243 static WORD_LIST *list_quote_escapes PARAMS((WORD_LIST *));
244 static WORD_LIST *list_dequote_escapes PARAMS((WORD_LIST *));
245
246 static char *make_quoted_char PARAMS((int));
247 static WORD_LIST *quote_list PARAMS((WORD_LIST *));
248
249 static int unquoted_substring PARAMS((char *, char *));
250 static int unquoted_member PARAMS((int, char *));
251
252 #if defined (ARRAY_VARS)
253 static SHELL_VAR *do_compound_assignment PARAMS((char *, char *, int));
254 #endif
255 static int do_assignment_internal PARAMS((const WORD_DESC *, int));
256
257 static char *string_extract_verbatim PARAMS((char *, size_t, int *, char *, int));
258 static char *string_extract PARAMS((char *, int *, char *, int));
259 static char *string_extract_double_quoted PARAMS((char *, int *, int));
260 static inline char *string_extract_single_quoted PARAMS((char *, int *));
261 static inline int skip_single_quoted PARAMS((const char *, size_t, int, int));
262 static int skip_double_quoted PARAMS((char *, size_t, int, int));
263 static char *extract_delimited_string PARAMS((char *, int *, char *, char *, char *, int));
264 static char *extract_dollar_brace_string PARAMS((char *, int *, int, int));
265 static int skip_matched_pair PARAMS((const char *, int, int, int, int));
266
267 static char *pos_params PARAMS((char *, int, int, int, int));
268
269 static unsigned char *mb_getcharlens PARAMS((char *, int));
270
271 static char *remove_upattern PARAMS((char *, char *, int));
272 #if defined (HANDLE_MULTIBYTE)
273 static wchar_t *remove_wpattern PARAMS((wchar_t *, size_t, wchar_t *, int));
274 #endif
275 static char *remove_pattern PARAMS((char *, char *, int));
276
277 static int match_upattern PARAMS((char *, char *, int, char **, char **));
278 #if defined (HANDLE_MULTIBYTE)
279 static int match_wpattern PARAMS((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
280 #endif
281 static int match_pattern PARAMS((char *, char *, int, char **, char **));
282 static int getpatspec PARAMS((int, char *));
283 static char *getpattern PARAMS((char *, int, int));
284 static char *variable_remove_pattern PARAMS((char *, char *, int, int));
285 static char *list_remove_pattern PARAMS((WORD_LIST *, char *, int, int, int));
286 static char *parameter_list_remove_pattern PARAMS((int, char *, int, int));
287 #ifdef ARRAY_VARS
288 static char *array_remove_pattern PARAMS((SHELL_VAR *, char *, int, int, int));
289 #endif
290 static char *parameter_brace_remove_pattern PARAMS((char *, char *, int, char *, int, int, int));
291
292 static char *string_var_assignment PARAMS((SHELL_VAR *, char *));
293 #if defined (ARRAY_VARS)
294 static char *array_var_assignment PARAMS((SHELL_VAR *, int, int, int));
295 #endif
296 static char *pos_params_assignment PARAMS((WORD_LIST *, int, int));
297 static char *string_transform PARAMS((int, SHELL_VAR *, char *));
298 static char *list_transform PARAMS((int, SHELL_VAR *, WORD_LIST *, int, int));
299 static char *parameter_list_transform PARAMS((int, int, int));
300 #if defined ARRAY_VARS
301 static char *array_transform PARAMS((int, SHELL_VAR *, int, int));
302 #endif
303 static char *parameter_brace_transform PARAMS((char *, char *, int, char *, int, int, int, int));
304 static int valid_parameter_transform PARAMS((char *));
305
306 static char *process_substitute PARAMS((char *, int));
307
308 static char *read_comsub PARAMS((int, int, int, int *));
309
310 #ifdef ARRAY_VARS
311 static arrayind_t array_length_reference PARAMS((char *));
312 #endif
313
314 static int valid_brace_expansion_word PARAMS((char *, int));
315 static int chk_atstar PARAMS((char *, int, int, int *, int *));
316 static int chk_arithsub PARAMS((const char *, int));
317
318 static WORD_DESC *parameter_brace_expand_word PARAMS((char *, int, int, int, arrayind_t *));
319 static char *parameter_brace_find_indir PARAMS((char *, int, int, int));
320 static WORD_DESC *parameter_brace_expand_indir PARAMS((char *, int, int, int, int *, int *));
321 static WORD_DESC *parameter_brace_expand_rhs PARAMS((char *, char *, int, int, int, int *, int *));
322 static void parameter_brace_expand_error PARAMS((char *, char *, int));
323
324 static int valid_length_expression PARAMS((char *));
325 static intmax_t parameter_brace_expand_length PARAMS((char *));
326
327 static char *skiparith PARAMS((char *, int));
328 static int verify_substring_values PARAMS((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
329 static int get_var_and_type PARAMS((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
330 static char *mb_substring PARAMS((char *, int, int));
331 static char *parameter_brace_substring PARAMS((char *, char *, int, char *, int, int, int));
332
333 static int shouldexp_replacement PARAMS((char *));
334
335 static char *pos_params_pat_subst PARAMS((char *, char *, char *, int));
336
337 static char *parameter_brace_patsub PARAMS((char *, char *, int, char *, int, int, int));
338
339 static char *pos_params_casemod PARAMS((char *, char *, int, int));
340 static char *parameter_brace_casemod PARAMS((char *, char *, int, int, char *, int, int, int));
341
342 static WORD_DESC *parameter_brace_expand PARAMS((char *, int *, int, int, int *, int *));
343 static WORD_DESC *param_expand PARAMS((char *, int *, int, int *, int *, int *, int *, int));
344
345 static WORD_LIST *expand_word_internal PARAMS((WORD_DESC *, int, int, int *, int *));
346
347 static WORD_LIST *word_list_split PARAMS((WORD_LIST *));
348
349 static void exp_jump_to_top_level PARAMS((int));
350
351 static WORD_LIST *separate_out_assignments PARAMS((WORD_LIST *));
352 static WORD_LIST *glob_expand_word_list PARAMS((WORD_LIST *, int));
353 #ifdef BRACE_EXPANSION
354 static WORD_LIST *brace_expand_word_list PARAMS((WORD_LIST *, int));
355 #endif
356 #if defined (ARRAY_VARS)
357 static int make_internal_declare PARAMS((char *, char *, char *));
358 static void expand_compound_assignment_word PARAMS((WORD_LIST *, int));
359 static WORD_LIST *expand_declaration_argument PARAMS((WORD_LIST *, WORD_LIST *));
360 #endif
361 static WORD_LIST *shell_expand_word_list PARAMS((WORD_LIST *, int));
362 static WORD_LIST *expand_word_list_internal PARAMS((WORD_LIST *, int));
363
364 /* **************************************************************** */
365 /* */
366 /* Utility Functions */
367 /* */
368 /* **************************************************************** */
369
370 #if defined (DEBUG)
371 void
372 dump_word_flags (flags)
373 int flags;
374 {
375 int f;
376
377 f = flags;
378 fprintf (stderr, "%d -> ", f);
379 if (f & W_ARRAYIND)
380 {
381 f &= ~W_ARRAYIND;
382 fprintf (stderr, "W_ARRAYIND%s", f ? "|" : "");
383 }
384 if (f & W_ASSIGNASSOC)
385 {
386 f &= ~W_ASSIGNASSOC;
387 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
388 }
389 if (f & W_ASSIGNARRAY)
390 {
391 f &= ~W_ASSIGNARRAY;
392 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
393 }
394 if (f & W_SAWQUOTEDNULL)
395 {
396 f &= ~W_SAWQUOTEDNULL;
397 fprintf (stderr, "W_SAWQUOTEDNULL%s", f ? "|" : "");
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 }
424 if (f & W_ASSNGLOBAL)
425 {
426 f &= ~W_ASSNGLOBAL;
427 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
428 }
429 if (f & W_COMPASSIGN)
430 {
431 f &= ~W_COMPASSIGN;
432 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
433 }
434 if (f & W_EXPANDRHS)
435 {
436 f &= ~W_EXPANDRHS;
437 fprintf (stderr, "W_EXPANDRHS%s", f ? "|" : "");
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 }
454 if (f & W_NOASSNTILDE)
455 {
456 f &= ~W_NOASSNTILDE;
457 fprintf (stderr, "W_NOASSNTILDE%s", f ? "|" : "");
458 }
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 }
484 if (f & W_NOSPLIT)
485 {
486 f &= ~W_NOSPLIT;
487 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
488 }
489 if (f & W_NOBRACE)
490 {
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 ? "|" : "");
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 }
519 if (f & W_COMPLETE)
520 {
521 f &= ~W_COMPLETE;
522 fprintf (stderr, "W_COMPLETE%s", f ? "|" : "");
523 }
524 if (f & W_CHKLOCAL)
525 {
526 f &= ~W_CHKLOCAL;
527 fprintf (stderr, "W_CHKLOCAL%s", f ? "|" : "");
528 }
529 if (f & W_FORCELOCAL)
530 {
531 f &= ~W_FORCELOCAL;
532 fprintf (stderr, "W_FORCELOCAL%s", f ? "|" : "");
533 }
534
535 fprintf (stderr, "\n");
536 fflush (stderr);
537 }
538 #endif
539
540 #ifdef INCLUDE_UNUSED
541 static char *
542 quoted_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 {
556 s++;
557 continue;
558 }
559 l++;
560 if (*s == 0)
561 break;
562 }
563
564 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
565
566 /* Copy LEN characters, including quote characters. */
567 s = string + l;
568 for (l = 0; l < len; s++)
569 {
570 if (*s == CTLESC)
571 *r++ = *s++;
572 *r++ = *s;
573 l++;
574 if (*s == 0)
575 break;
576 }
577 *r = '\0';
578 return result;
579 }
580 #endif
581
582 #ifdef INCLUDE_UNUSED
583 /* Return the length of S, skipping over quoted characters */
584 static int
585 quoted_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
606
607 #ifdef INCLUDE_UNUSED
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. */
612 static char *
613 quoted_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
635 /* Return 1 if CHARACTER appears in an unquoted portion of
636 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
637 static int
638 unquoted_member (character, string)
639 int character;
640 char *string;
641 {
642 size_t slen;
643 int sindex, c;
644 DECLARE_MBSTATE;
645
646 slen = strlen (string);
647 sindex = 0;
648 while (c = string[sindex])
649 {
650 if (c == character)
651 return (1);
652
653 switch (c)
654 {
655 default:
656 ADVANCE_CHAR (string, slen, sindex);
657 break;
658
659 case '\\':
660 sindex++;
661 if (string[sindex])
662 ADVANCE_CHAR (string, slen, sindex);
663 break;
664
665 case '\'':
666 sindex = skip_single_quoted (string, slen, ++sindex, 0);
667 break;
668
669 case '"':
670 sindex = skip_double_quoted (string, slen, ++sindex, 0);
671 break;
672 }
673 }
674 return (0);
675 }
676
677 /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
678 static int
679 unquoted_substring (substr, string)
680 char *substr, *string;
681 {
682 size_t slen;
683 int sindex, c, sublen;
684 DECLARE_MBSTATE;
685
686 if (substr == 0 || *substr == '\0')
687 return (0);
688
689 slen = strlen (string);
690 sublen = strlen (substr);
691 for (sindex = 0; c = string[sindex]; )
692 {
693 if (STREQN (string + sindex, substr, sublen))
694 return (1);
695
696 switch (c)
697 {
698 case '\\':
699 sindex++;
700 if (string[sindex])
701 ADVANCE_CHAR (string, slen, sindex);
702 break;
703
704 case '\'':
705 sindex = skip_single_quoted (string, slen, ++sindex, 0);
706 break;
707
708 case '"':
709 sindex = skip_double_quoted (string, slen, ++sindex, 0);
710 break;
711
712 default:
713 ADVANCE_CHAR (string, slen, sindex);
714 break;
715 }
716 }
717 return (0);
718 }
719 #endif
720
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. */
728
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. */
733 INLINE char *
734 sub_append_string (source, target, indx, size)
735 char *source, *target;
736 int *indx;
737 size_t *size;
738 {
739 if (source)
740 {
741 int n;
742 size_t srclen;
743
744 srclen = STRLEN (source);
745 if (srclen >= (int)(*size - *indx))
746 {
747 n = srclen + *indx;
748 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
749 target = (char *)xrealloc (target, (*size = n));
750 }
751
752 FASTCOPY (source, target + *indx, srclen);
753 *indx += srclen;
754 target[*indx] = '\0';
755
756 free (source);
757 }
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. */
765 char *
766 sub_append_number (number, target, indx, size)
767 intmax_t number;
768 char *target;
769 int *indx;
770 size_t *size;
771 {
772 char *temp;
773
774 temp = itos (number);
775 return (sub_append_string (temp, target, indx, size));
776 }
777 #endif
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.
782 Understand about backslashes in the string. If (flags & SX_VARNAME)
783 is non-zero, and array variables have been compiled into the shell,
784 everything between a `[' and a corresponding `]' is skipped over.
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
787 contain a closing character from CHARLIST. */
788 static char *
789 string_extract (string, sindex, charlist, flags)
790 char *string;
791 int *sindex;
792 char *charlist;
793 int flags;
794 {
795 register int c, i;
796 int found;
797 size_t slen;
798 char *temp;
799 DECLARE_MBSTATE;
800
801 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
802 i = *sindex;
803 found = 0;
804 while (c = string[i])
805 {
806 if (c == '\\')
807 {
808 if (string[i + 1])
809 i++;
810 else
811 break;
812 }
813 #if defined (ARRAY_VARS)
814 else if ((flags & SX_VARNAME) && c == LBRACK)
815 {
816 int ni;
817 /* If this is an array subscript, skip over it and continue. */
818 ni = skipsubscript (string, i, 0);
819 if (string[ni] == RBRACK)
820 i = ni;
821 }
822 #endif
823 else if (MEMBER (c, charlist))
824 {
825 found = 1;
826 break;
827 }
828
829 ADVANCE_CHAR (string, slen, i);
830 }
831
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. */
834 if ((flags & SX_REQMATCH) && found == 0)
835 {
836 *sindex = i;
837 return (&extract_string_error);
838 }
839
840 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
841 *sindex = i;
842
843 return (temp);
844 }
845
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. */
853 static char *
854 string_extract_double_quoted (string, sindex, flags)
855 char *string;
856 int *sindex, flags;
857 {
858 size_t slen;
859 char *send;
860 int j, i, t;
861 unsigned char c;
862 char *temp, *ret; /* The new string we return. */
863 int pass_next, backquote, si; /* State variables for the machine. */
864 int dquote;
865 int stripdq;
866 DECLARE_MBSTATE;
867
868 slen = strlen (string + *sindex) + *sindex;
869 send = string + slen;
870
871 stripdq = (flags & SX_STRIPDQ);
872
873 pass_next = backquote = dquote = 0;
874 temp = (char *)xmalloc (1 + slen - *sindex);
875
876 j = 0;
877 i = *sindex;
878 while (c = string[i])
879 {
880 /* Process a character that was quoted by a backslash. */
881 if (pass_next)
882 {
883 /* XXX - take another look at this in light of Interp 221 */
884 /* Posix.2 sez:
885
886 ``The backslash shall retain its special meaning as an escape
887 character only when followed by one of the characters:
888 $ ` " \ <newline>''.
889
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
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. */
906 if ((stripdq == 0 && c != '"') ||
907 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
908 temp[j++] = '\\';
909 pass_next = 0;
910
911 add_one_character:
912 COPY_CHAR_I (temp, j, string, send, i);
913 continue;
914 }
915
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 == '\\')
920 {
921 pass_next++;
922 i++;
923 continue;
924 }
925
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)
931 {
932 if (c == '`')
933 backquote = 0;
934 temp[j++] = c; /* COPY_CHAR_I? */
935 i++;
936 continue;
937 }
938
939 if (c == '`')
940 {
941 temp[j++] = c;
942 backquote++;
943 i++;
944 continue;
945 }
946
947 /* Pass everything between `$(' and the matching `)' or a quoted
948 ${ ... } pair through according to the Posix.2 specification. */
949 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
950 {
951 int free_ret = 1;
952
953 si = i + 2;
954 if (string[i + 1] == LPAREN)
955 ret = extract_command_subst (string, &si, (flags & SX_COMPLETE));
956 else
957 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
958
959 temp[j++] = '$';
960 temp[j++] = string[i + 1];
961
962 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
963 is set. */
964 if (ret == 0 && no_longjmp_on_fatal_error)
965 {
966 free_ret = 0;
967 ret = string + i + 2;
968 }
969
970 /* XXX - CHECK_STRING_OVERRUN here? */
971 for (t = 0; ret[t]; t++, j++)
972 temp[j] = ret[t];
973 temp[j] = string[si];
974
975 if (si < i + 2) /* we went back? */
976 i += 2;
977 else if (string[si])
978 {
979 j++;
980 i = si + 1;
981 }
982 else
983 i = si;
984
985 if (free_ret)
986 free (ret);
987 continue;
988 }
989
990 /* Add any character but a double quote to the quoted string we're
991 accumulating. */
992 if (c != '"')
993 goto add_one_character;
994
995 /* c == '"' */
996 if (stripdq)
997 {
998 dquote ^= 1;
999 i++;
1000 continue;
1001 }
1002
1003 break;
1004 }
1005 temp[j] = '\0';
1006
1007 /* Point to after the closing quote. */
1008 if (c)
1009 i++;
1010 *sindex = i;
1011
1012 return (temp);
1013 }
1014
1015 /* This should really be another option to string_extract_double_quoted. */
1016 static int
1017 skip_double_quoted (string, slen, sind, flags)
1018 char *string;
1019 size_t slen;
1020 int sind;
1021 int flags;
1022 {
1023 int c, i;
1024 char *ret;
1025 int pass_next, backquote, si;
1026 DECLARE_MBSTATE;
1027
1028 pass_next = backquote = 0;
1029 i = sind;
1030 while (c = string[i])
1031 {
1032 if (pass_next)
1033 {
1034 pass_next = 0;
1035 ADVANCE_CHAR (string, slen, i);
1036 continue;
1037 }
1038 else if (c == '\\')
1039 {
1040 pass_next++;
1041 i++;
1042 continue;
1043 }
1044 else if (backquote)
1045 {
1046 if (c == '`')
1047 backquote = 0;
1048 ADVANCE_CHAR (string, slen, i);
1049 continue;
1050 }
1051 else if (c == '`')
1052 {
1053 backquote++;
1054 i++;
1055 continue;
1056 }
1057 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
1058 {
1059 si = i + 2;
1060 if (string[i + 1] == LPAREN)
1061 ret = extract_command_subst (string, &si, SX_NOALLOC|(flags&SX_COMPLETE));
1062 else
1063 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
1064
1065 /* These can consume the entire string if they are unterminated */
1066 CHECK_STRING_OVERRUN (i, si, slen, c);
1067
1068 i = si + 1;
1069 continue;
1070 }
1071 else if (c != '"')
1072 {
1073 ADVANCE_CHAR (string, slen, i);
1074 continue;
1075 }
1076 else
1077 break;
1078 }
1079
1080 if (c)
1081 i++;
1082
1083 return (i);
1084 }
1085
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. */
1090 static inline char *
1091 string_extract_single_quoted (string, sindex)
1092 char *string;
1093 int *sindex;
1094 {
1095 register int i;
1096 size_t slen;
1097 char *t;
1098 DECLARE_MBSTATE;
1099
1100 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1101 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
1102 i = *sindex;
1103 while (string[i] && string[i] != '\'')
1104 ADVANCE_CHAR (string, slen, i);
1105
1106 t = substring (string, *sindex, i);
1107
1108 if (string[i])
1109 i++;
1110 *sindex = i;
1111
1112 return (t);
1113 }
1114
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. */
1118 static inline int
1119 skip_single_quoted (string, slen, sind, flags)
1120 const char *string;
1121 size_t slen;
1122 int sind;
1123 int flags;
1124 {
1125 register int c;
1126 DECLARE_MBSTATE;
1127
1128 c = sind;
1129 while (string[c] && string[c] != '\'')
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 }
1135
1136 if (string[c])
1137 c++;
1138 return c;
1139 }
1140
1141 /* Just like string_extract, but doesn't hack backslashes or any of
1142 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
1143 static char *
1144 string_extract_verbatim (string, slen, sindex, charlist, flags)
1145 char *string;
1146 size_t slen;
1147 int *sindex;
1148 char *charlist;
1149 int flags;
1150 {
1151 register int i;
1152 #if defined (HANDLE_MULTIBYTE)
1153 wchar_t *wcharlist;
1154 #endif
1155 int c;
1156 char *temp;
1157 DECLARE_MBSTATE;
1158
1159 if ((flags & SX_NOCTLESC) && charlist[0] == '\'' && charlist[1] == '\0')
1160 {
1161 temp = string_extract_single_quoted (string, sindex);
1162 --*sindex; /* leave *sindex at separator character */
1163 return temp;
1164 }
1165
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
1178 i = *sindex;
1179 #if defined (HANDLE_MULTIBYTE)
1180 wcharlist = 0;
1181 #endif
1182 while (c = string[i])
1183 {
1184 #if defined (HANDLE_MULTIBYTE)
1185 size_t mblength;
1186 #endif
1187 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1188 {
1189 i += 2;
1190 CHECK_STRING_OVERRUN (i, i, slen, c);
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)
1197 {
1198 i += 2;
1199 CHECK_STRING_OVERRUN (i, i, slen, c);
1200 continue;
1201 }
1202
1203 #if defined (HANDLE_MULTIBYTE)
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);
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;
1225 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1226 mbstowcs (wcharlist, charlist, len + 1);
1227 }
1228
1229 if (wcschr (wcharlist, wc))
1230 break;
1231 }
1232 }
1233 else
1234 #endif
1235 if (MEMBER (c, charlist))
1236 break;
1237
1238 ADVANCE_CHAR (string, slen, i);
1239 }
1240
1241 #if defined (HANDLE_MULTIBYTE)
1242 FREE (wcharlist);
1243 #endif
1244
1245 temp = substring (string, *sindex, i);
1246 *sindex = i;
1247
1248 return (temp);
1249 }
1250
1251 /* Extract the $( construct in STRING, and return a new string.
1252 Start extracting at (SINDEX) as if we had just seen "$(".
1253 Make (SINDEX) get the position of the matching ")". )
1254 XFLAGS is additional flags to pass to other extraction functions. */
1255 char *
1256 extract_command_subst (string, sindex, xflags)
1257 char *string;
1258 int *sindex;
1259 int xflags;
1260 {
1261 char *ret;
1262
1263 if (string[*sindex] == LPAREN || (xflags & SX_COMPLETE))
1264 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1265 else
1266 {
1267 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1268 ret = xparse_dolparen (string, string+*sindex, sindex, xflags);
1269 return ret;
1270 }
1271 }
1272
1273 /* Extract the $[ construct in STRING, and return a new string. (])
1274 Start extracting at (SINDEX) as if we had just seen "$[".
1275 Make (SINDEX) get the position of the matching "]". */
1276 char *
1277 extract_arithmetic_subst (string, sindex)
1278 char *string;
1279 int *sindex;
1280 {
1281 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
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 "<(".
1287 Make (SINDEX) get the position of the matching ")". */ /*))*/
1288 char *
1289 extract_process_subst (string, starter, sindex, xflags)
1290 char *string;
1291 char *starter;
1292 int *sindex;
1293 int xflags;
1294 {
1295 #if 0
1296 /* XXX - check xflags&SX_COMPLETE here? */
1297 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
1298 #else
1299 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1300 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1301 #endif
1302 }
1303 #endif /* PROCESS_SUBSTITUTION */
1304
1305 #if defined (ARRAY_VARS)
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. */
1309 char *
1310 extract_array_assignment_list (string, sindex)
1311 char *string;
1312 int *sindex;
1313 {
1314 int slen;
1315 char *ret;
1316
1317 slen = strlen (string);
1318 if (string[slen - 1] == RPAREN)
1319 {
1320 ret = substring (string, *sindex, slen - 1);
1321 *sindex = slen - 1;
1322 return ret;
1323 }
1324 return 0;
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. */
1336 static char *
1337 extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
1338 char *string;
1339 int *sindex;
1340 char *opener, *alt_opener, *closer;
1341 int flags;
1342 {
1343 int i, c, si;
1344 size_t slen;
1345 char *t, *result;
1346 int pass_character, nesting_level, in_comment;
1347 int len_closer, len_opener, len_alt_opener;
1348 DECLARE_MBSTATE;
1349
1350 slen = strlen (string + *sindex) + *sindex;
1351 len_opener = STRLEN (opener);
1352 len_alt_opener = STRLEN (alt_opener);
1353 len_closer = STRLEN (closer);
1354
1355 pass_character = in_comment = 0;
1356
1357 nesting_level = 1;
1358 i = *sindex;
1359
1360 while (nesting_level)
1361 {
1362 c = string[i];
1363
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
1373 if (c == 0)
1374 break;
1375
1376 if (in_comment)
1377 {
1378 if (c == '\n')
1379 in_comment = 0;
1380 ADVANCE_CHAR (string, slen, i);
1381 continue;
1382 }
1383
1384 if (pass_character) /* previous char was backslash */
1385 {
1386 pass_character = 0;
1387 ADVANCE_CHAR (string, slen, i);
1388 continue;
1389 }
1390
1391 /* Not exactly right yet; should handle shell metacharacters and
1392 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
1393 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
1394 {
1395 in_comment = 1;
1396 ADVANCE_CHAR (string, slen, i);
1397 continue;
1398 }
1399
1400 if (c == CTLESC || c == '\\')
1401 {
1402 pass_character++;
1403 i++;
1404 continue;
1405 }
1406
1407 /* Process a nested command substitution, but only if we're parsing an
1408 arithmetic substitution. */
1409 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1410 {
1411 si = i + 2;
1412 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1413 CHECK_STRING_OVERRUN (i, si, slen, c);
1414 i = si + 1;
1415 continue;
1416 }
1417
1418 /* Process a nested OPENER. */
1419 if (STREQN (string + i, opener, len_opener))
1420 {
1421 si = i + len_opener;
1422 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
1423 CHECK_STRING_OVERRUN (i, si, slen, c);
1424 i = si + 1;
1425 continue;
1426 }
1427
1428 /* Process a nested ALT_OPENER */
1429 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
1430 {
1431 si = i + len_alt_opener;
1432 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
1433 CHECK_STRING_OVERRUN (i, si, slen, c);
1434 i = si + 1;
1435 continue;
1436 }
1437
1438 /* If the current substring terminates the delimited string, decrement
1439 the nesting level. */
1440 if (STREQN (string + i, closer, len_closer))
1441 {
1442 i += len_closer - 1; /* move to last byte of the closer */
1443 nesting_level--;
1444 if (nesting_level == 0)
1445 break;
1446 }
1447
1448 /* Pass old-style command substitution through verbatim. */
1449 if (c == '`')
1450 {
1451 si = i + 1;
1452 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1453 CHECK_STRING_OVERRUN (i, si, slen, c);
1454 i = si + 1;
1455 continue;
1456 }
1457
1458 /* Pass single-quoted and double-quoted strings through verbatim. */
1459 if (c == '\'' || c == '"')
1460 {
1461 si = i + 1;
1462 i = (c == '\'') ? skip_single_quoted (string, slen, si, 0)
1463 : skip_double_quoted (string, slen, si, 0);
1464 continue;
1465 }
1466
1467 /* move past this character, which was not special. */
1468 ADVANCE_CHAR (string, slen, i);
1469 }
1470
1471 if (c == 0 && nesting_level)
1472 {
1473 if (no_longjmp_on_fatal_error == 0)
1474 {
1475 last_command_exit_value = EXECUTION_FAILURE;
1476 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
1477 exp_jump_to_top_level (DISCARD);
1478 }
1479 else
1480 {
1481 *sindex = i;
1482 return (char *)NULL;
1483 }
1484 }
1485
1486 si = i - *sindex - len_closer + 1;
1487 if (flags & SX_NOALLOC)
1488 result = (char *)NULL;
1489 else
1490 {
1491 result = (char *)xmalloc (1 + si);
1492 strncpy (result, string + *sindex, si);
1493 result[si] = '\0';
1494 }
1495 *sindex = i;
1496
1497 return (result);
1498 }
1499
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 */
1508 static char *
1509 extract_dollar_brace_string (string, sindex, quoted, flags)
1510 char *string;
1511 int *sindex, quoted, flags;
1512 {
1513 register int i, c;
1514 size_t slen;
1515 int pass_character, nesting_level, si, dolbrace_state;
1516 char *result, *t;
1517 DECLARE_MBSTATE;
1518
1519 pass_character = 0;
1520 nesting_level = 1;
1521 slen = strlen (string + *sindex) + *sindex;
1522
1523 /* The handling of dolbrace_state needs to agree with the code in parse.y:
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 ${param op word} (SX_WORD). */
1527 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1528 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1529 dolbrace_state = DOLBRACE_QUOTE;
1530
1531 i = *sindex;
1532 while (c = string[i])
1533 {
1534 if (pass_character)
1535 {
1536 pass_character = 0;
1537 ADVANCE_CHAR (string, slen, i);
1538 continue;
1539 }
1540
1541 /* CTLESCs and backslashes quote the next character. */
1542 if (c == CTLESC || c == '\\')
1543 {
1544 pass_character++;
1545 i++;
1546 continue;
1547 }
1548
1549 if (string[i] == '$' && string[i+1] == LBRACE)
1550 {
1551 nesting_level++;
1552 i += 2;
1553 continue;
1554 }
1555
1556 if (c == RBRACE)
1557 {
1558 nesting_level--;
1559 if (nesting_level == 0)
1560 break;
1561 i++;
1562 continue;
1563 }
1564
1565 /* Pass the contents of old-style command substitutions through
1566 verbatim. */
1567 if (c == '`')
1568 {
1569 si = i + 1;
1570 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1571
1572 CHECK_STRING_OVERRUN (i, si, slen, c);
1573
1574 i = si + 1;
1575 continue;
1576 }
1577
1578 /* Pass the contents of new-style command substitutions and
1579 arithmetic substitutions through verbatim. */
1580 if (string[i] == '$' && string[i+1] == LPAREN)
1581 {
1582 si = i + 2;
1583 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
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
1600 i = si + 1;
1601 continue;
1602 }
1603 #endif
1604
1605 /* Pass the contents of double-quoted strings through verbatim. */
1606 if (c == '"')
1607 {
1608 si = i + 1;
1609 i = skip_double_quoted (string, slen, si, 0);
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);*/
1617 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
1618 ADVANCE_CHAR (string, slen, i);
1619 else
1620 {
1621 si = i + 1;
1622 i = skip_single_quoted (string, slen, si, 0);
1623 }
1624
1625 continue;
1626 }
1627
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
1638 /* move past this character, which was not special. */
1639 ADVANCE_CHAR (string, slen, i);
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)
1648 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
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;
1653 /* This is intended to handle all of the [:]op expansions and the substring/
1654 length/pattern removal/pattern substitution expansions. */
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;
1659 }
1660
1661 if (c == 0 && nesting_level)
1662 {
1663 if (no_longjmp_on_fatal_error == 0)
1664 { /* { */
1665 last_command_exit_value = EXECUTION_FAILURE;
1666 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
1667 exp_jump_to_top_level (DISCARD);
1668 }
1669 else
1670 {
1671 *sindex = i;
1672 return ((char *)NULL);
1673 }
1674 }
1675
1676 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
1677 *sindex = i;
1678
1679 return (result);
1680 }
1681
1682 /* Remove backslashes which are quoting backquotes from STRING. Modifies
1683 STRING, and returns a pointer to it. */
1684 char *
1685 de_backslash (string)
1686 char *string;
1687 {
1688 register size_t slen;
1689 register int i, j, prev_i;
1690 DECLARE_MBSTATE;
1691
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] == '\\' ||
1699 string[i + 1] == '$'))
1700 i++;
1701 prev_i = i;
1702 ADVANCE_CHAR (string, slen, i);
1703 if (j < prev_i)
1704 do string[j++] = string[prev_i++]; while (prev_i < i);
1705 else
1706 j = i;
1707 }
1708 string[j] = '\0';
1709
1710 return (string);
1711 }
1712
1713 #if 0
1714 /*UNUSED*/
1715 /* Replace instances of \! in a string with !. */
1716 void
1717 unquote_bang (string)
1718 char *string;
1719 {
1720 register int i, j;
1721 register char *temp;
1722
1723 temp = (char *)xmalloc (1 + strlen (string));
1724
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);
1735 }
1736 #endif
1737
1738 #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = oldjmp; return (x); } while (0)
1739
1740 /* This function assumes s[i] == open; returns with s[ret] == close; used to
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
1744 assignment parsing (see arrayfunc.c:assign_compound_array_list()) or
1745 during execution by a builtin which has already undergone word expansion. */
1746 static int
1747 skip_matched_pair (string, start, open, close, flags)
1748 const char *string;
1749 int start, open, close, flags;
1750 {
1751 int i, pass_next, backq, si, c, count, oldjmp;
1752 size_t slen;
1753 char *temp, *ss;
1754 DECLARE_MBSTATE;
1755
1756 slen = strlen (string + start) + start;
1757 oldjmp = no_longjmp_on_fatal_error;
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 }
1774 else if ((flags & 1) == 0 && c == '\\')
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 }
1787 else if ((flags & 1) == 0 && c == '`')
1788 {
1789 backq = 1;
1790 i++;
1791 continue;
1792 }
1793 else if ((flags & 1) == 0 && c == open)
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 }
1807 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
1808 {
1809 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i, 0)
1810 : skip_double_quoted (ss, slen, ++i, 0);
1811 /* no increment, the skip functions increment past the closing quote. */
1812 }
1813 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
1814 {
1815 si = i + 2;
1816 if (string[si] == '\0')
1817 CQ_RETURN(si);
1818
1819 /* XXX - extract_command_subst here? */
1820 if (string[i+1] == LPAREN)
1821 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1822 else
1823 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
1824
1825 CHECK_STRING_OVERRUN (i, si, slen, c);
1826
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)
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. */
1844 int
1845 skipsubscript (string, start, flags)
1846 const char *string;
1847 int start, flags;
1848 {
1849 return (skip_matched_pair (string, start, '[', ']', flags));
1850 }
1851 #endif
1852
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. */
1859 int
1860 skip_to_delim (string, start, delims, flags)
1861 char *string;
1862 int start;
1863 char *delims;
1864 int flags;
1865 {
1866 int i, pass_next, backq, dquote, si, c, oldjmp;
1867 int invert, skipquote, skipcmd, noprocsub, completeflag;
1868 int arithexp, skipcol;
1869 size_t slen;
1870 char *temp, open[3];
1871 DECLARE_MBSTATE;
1872
1873 slen = strlen (string + start) + start;
1874 oldjmp = no_longjmp_on_fatal_error;
1875 if (flags & SD_NOJMP)
1876 no_longjmp_on_fatal_error = 1;
1877 invert = (flags & SD_INVERT);
1878 skipcmd = (flags & SD_NOSKIPCMD) == 0;
1879 noprocsub = (flags & SD_NOPROCSUB);
1880 completeflag = (flags & SD_COMPLETE) ? SX_COMPLETE : 0;
1881
1882 arithexp = (flags & SD_ARITHEXP);
1883 skipcol = 0;
1884
1885 i = start;
1886 pass_next = backq = dquote = 0;
1887 while (c = string[i])
1888 {
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 =='"'));
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 }
1921 else if (arithexp && skipcol && c == ':')
1922 {
1923 skipcol--;
1924 i++;
1925 continue;
1926 }
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 }
1960 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
1961 {
1962 si = i + 2;
1963 if (string[si] == '\0')
1964 CQ_RETURN(si);
1965
1966 if (string[i+1] == LPAREN)
1967 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1968 else
1969 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1970 CHECK_STRING_OVERRUN (i, si, slen, c);
1971 i = si;
1972 if (string[i] == '\0') /* don't increment i past EOS in loop */
1973 break;
1974 i++;
1975 continue;
1976 }
1977 #if defined (PROCESS_SUBSTITUTION)
1978 else if (skipcmd && noprocsub == 0 && (c == '<' || c == '>') && string[i+1] == LPAREN)
1979 {
1980 si = i + 2;
1981 if (string[si] == '\0')
1982 CQ_RETURN(si);
1983
1984 temp = extract_delimited_string (string, &si, (c == '<') ? "<(" : ">(", "(", ")", SX_COMMAND|SX_NOALLOC); /* )) */
1985 CHECK_STRING_OVERRUN (i, si, slen, c);
1986 i = si;
1987 if (string[i] == '\0')
1988 break;
1989 i++;
1990 continue;
1991 }
1992 #endif /* PROCESS_SUBSTITUTION */
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
2005 CHECK_STRING_OVERRUN (i, si, slen, c);
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
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 }
2027 else if ((skipquote || invert) && (member (c, delims) == 0))
2028 break;
2029 else
2030 ADVANCE_CHAR (string, slen, i);
2031 }
2032
2033 CQ_RETURN(i);
2034 }
2035
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 */
2041 int
2042 skip_to_histexp (string, start, delims, flags)
2043 char *string;
2044 int start;
2045 char *delims;
2046 int flags;
2047 {
2048 int i, pass_next, backq, dquote, c, oldjmp;
2049 int histexp_comsub, histexp_backq, old_dquote;
2050 size_t slen;
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
2157 #if defined (READLINE)
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
2160 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
2161 single and double-quoted string parsing functions should not return an
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. */
2165
2166 int
2167 char_is_quoted (string, eindex)
2168 char *string;
2169 int eindex;
2170 {
2171 int i, pass_next, c, oldjmp;
2172 size_t slen;
2173 DECLARE_MBSTATE;
2174
2175 slen = strlen (string);
2176 oldjmp = no_longjmp_on_fatal_error;
2177 no_longjmp_on_fatal_error = 1;
2178 i = pass_next = 0;
2179 while (i <= eindex)
2180 {
2181 c = string[i];
2182
2183 if (pass_next)
2184 {
2185 pass_next = 0;
2186 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
2187 CQ_RETURN(1);
2188 ADVANCE_CHAR (string, slen, i);
2189 continue;
2190 }
2191 else if (c == '\\')
2192 {
2193 pass_next = 1;
2194 i++;
2195 continue;
2196 }
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 }
2204 else if (c == '\'' || c == '"')
2205 {
2206 i = (c == '\'') ? skip_single_quoted (string, slen, ++i, 0)
2207 : skip_double_quoted (string, slen, ++i, SX_COMPLETE);
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);
2214 }
2215
2216 CQ_RETURN(0);
2217 }
2218
2219 int
2220 unclosed_pair (string, eindex, openstr)
2221 char *string;
2222 int eindex;
2223 char *openstr;
2224 {
2225 int i, pass_next, openc, olen;
2226 size_t slen;
2227 DECLARE_MBSTATE;
2228
2229 slen = strlen (string);
2230 olen = strlen (openstr);
2231 i = pass_next = openc = 0;
2232 while (i <= eindex)
2233 {
2234 if (pass_next)
2235 {
2236 pass_next = 0;
2237 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
2238 return 0;
2239 ADVANCE_CHAR (string, slen, i);
2240 continue;
2241 }
2242 else if (string[i] == '\\')
2243 {
2244 pass_next = 1;
2245 i++;
2246 continue;
2247 }
2248 else if (STREQN (string + i, openstr, olen))
2249 {
2250 openc = 1 - openc;
2251 i += olen;
2252 }
2253 /* XXX - may want to handle $'...' specially here */
2254 else if (string[i] == '\'' || string[i] == '"')
2255 {
2256 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i, 0)
2257 : skip_double_quoted (string, slen, i, SX_COMPLETE);
2258 if (i > eindex)
2259 return 0;
2260 }
2261 else
2262 ADVANCE_CHAR (string, slen, i);
2263 }
2264 return (openc);
2265 }
2266
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
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,
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
2273 DELIMS delimit separate fields. This is used by programmable completion. */
2274 WORD_LIST *
2275 split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
2276 char *string;
2277 int slen;
2278 char *delims;
2279 int sentinel, flags;
2280 int *nwp, *cwp;
2281 {
2282 int ts, te, i, nw, cw, ifs_split, dflags;
2283 char *token, *d, *d2;
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
2295 d = (delims == 0) ? ifs_value : delims;
2296 ifs_split = delims == 0;
2297
2298 /* Make d2 the non-whitespace characters in delims */
2299 d2 = 0;
2300 if (delims)
2301 {
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])
2312 {
2313 #if defined (HANDLE_MULTIBYTE)
2314 mbstate_t state_bak;
2315 state_bak = state;
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)
2329 d2[ts++] = delims[i];
2330
2331 i++;
2332 slength--;
2333 }
2334 d2[ts] = '\0';
2335 }
2336
2337 ret = (WORD_LIST *)NULL;
2338
2339 /* Remove sequences of whitespace characters at the start of the string, as
2340 long as those characters are delimiters. */
2341 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
2342 ;
2343 if (string[i] == '\0')
2344 {
2345 FREE (d2);
2346 return (ret);
2347 }
2348
2349 ts = i;
2350 nw = 0;
2351 cw = -1;
2352 dflags = flags|SD_NOJMP;
2353 while (1)
2354 {
2355 te = skip_to_delim (string, ts, d, dflags);
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. */
2360 if (ts == te && d2 && member (string[ts], d2))
2361 {
2362 te = ts + 1;
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)
2366 while (member (string[te], d) && spctabnl (string[te]) && ((flags&SD_NOQUOTEDELIM) == 0 || (string[te] != '\'' && string[te] != '"')))
2367 te++;
2368 else
2369 while (member (string[te], d2) && ((flags&SD_NOQUOTEDELIM) == 0 || (string[te] != '\'' && string[te] != '"')))
2370 te++;
2371 }
2372
2373 token = substring (string, ts, te);
2374
2375 ret = add_string_to_list (token, ret); /* XXX */
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
2383 sentinel word to the current word. */
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
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. */
2390 if (cwp && cw == -1 && sentinel < ts)
2391 {
2392 tl = make_word_list (make_word (""), ret->next);
2393 ret->next = tl;
2394 cw = nw;
2395 nw++;
2396 }
2397
2398 if (string[te] == 0)
2399 break;
2400
2401 i = te;
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] != '"')))
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
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))
2418 {
2419 if (whitespace (string[sentinel - 1]))
2420 {
2421 token = "";
2422 ret = add_string_to_list (token, ret);
2423 nw++;
2424 }
2425 cw = nw;
2426 }
2427
2428 if (nwp)
2429 *nwp = nw;
2430 if (cwp)
2431 *cwp = cw;
2432
2433 FREE (d2);
2434
2435 return (REVERSE_LIST (ret, WORD_LIST *));
2436 }
2437 #endif /* READLINE */
2438
2439 #if 0
2440 /* UNUSED */
2441 /* Extract the name of the variable to bind to from the assignment string. */
2442 char *
2443 assignment_name (string)
2444 char *string;
2445 {
2446 int offset;
2447 char *temp;
2448
2449 offset = assignment (string, 0);
2450 if (offset == 0)
2451 return (char *)NULL;
2452 temp = substring (string, 0, offset);
2453 return (temp);
2454 }
2455 #endif
2456
2457 /* **************************************************************** */
2458 /* */
2459 /* Functions to convert strings to WORD_LISTs and vice versa */
2460 /* */
2461 /* **************************************************************** */
2462
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. */
2465 char *
2466 string_list_internal (list, sep)
2467 WORD_LIST *list;
2468 char *sep;
2469 {
2470 register WORD_LIST *t;
2471 char *result, *r;
2472 size_t word_len, sep_len, result_size;
2473
2474 if (list == 0)
2475 return ((char *)NULL);
2476
2477 /* Short-circuit quickly if we don't need to separate anything. */
2478 if (list->next == 0)
2479 return (savestring (list->word->word));
2480
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
2492 r = result = (char *)xmalloc (result_size + 1);
2493
2494 for (t = list; t; t = t->next)
2495 {
2496 if (t != list && sep_len)
2497 {
2498 if (sep_len > 1)
2499 {
2500 FASTCOPY (sep, r, sep_len);
2501 r += sep_len;
2502 }
2503 else
2504 *r++ = sep[0];
2505 }
2506
2507 word_len = strlen (t->word->word);
2508 FASTCOPY (t->word->word, r, word_len);
2509 r += word_len;
2510 }
2511
2512 *r = '\0';
2513 return (result);
2514 }
2515
2516 /* Return a single string of all the words present in LIST, separating
2517 each word with a space. */
2518 char *
2519 string_list (list)
2520 WORD_LIST *list;
2521 {
2522 return (string_list_internal (list, " "));
2523 }
2524
2525 /* An external interface that can be used by the rest of the shell to
2526 obtain a string containing the first character in $IFS. Handles all
2527 the multibyte complications. If LENP is non-null, it is set to the
2528 length of the returned string. */
2529 char *
2530 ifs_firstchar (lenp)
2531 int *lenp;
2532 {
2533 char *ret;
2534 int len;
2535
2536 ret = xmalloc (MB_LEN_MAX + 1);
2537 #if defined (HANDLE_MULTIBYTE)
2538 if (ifs_firstc_len == 1)
2539 {
2540 ret[0] = ifs_firstc[0];
2541 ret[1] = '\0';
2542 len = ret[0] ? 1 : 0;
2543 }
2544 else
2545 {
2546 memcpy (ret, ifs_firstc, ifs_firstc_len);
2547 ret[len = ifs_firstc_len] = '\0';
2548 }
2549 #else
2550 ret[0] = ifs_firstc;
2551 ret[1] = '\0';
2552 len = ret[0] ? 0 : 1;
2553 #endif
2554
2555 if (lenp)
2556 *lenp = len;
2557
2558 return ret;
2559 }
2560
2561 /* Return a single string of all the words present in LIST, obeying the
2562 quoting rules for "$*", 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." */
2566 /* Posix interpretation 888 changes this when IFS is null by specifying
2567 that when unquoted, this expands to separate arguments */
2568 char *
2569 string_list_dollar_star (list, quoted, flags)
2570 WORD_LIST *list;
2571 int quoted, flags;
2572 {
2573 char *ret;
2574 #if defined (HANDLE_MULTIBYTE)
2575 # if defined (__GNUC__)
2576 char sep[MB_CUR_MAX + 1];
2577 # else
2578 char *sep = 0;
2579 # endif
2580 #else
2581 char sep[2];
2582 #endif
2583
2584 #if defined (HANDLE_MULTIBYTE)
2585 # if !defined (__GNUC__)
2586 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2587 # endif /* !__GNUC__ */
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
2599 sep[0] = ifs_firstc;
2600 sep[1] = '\0';
2601 #endif
2602
2603 ret = string_list_internal (list, sep);
2604 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2605 free (sep);
2606 #endif
2607 return ret;
2608 }
2609
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
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'. */
2625 char *
2626 string_list_dollar_at (list, quoted, flags)
2627 WORD_LIST *list;
2628 int quoted;
2629 int flags;
2630 {
2631 char *ifs, *ret;
2632 #if defined (HANDLE_MULTIBYTE)
2633 # if defined (__GNUC__)
2634 char sep[MB_CUR_MAX + 1];
2635 # else
2636 char *sep = 0;
2637 # endif /* !__GNUC__ */
2638 #else
2639 char sep[2];
2640 #endif
2641 WORD_LIST *tlist;
2642
2643 /* XXX this could just be ifs = ifs_value; */
2644 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
2645
2646 #if defined (HANDLE_MULTIBYTE)
2647 # if !defined (__GNUC__)
2648 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2649 # endif /* !__GNUC__ */
2650 /* XXX - testing PF_ASSIGNRHS to make sure positional parameters are
2651 separated with a space even when word splitting will not occur. */
2652 if (flags & PF_ASSIGNRHS)
2653 {
2654 sep[0] = ' ';
2655 sep[1] = '\0';
2656 }
2657 else if (ifs && *ifs)
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 }
2675 #else /* !HANDLE_MULTIBYTE */
2676 /* XXX - PF_ASSIGNRHS means no word splitting, so we want positional
2677 parameters separated by a space. */
2678 sep[0] = ((flags & PF_ASSIGNRHS) || ifs == 0 || *ifs == 0) ? ' ' : *ifs;
2679 sep[1] = '\0';
2680 #endif /* !HANDLE_MULTIBYTE */
2681
2682 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2683 it now that quote_escapes quotes spaces */
2684 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
2685 ? quote_list (list)
2686 : list_quote_escapes (list);
2687
2688 ret = string_list_internal (tlist, sep);
2689 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2690 free (sep);
2691 #endif
2692 return ret;
2693 }
2694
2695 /* Turn the positional parameters into a string, understanding quoting and
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. */
2699 /* This needs to fully understand the additional contexts where word
2700 splitting does not occur (W_ASSIGNRHS, etc.) */
2701 char *
2702 string_list_pos_params (pchar, list, quoted, pflags)
2703 int pchar;
2704 WORD_LIST *list;
2705 int quoted, pflags;
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);
2714 ret = string_list_dollar_star (tlist, 0, 0);
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 }
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 */
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 */
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. */
2731 ret = string_list_dollar_star (list, quoted, 0);
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. */
2741 ret = string_list_dollar_at (list, quoted, 0);
2742 else if (pchar == '@' && quoted == 0 && ifs_is_null) /* XXX */
2743 ret = string_list_dollar_at (list, quoted, 0); /* Posix interp 888 */
2744 else if (pchar == '@' && quoted == 0 && (pflags & PF_ASSIGNRHS))
2745 ret = string_list_dollar_at (list, quoted, pflags); /* Posix interp 888 */
2746 else if (pchar == '@')
2747 ret = string_list_dollar_star (list, quoted, 0);
2748 else
2749 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2750
2751 return ret;
2752 }
2753
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
2779 /* This performs word splitting and quoted null character removal on
2780 STRING. */
2781 #define issep(c) \
2782 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2783 : (c) == (separators)[0]) \
2784 : 0)
2785
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
2793 WORD_LIST *
2794 list_string (string, separators, quoted)
2795 register char *string, *separators;
2796 int quoted;
2797 {
2798 WORD_LIST *result;
2799 WORD_DESC *t;
2800 char *current_word, *s;
2801 int sindex, sh_style_split, whitesep, xflags, free_word;
2802 size_t slen;
2803
2804 if (!string || !*string)
2805 return ((WORD_LIST *)NULL);
2806
2807 sh_style_split = separators && separators[0] == ' ' &&
2808 separators[1] == '\t' &&
2809 separators[2] == '\n' &&
2810 separators[3] == '\0';
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 }
2816
2817 slen = 0;
2818 /* Remove sequences of whitespace at the beginning of STRING, as
2819 long as those characters appear in IFS. Do not do this if
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. */
2823 #if 0
2824 if (!quoted || !separators || !*separators)
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
2831 {
2832 for (s = string; *s && issep (*s) && ifs_whitespace (*s); s++);
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:
2842 extract a word, stopping at a separator
2843 skip sequences of whitespace characters as long as they are separators
2844 This obeys the field splitting rules in Posix.2. */
2845 slen = STRLEN (string);
2846 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
2847 {
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 */
2850 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
2851 if (current_word == 0)
2852 break;
2853
2854 free_word = 1; /* If non-zero, we free current_word */
2855
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 {
2862 t = alloc_word_desc ();
2863 t->word = make_quoted_char ('\0');
2864 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2865 result = make_word_list (t, result);
2866 }
2867 else if (current_word[0] != '\0')
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);
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;
2881 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
2882 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2883 result->word->flags |= W_QUOTED;
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;
2888 }
2889
2890 /* If we're not doing sequences of separators in the traditional
2891 Bourne shell style, then add a quoted null argument. */
2892 else if (!sh_style_split && !ifs_whitespace (string[sindex]))
2893 {
2894 t = alloc_word_desc ();
2895 t->word = make_quoted_char ('\0');
2896 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2897 result = make_word_list (t, result);
2898 }
2899
2900 if (free_word)
2901 free (current_word);
2902
2903 /* Note whether or not the separator is IFS whitespace, used later. */
2904 whitesep = string[sindex] && ifs_whitesep (string[sindex]);
2905
2906 /* Move past the current separator character. */
2907 if (string[sindex])
2908 {
2909 DECLARE_MBSTATE;
2910 ADVANCE_CHAR (string, slen, sindex);
2911 }
2912
2913 /* Now skip sequences of whitespace characters if they are
2914 in the list of separators. */
2915 while (string[sindex] && ifs_whitesep (string[sindex]) && issep (string[sindex]))
2916 sindex++;
2917
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). */
2922 if (string[sindex] && whitesep && issep (string[sindex]) && !ifs_whitesep (string[sindex]))
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) */
2927 while (string[sindex] && ifs_whitesep (string[sindex]) && isifs (string[sindex]))
2928 sindex++;
2929 }
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
2936 the `read' builtin.
2937
2938 This is never called with SEPARATORS != $IFS, and takes advantage of that.
2939
2940 XXX - this function is very similar to list_string; they should be
2941 combined - XXX */
2942
2943 /* character is in $IFS */
2944 #define islocalsep(c) (local_cmap[(unsigned char)(c)] != 0)
2945
2946 char *
2947 get_word_from_string (stringp, separators, endptr)
2948 char **stringp, *separators, **endptr;
2949 {
2950 register char *s;
2951 char *current_word;
2952 int sindex, sh_style_split, whitesep, xflags;
2953 unsigned char local_cmap[UCHAR_MAX+1]; /* really only need single-byte chars here */
2954 size_t slen;
2955
2956 if (!stringp || !*stringp || !**stringp)
2957 return ((char *)NULL);
2958
2959 sh_style_split = separators && separators[0] == ' ' &&
2960 separators[1] == '\t' &&
2961 separators[2] == '\n' &&
2962 separators[3] == '\0';
2963 memset (local_cmap, '\0', sizeof (local_cmap));
2964 for (xflags = 0, s = separators; s && *s; s++)
2965 {
2966 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2967 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2968 local_cmap[(unsigned char)*s] = 1; /* local charmap of separators */
2969 }
2970
2971 s = *stringp;
2972 slen = 0;
2973
2974 /* Remove sequences of whitespace at the beginning of STRING, as
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)
2978 for (; *s && spctabnl (*s) && islocalsep (*s); s++);
2979 else
2980 for (; *s && ifs_whitespace (*s) && islocalsep (*s); s++);
2981
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);
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.
2995
2996 This obeys the field splitting rules in Posix.2. */
2997 sindex = 0;
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);
3001 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
3002
3003 /* Set ENDPTR to the first character after the end of the word. */
3004 if (endptr)
3005 *endptr = s + sindex;
3006
3007 /* Note whether or not the separator is IFS whitespace, used later. */
3008 whitesep = s[sindex] && ifs_whitesep (s[sindex]);
3009
3010 /* Move past the current separator character. */
3011 if (s[sindex])
3012 {
3013 DECLARE_MBSTATE;
3014 ADVANCE_CHAR (s, slen, sindex);
3015 }
3016
3017 /* Now skip sequences of space, tab, or newline characters if they are
3018 in the list of separators. */
3019 while (s[sindex] && spctabnl (s[sindex]) && islocalsep (s[sindex]))
3020 sindex++;
3021
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). */
3026 if (s[sindex] && whitesep && islocalsep (s[sindex]) && !ifs_whitesep (s[sindex]))
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. */
3031 while (s[sindex] && ifs_whitesep (s[sindex]) && islocalsep(s[sindex]))
3032 sindex++;
3033 }
3034
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. */
3045 char *
3046 strip_trailing_ifs_whitespace (string, separators, saw_escape)
3047 char *string, *separators;
3048 int saw_escape;
3049 {
3050 char *s;
3051
3052 s = string + STRLEN (string) - 1;
3053 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
3054 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
3055 s--;
3056 *++s = '\0';
3057 return string;
3058 }
3059
3060 #if 0
3061 /* UNUSED */
3062 /* Split STRING into words at whitespace. Obeys shell-style quoting with
3063 backslashes, single and double quotes. */
3064 WORD_LIST *
3065 list_string_with_quotes (string)
3066 char *string;
3067 {
3068 WORD_LIST *list;
3069 char *token, *s;
3070 size_t s_len;
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
3078 s_len = strlen (s);
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 == '\'')
3091 i = skip_single_quoted (s, s_len, ++i, 0);
3092 else if (c == '"')
3093 i = skip_double_quoted (s, s_len, ++i, 0);
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. */
3098 token = substring (s, tokstart, i);
3099 list = add_string_to_list (token, list);
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 }
3113 #endif
3114
3115 /********************************************************/
3116 /* */
3117 /* Functions to perform assignment statements */
3118 /* */
3119 /********************************************************/
3120
3121 #if defined (ARRAY_VARS)
3122 static SHELL_VAR *
3123 do_compound_assignment (name, value, flags)
3124 char *name, *value;
3125 int flags;
3126 {
3127 SHELL_VAR *v;
3128 int mklocal, mkassoc, mkglobal, chklocal;
3129 WORD_LIST *list;
3130 char *newname; /* used for local nameref references */
3131
3132 mklocal = flags & ASS_MKLOCAL;
3133 mkassoc = flags & ASS_MKASSOC;
3134 mkglobal = flags & ASS_MKGLOBAL;
3135 chklocal = flags & ASS_CHKLOCAL;
3136
3137 if (mklocal && variable_context)
3138 {
3139 v = find_variable (name); /* follows namerefs */
3140 newname = (v == 0) ? nameref_transform_name (name, flags) : v->name;
3141 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3142 {
3143 if (readonly_p (v))
3144 err_readonly (name);
3145 return (v); /* XXX */
3146 }
3147 list = expand_compound_array_assignment (v, value, flags);
3148 if (mkassoc)
3149 v = make_local_assoc_variable (newname, 0);
3150 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
3151 v = make_local_array_variable (newname, 0);
3152 if (v)
3153 assign_compound_array_list (v, list, flags);
3154 if (list)
3155 dispose_words (list);
3156 }
3157 /* In a function but forcing assignment in global context. CHKLOCAL means to
3158 check for an existing local variable first. */
3159 else if (mkglobal && variable_context)
3160 {
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);
3166 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3167 {
3168 if (readonly_p (v))
3169 err_readonly (name);
3170 return (v); /* XXX */
3171 }
3172 /* sanity check */
3173 newname = (v == 0) ? nameref_transform_name (name, flags) : name;
3174 list = expand_compound_array_assignment (v, value, flags);
3175 if (v == 0 && mkassoc)
3176 v = make_new_assoc_variable (newname);
3177 else if (v && mkassoc && assoc_p (v) == 0)
3178 v = convert_var_to_assoc (v);
3179 else if (v == 0)
3180 v = make_new_array_variable (newname);
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);
3185 if (list)
3186 dispose_words (list);
3187 }
3188 else
3189 {
3190 v = assign_array_from_string (name, value, flags);
3191 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3192 {
3193 if (readonly_p (v))
3194 err_readonly (name);
3195 return (v); /* XXX */
3196 }
3197 }
3198
3199 return (v);
3200 }
3201 #endif
3202
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. */
3208 static int
3209 do_assignment_internal (word, expand)
3210 const WORD_DESC *word;
3211 int expand;
3212 {
3213 int offset, appendop, assign_list, aflags, retval;
3214 char *name, *value, *temp;
3215 SHELL_VAR *entry;
3216 #if defined (ARRAY_VARS)
3217 char *t;
3218 int ni;
3219 #endif
3220 const char *string;
3221
3222 if (word == 0 || word->word == 0)
3223 return 0;
3224
3225 appendop = assign_list = aflags = 0;
3226 string = word->word;
3227 offset = assignment (string, 0);
3228 name = savestring (string);
3229 value = (char *)NULL;
3230
3231 if (name[offset] == '=')
3232 {
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 */
3240 temp = name + offset + 1;
3241
3242 #if defined (ARRAY_VARS)
3243 if (expand && (word->flags & W_COMPASSIGN))
3244 {
3245 assign_list = ni = 1;
3246 value = extract_array_assignment_list (temp, &ni);
3247 }
3248 else
3249 #endif
3250 if (expand && temp[0])
3251 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
3252 else
3253 value = savestring (temp);
3254 }
3255
3256 if (value == 0)
3257 {
3258 value = (char *)xmalloc (1);
3259 value[0] = '\0';
3260 }
3261
3262 if (echo_command_at_execute)
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 }
3270
3271 #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
3272
3273 if (appendop)
3274 aflags |= ASS_APPEND;
3275
3276 #if defined (ARRAY_VARS)
3277 if (t = mbschr (name, LBRACK))
3278 {
3279 if (assign_list)
3280 {
3281 report_error (_("%s: cannot assign list to array member"), name);
3282 ASSIGN_RETURN (0);
3283 }
3284 entry = assign_array_element (name, value, aflags);
3285 if (entry == 0)
3286 ASSIGN_RETURN (0);
3287 }
3288 else if (assign_list)
3289 {
3290 if ((word->flags & W_ASSIGNARG) && (word->flags & W_CHKLOCAL))
3291 aflags |= ASS_CHKLOCAL;
3292 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
3293 aflags |= ASS_MKLOCAL;
3294 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
3295 aflags |= ASS_MKGLOBAL;
3296 if (word->flags & W_ASSIGNASSOC)
3297 aflags |= ASS_MKASSOC;
3298 entry = do_compound_assignment (name, value, aflags);
3299 }
3300 else
3301 #endif /* ARRAY_VARS */
3302 entry = bind_variable (name, value, aflags);
3303
3304 if (entry)
3305 stupidly_hack_special_variables (entry->name); /* might be a nameref */
3306 else
3307 stupidly_hack_special_variables (name);
3308
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 {
3314 set_exit_status (EXECUTION_FAILURE);
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);
3324 }
3325
3326 /* Perform the assignment statement in STRING, and expand the
3327 right side by doing tilde, command and parameter expansion. */
3328 int
3329 do_assignment (string)
3330 char *string;
3331 {
3332 WORD_DESC td;
3333
3334 td.flags = W_ASSIGNMENT;
3335 td.word = string;
3336
3337 return do_assignment_internal (&td, 1);
3338 }
3339
3340 int
3341 do_word_assignment (word, flags)
3342 WORD_DESC *word;
3343 int flags;
3344 {
3345 return do_assignment_internal (word, 1);
3346 }
3347
3348 /* Given STRING, an assignment string, get the value of the right side
3349 of the `=', and bind it to the left side. Do not perform any word
3350 expansions on the right hand side. */
3351 int
3352 do_assignment_no_expand (string)
3353 char *string;
3354 {
3355 WORD_DESC td;
3356
3357 td.flags = W_ASSIGNMENT;
3358 td.word = string;
3359
3360 return (do_assignment_internal (&td, 0));
3361 }
3362
3363 /***************************************************
3364 * *
3365 * Functions to manage the positional parameters *
3366 * *
3367 ***************************************************/
3368
3369 /* Return the word list that corresponds to `$*'. */
3370 WORD_LIST *
3371 list_rest_of_args ()
3372 {
3373 register WORD_LIST *list, *args;
3374 int i;
3375
3376 /* Break out of the loop as soon as one of the dollar variables is null. */
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);
3382
3383 return (REVERSE_LIST (list, WORD_LIST *));
3384 }
3385
3386 /* Return the value of a positional parameter. This handles values > 10. */
3387 char *
3388 get_dollar_var_value (ind)
3389 intmax_t ind;
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 ${11} */
3397 {
3398 ind -= 10;
3399 for (p = rest_of_args; p && ind--; p = p->next)
3400 ;
3401 temp = p ? savestring (p->word->word) : (char *)NULL;
3402 }
3403 return (temp);
3404 }
3405
3406 /* Make a single large string out of the dollar digit variables,
3407 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
3408 case of "$*" with respect to IFS. */
3409 char *
3410 string_rest_of_args (dollar_star)
3411 int dollar_star;
3412 {
3413 register WORD_LIST *list;
3414 char *string;
3415
3416 list = list_rest_of_args ();
3417 string = dollar_star ? string_list_dollar_star (list, 0, 0) : string_list (list);
3418 dispose_words (list);
3419 return (string);
3420 }
3421
3422 /* Return a string containing the positional parameters from START to
3423 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
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. */
3427 static char *
3428 pos_params (string, start, end, quoted, pflags)
3429 char *string;
3430 int start, end, quoted, pflags;
3431 {
3432 WORD_LIST *save, *params, *h, *t;
3433 char *ret;
3434 int i;
3435
3436 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3437 if (start == end)
3438 return ((char *)NULL);
3439
3440 save = params = list_rest_of_args ();
3441 if (save == 0 && start > 0)
3442 return ((char *)NULL);
3443
3444 if (start == 0) /* handle ${@:0[:x]} specially */
3445 {
3446 t = make_word_list (make_word (dollar_vars[0]), params);
3447 save = params = t;
3448 }
3449
3450 for (i = start ? 1 : 0; params && i < start; i++)
3451 params = params->next;
3452 if (params == 0)
3453 {
3454 dispose_words (save);
3455 return ((char *)NULL);
3456 }
3457 for (h = t = params; params && i < end; i++)
3458 {
3459 t = params;
3460 params = params->next;
3461 }
3462 t->next = (WORD_LIST *)NULL;
3463
3464 ret = string_list_pos_params (string[0], h, quoted, pflags);
3465
3466 if (t != params)
3467 t->next = params;
3468
3469 dispose_words (save);
3470 return (ret);
3471 }
3472
3473 /******************************************************************/
3474 /* */
3475 /* Functions to expand strings to strings or WORD_LISTs */
3476 /* */
3477 /******************************************************************/
3478
3479 #if defined (PROCESS_SUBSTITUTION)
3480 #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
3481 #else
3482 #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
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. */
3488 static char *
3489 expand_string_if_necessary (string, quoted, func)
3490 char *string;
3491 int quoted;
3492 EXPFUNC *func;
3493 {
3494 WORD_LIST *list;
3495 size_t slen;
3496 int i, saw_quote;
3497 char *ret;
3498 DECLARE_MBSTATE;
3499
3500 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3501 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3502 i = saw_quote = 0;
3503 while (string[i])
3504 {
3505 if (EXP_CHAR (string[i]))
3506 break;
3507 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3508 saw_quote = 1;
3509 ADVANCE_CHAR (string, slen, i);
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);
3527
3528 return ret;
3529 }
3530
3531 static inline char *
3532 expand_string_to_string_internal (string, quoted, func)
3533 char *string;
3534 int quoted;
3535 EXPFUNC *func;
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
3555 char *
3556 expand_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
3563 char *
3564 expand_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
3571 char *
3572 expand_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
3579 char *
3580 expand_arith_string (string, quoted)
3581 char *string;
3582 int quoted;
3583 {
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 */
3607 td.flags = W_NOPROCSUB|W_NOTILDE; /* don't want process substitution or tilde expansion */
3608 #if 0 /* TAG: bash-5.2 */
3609 if (quoted & Q_ARRAYSUB)
3610 td.flags |= W_NOCOMSUB;
3611 #endif
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 }
3634 else if (saw_quote && (quoted & Q_ARITH))
3635 ret = string_quote_removal (string, quoted);
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;
3642 }
3643
3644 #if defined (COND_COMMAND)
3645 /* Just remove backslashes in STRING. Returns a new string. */
3646 char *
3647 remove_backslashes (string)
3648 char *string;
3649 {
3650 char *r, *ret, *s;
3651
3652 r = ret = (char *)xmalloc (strlen (string) + 1);
3653 for (s = string; s && *s; )
3654 {
3655 if (*s == '\\')
3656 s++;
3657 if (*s == 0)
3658 break;
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
3667 [[...]] expression. If SPECIAL is 1, this is the rhs argument
3668 to the != or == operator, and should be treated as a pattern. In
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
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. */
3675 char *
3676 cond_expand_word (w, special)
3677 WORD_DESC *w;
3678 int special;
3679 {
3680 char *r, *p;
3681 WORD_LIST *l;
3682 int qflags;
3683
3684 if (w->word == 0 || w->word[0] == '\0')
3685 return ((char *)NULL);
3686
3687 expand_no_split_dollar_star = 1;
3688 w->flags |= W_NOSPLIT2;
3689 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3690 expand_no_split_dollar_star = 0;
3691 if (l)
3692 {
3693 if (special == 0) /* LHS */
3694 {
3695 if (l->word)
3696 word_list_remove_quoted_nulls (l);
3697 dequote_list (l);
3698 r = string_list (l);
3699 }
3700 else
3701 {
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. */
3705 qflags = QGLOB_CVTNULL|QGLOB_CTLESC;
3706 if (special == 2)
3707 qflags |= QGLOB_REGEXP;
3708 word_list_remove_quoted_nulls (l);
3709 p = string_list (l);
3710 r = quote_string_for_globbing (p, qflags);
3711 free (p);
3712 }
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. */
3725 static WORD_LIST *
3726 call_expand_word_internal (w, q, i, c, e)
3727 WORD_DESC *w;
3728 int q, i, *c, *e;
3729 {
3730 WORD_LIST *result;
3731
3732 result = expand_word_internal (w, q, i, c, e);
3733 if (result == &expand_word_error || result == &expand_word_fatal)
3734 {
3735 /* By convention, each time this error is returned, w->word has
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). */
3739 w->word = (char *)NULL;
3740 last_command_exit_value = EXECUTION_FAILURE;
3741 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
3742 /* NOTREACHED */
3743 return (NULL);
3744 }
3745 else
3746 return (result);
3747 }
3748
3749 /* Perform parameter expansion, command substitution, and arithmetic
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. */
3753 static WORD_LIST *
3754 expand_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
3764 td.flags = 0;
3765 td.word = savestring (string);
3766
3767 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3768
3769 FREE (td.word);
3770 return (tresult);
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. */
3778 WORD_LIST *
3779 expand_string_unsplit (string, quoted)
3780 char *string;
3781 int quoted;
3782 {
3783 WORD_LIST *value;
3784
3785 if (string == 0 || *string == '\0')
3786 return ((WORD_LIST *)NULL);
3787
3788 expand_no_split_dollar_star = 1;
3789 value = expand_string_internal (string, quoted);
3790 expand_no_split_dollar_star = 0;
3791
3792 if (value)
3793 {
3794 if (value->word)
3795 {
3796 remove_quoted_nulls (value->word->word); /* XXX */
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 */
3805 WORD_LIST *
3806 expand_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
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
3826 td.flags = W_ASSIGNRHS;
3827 #endif
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 {
3838 remove_quoted_nulls (value->word->word); /* XXX */
3839 value->word->flags &= ~W_HASQUOTEDNULL;
3840 }
3841 dequote_list (value);
3842 }
3843 return (value);
3844 }
3845
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. */
3851 WORD_LIST *
3852 expand_prompt_string (string, quoted, wflags)
3853 char *string;
3854 int quoted;
3855 int wflags;
3856 {
3857 WORD_LIST *value;
3858 WORD_DESC td;
3859
3860 if (string == 0 || *string == 0)
3861 return ((WORD_LIST *)NULL);
3862
3863 td.flags = wflags;
3864 td.word = savestring (string);
3865
3866 no_longjmp_on_fatal_error = 1;
3867 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3868 no_longjmp_on_fatal_error = 0;
3869
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)
3879 {
3880 remove_quoted_nulls (value->word->word); /* XXX */
3881 value->word->flags &= ~W_HASQUOTEDNULL;
3882 }
3883 dequote_list (value);
3884 }
3885 return (value);
3886 }
3887
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
3892 substitution, arithmetic expansion, and word splitting. */
3893 static WORD_LIST *
3894 expand_string_leave_quoted (string, quoted)
3895 char *string;
3896 int quoted;
3897 {
3898 WORD_LIST *tlist;
3899 WORD_LIST *tresult;
3900
3901 if (string == 0 || *string == '\0')
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
3915 /* This does not perform word splitting or dequote the WORD_LIST
3916 it returns. */
3917 static WORD_LIST *
3918 expand_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
3937 note this to callees.
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 */
3945 td.flags = W_EXPANDRHS; /* expanding RHS of ${paramOPword} */
3946 td.flags |= W_NOSPLIT2; /* no splitting, remove "" and '' */
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. */
3964 static WORD_LIST *
3965 expand_string_for_pat (string, quoted, dollar_at_p, expanded_p)
3966 char *string;
3967 int quoted, *dollar_at_p, *expanded_p;
3968 {
3969 WORD_DESC td;
3970 WORD_LIST *tresult;
3971 int oexp;
3972
3973 if (string == 0 || *string == '\0')
3974 return (WORD_LIST *)NULL;
3975
3976 oexp = expand_no_split_dollar_star;
3977 expand_no_split_dollar_star = 1;
3978 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
3979 td.word = string;
3980 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, expanded_p);
3981 expand_no_split_dollar_star = oexp;
3982
3983 return (tresult);
3984 }
3985
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. */
3991 WORD_LIST *
3992 expand_string (string, quoted)
3993 char *string;
3994 int quoted;
3995 {
3996 WORD_LIST *result;
3997
3998 if (string == 0 || *string == '\0')
3999 return ((WORD_LIST *)NULL);
4000
4001 result = expand_string_leave_quoted (string, quoted);
4002 return (result ? dequote_list (result) : result);
4003 }
4004
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
4015 WORD_LIST *
4016 expand_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. */
4031 WORD_LIST *
4032 expand_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. */
4045 WORD_LIST *
4046 expand_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
4062 /***************************************************
4063 * *
4064 * Functions to handle quoting chars *
4065 * *
4066 ***************************************************/
4067
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
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
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
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. */
4086 static char *
4087 quote_escapes_internal (string, flags)
4088 const char *string;
4089 int flags;
4090 {
4091 const char *s, *send;
4092 char *t, *result;
4093 size_t slen;
4094 int quote_spaces, skip_ctlesc, skip_ctlnul, nosplit;
4095 DECLARE_MBSTATE;
4096
4097 slen = strlen (string);
4098 send = string + slen;
4099
4100 quote_spaces = (ifs_value && *ifs_value == 0);
4101 nosplit = (flags & PF_NOSPLIT2);
4102
4103 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
4104 {
4105 skip_ctlesc |= (nosplit == 0 && *s == CTLESC);
4106 skip_ctlnul |= (nosplit == 0 && *s == CTLNUL);
4107 }
4108
4109 t = result = (char *)xmalloc ((slen * 2) + 1);
4110 s = string;
4111
4112 while (*s)
4113 {
4114 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
4115 *t++ = CTLESC;
4116 COPY_CHAR_P (t, s, send);
4117 }
4118 *t = '\0';
4119
4120 return (result);
4121 }
4122
4123 char *
4124 quote_escapes (string)
4125 const char *string;
4126 {
4127 return (quote_escapes_internal (string, 0));
4128 }
4129
4130 char *
4131 quote_rhs (string)
4132 const char *string;
4133 {
4134 return (quote_escapes_internal (string, PF_NOSPLIT2));
4135 }
4136
4137 static WORD_LIST *
4138 list_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
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. */
4164 char *
4165 dequote_escapes (string)
4166 const char *string;
4167 {
4168 const char *s, *send;
4169 char *t, *result;
4170 size_t slen;
4171 int quote_spaces;
4172 DECLARE_MBSTATE;
4173
4174 if (string == 0)
4175 return (char *)0;
4176
4177 slen = strlen (string);
4178 send = string + slen;
4179
4180 t = result = (char *)xmalloc (slen + 1);
4181
4182 if (strchr (string, CTLESC) == 0)
4183 return (strcpy (result, string));
4184
4185 quote_spaces = (ifs_value && *ifs_value == 0);
4186
4187 s = string;
4188 while (*s)
4189 {
4190 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
4191 {
4192 s++;
4193 if (*s == '\0')
4194 break;
4195 }
4196 COPY_CHAR_P (t, s, send);
4197 }
4198 *t = '\0';
4199
4200 return result;
4201 }
4202
4203 #if defined (INCLUDE_UNUSED)
4204 static WORD_LIST *
4205 list_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 }
4219 #endif
4220
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. */
4224 static char *
4225 make_quoted_char (c)
4226 int c;
4227 {
4228 char *temp;
4229
4230 temp = (char *)xmalloc (3);
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
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. */
4248 char *
4249 quote_string (string)
4250 char *string;
4251 {
4252 register char *t;
4253 size_t slen;
4254 char *result, *send;
4255
4256 if (*string == 0)
4257 {
4258 result = (char *)xmalloc (2);
4259 result[0] = CTLNUL;
4260 result[1] = '\0';
4261 }
4262 else
4263 {
4264 DECLARE_MBSTATE;
4265
4266 slen = strlen (string);
4267 send = string + slen;
4268
4269 result = (char *)xmalloc ((slen * 2) + 1);
4270
4271 for (t = result; string < send; )
4272 {
4273 *t++ = CTLESC;
4274 COPY_CHAR_P (t, string, send);
4275 }
4276 *t = '\0';
4277 }
4278 return (result);
4279 }
4280
4281 /* De-quote quoted characters in STRING. */
4282 char *
4283 dequote_string (string)
4284 char *string;
4285 {
4286 register char *s, *t;
4287 size_t slen;
4288 char *result, *send;
4289 DECLARE_MBSTATE;
4290
4291 #if defined (DEBUG)
4292 if (string[0] == CTLESC && string[1] == 0)
4293 internal_inform ("dequote_string: string with bare CTLESC");
4294 #endif
4295
4296 slen = STRLEN (string);
4297
4298 t = result = (char *)xmalloc (slen + 1);
4299
4300 if (QUOTED_NULL (string))
4301 {
4302 result[0] = '\0';
4303 return (result);
4304 }
4305
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
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. */
4316 if (strchr (string, CTLESC) == NULL)
4317 return (strcpy (result, string));
4318
4319 send = string + slen;
4320 s = string;
4321 while (*s)
4322 {
4323 if (*s == CTLESC)
4324 {
4325 s++;
4326 if (*s == '\0')
4327 break;
4328 }
4329 COPY_CHAR_P (t, s, send);
4330 }
4331
4332 *t = '\0';
4333 return (result);
4334 }
4335
4336 /* Quote the entire WORD_LIST list. */
4337 static WORD_LIST *
4338 quote_list (list)
4339 WORD_LIST *list;
4340 {
4341 register WORD_LIST *w;
4342 char *t;
4343
4344 for (w = list; w; w = w->next)
4345 {
4346 t = w->word->word;
4347 w->word->word = quote_string (t);
4348 if (*t == 0)
4349 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
4350 w->word->flags |= W_QUOTED;
4351 free (t);
4352 }
4353 return list;
4354 }
4355
4356 WORD_DESC *
4357 dequote_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
4371 /* De-quote quoted characters in each word in LIST. */
4372 WORD_LIST *
4373 dequote_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);
4382 if (QUOTED_NULL (tlist->word->word))
4383 tlist->word->flags &= ~W_HASQUOTEDNULL;
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. */
4392 char *
4393 remove_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
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(). */
4412 char *
4413 remove_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
4445 char *
4446 remove_quoted_nulls (string)
4447 char *string;
4448 {
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)
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. */
4467 i++;
4468 string[j++] = CTLESC;
4469 if (i == slen)
4470 break;
4471 }
4472 else if (string[i] == CTLNUL)
4473 {
4474 i++;
4475 continue;
4476 }
4477
4478 prev_i = i;
4479 ADVANCE_CHAR (string, slen, i); /* COPY_CHAR_I? */
4480 if (j < prev_i)
4481 {
4482 do string[j++] = string[prev_i++]; while (prev_i < i);
4483 }
4484 else
4485 j = i;
4486 }
4487 string[j] = '\0';
4488
4489 return (string);
4490 }
4491
4492 /* Perform quoted null character removal on each element of LIST.
4493 This modifies LIST. */
4494 void
4495 word_list_remove_quoted_nulls (list)
4496 WORD_LIST *list;
4497 {
4498 register WORD_LIST *t;
4499
4500 for (t = list; t; t = t->next)
4501 {
4502 remove_quoted_nulls (t->word->word);
4503 t->word->flags &= ~W_HASQUOTEDNULL;
4504 }
4505 }
4506
4507 /* **************************************************************** */
4508 /* */
4509 /* Functions for Matching and Removing Patterns */
4510 /* */
4511 /* **************************************************************** */
4512
4513 #if defined (HANDLE_MULTIBYTE)
4514 # ifdef INCLUDE_UNUSED
4515 static unsigned char *
4516 mb_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 }
4537 # endif
4538 #endif
4539
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
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
4553 /* Returns its first argument if nothing matched; new memory otherwise */
4554 static char *
4555 remove_upattern (param, pattern, op)
4556 char *param, *pattern;
4557 int op;
4558 {
4559 register size_t len;
4560 register char *end;
4561 register char *p, *ret, c;
4562
4563 len = STRLEN (param);
4564 end = param + len;
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';
4572 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4573 {
4574 *p = c;
4575 return (savestring (p));
4576 }
4577 *p = c;
4578
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';
4586 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4587 {
4588 *p = c;
4589 return (savestring (p));
4590 }
4591 *p = c;
4592 }
4593 break;
4594
4595 case RP_LONG_RIGHT: /* remove longest match at end */
4596 for (p = param; p <= end; p++)
4597 {
4598 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
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 {
4611 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4612 {
4613 c = *p; *p = '\0';
4614 ret = savestring (param);
4615 *p = c;
4616 return (ret);
4617 }
4618 }
4619 break;
4620 }
4621
4622 return (param); /* no match, return original string */
4623 }
4624
4625 #if defined (HANDLE_MULTIBYTE)
4626 /* Returns its first argument if nothing matched; new memory otherwise */
4627 static wchar_t *
4628 remove_wpattern (wparam, wstrlen, wpattern, op)
4629 wchar_t *wparam;
4630 size_t wstrlen;
4631 wchar_t *wpattern;
4632 int op;
4633 {
4634 wchar_t wc, *ret;
4635 int n;
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
4692 return (wparam); /* no match, return original string */
4693 }
4694 #endif /* HANDLE_MULTIBYTE */
4695
4696 static char *
4697 remove_pattern (param, pattern, op)
4698 char *param, *pattern;
4699 int op;
4700 {
4701 char *xret;
4702
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;
4715
4716 /* XXX - could optimize here by checking param and pattern for multibyte
4717 chars with mbsmbchar and calling remove_upattern. */
4718
4719 n = xdupmbstowcs (&wpattern, NULL, pattern);
4720 if (n == (size_t)-1)
4721 {
4722 xret = remove_upattern (param, pattern, op);
4723 return ((xret == param) ? savestring (param) : xret);
4724 }
4725 n = xdupmbstowcs (&wparam, NULL, param);
4726
4727 if (n == (size_t)-1)
4728 {
4729 free (wpattern);
4730 xret = remove_upattern (param, pattern, op);
4731 return ((xret == param) ? savestring (param) : xret);
4732 }
4733 oret = ret = remove_wpattern (wparam, n, wpattern, op);
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 }
4742
4743 free (wparam);
4744 free (wpattern);
4745
4746 n = strlen (param);
4747 xret = (char *)xmalloc (n + 1);
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
4756 {
4757 xret = remove_upattern (param, pattern, op);
4758 return ((xret == param) ? savestring (param) : xret);
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. */
4768 static int
4769 match_upattern (string, pat, mtype, sp, ep)
4770 char *string, *pat;
4771 int mtype;
4772 char **sp, **ep;
4773 {
4774 int c, mlen;
4775 size_t len;
4776 register char *p, *p1, *npat;
4777 char *end;
4778
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);
4788 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
4789 {
4790 int unescaped_backslash;
4791 char *pp;
4792
4793 p = npat = (char *)xmalloc (len + 3);
4794 p1 = pat;
4795 if ((mtype != MATCH_BEG) && (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob)))
4796 *p++ = '*';
4797 while (*p1)
4798 *p++ = *p1++;
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 `*'. */
4806 if ((mtype != MATCH_END) && (p1[-1] == '*' && (unescaped_backslash = p1[-2] == '\\')))
4807 {
4808 pp = p1 - 3;
4809 while (pp >= pat && *pp-- == '\\')
4810 unescaped_backslash = 1 - unescaped_backslash;
4811 if (unescaped_backslash)
4812 *p++ = '*';
4813 }
4814 else if (mtype != MATCH_END && p1[-1] != '*')
4815 *p++ = '*';
4816 #else
4817 if (p1[-1] != '*' || p1[-2] == '\\')
4818 *p++ = '*';
4819 #endif
4820 *p = '\0';
4821 }
4822 else
4823 npat = pat;
4824 c = strmatch (npat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
4825 if (npat != pat)
4826 free (npat);
4827 if (c == FNM_NOMATCH)
4828 return (0);
4829
4830 len = STRLEN (string);
4831 end = string + len;
4832
4833 mlen = umatchlen (pat, len);
4834 if (mlen > (int)len)
4835 return (0);
4836
4837 switch (mtype)
4838 {
4839 case MATCH_ANY:
4840 for (p = string; p <= end; p++)
4841 {
4842 if (match_pattern_char (pat, p, FNMATCH_IGNCASE))
4843 {
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--)
4856 {
4857 c = *p1; *p1 = '\0';
4858 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4859 {
4860 *p1 = c;
4861 *sp = p;
4862 *ep = p1;
4863 return 1;
4864 }
4865 *p1 = c;
4866 #if 1
4867 /* If MLEN != -1, we have a fixed length pattern. */
4868 if (mlen != -1)
4869 break;
4870 #endif
4871 }
4872 }
4873 }
4874
4875 return (0);
4876
4877 case MATCH_BEG:
4878 if (match_pattern_char (pat, string, FNMATCH_IGNCASE) == 0)
4879 return (0);
4880
4881 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
4882 {
4883 c = *p; *p = '\0';
4884 if (strmatch (pat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4885 {
4886 *p = c;
4887 *sp = string;
4888 *ep = p;
4889 return 1;
4890 }
4891 *p = c;
4892 /* If MLEN != -1, we have a fixed length pattern. */
4893 if (mlen != -1)
4894 break;
4895 }
4896
4897 return (0);
4898
4899 case MATCH_END:
4900 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
4901 {
4902 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4903 {
4904 *sp = p;
4905 *ep = end;
4906 return 1;
4907 }
4908 /* If MLEN != -1, we have a fixed length pattern. */
4909 if (mlen != -1)
4910 break;
4911 }
4912
4913 return (0);
4914 }
4915
4916 return (0);
4917 }
4918
4919 #if defined (HANDLE_MULTIBYTE)
4920
4921 #define WFOLD(c) (match_ignore_case && iswupper (c) ? towlower (c) : (c))
4922
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. */
4926 static int
4927 match_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 {
4935 wchar_t wc, *wp, *nwpat, *wp1;
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)
4943 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
4944 #endif
4945
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. */
4952 len = wcslen (wpat);
4953 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
4954 {
4955 int unescaped_backslash;
4956 wchar_t *wpp;
4957
4958 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
4959 wp1 = wpat;
4960 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
4961 *wp++ = L'*';
4962 while (*wp1 != L'\0')
4963 *wp++ = *wp1++;
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
4977 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4978 *wp++ = L'*';
4979 #endif
4980 *wp = '\0';
4981 }
4982 else
4983 nwpat = wpat;
4984 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
4985 if (nwpat != wpat)
4986 free (nwpat);
4987 if (len == FNM_NOMATCH)
4988 return (0);
4989
4990 mlen = wmatchlen (wpat, wstrlen);
4991 if (mlen > (int)wstrlen)
4992 return (0);
4993
4994 /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
4995 switch (mtype)
4996 {
4997 case MATCH_ANY:
4998 for (n = 0; n <= wstrlen; n++)
4999 {
5000 n2 = simple ? (WFOLD(*wpat) == WFOLD(wstring[n])) : match_pattern_wchar (wpat, wstring + n, FNMATCH_IGNCASE);
5001 if (n2)
5002 {
5003 n1 = (mlen == -1) ? wstrlen : n + mlen;
5004 if (n1 > wstrlen)
5005 break;
5006
5007 for ( ; n1 >= n; n1--)
5008 {
5009 wc = wstring[n1]; wstring[n1] = L'\0';
5010 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
5011 {
5012 wstring[n1] = wc;
5013 *sp = indices[n];
5014 *ep = indices[n1];
5015 return 1;
5016 }
5017 wstring[n1] = wc;
5018 /* If MLEN != -1, we have a fixed length pattern. */
5019 if (mlen != -1)
5020 break;
5021 }
5022 }
5023 }
5024
5025 return (0);
5026
5027 case MATCH_BEG:
5028 if (match_pattern_wchar (wpat, wstring, FNMATCH_IGNCASE) == 0)
5029 return (0);
5030
5031 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
5032 {
5033 wc = wstring[n]; wstring[n] = L'\0';
5034 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
5035 {
5036 wstring[n] = wc;
5037 *sp = indices[0];
5038 *ep = indices[n];
5039 return 1;
5040 }
5041 wstring[n] = wc;
5042 /* If MLEN != -1, we have a fixed length pattern. */
5043 if (mlen != -1)
5044 break;
5045 }
5046
5047 return (0);
5048
5049 case MATCH_END:
5050 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
5051 {
5052 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
5053 {
5054 *sp = indices[n];
5055 *ep = indices[wstrlen];
5056 return 1;
5057 }
5058 /* If MLEN != -1, we have a fixed length pattern. */
5059 if (mlen != -1)
5060 break;
5061 }
5062
5063 return (0);
5064 }
5065
5066 return (0);
5067 }
5068 #undef WFOLD
5069 #endif /* HANDLE_MULTIBYTE */
5070
5071 static int
5072 match_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
5084 if (string == 0 || pat == 0 || *pat == 0)
5085 return (0);
5086
5087 #if defined (HANDLE_MULTIBYTE)
5088 if (MB_CUR_MAX > 1)
5089 {
5090 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
5091 return (match_upattern (string, pat, mtype, sp, ep));
5092
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 }
5114
5115 static int
5116 getpatspec (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
5129 to be called to fix it up for strmatch (). If QUOTED is non-zero,
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#'*'}". */
5134 static char *
5135 getpattern (value, quoted, expandpat)
5136 char *value;
5137 int quoted, expandpat;
5138 {
5139 char *pat, *tword;
5140 WORD_LIST *l;
5141 #if 0
5142 int i;
5143 #endif
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
5150 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
5151 {
5152 i = 0;
5153 pat = string_extract_double_quoted (tword, &i, SX_STRIPDQ);
5154 free (tword);
5155 tword = pat;
5156 }
5157 #endif
5158
5159 /* expand_string_for_pat () leaves WORD quoted and does not perform
5160 word splitting. */
5161 l = *value ? expand_string_for_pat (value,
5162 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
5163 (int *)NULL, (int *)NULL)
5164 : (WORD_LIST *)0;
5165 if (l)
5166 word_list_remove_quoted_nulls (l);
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
5178 #if 0
5179 /* Handle removing a pattern from a string as a result of ${name%[%]value}
5180 or ${name#[#]value}. */
5181 static char *
5182 variable_remove_pattern (value, pattern, patspec, quoted)
5183 char *value, *pattern;
5184 int patspec, quoted;
5185 {
5186 char *tword;
5187
5188 tword = remove_pattern (value, pattern, patspec);
5189
5190 return (tword);
5191 }
5192 #endif
5193
5194 static char *
5195 list_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 *);
5213 tword = string_list_pos_params (itype, l, quoted, 0);
5214 dispose_words (l);
5215
5216 return (tword);
5217 }
5218
5219 static char *
5220 parameter_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)
5237 static char *
5238 array_remove_pattern (var, pattern, patspec, starsub, quoted)
5239 SHELL_VAR *var;
5240 char *pattern;
5241 int patspec;
5242 int starsub; /* so we can figure out how it's indexed */
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
5252 v = var; /* XXX - for now */
5253
5254 itype = starsub ? '*' : '@';
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
5269 static char *
5270 parameter_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;
5277 char *temp1, *val, *pattern, *oname;
5278 SHELL_VAR *v;
5279
5280 if (value == 0)
5281 return ((char *)NULL);
5282
5283 oname = this_command_name;
5284 this_command_name = varname;
5285
5286 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
5287 if (vtype == -1)
5288 {
5289 this_command_name = oname;
5290 return ((char *)NULL);
5291 }
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:
5325 temp1 = array_remove_pattern (v, pattern, patspec, starsub, quoted);
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);
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))
5341 {
5342 val = quote_escapes (temp1);
5343 free (temp1);
5344 temp1 = val;
5345 }
5346 break;
5347 }
5348
5349 this_command_name = oname;
5350
5351 FREE (pattern);
5352 return temp1;
5353 }
5354
5355 #if defined (PROCESS_SUBSTITUTION)
5356
5357 static void reap_some_procsubs PARAMS((int));
5358
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
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. */
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. */
5377 struct temp_fifo {
5378 char *file;
5379 pid_t proc;
5380 };
5381
5382 static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
5383 static int nfifo;
5384 static int fifo_list_size;
5385
5386 void
5387 clear_fifo_list ()
5388 {
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;
5399 }
5400
5401 void *
5402 copy_fifo_list (sizep)
5403 int *sizep;
5404 {
5405 if (sizep)
5406 *sizep = 0;
5407 return (void *)NULL;
5408 }
5409
5410 static void
5411 add_fifo_list (pathname)
5412 char *pathname;
5413 {
5414 int osize, i;
5415
5416 if (nfifo >= fifo_list_size - 1)
5417 {
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 }
5427 }
5428
5429 fifo_list[nfifo].file = savestring (pathname);
5430 nfifo++;
5431 }
5432
5433 void
5434 unlink_fifo (i)
5435 int i;
5436 {
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 }
5444 }
5445
5446 void
5447 unlink_fifo_list ()
5448 {
5449 int saved, i, j;
5450
5451 if (nfifo == 0)
5452 return;
5453
5454 for (i = saved = 0; i < nfifo; i++)
5455 {
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++;
5465 }
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 {
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 }
5480 j++;
5481 }
5482 nfifo = j;
5483 }
5484 else
5485 nfifo = 0;
5486 }
5487
5488 void
5489 unlink_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
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). */
5513 void
5514 close_new_fifos (list, lsize)
5515 void *list;
5516 int lsize;
5517 {
5518 int i;
5519 char *plist;
5520
5521 if (list == 0)
5522 {
5523 unlink_fifo_list ();
5524 return;
5525 }
5526
5527 for (plist = (char *)list, i = 0; i < lsize; i++)
5528 if (plist[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
5529 unlink_fifo (i);
5530
5531 for (i = lsize; i < fifo_list_size; i++)
5532 unlink_fifo (i);
5533 }
5534
5535 int
5536 find_procsub_child (pid)
5537 pid_t pid;
5538 {
5539 int i;
5540
5541 for (i = 0; i < nfifo; i++)
5542 if (fifo_list[i].proc == pid)
5543 return i;
5544 return -1;
5545 }
5546
5547 void
5548 set_procsub_status (ind, pid, status)
5549 int ind;
5550 pid_t pid;
5551 int status;
5552 {
5553 if (ind >= 0 && ind < nfifo)
5554 fifo_list[ind].proc = (pid_t)-1; /* sentinel */
5555 }
5556
5557 /* If we've marked the process for this procsub as dead, close the
5558 associated file descriptor and delete the FIFO. */
5559 static void
5560 reap_some_procsubs (max)
5561 int max;
5562 {
5563 int i;
5564
5565 for (i = 0; i < max; i++)
5566 if (fifo_list[i].proc == (pid_t)-1) /* reaped */
5567 unlink_fifo (i);
5568 }
5569
5570 void
5571 reap_procsubs ()
5572 {
5573 reap_some_procsubs (nfifo);
5574 }
5575
5576 #if 0
5577 /* UNUSED */
5578 void
5579 wait_procsubs ()
5580 {
5581 int i, r;
5582
5583 for (i = 0; i < nfifo; i++)
5584 {
5585 if (fifo_list[i].proc != (pid_t)-1 && fifo_list[i].proc > 0)
5586 {
5587 r = wait_for (fifo_list[i].proc, 0);
5588 save_proc_status (fifo_list[i].proc, r);
5589 fifo_list[i].proc = (pid_t)-1;
5590 }
5591 }
5592 }
5593 #endif
5594
5595 int
5596 fifos_pending ()
5597 {
5598 return nfifo;
5599 }
5600
5601 int
5602 num_fifos ()
5603 {
5604 return nfifo;
5605 }
5606
5607 static char *
5608 make_named_pipe ()
5609 {
5610 char *tname;
5611
5612 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
5613 if (mkfifo (tname, 0600) < 0)
5614 {
5615 free (tname);
5616 return ((char *)NULL);
5617 }
5618
5619 add_fifo_list (tname);
5620 return (tname);
5621 }
5622
5623 #else /* HAVE_DEV_FD */
5624
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. */
5631
5632 static pid_t *dev_fd_list = (pid_t *)NULL;
5633 static int nfds;
5634 static int totfds; /* The highest possible number of open files. */
5635
5636 void
5637 clear_fifo (i)
5638 int i;
5639 {
5640 if (dev_fd_list[i])
5641 {
5642 dev_fd_list[i] = 0;
5643 nfds--;
5644 }
5645 }
5646
5647 void
5648 clear_fifo_list ()
5649 {
5650 register int i;
5651
5652 if (nfds == 0)
5653 return;
5654
5655 for (i = 0; nfds && i < totfds; i++)
5656 clear_fifo (i);
5657
5658 nfds = 0;
5659 }
5660
5661 void *
5662 copy_fifo_list (sizep)
5663 int *sizep;
5664 {
5665 void *ret;
5666
5667 if (nfds == 0 || totfds == 0)
5668 {
5669 if (sizep)
5670 *sizep = 0;
5671 return (void *)NULL;
5672 }
5673
5674 if (sizep)
5675 *sizep = totfds;
5676 ret = xmalloc (totfds * sizeof (pid_t));
5677 return (memcpy (ret, dev_fd_list, totfds * sizeof (pid_t)));
5678 }
5679
5680 static void
5681 add_fifo_list (fd)
5682 int fd;
5683 {
5684 if (dev_fd_list == 0 || fd >= totfds)
5685 {
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));
5698 }
5699
5700 dev_fd_list[fd] = 1; /* marker; updated later */
5701 nfds++;
5702 }
5703
5704 int
5705 fifos_pending ()
5706 {
5707 return 0; /* used for cleanup; not needed with /dev/fd */
5708 }
5709
5710 int
5711 num_fifos ()
5712 {
5713 return nfds;
5714 }
5715
5716 void
5717 unlink_fifo (fd)
5718 int fd;
5719 {
5720 if (dev_fd_list[fd])
5721 {
5722 close (fd);
5723 dev_fd_list[fd] = 0;
5724 nfds--;
5725 }
5726 }
5727
5728 void
5729 unlink_fifo_list ()
5730 {
5731 register int i;
5732
5733 if (nfds == 0)
5734 return;
5735
5736 for (i = totfds-1; nfds && i >= 0; i--)
5737 unlink_fifo (i);
5738
5739 nfds = 0;
5740 }
5741
5742 void
5743 unlink_all_fifos ()
5744 {
5745 unlink_fifo_list ();
5746 }
5747
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). */
5753 void
5754 close_new_fifos (list, lsize)
5755 void *list;
5756 int lsize;
5757 {
5758 int i;
5759 pid_t *plist;
5760
5761 if (list == 0)
5762 {
5763 unlink_fifo_list ();
5764 return;
5765 }
5766
5767 for (plist = (pid_t *)list, i = 0; i < lsize; i++)
5768 if (plist[i] == 0 && i < totfds && dev_fd_list[i])
5769 unlink_fifo (i);
5770
5771 for (i = lsize; i < totfds; i++)
5772 unlink_fifo (i);
5773 }
5774
5775 int
5776 find_procsub_child (pid)
5777 pid_t pid;
5778 {
5779 int i;
5780
5781 if (nfds == 0)
5782 return -1;
5783
5784 for (i = 0; i < totfds; i++)
5785 if (dev_fd_list[i] == pid)
5786 return i;
5787
5788 return -1;
5789 }
5790
5791 void
5792 set_procsub_status (ind, pid, status)
5793 int ind;
5794 pid_t pid;
5795 int status;
5796 {
5797 if (ind >= 0 && ind < totfds)
5798 dev_fd_list[ind] = (pid_t)-1; /* sentinel */
5799 }
5800
5801 /* If we've marked the process for this procsub as dead, close the
5802 associated file descriptor. */
5803 static void
5804 reap_some_procsubs (max)
5805 int max;
5806 {
5807 int i;
5808
5809 for (i = 0; nfds > 0 && i < max; i++)
5810 if (dev_fd_list[i] == (pid_t)-1)
5811 unlink_fifo (i);
5812 }
5813
5814 void
5815 reap_procsubs ()
5816 {
5817 reap_some_procsubs (totfds);
5818 }
5819
5820 #if 0
5821 /* UNUSED */
5822 void
5823 wait_procsubs ()
5824 {
5825 int i, r;
5826
5827 for (i = 0; nfds > 0 && i < totfds; i++)
5828 {
5829 if (dev_fd_list[i] != (pid_t)-1 && dev_fd_list[i] > 0)
5830 {
5831 r = wait_for (dev_fd_list[i], 0);
5832 save_proc_status (dev_fd_list[i], r);
5833 dev_fd_list[i] = (pid_t)-1;
5834 }
5835 }
5836 }
5837 #endif
5838
5839 #if defined (NOTDEF)
5840 print_dev_fd_list ()
5841 {
5842 register int i;
5843
5844 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
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
5856 static char *
5857 make_dev_fd_filename (fd)
5858 int fd;
5859 {
5860 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
5861
5862 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
5863
5864 strcpy (ret, DEV_FD_PREFIX);
5865 p = inttostr (fd, intbuf, sizeof (intbuf));
5866 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5867
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
5886 static char *
5887 process_substitute (string, open_for_read_in_child)
5888 char *string;
5889 int open_for_read_in_child;
5890 {
5891 char *pathname;
5892 int fd, result, rc, function_value;
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;
5900 #endif
5901
5902 if (!string || !*string || wordexp_only)
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 {
5910 sys_error ("%s", _("cannot make pipe for process substitution"));
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];
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
5921 pathname = make_dev_fd_filename (parent_pipe_fd);
5922 #endif /* HAVE_DEV_FD */
5923
5924 if (pathname == 0)
5925 {
5926 sys_error ("%s", _("cannot make pipe for process substitution"));
5927 return ((char *)NULL);
5928 }
5929
5930 old_pid = last_made_pid;
5931
5932 #if defined (JOB_CONTROL)
5933 old_pipeline_pgrp = pipeline_pgrp;
5934 if (pipeline_pgrp == 0 || (subshell_environment & (SUBSHELL_PIPE|SUBSHELL_FORK|SUBSHELL_ASYNC)) == 0)
5935 pipeline_pgrp = shell_pgrp;
5936 save_pipeline (1);
5937 #endif /* JOB_CONTROL */
5938
5939 pid = make_child ((char *)NULL, FORK_ASYNC);
5940 if (pid == 0)
5941 {
5942 #if 0
5943 int old_interactive;
5944
5945 old_interactive = interactive;
5946 #endif
5947 /* The currently-executing shell is not interactive */
5948 interactive = 0;
5949
5950 reset_terminating_signals (); /* XXX */
5951 free_pushed_string_input ();
5952 /* Cancel traps, in trap.c. */
5953 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
5954 QUIT; /* catch any interrupts we got post-fork */
5955 setup_async_signals ();
5956 #if 0
5957 if (open_for_read_in_child == 0 && old_interactive && (bash_input.type == st_stdin || bash_input.type == st_stream))
5958 async_redirect_stdin ();
5959 #endif
5960
5961 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB|SUBSHELL_ASYNC;
5962
5963 /* We don't inherit the verbose option for command substitutions now, so
5964 let's try it for process substitutions. */
5965 change_flag ('v', FLAG_OFF);
5966
5967 /* if we're expanding a redirection, we shouldn't have access to the
5968 temporary environment, but commands in the subshell should have
5969 access to their own temporary environment. */
5970 if (expanding_redir)
5971 flush_temporary_env ();
5972 }
5973
5974 #if defined (JOB_CONTROL)
5975 set_sigchld_handler ();
5976 stop_making_children ();
5977 /* XXX - should we only do this in the parent? (as in command subst) */
5978 pipeline_pgrp = old_pipeline_pgrp;
5979 #else
5980 stop_making_children ();
5981 #endif /* JOB_CONTROL */
5982
5983 if (pid < 0)
5984 {
5985 sys_error ("%s", _("cannot make child for process substitution"));
5986 free (pathname);
5987 #if defined (HAVE_DEV_FD)
5988 close (parent_pipe_fd);
5989 close (child_pipe_fd);
5990 #endif /* HAVE_DEV_FD */
5991 #if defined (JOB_CONTROL)
5992 restore_pipeline (1);
5993 #endif
5994 return ((char *)NULL);
5995 }
5996
5997 if (pid > 0)
5998 {
5999 #if defined (JOB_CONTROL)
6000 last_procsub_child = restore_pipeline (0);
6001 /* We assume that last_procsub_child->next == last_procsub_child because
6002 of how jobs.c:add_process() works. */
6003 last_procsub_child->next = 0;
6004 procsub_add (last_procsub_child);
6005 #endif
6006
6007 #if defined (HAVE_DEV_FD)
6008 dev_fd_list[parent_pipe_fd] = pid;
6009 #else
6010 fifo_list[nfifo-1].proc = pid;
6011 #endif
6012
6013 last_made_pid = old_pid;
6014
6015 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6016 close_pgrp_pipe ();
6017 #endif /* JOB_CONTROL && PGRP_PIPE */
6018
6019 #if defined (HAVE_DEV_FD)
6020 close (child_pipe_fd);
6021 #endif /* HAVE_DEV_FD */
6022
6023 return (pathname);
6024 }
6025
6026 set_sigint_handler ();
6027
6028 #if defined (JOB_CONTROL)
6029 /* make sure we don't have any job control */
6030 set_job_control (0);
6031
6032 /* Clear out any existing list of process substitutions */
6033 procsub_clear ();
6034
6035 /* The idea is that we want all the jobs we start from an async process
6036 substitution to be in the same process group, but not the same pgrp
6037 as our parent shell, since we don't want to affect our parent shell's
6038 jobs if we get a SIGHUP and end up calling hangup_all_jobs, for example.
6039 If pipeline_pgrp != shell_pgrp, we assume that there is a job control
6040 shell somewhere in our parent process chain (since make_child initializes
6041 pipeline_pgrp to shell_pgrp if job_control == 0). What we do in this
6042 case is to set pipeline_pgrp to our PID, so all jobs started by this
6043 process have that same pgrp and we are basically the process group leader.
6044 This should not have negative effects on child processes surviving
6045 after we exit, since we wait for the children we create, but that is
6046 something to watch for. */
6047
6048 if (pipeline_pgrp != shell_pgrp)
6049 pipeline_pgrp = getpid ();
6050 #endif /* JOB_CONTROL */
6051
6052 #if !defined (HAVE_DEV_FD)
6053 /* Open the named pipe in the child. */
6054 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
6055 if (fd < 0)
6056 {
6057 /* Two separate strings for ease of translation. */
6058 if (open_for_read_in_child)
6059 sys_error (_("cannot open named pipe %s for reading"), pathname);
6060 else
6061 sys_error (_("cannot open named pipe %s for writing"), pathname);
6062
6063 exit (127);
6064 }
6065 if (open_for_read_in_child)
6066 {
6067 if (sh_unset_nodelay_mode (fd) < 0)
6068 {
6069 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
6070 exit (127);
6071 }
6072 }
6073 #else /* HAVE_DEV_FD */
6074 fd = child_pipe_fd;
6075 #endif /* HAVE_DEV_FD */
6076
6077 /* Discard buffered stdio output before replacing the underlying file
6078 descriptor. */
6079 if (open_for_read_in_child == 0)
6080 fpurge (stdout);
6081
6082 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
6083 {
6084 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
6085 open_for_read_in_child ? 0 : 1);
6086 exit (127);
6087 }
6088
6089 if (fd != (open_for_read_in_child ? 0 : 1))
6090 close (fd);
6091
6092 /* Need to close any files that this process has open to pipes inherited
6093 from its parent. */
6094 if (current_fds_to_close)
6095 {
6096 close_fd_bitmap (current_fds_to_close);
6097 current_fds_to_close = (struct fd_bitmap *)NULL;
6098 }
6099
6100 #if defined (HAVE_DEV_FD)
6101 /* Make sure we close the parent's end of the pipe and clear the slot
6102 in the fd list so it is not closed later, if reallocated by, for
6103 instance, pipe(2). */
6104 close (parent_pipe_fd);
6105 dev_fd_list[parent_pipe_fd] = 0;
6106 #endif /* HAVE_DEV_FD */
6107
6108 /* subshells shouldn't have this flag, which controls using the temporary
6109 environment for variable lookups. We have already flushed the temporary
6110 environment above in the case we're expanding a redirection, so processes
6111 executed by this command need to be able to set it independently of their
6112 parent. */
6113 expanding_redir = 0;
6114
6115 remove_quoted_escapes (string);
6116
6117 #if 0 /* TAG: bash-5.2 */
6118 startup_state = 2; /* see if we can avoid a fork */
6119 parse_and_execute_level = 0;
6120 #endif
6121
6122 /* Give process substitution a place to jump back to on failure,
6123 so we don't go back up to main (). */
6124 result = setjmp_nosigs (top_level);
6125
6126 /* If we're running a process substitution inside a shell function,
6127 trap `return' so we don't return from the function in the subshell
6128 and go off to never-never land. */
6129 if (result == 0 && return_catch_flag)
6130 function_value = setjmp_nosigs (return_catch);
6131 else
6132 function_value = 0;
6133
6134 if (result == ERREXIT)
6135 rc = last_command_exit_value;
6136 else if (result == EXITPROG)
6137 rc = last_command_exit_value;
6138 else if (result)
6139 rc = EXECUTION_FAILURE;
6140 else if (function_value)
6141 rc = return_catch_value;
6142 else
6143 {
6144 subshell_level++;
6145 rc = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
6146 /* leave subshell level intact for any exit trap */
6147 }
6148
6149 #if !defined (HAVE_DEV_FD)
6150 /* Make sure we close the named pipe in the child before we exit. */
6151 close (open_for_read_in_child ? 0 : 1);
6152 #endif /* !HAVE_DEV_FD */
6153
6154 last_command_exit_value = rc;
6155 rc = run_exit_trap ();
6156 exit (rc);
6157 /*NOTREACHED*/
6158 }
6159 #endif /* PROCESS_SUBSTITUTION */
6160
6161 /***********************************/
6162 /* */
6163 /* Command Substitution */
6164 /* */
6165 /***********************************/
6166
6167 static char *
6168 read_comsub (fd, quoted, flags, rflag)
6169 int fd, quoted, flags;
6170 int *rflag;
6171 {
6172 char *istring, buf[512], *bufp;
6173 int istring_index, c, tflag, skip_ctlesc, skip_ctlnul;
6174 int mb_cur_max;
6175 size_t istring_size;
6176 ssize_t bufn;
6177 int nullbyte;
6178 #if defined (HANDLE_MULTIBYTE)
6179 mbstate_t ps;
6180 wchar_t wc;
6181 size_t mblen;
6182 int i;
6183 #endif
6184
6185 istring = (char *)NULL;
6186 istring_index = istring_size = bufn = tflag = 0;
6187
6188 skip_ctlesc = ifs_cmap[CTLESC];
6189 skip_ctlnul = ifs_cmap[CTLNUL];
6190
6191 mb_cur_max = MB_CUR_MAX;
6192 nullbyte = 0;
6193
6194 /* Read the output of the command through the pipe. */
6195 while (1)
6196 {
6197 if (fd < 0)
6198 break;
6199 if (--bufn <= 0)
6200 {
6201 bufn = zread (fd, buf, sizeof (buf));
6202 if (bufn <= 0)
6203 break;
6204 bufp = buf;
6205 }
6206 c = *bufp++;
6207
6208 if (c == 0)
6209 {
6210 #if 1
6211 if (nullbyte == 0)
6212 {
6213 internal_warning ("%s", _("command substitution: ignored null byte in input"));
6214 nullbyte = 1;
6215 }
6216 #endif
6217 continue;
6218 }
6219
6220 /* Add the character to ISTRING, possibly after resizing it. */
6221 RESIZE_MALLOCED_BUFFER (istring, istring_index, mb_cur_max+1, istring_size, 512);
6222
6223 /* This is essentially quote_string inline */
6224 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
6225 istring[istring_index++] = CTLESC;
6226 else if ((flags & PF_ASSIGNRHS) && skip_ctlesc && c == CTLESC)
6227 istring[istring_index++] = CTLESC;
6228 /* Escape CTLESC and CTLNUL in the output to protect those characters
6229 from the rest of the word expansions (word splitting and globbing.)
6230 This is essentially quote_escapes inline. */
6231 else if (skip_ctlesc == 0 && c == CTLESC)
6232 istring[istring_index++] = CTLESC;
6233 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
6234 istring[istring_index++] = CTLESC;
6235
6236 #if defined (HANDLE_MULTIBYTE)
6237 if ((locale_utf8locale && (c & 0x80)) ||
6238 (locale_utf8locale == 0 && mb_cur_max > 1 && (unsigned char)c > 127))
6239 {
6240 /* read a multibyte character from buf */
6241 /* punt on the hard case for now */
6242 memset (&ps, '\0', sizeof (mbstate_t));
6243 mblen = mbrtowc (&wc, bufp-1, bufn+1, &ps);
6244 if (MB_INVALIDCH (mblen) || mblen == 0 || mblen == 1)
6245 istring[istring_index++] = c;
6246 else
6247 {
6248 istring[istring_index++] = c;
6249 for (i = 0; i < mblen-1; i++)
6250 istring[istring_index++] = *bufp++;
6251 bufn -= mblen - 1;
6252 }
6253 continue;
6254 }
6255 #endif
6256
6257 istring[istring_index++] = c;
6258 }
6259
6260 if (istring)
6261 istring[istring_index] = '\0';
6262
6263 /* If we read no output, just return now and save ourselves some
6264 trouble. */
6265 if (istring_index == 0)
6266 {
6267 FREE (istring);
6268 if (rflag)
6269 *rflag = tflag;
6270 return (char *)NULL;
6271 }
6272
6273 /* Strip trailing newlines from the output of the command. */
6274 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6275 {
6276 while (istring_index > 0)
6277 {
6278 if (istring[istring_index - 1] == '\n')
6279 {
6280 --istring_index;
6281
6282 /* If the newline was quoted, remove the quoting char. */
6283 if (istring[istring_index - 1] == CTLESC)
6284 --istring_index;
6285 }
6286 else
6287 break;
6288 }
6289 istring[istring_index] = '\0';
6290 }
6291 else
6292 strip_trailing (istring, istring_index - 1, 1);
6293
6294 if (rflag)
6295 *rflag = tflag;
6296 return istring;
6297 }
6298
6299 /* Perform command substitution on STRING. This returns a WORD_DESC * with the
6300 contained string possibly quoted. */
6301 WORD_DESC *
6302 command_substitute (string, quoted, flags)
6303 char *string;
6304 int quoted;
6305 int flags;
6306 {
6307 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
6308 char *istring, *s;
6309 int result, fildes[2], function_value, pflags, rc, tflag, fork_flags;
6310 WORD_DESC *ret;
6311 sigset_t set, oset;
6312
6313 istring = (char *)NULL;
6314
6315 /* Don't fork () if there is no need to. In the case of no command to
6316 run, just return NULL. */
6317 #if 1
6318 for (s = string; s && *s && (shellblank (*s) || *s == '\n'); s++)
6319 ;
6320 if (s == 0 || *s == 0)
6321 return ((WORD_DESC *)NULL);
6322 #else
6323 if (!string || !*string || (string[0] == '\n' && !string[1]))
6324 return ((WORD_DESC *)NULL);
6325 #endif
6326
6327 if (wordexp_only && read_but_dont_execute)
6328 {
6329 last_command_exit_value = EX_WEXPCOMSUB;
6330 jump_to_top_level (EXITPROG);
6331 }
6332
6333 /* We're making the assumption here that the command substitution will
6334 eventually run a command from the file system. Since we'll run
6335 maybe_make_export_env in this subshell before executing that command,
6336 the parent shell and any other shells it starts will have to remake
6337 the environment. If we make it before we fork, other shells won't
6338 have to. Don't bother if we have any temporary variable assignments,
6339 though, because the export environment will be remade after this
6340 command completes anyway, but do it if all the words to be expanded
6341 are variable assignments. */
6342 if (subst_assign_varlist == 0 || garglist == 0)
6343 maybe_make_export_env (); /* XXX */
6344
6345 /* Flags to pass to parse_and_execute() */
6346 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
6347
6348 old_pid = last_made_pid;
6349
6350 /* Pipe the output of executing STRING into the current shell. */
6351 if (pipe (fildes) < 0)
6352 {
6353 sys_error ("%s", _("cannot make pipe for command substitution"));
6354 goto error_exit;
6355 }
6356
6357 #if defined (JOB_CONTROL)
6358 old_pipeline_pgrp = pipeline_pgrp;
6359 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline or
6360 we've already forked to run a disk command (and are expanding redirections,
6361 for example). */
6362 if ((subshell_environment & (SUBSHELL_FORK|SUBSHELL_PIPE)) == 0)
6363 pipeline_pgrp = shell_pgrp;
6364 cleanup_the_pipeline ();
6365 #endif /* JOB_CONTROL */
6366
6367 old_async_pid = last_asynchronous_pid;
6368 fork_flags = (subshell_environment&SUBSHELL_ASYNC) ? FORK_ASYNC : 0;
6369 pid = make_child ((char *)NULL, fork_flags|FORK_NOTERM);
6370 last_asynchronous_pid = old_async_pid;
6371
6372 if (pid == 0)
6373 {
6374 /* Reset the signal handlers in the child, but don't free the
6375 trap strings. Set a flag noting that we have to free the
6376 trap strings if we run trap to change a signal disposition. */
6377 reset_signal_handlers ();
6378 if (ISINTERRUPT)
6379 {
6380 kill (getpid (), SIGINT);
6381 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6382 }
6383 QUIT; /* catch any interrupts we got post-fork */
6384 subshell_environment |= SUBSHELL_RESETTRAP;
6385 }
6386
6387 #if defined (JOB_CONTROL)
6388 /* XXX DO THIS ONLY IN PARENT ? XXX */
6389 set_sigchld_handler ();
6390 stop_making_children ();
6391 if (pid != 0)
6392 pipeline_pgrp = old_pipeline_pgrp;
6393 #else
6394 stop_making_children ();
6395 #endif /* JOB_CONTROL */
6396
6397 if (pid < 0)
6398 {
6399 sys_error (_("cannot make child for command substitution"));
6400 error_exit:
6401
6402 last_made_pid = old_pid;
6403
6404 FREE (istring);
6405 close (fildes[0]);
6406 close (fildes[1]);
6407 return ((WORD_DESC *)NULL);
6408 }
6409
6410 if (pid == 0)
6411 {
6412 /* The currently executing shell is not interactive. */
6413 interactive = 0;
6414
6415 set_sigint_handler (); /* XXX */
6416
6417 free_pushed_string_input ();
6418
6419 /* Discard buffered stdio output before replacing the underlying file
6420 descriptor. */
6421 fpurge (stdout);
6422
6423 if (dup2 (fildes[1], 1) < 0)
6424 {
6425 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
6426 exit (EXECUTION_FAILURE);
6427 }
6428
6429 /* If standard output is closed in the parent shell
6430 (such as after `exec >&-'), file descriptor 1 will be
6431 the lowest available file descriptor, and end up in
6432 fildes[0]. This can happen for stdin and stderr as well,
6433 but stdout is more important -- it will cause no output
6434 to be generated from this command. */
6435 if ((fildes[1] != fileno (stdin)) &&
6436 (fildes[1] != fileno (stdout)) &&
6437 (fildes[1] != fileno (stderr)))
6438 close (fildes[1]);
6439
6440 if ((fildes[0] != fileno (stdin)) &&
6441 (fildes[0] != fileno (stdout)) &&
6442 (fildes[0] != fileno (stderr)))
6443 close (fildes[0]);
6444
6445 #ifdef __CYGWIN__
6446 /* Let stdio know the fd may have changed from text to binary mode, and
6447 make sure to preserve stdout line buffering. */
6448 freopen (NULL, "w", stdout);
6449 sh_setlinebuf (stdout);
6450 #endif /* __CYGWIN__ */
6451
6452 /* This is a subshell environment. */
6453 subshell_environment |= SUBSHELL_COMSUB;
6454
6455 /* Many shells do not appear to inherit the -v option for command
6456 substitutions. */
6457 change_flag ('v', FLAG_OFF);
6458
6459 /* When inherit_errexit option is not enabled, command substitution does
6460 not inherit the -e flag. It is enabled when Posix mode is enabled */
6461 if (inherit_errexit == 0)
6462 {
6463 builtin_ignoring_errexit = 0;
6464 change_flag ('e', FLAG_OFF);
6465 }
6466 set_shellopts ();
6467
6468 /* If we are expanding a redirection, we can dispose of any temporary
6469 environment we received, since redirections are not supposed to have
6470 access to the temporary environment. We will have to see whether this
6471 affects temporary environments supplied to `eval', but the temporary
6472 environment gets copied to builtin_env at some point. */
6473 if (expanding_redir)
6474 {
6475 flush_temporary_env ();
6476 expanding_redir = 0;
6477 }
6478
6479 remove_quoted_escapes (string);
6480
6481 startup_state = 2; /* see if we can avoid a fork */
6482 parse_and_execute_level = 0;
6483
6484 /* Give command substitution a place to jump back to on failure,
6485 so we don't go back up to main (). */
6486 result = setjmp_nosigs (top_level);
6487
6488 /* If we're running a command substitution inside a shell function,
6489 trap `return' so we don't return from the function in the subshell
6490 and go off to never-never land. */
6491 if (result == 0 && return_catch_flag)
6492 function_value = setjmp_nosigs (return_catch);
6493 else
6494 function_value = 0;
6495
6496 if (result == ERREXIT)
6497 rc = last_command_exit_value;
6498 else if (result == EXITPROG)
6499 rc = last_command_exit_value;
6500 else if (result)
6501 rc = EXECUTION_FAILURE;
6502 else if (function_value)
6503 rc = return_catch_value;
6504 else
6505 {
6506 subshell_level++;
6507 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
6508 /* leave subshell level intact for any exit trap */
6509 }
6510
6511 last_command_exit_value = rc;
6512 rc = run_exit_trap ();
6513 #if defined (PROCESS_SUBSTITUTION)
6514 unlink_fifo_list ();
6515 #endif
6516 exit (rc);
6517 }
6518 else
6519 {
6520 int dummyfd;
6521
6522 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6523 close_pgrp_pipe ();
6524 #endif /* JOB_CONTROL && PGRP_PIPE */
6525
6526 close (fildes[1]);
6527
6528 begin_unwind_frame ("read-comsub");
6529 dummyfd = fildes[0];
6530 add_unwind_protect (close, dummyfd);
6531
6532 /* Block SIGINT while we're reading from the pipe. If the child
6533 process gets a SIGINT, it will either handle it or die, and the
6534 read will return. */
6535 BLOCK_SIGNAL (SIGINT, set, oset);
6536 tflag = 0;
6537 istring = read_comsub (fildes[0], quoted, flags, &tflag);
6538
6539 close (fildes[0]);
6540 discard_unwind_frame ("read-comsub");
6541 UNBLOCK_SIGNAL (oset);
6542
6543 current_command_subst_pid = pid;
6544 last_command_exit_value = wait_for (pid, JWAIT_NOTERM);
6545 last_command_subst_pid = pid;
6546 last_made_pid = old_pid;
6547
6548 #if defined (JOB_CONTROL)
6549 /* If last_command_exit_value > 128, then the substituted command
6550 was terminated by a signal. If that signal was SIGINT, then send
6551 SIGINT to ourselves. This will break out of loops, for instance. */
6552 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
6553 kill (getpid (), SIGINT);
6554 #endif /* JOB_CONTROL */
6555
6556 ret = alloc_word_desc ();
6557 ret->word = istring;
6558 ret->flags = tflag;
6559
6560 return ret;
6561 }
6562 }
6563
6564 /********************************************************
6565 * *
6566 * Utility functions for parameter expansion *
6567 * *
6568 ********************************************************/
6569
6570 #if defined (ARRAY_VARS)
6571
6572 static arrayind_t
6573 array_length_reference (s)
6574 char *s;
6575 {
6576 int len;
6577 arrayind_t ind;
6578 char *akey;
6579 char *t, c;
6580 ARRAY *array;
6581 HASH_TABLE *h;
6582 SHELL_VAR *var;
6583
6584 var = array_variable_part (s, 0, &t, &len);
6585
6586 /* If unbound variables should generate an error, report one and return
6587 failure. */
6588 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
6589 {
6590 c = *--t;
6591 *t = '\0';
6592 set_exit_status (EXECUTION_FAILURE);
6593 err_unboundvar (s);
6594 *t = c;
6595 return (-1);
6596 }
6597 else if (var == 0 || invisible_p (var))
6598 return 0;
6599
6600 /* We support a couple of expansions for variables that are not arrays.
6601 We'll return the length of the value for v[0], and 1 for v[@] or
6602 v[*]. Return 0 for everything else. */
6603
6604 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
6605 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
6606
6607 if (ALL_ELEMENT_SUB (t[0]) && t[1] == RBRACK)
6608 {
6609 if (assoc_p (var))
6610 return (h ? assoc_num_elements (h) : 0);
6611 else if (array_p (var))
6612 return (array ? array_num_elements (array) : 0);
6613 else
6614 return (var_isset (var) ? 1 : 0);
6615 }
6616
6617 if (assoc_p (var))
6618 {
6619 t[len - 1] = '\0';
6620 akey = expand_assignment_string_to_string (t, 0); /* [ */
6621 t[len - 1] = RBRACK;
6622 if (akey == 0 || *akey == 0)
6623 {
6624 err_badarraysub (t);
6625 FREE (akey);
6626 return (-1);
6627 }
6628 t = assoc_reference (assoc_cell (var), akey);
6629 free (akey);
6630 }
6631 else
6632 {
6633 ind = array_expand_index (var, t, len, 0);
6634 /* negative subscripts to indexed arrays count back from end */
6635 if (var && array_p (var) && ind < 0)
6636 ind = array_max_index (array_cell (var)) + 1 + ind;
6637 if (ind < 0)
6638 {
6639 err_badarraysub (t);
6640 return (-1);
6641 }
6642 if (array_p (var))
6643 t = array_reference (array, ind);
6644 else
6645 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6646 }
6647
6648 len = MB_STRLEN (t);
6649 return (len);
6650 }
6651 #endif /* ARRAY_VARS */
6652
6653 static int
6654 valid_brace_expansion_word (name, var_is_special)
6655 char *name;
6656 int var_is_special;
6657 {
6658 if (DIGIT (*name) && all_digits (name))
6659 return 1;
6660 else if (var_is_special)
6661 return 1;
6662 #if defined (ARRAY_VARS)
6663 else if (valid_array_reference (name, 0))
6664 return 1;
6665 #endif /* ARRAY_VARS */
6666 else if (legal_identifier (name))
6667 return 1;
6668 else
6669 return 0;
6670 }
6671
6672 static int
6673 chk_atstar (name, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
6674 char *name;
6675 int quoted, pflags;
6676 int *quoted_dollar_atp, *contains_dollar_at;
6677 {
6678 char *temp1;
6679
6680 if (name == 0)
6681 {
6682 if (quoted_dollar_atp)
6683 *quoted_dollar_atp = 0;
6684 if (contains_dollar_at)
6685 *contains_dollar_at = 0;
6686 return 0;
6687 }
6688
6689 /* check for $@ and $* */
6690 if (name[0] == '@' && name[1] == 0)
6691 {
6692 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6693 *quoted_dollar_atp = 1;
6694 if (contains_dollar_at)
6695 *contains_dollar_at = 1;
6696 return 1;
6697 }
6698 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6699 {
6700 /* Need more checks here that parallel what string_list_pos_params and
6701 param_expand do. Check expand_no_split_dollar_star and ??? */
6702 if (contains_dollar_at && expand_no_split_dollar_star == 0)
6703 *contains_dollar_at = 1;
6704 return 1;
6705 }
6706
6707 /* Now check for ${array[@]} and ${array[*]} */
6708 #if defined (ARRAY_VARS)
6709 else if (valid_array_reference (name, 0))
6710 {
6711 temp1 = mbschr (name, LBRACK);
6712 if (temp1 && temp1[1] == '@' && temp1[2] == RBRACK)
6713 {
6714 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6715 *quoted_dollar_atp = 1;
6716 if (contains_dollar_at)
6717 *contains_dollar_at = 1;
6718 return 1;
6719 }
6720 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6721 which should result in separate words even when IFS is unset. */
6722 if (temp1 && temp1[1] == '*' && temp1[2] == RBRACK && quoted == 0)
6723 {
6724 if (contains_dollar_at)
6725 *contains_dollar_at = 1;
6726 return 1;
6727 }
6728 }
6729 #endif
6730 return 0;
6731 }
6732
6733 /* Parameter expand NAME, and return a new string which is the expansion,
6734 or NULL if there was no expansion. NAME is as given in ${NAMEcWORD}.
6735 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6736 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6737 NAME was found inside of a double-quoted expression. */
6738 static WORD_DESC *
6739 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
6740 char *name;
6741 int var_is_special, quoted, pflags;
6742 arrayind_t *indp;
6743 {
6744 WORD_DESC *ret;
6745 char *temp, *tt;
6746 intmax_t arg_index;
6747 SHELL_VAR *var;
6748 int atype, rflags;
6749 arrayind_t ind;
6750
6751 ret = 0;
6752 temp = 0;
6753 rflags = 0;
6754
6755 if (indp)
6756 *indp = INTMAX_MIN;
6757
6758 /* Handle multiple digit arguments, as in ${11}. */
6759 if (legal_number (name, &arg_index))
6760 {
6761 tt = get_dollar_var_value (arg_index);
6762 if (tt)
6763 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6764 ? quote_string (tt)
6765 : quote_escapes (tt);
6766 else
6767 temp = (char *)NULL;
6768 FREE (tt);
6769 }
6770 else if (var_is_special) /* ${@} */
6771 {
6772 int sindex;
6773 tt = (char *)xmalloc (2 + strlen (name));
6774 tt[sindex = 0] = '$';
6775 strcpy (tt + 1, name);
6776
6777 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
6778 (int *)NULL, (int *)NULL, pflags);
6779 free (tt);
6780 }
6781 #if defined (ARRAY_VARS)
6782 else if (valid_array_reference (name, 0))
6783 {
6784 expand_arrayref:
6785 var = array_variable_part (name, 0, &tt, (int *)0);
6786 /* These are the cases where word splitting will not be performed */
6787 if (pflags & PF_ASSIGNRHS)
6788 {
6789 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == RBRACK)
6790 {
6791 /* Only treat as double quoted if array variable */
6792 if (var && (array_p (var) || assoc_p (var)))
6793 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6794 else
6795 temp = array_value (name, quoted, 0, &atype, &ind);
6796 }
6797 else
6798 temp = array_value (name, quoted, 0, &atype, &ind);
6799 }
6800 /* Posix interp 888 */
6801 else if (pflags & PF_NOSPLIT2)
6802 {
6803 /* Special cases, then general case, for each of A[@], A[*], A[n] */
6804 #if defined (HANDLE_MULTIBYTE)
6805 if (tt[0] == '@' && tt[1] == RBRACK && var && quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc[0] != ' ')
6806 #else
6807 if (tt[0] == '@' && tt[1] == RBRACK && var && quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc != ' ')
6808 #endif
6809 temp = array_value (name, Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6810 else if (tt[0] == '@' && tt[1] == RBRACK)
6811 temp = array_value (name, quoted, 0, &atype, &ind);
6812 else if (tt[0] == '*' && tt[1] == RBRACK && expand_no_split_dollar_star && ifs_is_null)
6813 temp = array_value (name, Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT, 0, &atype, &ind);
6814 else if (tt[0] == '*' && tt[1] == RBRACK)
6815 temp = array_value (name, quoted, 0, &atype, &ind);
6816 else
6817 temp = array_value (name, quoted, 0, &atype, &ind);
6818 }
6819 else if (tt[0] == '*' && tt[1] == RBRACK && expand_no_split_dollar_star && ifs_is_null)
6820 temp = array_value (name, Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT, 0, &atype, &ind);
6821 else
6822 temp = array_value (name, quoted, 0, &atype, &ind);
6823 if (atype == 0 && temp)
6824 {
6825 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6826 ? quote_string (temp)
6827 : quote_escapes (temp);
6828 rflags |= W_ARRAYIND;
6829 if (indp)
6830 *indp = ind;
6831 }
6832 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6833 rflags |= W_HASQUOTEDNULL;
6834 }
6835 #endif
6836 else if (var = find_variable (name))
6837 {
6838 if (var_isset (var) && invisible_p (var) == 0)
6839 {
6840 #if defined (ARRAY_VARS)
6841 /* We avoid a memory leak by saving TT as the memory allocated by
6842 assoc_to_string or array_to_string and leaving it 0 otherwise,
6843 then freeing TT after quoting temp. */
6844 tt = (char *)NULL;
6845 if ((pflags & PF_ALLINDS) && assoc_p (var))
6846 tt = temp = assoc_empty (assoc_cell (var)) ? (char *)NULL : assoc_to_string (assoc_cell (var), " ", quoted);
6847 else if ((pflags & PF_ALLINDS) && array_p (var))
6848 tt = temp = array_empty (array_cell (var)) ? (char *)NULL : array_to_string (array_cell (var), " ", quoted);
6849 else if (assoc_p (var))
6850 temp = assoc_reference (assoc_cell (var), "0");
6851 else if (array_p (var))
6852 temp = array_reference (array_cell (var), 0);
6853 else
6854 temp = value_cell (var);
6855 #else
6856 temp = value_cell (var);
6857 #endif
6858
6859 if (temp)
6860 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6861 ? quote_string (temp)
6862 : ((pflags & PF_ASSIGNRHS) ? quote_rhs (temp)
6863 : quote_escapes (temp));
6864 FREE (tt);
6865 }
6866 else
6867 temp = (char *)NULL;
6868 }
6869 else if (var = find_variable_last_nameref (name, 0))
6870 {
6871 temp = nameref_cell (var);
6872 #if defined (ARRAY_VARS)
6873 /* Handle expanding nameref whose value is x[n] */
6874 if (temp && *temp && valid_array_reference (temp, 0))
6875 {
6876 name = temp;
6877 goto expand_arrayref;
6878 }
6879 else
6880 #endif
6881 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6882 if (temp && *temp && legal_identifier (temp) == 0)
6883 {
6884 set_exit_status (EXECUTION_FAILURE);
6885 report_error (_("%s: invalid variable name for name reference"), temp);
6886 temp = &expand_param_error;
6887 }
6888 else
6889 temp = (char *)NULL;
6890 }
6891 else
6892 temp = (char *)NULL;
6893
6894 if (ret == 0)
6895 {
6896 ret = alloc_word_desc ();
6897 ret->word = temp;
6898 ret->flags |= rflags;
6899 }
6900 return ret;
6901 }
6902
6903 static char *
6904 parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
6905 char *name;
6906 int var_is_special, quoted, find_nameref;
6907 {
6908 char *temp, *t;
6909 WORD_DESC *w;
6910 SHELL_VAR *v;
6911 int pflags, oldex;
6912
6913 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
6914 nameref_p (v) && (t = nameref_cell (v)) && *t)
6915 return (savestring (t));
6916
6917 /* If var_is_special == 0, and name is not an array reference, this does
6918 more expansion than necessary. It should really look up the variable's
6919 value and not try to expand it. */
6920 pflags = PF_IGNUNBOUND;
6921 /* Note that we're not going to be doing word splitting here */
6922 if (var_is_special)
6923 {
6924 pflags |= PF_ASSIGNRHS; /* suppresses word splitting */
6925 oldex = expand_no_split_dollar_star;
6926 expand_no_split_dollar_star = 1;
6927 }
6928 w = parameter_brace_expand_word (name, var_is_special, quoted, pflags, 0);
6929 if (var_is_special)
6930 expand_no_split_dollar_star = oldex;
6931
6932 t = w->word;
6933 /* Have to dequote here if necessary */
6934 if (t)
6935 {
6936 temp = ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || var_is_special)
6937 ? dequote_string (t)
6938 : dequote_escapes (t);
6939 free (t);
6940 t = temp;
6941 }
6942 dispose_word_desc (w);
6943
6944 return t;
6945 }
6946
6947 /* Expand an indirect reference to a variable: ${!NAME} expands to the
6948 value of the variable whose name is the value of NAME. */
6949 static WORD_DESC *
6950 parameter_brace_expand_indir (name, var_is_special, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
6951 char *name;
6952 int var_is_special, quoted, pflags;
6953 int *quoted_dollar_atp, *contains_dollar_at;
6954 {
6955 char *t;
6956 WORD_DESC *w;
6957 SHELL_VAR *v;
6958
6959 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6960 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6961 bash performs an indirect lookup on foo[0] and expands the result;
6962 ksh93 expands bar[0]. We could do that here -- there are enough usable
6963 primitives to do that -- but do not at this point. */
6964 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
6965 {
6966 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6967 {
6968 w = alloc_word_desc ();
6969 w->word = savestring (t);
6970 w->flags = 0;
6971 return w;
6972 }
6973 }
6974
6975 /* An indirect reference to a positional parameter or a special parameter
6976 is ok. Indirect references to array references, as explained above, are
6977 ok (currently). Only references to unset variables are errors at this
6978 point. */
6979 if (legal_identifier (name) && v == 0)
6980 {
6981 report_error (_("%s: invalid indirect expansion"), name);
6982 w = alloc_word_desc ();
6983 w->word = &expand_param_error;
6984 w->flags = 0;
6985 return (w);
6986 }
6987
6988 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6989
6990 chk_atstar (t, quoted, pflags, quoted_dollar_atp, contains_dollar_at);
6991
6992 #if defined (ARRAY_VARS)
6993 /* Array references to unset variables are also an error */
6994 if (t == 0 && valid_array_reference (name, 0))
6995 {
6996 v = array_variable_part (name, 0, (char **)0, (int *)0);
6997 if (v == 0)
6998 {
6999 report_error (_("%s: invalid indirect expansion"), name);
7000 w = alloc_word_desc ();
7001 w->word = &expand_param_error;
7002 w->flags = 0;
7003 return (w);
7004 }
7005 else
7006 return (WORD_DESC *)NULL;
7007 }
7008 #endif
7009
7010 if (t == 0)
7011 return (WORD_DESC *)NULL;
7012
7013 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
7014 {
7015 report_error (_("%s: invalid variable name"), t);
7016 free (t);
7017 w = alloc_word_desc ();
7018 w->word = &expand_param_error;
7019 w->flags = 0;
7020 return (w);
7021 }
7022
7023 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, pflags, 0);
7024 free (t);
7025
7026 return w;
7027 }
7028
7029 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
7030 depending on the value of C, the separating character. C can be one of
7031 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
7032 between double quotes. */
7033 static WORD_DESC *
7034 parameter_brace_expand_rhs (name, value, op, quoted, pflags, qdollaratp, hasdollarat)
7035 char *name, *value;
7036 int op, quoted, pflags, *qdollaratp, *hasdollarat;
7037 {
7038 WORD_DESC *w;
7039 WORD_LIST *l, *tl;
7040 char *t, *t1, *temp, *vname;
7041 int l_hasdollat, sindex;
7042 SHELL_VAR *v;
7043
7044 /*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
7045 /* If the entire expression is between double quotes, we want to treat
7046 the value as a double-quoted string, with the exception that we strip
7047 embedded unescaped double quotes (for sh backwards compatibility). */
7048 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
7049 {
7050 sindex = 0;
7051 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
7052 }
7053 else
7054 temp = value;
7055
7056 w = alloc_word_desc ();
7057 l_hasdollat = 0;
7058 l = *temp ? expand_string_for_rhs (temp, quoted, op, pflags, &l_hasdollat, (int *)NULL)
7059 : (WORD_LIST *)0;
7060 if (hasdollarat)
7061 *hasdollarat = l_hasdollat || (l && l->next);
7062 if (temp != value)
7063 free (temp);
7064
7065 /* list_string takes multiple CTLNULs and turns them into an empty word
7066 with W_SAWQUOTEDNULL set. Turn it back into a single CTLNUL for the
7067 rest of this function and the caller. */
7068 for (tl = l; tl; tl = tl->next)
7069 {
7070 if (tl->word && (tl->word->word == 0 || tl->word->word[0] == 0) &&
7071 (tl->word->flags | W_SAWQUOTEDNULL))
7072 {
7073 t = make_quoted_char ('\0');
7074 FREE (tl->word->word);
7075 tl->word->word = t;
7076 tl->word->flags |= W_QUOTED|W_HASQUOTEDNULL;
7077 tl->word->flags &= ~W_SAWQUOTEDNULL;
7078 }
7079 }
7080
7081 if (l)
7082 {
7083 /* If l->next is not null, we know that TEMP contained "$@", since that
7084 is the only expansion that creates more than one word. */
7085 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
7086 {
7087 /*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
7088 *qdollaratp = 1;
7089 }
7090
7091 /* The expansion of TEMP returned something. We need to treat things
7092 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
7093 the individual words have already been quoted. We need to turn them
7094 into a string with the words separated by the first character of
7095 $IFS without any additional quoting, so string_list_dollar_at won't
7096 do the right thing. If IFS is null, we want "$@" to split into
7097 separate arguments, not be concatenated, so we use string_list_internal
7098 and mark the word to be split on spaces later. We use
7099 string_list_dollar_star for "$@" otherwise. */
7100 if (l->next && ifs_is_null)
7101 {
7102 temp = string_list_internal (l, " ");
7103 w->flags |= W_SPLITSPACE;
7104 }
7105 else if (l_hasdollat || l->next)
7106 temp = string_list_dollar_star (l, quoted, 0);
7107 else
7108 {
7109 temp = string_list (l);
7110 if (temp && (QUOTED_NULL (temp) == 0) && (l->word->flags & W_SAWQUOTEDNULL))
7111 w->flags |= W_SAWQUOTEDNULL; /* XXX */
7112 }
7113
7114 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
7115 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
7116 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
7117 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7118 (which is more paranoia than anything else), we need to return the
7119 quoted null string and set the flags to indicate it. */
7120 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
7121 {
7122 w->flags |= W_HASQUOTEDNULL;
7123 /*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
7124 /* If we return a quoted null with L_HASDOLLARAT, we either have a
7125 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
7126 parameters or a quoted expansion of "$@" with $1 == ''. In either
7127 case, we don't want to enable special handling of $@. */
7128 if (qdollaratp && l_hasdollat)
7129 *qdollaratp = 0;
7130 }
7131 dispose_words (l);
7132 }
7133 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
7134 {
7135 /* Posix interp 221 changed the rules on this. The idea is that
7136 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
7137 when foo and xxx are unset. The problem is that it's not in any
7138 way backwards compatible and few other shells do it. We're eventually
7139 going to try and split the difference (heh) a little bit here. */
7140 /* l_hasdollat == 1 means we saw a quoted dollar at. */
7141
7142 /* The brace expansion occurred between double quotes and there was
7143 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
7144 it does not expand to anything. In this case, we want to return
7145 a quoted empty string. Posix interp 888 */
7146 temp = make_quoted_char ('\0');
7147 w->flags |= W_HASQUOTEDNULL;
7148 /*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
7149 }
7150 else
7151 temp = (char *)NULL;
7152
7153 if (op == '-' || op == '+')
7154 {
7155 w->word = temp;
7156 return w;
7157 }
7158
7159 /* op == '=' */
7160 t1 = temp ? dequote_string (temp) : savestring ("");
7161 free (temp);
7162
7163 /* bash-4.4/5.0 */
7164 vname = name;
7165 if (*name == '!' &&
7166 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
7167 {
7168 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
7169 if (vname == 0 || *vname == 0)
7170 {
7171 report_error (_("%s: invalid indirect expansion"), name);
7172 free (vname);
7173 free (t1);
7174 dispose_word (w);
7175 return &expand_wdesc_error;
7176 }
7177 if (legal_identifier (vname) == 0)
7178 {
7179 report_error (_("%s: invalid variable name"), vname);
7180 free (vname);
7181 free (t1);
7182 dispose_word (w);
7183 return &expand_wdesc_error;
7184 }
7185 }
7186
7187 #if defined (ARRAY_VARS)
7188 if (valid_array_reference (vname, 0))
7189 v = assign_array_element (vname, t1, 0);
7190 else
7191 #endif /* ARRAY_VARS */
7192 v = bind_variable (vname, t1, 0);
7193
7194 if (v == 0 || readonly_p (v) || noassign_p (v)) /* expansion error */
7195 {
7196 if ((v == 0 || readonly_p (v)) && interactive_shell == 0 && posixly_correct)
7197 {
7198 last_command_exit_value = EXECUTION_FAILURE;
7199 exp_jump_to_top_level (FORCE_EOF);
7200 }
7201 else
7202 {
7203 if (vname != name)
7204 free (vname);
7205 last_command_exit_value = EX_BADUSAGE;
7206 exp_jump_to_top_level (DISCARD);
7207 }
7208 }
7209
7210 stupidly_hack_special_variables (vname);
7211
7212 if (vname != name)
7213 free (vname);
7214
7215 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
7216
7217 /* If we are double-quoted or if we are not going to be performing word
7218 splitting, we want to quote the value we return appropriately, like
7219 the other expansions this function handles. */
7220 w->word = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (t1) : quote_escapes (t1);
7221 /* If we have something that's non-null, that's not a quoted null string,
7222 and we're not going to be performing word splitting (we know we're not
7223 because the operator is `='), we can forget we saw a quoted null. */
7224 if (w->word && w->word[0] && QUOTED_NULL (w->word) == 0)
7225 w->flags &= ~W_SAWQUOTEDNULL;
7226 free (t1);
7227
7228 /* If we convert a null string into a quoted null, make sure the caller
7229 knows it. */
7230 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && QUOTED_NULL (w->word))
7231 w->flags |= W_HASQUOTEDNULL;
7232
7233 return w;
7234 }
7235
7236 /* Deal with the right hand side of a ${name:?value} expansion in the case
7237 that NAME is null or not set. If VALUE is non-null it is expanded and
7238 used as the error message to print, otherwise a standard message is
7239 printed. */
7240 static void
7241 parameter_brace_expand_error (name, value, check_null)
7242 char *name, *value;
7243 int check_null;
7244 {
7245 WORD_LIST *l;
7246 char *temp;
7247
7248 set_exit_status (EXECUTION_FAILURE); /* ensure it's non-zero */
7249 if (value && *value)
7250 {
7251 l = expand_string (value, 0);
7252 temp = string_list (l);
7253 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
7254 FREE (temp);
7255 dispose_words (l);
7256 }
7257 else if (check_null == 0)
7258 report_error (_("%s: parameter not set"), name);
7259 else
7260 report_error (_("%s: parameter null or not set"), name);
7261
7262 /* Free the data we have allocated during this expansion, since we
7263 are about to longjmp out. */
7264 free (name);
7265 FREE (value);
7266 }
7267
7268 /* Return 1 if NAME is something for which parameter_brace_expand_length is
7269 OK to do. */
7270 static int
7271 valid_length_expression (name)
7272 char *name;
7273 {
7274 return (name[1] == '\0' || /* ${#} */
7275 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
7276 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
7277 #if defined (ARRAY_VARS)
7278 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
7279 #endif
7280 legal_identifier (name + 1)); /* ${#PS1} */
7281 }
7282
7283 /* Handle the parameter brace expansion that requires us to return the
7284 length of a parameter. */
7285 static intmax_t
7286 parameter_brace_expand_length (name)
7287 char *name;
7288 {
7289 char *t, *newname;
7290 intmax_t number, arg_index;
7291 WORD_LIST *list;
7292 SHELL_VAR *var;
7293
7294 var = (SHELL_VAR *)NULL;
7295
7296 if (name[1] == '\0') /* ${#} */
7297 number = number_of_args ();
7298 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
7299 number = number_of_args ();
7300 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
7301 {
7302 /* Take the lengths of some of the shell's special parameters. */
7303 switch (name[1])
7304 {
7305 case '-':
7306 t = which_set_flags ();
7307 break;
7308 case '?':
7309 t = itos (last_command_exit_value);
7310 break;
7311 case '$':
7312 t = itos (dollar_dollar_pid);
7313 break;
7314 case '!':
7315 if (last_asynchronous_pid == NO_PID)
7316 t = (char *)NULL; /* XXX - error if set -u set? */
7317 else
7318 t = itos (last_asynchronous_pid);
7319 break;
7320 case '#':
7321 t = itos (number_of_args ());
7322 break;
7323 }
7324 number = STRLEN (t);
7325 FREE (t);
7326 }
7327 #if defined (ARRAY_VARS)
7328 else if (valid_array_reference (name + 1, 0))
7329 number = array_length_reference (name + 1);
7330 #endif /* ARRAY_VARS */
7331 else
7332 {
7333 number = 0;
7334
7335 if (legal_number (name + 1, &arg_index)) /* ${#1} */
7336 {
7337 t = get_dollar_var_value (arg_index);
7338 if (t == 0 && unbound_vars_is_error)
7339 return INTMAX_MIN;
7340 number = MB_STRLEN (t);
7341 FREE (t);
7342 }
7343 #if defined (ARRAY_VARS)
7344 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
7345 {
7346 if (assoc_p (var))
7347 t = assoc_reference (assoc_cell (var), "0");
7348 else
7349 t = array_reference (array_cell (var), 0);
7350 if (t == 0 && unbound_vars_is_error)
7351 return INTMAX_MIN;
7352 number = MB_STRLEN (t);
7353 }
7354 #endif
7355 /* Fast path for the common case of taking the length of a non-dynamic
7356 scalar variable value. */
7357 else if ((var || (var = find_variable (name + 1))) &&
7358 invisible_p (var) == 0 &&
7359 array_p (var) == 0 && assoc_p (var) == 0 &&
7360 var->dynamic_value == 0)
7361 number = value_cell (var) ? MB_STRLEN (value_cell (var)) : 0;
7362 else if (var == 0 && unbound_vars_is_error == 0)
7363 number = 0;
7364 else /* ${#PS1} */
7365 {
7366 newname = savestring (name);
7367 newname[0] = '$';
7368 list = expand_string (newname, Q_DOUBLE_QUOTES);
7369 t = list ? string_list (list) : (char *)NULL;
7370 free (newname);
7371 if (list)
7372 dispose_words (list);
7373
7374 number = t ? MB_STRLEN (t) : 0;
7375 FREE (t);
7376 }
7377 }
7378
7379 return (number);
7380 }
7381
7382 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
7383 so we do some ad-hoc parsing of an arithmetic expression to find
7384 the first DELIM, instead of using strchr(3). Two rules:
7385 1. If the substring contains a `(', read until closing `)'.
7386 2. If the substring contains a `?', read past one `:' for each `?'.
7387 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
7388 */
7389
7390 static char *
7391 skiparith (substr, delim)
7392 char *substr;
7393 int delim;
7394 {
7395 int i;
7396 char delims[2];
7397
7398 delims[0] = delim;
7399 delims[1] = '\0';
7400
7401 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
7402 return (substr + i);
7403 }
7404
7405 /* Verify and limit the start and end of the desired substring. If
7406 VTYPE == 0, a regular shell variable is being used; if it is 1,
7407 then the positional parameters are being used; if it is 2, then
7408 VALUE is really a pointer to an array variable that should be used.
7409 Return value is 1 if both values were OK, 0 if there was a problem
7410 with an invalid expression, or -1 if the values were out of range. */
7411 static int
7412 verify_substring_values (v, value, substr, vtype, e1p, e2p)
7413 SHELL_VAR *v;
7414 char *value, *substr;
7415 int vtype;
7416 intmax_t *e1p, *e2p;
7417 {
7418 char *t, *temp1, *temp2;
7419 arrayind_t len;
7420 int expok;
7421 #if defined (ARRAY_VARS)
7422 ARRAY *a;
7423 HASH_TABLE *h;
7424 #endif
7425
7426 /* duplicate behavior of strchr(3) */
7427 t = skiparith (substr, ':');
7428 if (*t && *t == ':')
7429 *t = '\0';
7430 else
7431 t = (char *)0;
7432
7433 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
7434 *e1p = evalexp (temp1, 0, &expok); /* XXX - EXP_EXPANDED? */
7435 free (temp1);
7436 if (expok == 0)
7437 return (0);
7438
7439 len = -1; /* paranoia */
7440 switch (vtype)
7441 {
7442 case VT_VARIABLE:
7443 case VT_ARRAYMEMBER:
7444 len = MB_STRLEN (value);
7445 break;
7446 case VT_POSPARMS:
7447 len = number_of_args () + 1;
7448 if (*e1p == 0)
7449 len++; /* add one arg if counting from $0 */
7450 break;
7451 #if defined (ARRAY_VARS)
7452 case VT_ARRAYVAR:
7453 /* For arrays, the first value deals with array indices. Negative
7454 offsets count from one past the array's maximum index. Associative
7455 arrays treat the number of elements as the maximum index. */
7456 if (assoc_p (v))
7457 {
7458 h = assoc_cell (v);
7459 len = assoc_num_elements (h) + (*e1p < 0);
7460 }
7461 else
7462 {
7463 a = (ARRAY *)value;
7464 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7465 }
7466 break;
7467 #endif
7468 }
7469
7470 if (len == -1) /* paranoia */
7471 return -1;
7472
7473 if (*e1p < 0) /* negative offsets count from end */
7474 *e1p += len;
7475
7476 if (*e1p > len || *e1p < 0)
7477 return (-1);
7478
7479 #if defined (ARRAY_VARS)
7480 /* For arrays, the second offset deals with the number of elements. */
7481 if (vtype == VT_ARRAYVAR)
7482 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
7483 #endif
7484
7485 if (t)
7486 {
7487 t++;
7488 temp2 = savestring (t);
7489 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7490 free (temp2);
7491 t[-1] = ':';
7492 *e2p = evalexp (temp1, 0, &expok); /* XXX - EXP_EXPANDED? */
7493 free (temp1);
7494 if (expok == 0)
7495 return (0);
7496
7497 /* Should we allow positional parameter length < 0 to count backwards
7498 from end of positional parameters? */
7499 #if 1
7500 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
7501 #else /* TAG: bash-5.2 */
7502 if (vtype == VT_ARRAYVAR && *e2p < 0)
7503 #endif
7504 {
7505 internal_error (_("%s: substring expression < 0"), t);
7506 return (0);
7507 }
7508 #if defined (ARRAY_VARS)
7509 /* In order to deal with sparse arrays, push the intelligence about how
7510 to deal with the number of elements desired down to the array-
7511 specific functions. */
7512 if (vtype != VT_ARRAYVAR)
7513 #endif
7514 {
7515 if (*e2p < 0)
7516 {
7517 *e2p += len;
7518 if (*e2p < 0 || *e2p < *e1p)
7519 {
7520 internal_error (_("%s: substring expression < 0"), t);
7521 return (0);
7522 }
7523 }
7524 else
7525 *e2p += *e1p; /* want E2 chars starting at E1 */
7526 if (*e2p > len)
7527 *e2p = len;
7528 }
7529 }
7530 else
7531 *e2p = len;
7532
7533 return (1);
7534 }
7535
7536 /* Return the type of variable specified by VARNAME (simple variable,
7537 positional param, or array variable). Also return the value specified
7538 by VARNAME (value of a variable or a reference to an array element).
7539 QUOTED is the standard description of quoting state, using Q_* defines.
7540 FLAGS is currently a set of flags to pass to array_value. If IND is
7541 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7542 passed to array_value so the array index is not computed again.
7543 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7544 characters in the value are quoted with CTLESC and takes appropriate
7545 steps. For convenience, *VALP is set to the dequoted VALUE. */
7546 static int
7547 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
7548 char *varname, *value;
7549 arrayind_t ind;
7550 int quoted, flags;
7551 SHELL_VAR **varp;
7552 char **valp;
7553 {
7554 int vtype, want_indir;
7555 char *temp, *vname;
7556 SHELL_VAR *v;
7557 arrayind_t lind;
7558
7559 want_indir = *varname == '!' &&
7560 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7561 || VALID_INDIR_PARAM (varname[1]));
7562 if (want_indir)
7563 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
7564 /* XXX - what if vname == 0 || *vname == 0 ? */
7565 else
7566 vname = varname;
7567
7568 if (vname == 0)
7569 {
7570 vtype = VT_VARIABLE;
7571 *varp = (SHELL_VAR *)NULL;
7572 *valp = (char *)NULL;
7573 return (vtype);
7574 }
7575
7576 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
7577 vtype = STR_DOLLAR_AT_STAR (vname);
7578 if (vtype == VT_POSPARMS && vname[0] == '*')
7579 vtype |= VT_STARSUB;
7580 *varp = (SHELL_VAR *)NULL;
7581
7582 #if defined (ARRAY_VARS)
7583 if (valid_array_reference (vname, 0))
7584 {
7585 v = array_variable_part (vname, 0, &temp, (int *)0);
7586 /* If we want to signal array_value to use an already-computed index,
7587 set LIND to that index */
7588 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
7589 if (v && invisible_p (v))
7590 {
7591 vtype = VT_ARRAYMEMBER;
7592 *varp = (SHELL_VAR *)NULL;
7593 *valp = (char *)NULL;
7594 }
7595 if (v && (array_p (v) || assoc_p (v)))
7596 {
7597 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == RBRACK)
7598 {
7599 /* Callers have to differentiate between indexed and associative */
7600 vtype = VT_ARRAYVAR;
7601 if (temp[0] == '*')
7602 vtype |= VT_STARSUB;
7603 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
7604 }
7605 else
7606 {
7607 vtype = VT_ARRAYMEMBER;
7608 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7609 }
7610 *varp = v;
7611 }
7612 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == RBRACK))
7613 {
7614 vtype = VT_VARIABLE;
7615 *varp = v;
7616 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7617 *valp = value ? dequote_string (value) : (char *)NULL;
7618 else
7619 *valp = value ? dequote_escapes (value) : (char *)NULL;
7620 }
7621 else
7622 {
7623 vtype = VT_ARRAYMEMBER;
7624 *varp = v;
7625 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7626 }
7627 }
7628 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
7629 {
7630 vtype = VT_ARRAYMEMBER;
7631 *varp = v;
7632 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
7633 }
7634 else
7635 #endif
7636 {
7637 if (value && vtype == VT_VARIABLE)
7638 {
7639 *varp = find_variable (vname);
7640 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7641 *valp = dequote_string (value);
7642 else
7643 *valp = dequote_escapes (value);
7644 }
7645 else
7646 *valp = value;
7647 }
7648
7649 if (want_indir)
7650 free (vname);
7651
7652 return vtype;
7653 }
7654
7655 /***********************************************************/
7656 /* */
7657 /* Functions to perform transformations on variable values */
7658 /* */
7659 /***********************************************************/
7660
7661 static char *
7662 string_var_assignment (v, s)
7663 SHELL_VAR *v;
7664 char *s;
7665 {
7666 char flags[MAX_ATTRIBUTES], *ret, *val;
7667 int i;
7668
7669 val = (v && (invisible_p (v) || var_isset (v) == 0)) ? (char *)NULL : sh_quote_reusable (s, 0);
7670 i = var_attribute_string (v, 0, flags);
7671 if (i == 0 && val == 0)
7672 return (char *)NULL;
7673
7674 ret = (char *)xmalloc (i + STRLEN (val) + strlen (v->name) + 16 + MAX_ATTRIBUTES);
7675 if (i > 0 && val == 0)
7676 sprintf (ret, "declare -%s %s", flags, v->name);
7677 else if (i > 0)
7678 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
7679 else
7680 sprintf (ret, "%s=%s", v->name, val);
7681 free (val);
7682 return ret;
7683 }
7684
7685 #if defined (ARRAY_VARS)
7686 static char *
7687 array_var_assignment (v, itype, quoted, atype)
7688 SHELL_VAR *v;
7689 int itype, quoted, atype;
7690 {
7691 char *ret, *val, flags[MAX_ATTRIBUTES];
7692 int i;
7693
7694 if (v == 0)
7695 return (char *)NULL;
7696 if (atype == 2)
7697 val = array_p (v) ? array_to_kvpair (array_cell (v), 0)
7698 : assoc_to_kvpair (assoc_cell (v), 0);
7699 else
7700 val = array_p (v) ? array_to_assign (array_cell (v), 0)
7701 : assoc_to_assign (assoc_cell (v), 0);
7702
7703 if (val == 0 && (invisible_p (v) || var_isset (v) == 0))
7704 ; /* placeholder */
7705 else if (val == 0)
7706 {
7707 val = (char *)xmalloc (3);
7708 val[0] = LPAREN;
7709 val[1] = RPAREN;
7710 val[2] = 0;
7711 }
7712 else
7713 {
7714 ret = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (val) : quote_escapes (val);
7715 free (val);
7716 val = ret;
7717 }
7718
7719 if (atype == 2)
7720 return val;
7721
7722 i = var_attribute_string (v, 0, flags);
7723 ret = (char *)xmalloc (i + STRLEN (val) + strlen (v->name) + 16);
7724 if (val)
7725 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
7726 else
7727 sprintf (ret, "declare -%s %s", flags, v->name);
7728 free (val);
7729 return ret;
7730 }
7731 #endif
7732
7733 static char *
7734 pos_params_assignment (list, itype, quoted)
7735 WORD_LIST *list;
7736 int itype;
7737 int quoted;
7738 {
7739 char *temp, *ret;
7740
7741 /* first, we transform the list to quote each word. */
7742 temp = list_transform ('Q', (SHELL_VAR *)0, list, itype, quoted);
7743 ret = (char *)xmalloc (strlen (temp) + 8);
7744 strcpy (ret, "set -- ");
7745 strcpy (ret + 7, temp);
7746 free (temp);
7747 return ret;
7748 }
7749
7750 static char *
7751 string_transform (xc, v, s)
7752 int xc;
7753 SHELL_VAR *v;
7754 char *s;
7755 {
7756 char *ret, flags[MAX_ATTRIBUTES], *t;
7757 int i;
7758
7759 if (((xc == 'A' || xc == 'a') && v == 0))
7760 return (char *)NULL;
7761 else if (xc != 'a' && xc != 'A' && s == 0)
7762 return (char *)NULL;
7763
7764 switch (xc)
7765 {
7766 /* Transformations that interrogate the variable */
7767 case 'a':
7768 i = var_attribute_string (v, 0, flags);
7769 ret = (i > 0) ? savestring (flags) : (char *)NULL;
7770 break;
7771 case 'A':
7772 ret = string_var_assignment (v, s);
7773 break;
7774 case 'K':
7775 ret = sh_quote_reusable (s, 0);
7776 break;
7777 /* Transformations that modify the variable's value */
7778 case 'E':
7779 t = ansiexpand (s, 0, strlen (s), (int *)0);
7780 ret = dequote_escapes (t);
7781 free (t);
7782 break;
7783 case 'P':
7784 ret = decode_prompt_string (s);
7785 break;
7786 case 'Q':
7787 ret = sh_quote_reusable (s, 0);
7788 break;
7789 case 'U':
7790 ret = sh_modcase (s, 0, CASE_UPPER);
7791 break;
7792 case 'u':
7793 ret = sh_modcase (s, 0, CASE_UPFIRST); /* capitalize */
7794 break;
7795 case 'L':
7796 ret = sh_modcase (s, 0, CASE_LOWER);
7797 break;
7798 default:
7799 ret = (char *)NULL;
7800 break;
7801 }
7802 return ret;
7803 }
7804
7805 static char *
7806 list_transform (xc, v, list, itype, quoted)
7807 int xc;
7808 SHELL_VAR *v;
7809 WORD_LIST *list;
7810 int itype, quoted;
7811 {
7812 WORD_LIST *new, *l;
7813 WORD_DESC *w;
7814 char *tword;
7815 int qflags;
7816
7817 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
7818 {
7819 tword = string_transform (xc, v, l->word->word);
7820 w = alloc_word_desc ();
7821 w->word = tword ? tword : savestring (""); /* XXX */
7822 new = make_word_list (w, new);
7823 }
7824 l = REVERSE_LIST (new, WORD_LIST *);
7825
7826 qflags = quoted;
7827 /* If we are expanding in a context where word splitting will not be
7828 performed, treat as quoted. This changes how $* will be expanded. */
7829 if (itype == '*' && expand_no_split_dollar_star && ifs_is_null)
7830 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
7831
7832 tword = string_list_pos_params (itype, l, qflags, 0);
7833 dispose_words (l);
7834
7835 return (tword);
7836 }
7837
7838 static char *
7839 parameter_list_transform (xc, itype, quoted)
7840 int xc;
7841 int itype;
7842 int quoted;
7843 {
7844 char *ret;
7845 WORD_LIST *list;
7846
7847 list = list_rest_of_args ();
7848 if (list == 0)
7849 return ((char *)NULL);
7850 if (xc == 'A')
7851 ret = pos_params_assignment (list, itype, quoted);
7852 else
7853 ret = list_transform (xc, (SHELL_VAR *)0, list, itype, quoted);
7854 dispose_words (list);
7855 return (ret);
7856 }
7857
7858 #if defined (ARRAY_VARS)
7859 static char *
7860 array_transform (xc, var, starsub, quoted)
7861 int xc;
7862 SHELL_VAR *var;
7863 int starsub; /* so we can figure out how it's indexed */
7864 int quoted;
7865 {
7866 ARRAY *a;
7867 HASH_TABLE *h;
7868 int itype;
7869 char *ret;
7870 WORD_LIST *list;
7871 SHELL_VAR *v;
7872
7873 v = var; /* XXX - for now */
7874
7875 itype = starsub ? '*' : '@';
7876
7877 if (xc == 'A')
7878 return (array_var_assignment (v, itype, quoted, 1));
7879 else if (xc == 'K')
7880 return (array_var_assignment (v, itype, quoted, 2));
7881
7882 /* special case for unset arrays and attributes */
7883 if (xc == 'a' && (invisible_p (v) || var_isset (v) == 0))
7884 {
7885 char flags[MAX_ATTRIBUTES];
7886 int i;
7887
7888 i = var_attribute_string (v, 0, flags);
7889 return ((i > 0) ? savestring (flags) : (char *)NULL);
7890 }
7891
7892 a = (v && array_p (v)) ? array_cell (v) : 0;
7893 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
7894
7895 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
7896 if (list == 0)
7897 return ((char *)NULL);
7898 ret = list_transform (xc, v, list, itype, quoted);
7899 dispose_words (list);
7900
7901 return ret;
7902 }
7903 #endif /* ARRAY_VARS */
7904
7905 static int
7906 valid_parameter_transform (xform)
7907 char *xform;
7908 {
7909 if (xform[1])
7910 return 0;
7911
7912 /* check for valid values of xform[0] */
7913 switch (xform[0])
7914 {
7915 case 'a': /* expand to a string with just attributes */
7916 case 'A': /* expand as an assignment statement with attributes */
7917 case 'K': /* expand assoc array to list of key/value pairs */
7918 case 'E': /* expand like $'...' */
7919 case 'P': /* expand like prompt string */
7920 case 'Q': /* quote reusably */
7921 case 'U': /* transform to uppercase */
7922 case 'u': /* tranform by capitalizing */
7923 case 'L': /* transform to lowercase */
7924 return 1;
7925 default:
7926 return 0;
7927 }
7928 }
7929
7930 static char *
7931 parameter_brace_transform (varname, value, ind, xform, rtype, quoted, pflags, flags)
7932 char *varname, *value;
7933 int ind;
7934 char *xform;
7935 int rtype, quoted, pflags, flags;
7936 {
7937 int vtype, xc, starsub;
7938 char *temp1, *val, *oname;
7939 SHELL_VAR *v;
7940
7941 xc = xform[0];
7942 if (value == 0 && xc != 'A' && xc != 'a')
7943 return ((char *)NULL);
7944
7945 oname = this_command_name;
7946 this_command_name = varname;
7947
7948 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7949 if (vtype == -1)
7950 {
7951 this_command_name = oname;
7952 return ((char *)NULL);
7953 }
7954
7955 if (valid_parameter_transform (xform) == 0)
7956 {
7957 this_command_name = oname;
7958 #if 0 /* TAG: bash-5.2 Martin Schulte <gnu@schrader-schulte.de> 10/2020 */
7959 return (interactive_shell ? &expand_param_error : &expand_param_fatal);
7960 #else
7961 return &expand_param_error;
7962 #endif
7963 }
7964
7965 starsub = vtype & VT_STARSUB;
7966 vtype &= ~VT_STARSUB;
7967
7968 /* If we are asked to display the attributes of an unset variable, V will
7969 be NULL after the call to get_var_and_type. Double-check here. */
7970 if ((xc == 'a' || xc == 'A') && vtype == VT_VARIABLE && varname && v == 0)
7971 v = find_variable (varname);
7972
7973 temp1 = (char *)NULL; /* shut up gcc */
7974 switch (vtype)
7975 {
7976 case VT_VARIABLE:
7977 case VT_ARRAYMEMBER:
7978 temp1 = string_transform (xc, v, val);
7979 if (vtype == VT_VARIABLE)
7980 FREE (val);
7981 if (temp1)
7982 {
7983 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7984 ? quote_string (temp1)
7985 : quote_escapes (temp1);
7986 free (temp1);
7987 temp1 = val;
7988 }
7989 break;
7990 #if defined (ARRAY_VARS)
7991 case VT_ARRAYVAR:
7992 temp1 = array_transform (xc, v, starsub, quoted);
7993 if (temp1 && quoted == 0 && ifs_is_null)
7994 {
7995 /* Posix interp 888 */
7996 }
7997 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
7998 {
7999 val = quote_escapes (temp1);
8000 free (temp1);
8001 temp1 = val;
8002 }
8003 break;
8004 #endif
8005 case VT_POSPARMS:
8006 temp1 = parameter_list_transform (xc, varname[0], quoted);
8007 if (temp1 && quoted == 0 && ifs_is_null)
8008 {
8009 /* Posix interp 888 */
8010 }
8011 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
8012 {
8013 val = quote_escapes (temp1);
8014 free (temp1);
8015 temp1 = val;
8016 }
8017 break;
8018 }
8019
8020 this_command_name = oname;
8021 return temp1;
8022 }
8023
8024 /******************************************************/
8025 /* */
8026 /* Functions to extract substrings of variable values */
8027 /* */
8028 /******************************************************/
8029
8030 #if defined (HANDLE_MULTIBYTE)
8031 /* Character-oriented rather than strictly byte-oriented substrings. S and
8032 E, rather being strict indices into STRING, indicate character (possibly
8033 multibyte character) positions that require calculation.
8034 Used by the ${param:offset[:length]} expansion. */
8035 static char *
8036 mb_substring (string, s, e)
8037 char *string;
8038 int s, e;
8039 {
8040 char *tt;
8041 int start, stop, i;
8042 size_t slen;
8043 DECLARE_MBSTATE;
8044
8045 start = 0;
8046 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
8047 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
8048
8049 i = s;
8050 while (string[start] && i--)
8051 ADVANCE_CHAR (string, slen, start);
8052 stop = start;
8053 i = e - s;
8054 while (string[stop] && i--)
8055 ADVANCE_CHAR (string, slen, stop);
8056 tt = substring (string, start, stop);
8057 return tt;
8058 }
8059 #endif
8060
8061 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
8062 is `@', use the positional parameters; otherwise, use the value of
8063 VARNAME. If VARNAME is an array variable, use the array elements. */
8064
8065 static char *
8066 parameter_brace_substring (varname, value, ind, substr, quoted, pflags, flags)
8067 char *varname, *value;
8068 int ind;
8069 char *substr;
8070 int quoted, pflags, flags;
8071 {
8072 intmax_t e1, e2;
8073 int vtype, r, starsub;
8074 char *temp, *val, *tt, *oname;
8075 SHELL_VAR *v;
8076
8077 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
8078 return ((char *)NULL);
8079
8080 oname = this_command_name;
8081 this_command_name = varname;
8082
8083 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
8084 if (vtype == -1)
8085 {
8086 this_command_name = oname;
8087 return ((char *)NULL);
8088 }
8089
8090 starsub = vtype & VT_STARSUB;
8091 vtype &= ~VT_STARSUB;
8092
8093 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
8094 this_command_name = oname;
8095 if (r <= 0)
8096 {
8097 if (vtype == VT_VARIABLE)
8098 FREE (val);
8099 return ((r == 0) ? &expand_param_error : (char *)NULL);
8100 }
8101
8102 switch (vtype)
8103 {
8104 case VT_VARIABLE:
8105 case VT_ARRAYMEMBER:
8106 #if defined (HANDLE_MULTIBYTE)
8107 if (MB_CUR_MAX > 1)
8108 tt = mb_substring (val, e1, e2);
8109 else
8110 #endif
8111 tt = substring (val, e1, e2);
8112
8113 if (vtype == VT_VARIABLE)
8114 FREE (val);
8115 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
8116 temp = quote_string (tt);
8117 else
8118 temp = tt ? quote_escapes (tt) : (char *)NULL;
8119 FREE (tt);
8120 break;
8121 case VT_POSPARMS:
8122 case VT_ARRAYVAR:
8123 if (vtype == VT_POSPARMS)
8124 tt = pos_params (varname, e1, e2, quoted, pflags);
8125 #if defined (ARRAY_VARS)
8126 /* assoc_subrange and array_subrange both call string_list_pos_params,
8127 so we can treat this case just like VT_POSPARAMS. */
8128 else if (assoc_p (v))
8129 /* we convert to list and take first e2 elements starting at e1th
8130 element -- officially undefined for now */
8131 tt = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted, pflags);
8132 else
8133 /* We want E2 to be the number of elements desired (arrays can be
8134 sparse, so verify_substring_values just returns the numbers
8135 specified and we rely on array_subrange to understand how to
8136 deal with them). */
8137 tt = array_subrange (array_cell (v), e1, e2, starsub, quoted, pflags);
8138 #endif
8139 /* We want to leave this alone in every case where pos_params/
8140 string_list_pos_params quotes the list members */
8141 if (tt && quoted == 0 && ifs_is_null)
8142 {
8143 temp = tt; /* Posix interp 888 */
8144 }
8145 else if (tt && quoted == 0 && (pflags & PF_ASSIGNRHS))
8146 {
8147 temp = tt; /* Posix interp 888 */
8148 }
8149 else if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
8150 {
8151 temp = tt ? quote_escapes (tt) : (char *)NULL;
8152 FREE (tt);
8153 }
8154 else
8155 temp = tt;
8156 break;
8157
8158 default:
8159 temp = (char *)NULL;
8160 }
8161
8162 return temp;
8163 }
8164
8165 /****************************************************************/
8166 /* */
8167 /* Functions to perform pattern substitution on variable values */
8168 /* */
8169 /****************************************************************/
8170
8171 #ifdef INCLUDE_UNUSED
8172 static int
8173 shouldexp_replacement (s)
8174 char *s;
8175 {
8176 register char *p;
8177
8178 for (p = s; p && *p; p++)
8179 {
8180 if (*p == '\\')
8181 p++;
8182 else if (*p == '&')
8183 return 1;
8184 }
8185 return 0;
8186 }
8187 #endif
8188
8189 char *
8190 pat_subst (string, pat, rep, mflags)
8191 char *string, *pat, *rep;
8192 int mflags;
8193 {
8194 char *ret, *s, *e, *str, *rstr, *mstr, *send;
8195 int rptr, mtype, rxpand, mlen;
8196 size_t rsize, l, replen, rslen;
8197 DECLARE_MBSTATE;
8198
8199 if (string == 0)
8200 return (savestring (""));
8201
8202 mtype = mflags & MATCH_TYPEMASK;
8203
8204 #if 0 /* TAG: bash-5.2? */
8205 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
8206 #else
8207 rxpand = 0;
8208 #endif
8209
8210 /* Special cases:
8211 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
8212 * with REP and return the result.
8213 * 2. A null pattern with mtype == MATCH_END means to append REP to
8214 * STRING and return the result.
8215 * 3. A null STRING with a matching pattern means to append REP to
8216 * STRING and return the result.
8217 * These don't understand or process `&' in the replacement string.
8218 */
8219 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
8220 {
8221 replen = STRLEN (rep);
8222 l = STRLEN (string);
8223 ret = (char *)xmalloc (replen + l + 2);
8224 if (replen == 0)
8225 strcpy (ret, string);
8226 else if (mtype == MATCH_BEG)
8227 {
8228 strcpy (ret, rep);
8229 strcpy (ret + replen, string);
8230 }
8231 else
8232 {
8233 strcpy (ret, string);
8234 strcpy (ret + l, rep);
8235 }
8236 return (ret);
8237 }
8238 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
8239 {
8240 replen = STRLEN (rep);
8241 ret = (char *)xmalloc (replen + 1);
8242 if (replen == 0)
8243 ret[0] = '\0';
8244 else
8245 strcpy (ret, rep);
8246 return (ret);
8247 }
8248
8249 ret = (char *)xmalloc (rsize = 64);
8250 ret[0] = '\0';
8251 send = string + strlen (string);
8252
8253 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
8254 {
8255 if (match_pattern (str, pat, mtype, &s, &e) == 0)
8256 break;
8257 l = s - str;
8258
8259 if (rep && rxpand)
8260 {
8261 int x;
8262 mlen = e - s;
8263 mstr = xmalloc (mlen + 1);
8264 for (x = 0; x < mlen; x++)
8265 mstr[x] = s[x];
8266 mstr[mlen] = '\0';
8267 rstr = strcreplace (rep, '&', mstr, 0);
8268 free (mstr);
8269 rslen = strlen (rstr);
8270 }
8271 else
8272 {
8273 rstr = rep;
8274 rslen = replen;
8275 }
8276
8277 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
8278
8279 /* OK, now copy the leading unmatched portion of the string (from
8280 str to s) to ret starting at rptr (the current offset). Then copy
8281 the replacement string at ret + rptr + (s - str). Increment
8282 rptr (if necessary) and str and go on. */
8283 if (l)
8284 {
8285 strncpy (ret + rptr, str, l);
8286 rptr += l;
8287 }
8288 if (replen)
8289 {
8290 strncpy (ret + rptr, rstr, rslen);
8291 rptr += rslen;
8292 }
8293 str = e; /* e == end of match */
8294
8295 if (rstr != rep)
8296 free (rstr);
8297
8298 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
8299 break;
8300
8301 if (s == e)
8302 {
8303 /* On a zero-length match, make sure we copy one character, since
8304 we increment one character to avoid infinite recursion. */
8305 char *p, *origp, *origs;
8306 size_t clen;
8307
8308 RESIZE_MALLOCED_BUFFER (ret, rptr, locale_mb_cur_max, rsize, 64);
8309 #if defined (HANDLE_MULTIBYTE)
8310 p = origp = ret + rptr;
8311 origs = str;
8312 COPY_CHAR_P (p, str, send);
8313 rptr += p - origp;
8314 e += str - origs;
8315 #else
8316 ret[rptr++] = *str++;
8317 e++; /* avoid infinite recursion on zero-length match */
8318 #endif
8319 }
8320 }
8321
8322 /* Now copy the unmatched portion of the input string */
8323 if (str && *str)
8324 {
8325 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
8326 strcpy (ret + rptr, str);
8327 }
8328 else
8329 ret[rptr] = '\0';
8330
8331 return ret;
8332 }
8333
8334 /* Do pattern match and replacement on the positional parameters. */
8335 static char *
8336 pos_params_pat_subst (string, pat, rep, mflags)
8337 char *string, *pat, *rep;
8338 int mflags;
8339 {
8340 WORD_LIST *save, *params;
8341 WORD_DESC *w;
8342 char *ret;
8343 int pchar, qflags, pflags;
8344
8345 save = params = list_rest_of_args ();
8346 if (save == 0)
8347 return ((char *)NULL);
8348
8349 for ( ; params; params = params->next)
8350 {
8351 ret = pat_subst (params->word->word, pat, rep, mflags);
8352 w = alloc_word_desc ();
8353 w->word = ret ? ret : savestring ("");
8354 dispose_word (params->word);
8355 params->word = w;
8356 }
8357
8358 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
8359 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
8360 pflags = (mflags & MATCH_ASSIGNRHS) == MATCH_ASSIGNRHS ? PF_ASSIGNRHS : 0;
8361
8362 /* If we are expanding in a context where word splitting will not be
8363 performed, treat as quoted. This changes how $* will be expanded. */
8364 if (pchar == '*' && (mflags & MATCH_ASSIGNRHS) && expand_no_split_dollar_star && ifs_is_null)
8365 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
8366
8367 ret = string_list_pos_params (pchar, save, qflags, pflags);
8368 dispose_words (save);
8369
8370 return (ret);
8371 }
8372
8373 /* Perform pattern substitution on VALUE, which is the expansion of
8374 VARNAME. PATSUB is an expression supplying the pattern to match
8375 and the string to substitute. QUOTED is a flags word containing
8376 the type of quoting currently in effect. */
8377 static char *
8378 parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
8379 char *varname, *value;
8380 int ind;
8381 char *patsub;
8382 int quoted, pflags, flags;
8383 {
8384 int vtype, mflags, starsub, delim;
8385 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt, *oname;
8386 SHELL_VAR *v;
8387
8388 if (value == 0)
8389 return ((char *)NULL);
8390
8391 oname = this_command_name;
8392 this_command_name = varname; /* error messages */
8393
8394 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
8395 if (vtype == -1)
8396 {
8397 this_command_name = oname;
8398 return ((char *)NULL);
8399 }
8400
8401 starsub = vtype & VT_STARSUB;
8402 vtype &= ~VT_STARSUB;
8403
8404 mflags = 0;
8405 /* PATSUB is never NULL when this is called. */
8406 if (*patsub == '/')
8407 {
8408 mflags |= MATCH_GLOBREP;
8409 patsub++;
8410 }
8411
8412 /* Malloc this because expand_string_if_necessary or one of the expansion
8413 functions in its call chain may free it on a substitution error. */
8414 lpatsub = savestring (patsub);
8415
8416 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8417 mflags |= MATCH_QUOTED;
8418
8419 if (starsub)
8420 mflags |= MATCH_STARSUB;
8421
8422 if (pflags & PF_ASSIGNRHS)
8423 mflags |= MATCH_ASSIGNRHS;
8424
8425 /* If the pattern starts with a `/', make sure we skip over it when looking
8426 for the replacement delimiter. */
8427 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
8428 if (lpatsub[delim] == '/')
8429 {
8430 lpatsub[delim] = 0;
8431 rep = lpatsub + delim + 1;
8432 }
8433 else
8434 rep = (char *)NULL;
8435
8436 if (rep && *rep == '\0')
8437 rep = (char *)NULL;
8438
8439 /* Perform the same expansions on the pattern as performed by the
8440 pattern removal expansions. */
8441 pat = getpattern (lpatsub, quoted, 1);
8442
8443 if (rep)
8444 {
8445 /* We want to perform quote removal on the expanded replacement even if
8446 the entire expansion is double-quoted because the parser and string
8447 extraction functions treated quotes in the replacement string as
8448 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
8449 if (shell_compatibility_level > 42)
8450 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
8451 /* This is the bash-4.2 code. */
8452 else if ((mflags & MATCH_QUOTED) == 0)
8453 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
8454 else
8455 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
8456 }
8457
8458 /* ksh93 doesn't allow the match specifier to be a part of the expanded
8459 pattern. This is an extension. Make sure we don't anchor the pattern
8460 at the beginning or end of the string if we're doing global replacement,
8461 though. */
8462 p = pat;
8463 if (mflags & MATCH_GLOBREP)
8464 mflags |= MATCH_ANY;
8465 else if (pat && pat[0] == '#')
8466 {
8467 mflags |= MATCH_BEG;
8468 p++;
8469 }
8470 else if (pat && pat[0] == '%')
8471 {
8472 mflags |= MATCH_END;
8473 p++;
8474 }
8475 else
8476 mflags |= MATCH_ANY;
8477
8478 /* OK, we now want to substitute REP for PAT in VAL. If
8479 flags & MATCH_GLOBREP is non-zero, the substitution is done
8480 everywhere, otherwise only the first occurrence of PAT is
8481 replaced. The pattern matching code doesn't understand
8482 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
8483 values passed in (VT_VARIABLE) so the pattern substitution
8484 code works right. We need to requote special chars after
8485 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
8486 other cases if QUOTED == 0, since the posparams and arrays
8487 indexed by * or @ do special things when QUOTED != 0. */
8488
8489 switch (vtype)
8490 {
8491 case VT_VARIABLE:
8492 case VT_ARRAYMEMBER:
8493 temp = pat_subst (val, p, rep, mflags);
8494 if (vtype == VT_VARIABLE)
8495 FREE (val);
8496 if (temp)
8497 {
8498 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
8499 free (temp);
8500 temp = tt;
8501 }
8502 break;
8503 case VT_POSPARMS:
8504 /* This does the right thing for the case where we are not performing
8505 word splitting. MATCH_STARSUB restricts it to ${* /foo/bar}, and
8506 pos_params_pat_subst/string_list_pos_params will do the right thing
8507 in turn for the case where ifs_is_null. Posix interp 888 */
8508 if ((pflags & PF_NOSPLIT2) && (mflags & MATCH_STARSUB))
8509 mflags |= MATCH_ASSIGNRHS;
8510 temp = pos_params_pat_subst (val, p, rep, mflags);
8511 if (temp && quoted == 0 && ifs_is_null)
8512 {
8513 /* Posix interp 888 */
8514 }
8515 else if (temp && quoted == 0 && (pflags & PF_ASSIGNRHS))
8516 {
8517 /* Posix interp 888 */
8518 }
8519 else if (temp && (mflags & MATCH_QUOTED) == 0)
8520 {
8521 tt = quote_escapes (temp);
8522 free (temp);
8523 temp = tt;
8524 }
8525 break;
8526 #if defined (ARRAY_VARS)
8527 case VT_ARRAYVAR:
8528 /* If we are expanding in a context where word splitting will not be
8529 performed, treat as quoted. This changes how ${A[*]} will be
8530 expanded to make it identical to $*. */
8531 if ((mflags & MATCH_STARSUB) && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8532 mflags |= MATCH_QUOTED; /* Posix interp 888 */
8533
8534 /* these eventually call string_list_pos_params */
8535 if (assoc_p (v))
8536 temp = assoc_patsub (assoc_cell (v), p, rep, mflags);
8537 else
8538 temp = array_patsub (array_cell (v), p, rep, mflags);
8539
8540 if (temp && quoted == 0 && ifs_is_null)
8541 {
8542 /* Posix interp 888 */
8543 }
8544 else if (temp && (mflags & MATCH_QUOTED) == 0)
8545 {
8546 tt = quote_escapes (temp);
8547 free (temp);
8548 temp = tt;
8549 }
8550 break;
8551 #endif
8552 }
8553
8554 FREE (pat);
8555 FREE (rep);
8556 free (lpatsub);
8557
8558 this_command_name = oname;
8559
8560 return temp;
8561 }
8562
8563 /****************************************************************/
8564 /* */
8565 /* Functions to perform case modification on variable values */
8566 /* */
8567 /****************************************************************/
8568
8569 /* Do case modification on the positional parameters. */
8570
8571 static char *
8572 pos_params_modcase (string, pat, modop, mflags)
8573 char *string, *pat;
8574 int modop;
8575 int mflags;
8576 {
8577 WORD_LIST *save, *params;
8578 WORD_DESC *w;
8579 char *ret;
8580 int pchar, qflags, pflags;
8581
8582 save = params = list_rest_of_args ();
8583 if (save == 0)
8584 return ((char *)NULL);
8585
8586 for ( ; params; params = params->next)
8587 {
8588 ret = sh_modcase (params->word->word, pat, modop);
8589 w = alloc_word_desc ();
8590 w->word = ret ? ret : savestring ("");
8591 dispose_word (params->word);
8592 params->word = w;
8593 }
8594
8595 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
8596 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
8597 pflags = (mflags & MATCH_ASSIGNRHS) == MATCH_ASSIGNRHS ? PF_ASSIGNRHS : 0;
8598
8599 /* If we are expanding in a context where word splitting will not be
8600 performed, treat as quoted. This changes how $* will be expanded. */
8601 if (pchar == '*' && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8602 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
8603
8604 ret = string_list_pos_params (pchar, save, qflags, pflags);
8605 dispose_words (save);
8606
8607 return (ret);
8608 }
8609
8610 /* Perform case modification on VALUE, which is the expansion of
8611 VARNAME. MODSPEC is an expression supplying the type of modification
8612 to perform. QUOTED is a flags word containing the type of quoting
8613 currently in effect. */
8614 static char *
8615 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, pflags, flags)
8616 char *varname, *value;
8617 int ind, modspec;
8618 char *patspec;
8619 int quoted, pflags, flags;
8620 {
8621 int vtype, starsub, modop, mflags, x;
8622 char *val, *temp, *pat, *p, *lpat, *tt, *oname;
8623 SHELL_VAR *v;
8624
8625 if (value == 0)
8626 return ((char *)NULL);
8627
8628 oname = this_command_name;
8629 this_command_name = varname;
8630
8631 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
8632 if (vtype == -1)
8633 {
8634 this_command_name = oname;
8635 return ((char *)NULL);
8636 }
8637
8638 starsub = vtype & VT_STARSUB;
8639 vtype &= ~VT_STARSUB;
8640
8641 modop = 0;
8642 mflags = 0;
8643 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8644 mflags |= MATCH_QUOTED;
8645 if (starsub)
8646 mflags |= MATCH_STARSUB;
8647 if (pflags & PF_ASSIGNRHS)
8648 mflags |= MATCH_ASSIGNRHS;
8649
8650 p = patspec;
8651 if (modspec == '^')
8652 {
8653 x = p && p[0] == modspec;
8654 modop = x ? CASE_UPPER : CASE_UPFIRST;
8655 p += x;
8656 }
8657 else if (modspec == ',')
8658 {
8659 x = p && p[0] == modspec;
8660 modop = x ? CASE_LOWER : CASE_LOWFIRST;
8661 p += x;
8662 }
8663 else if (modspec == '~')
8664 {
8665 x = p && p[0] == modspec;
8666 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
8667 p += x;
8668 }
8669
8670 lpat = p ? savestring (p) : 0;
8671 /* Perform the same expansions on the pattern as performed by the
8672 pattern removal expansions. */
8673 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
8674
8675 /* OK, now we do the case modification. */
8676 switch (vtype)
8677 {
8678 case VT_VARIABLE:
8679 case VT_ARRAYMEMBER:
8680 temp = sh_modcase (val, pat, modop);
8681 if (vtype == VT_VARIABLE)
8682 FREE (val);
8683 if (temp)
8684 {
8685 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
8686 free (temp);
8687 temp = tt;
8688 }
8689 break;
8690
8691 case VT_POSPARMS:
8692 temp = pos_params_modcase (val, pat, modop, mflags);
8693 if (temp && quoted == 0 && ifs_is_null)
8694 {
8695 /* Posix interp 888 */
8696 }
8697 else if (temp && (mflags & MATCH_QUOTED) == 0)
8698 {
8699 tt = quote_escapes (temp);
8700 free (temp);
8701 temp = tt;
8702 }
8703 break;
8704
8705 #if defined (ARRAY_VARS)
8706 case VT_ARRAYVAR:
8707 /* If we are expanding in a context where word splitting will not be
8708 performed, treat as quoted. This changes how ${A[*]} will be
8709 expanded to make it identical to $*. */
8710 if ((mflags & MATCH_STARSUB) && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8711 mflags |= MATCH_QUOTED; /* Posix interp 888 */
8712
8713 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
8714 : array_modcase (array_cell (v), pat, modop, mflags);
8715
8716 if (temp && quoted == 0 && ifs_is_null)
8717 {
8718 /* Posix interp 888 */
8719 }
8720 else if (temp && (mflags & MATCH_QUOTED) == 0)
8721 {
8722 tt = quote_escapes (temp);
8723 free (temp);
8724 temp = tt;
8725 }
8726
8727 break;
8728 #endif
8729 }
8730
8731 FREE (pat);
8732 free (lpat);
8733
8734 this_command_name = oname;
8735
8736 return temp;
8737 }
8738
8739 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
8740 any occur, this must be a nested command substitution, so return 0.
8741 Otherwise, return 1. A valid arithmetic expression must always have a
8742 ( before a matching ), so any cases where there are more right parens
8743 means that this must not be an arithmetic expression, though the parser
8744 will not accept it without a balanced total number of parens. */
8745 static int
8746 chk_arithsub (s, len)
8747 const char *s;
8748 int len;
8749 {
8750 int i, count;
8751 DECLARE_MBSTATE;
8752
8753 i = count = 0;
8754 while (i < len)
8755 {
8756 if (s[i] == LPAREN)
8757 count++;
8758 else if (s[i] == RPAREN)
8759 {
8760 count--;
8761 if (count < 0)
8762 return 0;
8763 }
8764
8765 switch (s[i])
8766 {
8767 default:
8768 ADVANCE_CHAR (s, len, i);
8769 break;
8770
8771 case '\\':
8772 i++;
8773 if (s[i])
8774 ADVANCE_CHAR (s, len, i);
8775 break;
8776
8777 case '\'':
8778 i = skip_single_quoted (s, len, ++i, 0);
8779 break;
8780
8781 case '"':
8782 i = skip_double_quoted ((char *)s, len, ++i, 0);
8783 break;
8784 }
8785 }
8786
8787 return (count == 0);
8788 }
8789
8790 /****************************************************************/
8791 /* */
8792 /* Functions to perform parameter expansion on a string */
8793 /* */
8794 /****************************************************************/
8795
8796 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
8797 static WORD_DESC *
8798 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
8799 char *string;
8800 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
8801 {
8802 int check_nullness, var_is_set, var_is_null, var_is_special;
8803 int want_substring, want_indir, want_patsub, want_casemod, want_attributes;
8804 char *name, *value, *temp, *temp1;
8805 WORD_DESC *tdesc, *ret;
8806 int t_index, sindex, c, tflag, modspec, local_pflags, all_element_arrayref;
8807 intmax_t number;
8808 arrayind_t ind;
8809
8810 temp = temp1 = value = (char *)NULL;
8811 var_is_set = var_is_null = var_is_special = check_nullness = 0;
8812 want_substring = want_indir = want_patsub = want_casemod = want_attributes = 0;
8813
8814 local_pflags = 0;
8815 all_element_arrayref = 0;
8816
8817 sindex = *indexp;
8818 t_index = ++sindex;
8819 /* ${#var} doesn't have any of the other parameter expansions on it. */
8820 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
8821 name = string_extract (string, &t_index, "}", SX_VARNAME);
8822 else
8823 #if defined (CASEMOD_EXPANSIONS)
8824 /* To enable case-toggling expansions using the `~' operator character
8825 define CASEMOD_TOGGLECASE in config-top.h */
8826 # if defined (CASEMOD_TOGGLECASE)
8827 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
8828 # else
8829 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
8830 # endif /* CASEMOD_TOGGLECASE */
8831 #else
8832 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
8833 #endif /* CASEMOD_EXPANSIONS */
8834
8835 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
8836 the cleanest code ever. */
8837 if (*name == 0 && sindex == t_index && string[sindex] == '@')
8838 {
8839 name = (char *)xrealloc (name, 2);
8840 name[0] = '@';
8841 name[1] = '\0';
8842 t_index++;
8843 }
8844 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == RBRACE)
8845 {
8846 name = (char *)xrealloc (name, t_index - sindex + 2);
8847 name[t_index - sindex] = '@';
8848 name[t_index - sindex + 1] = '\0';
8849 t_index++;
8850 }
8851
8852 ret = 0;
8853 tflag = 0;
8854
8855 ind = INTMAX_MIN;
8856
8857 /* If the name really consists of a special variable, then make sure
8858 that we have the entire name. We don't allow indirect references
8859 to special variables except `#', `?', `@' and `*'. This clause is
8860 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
8861 general. */
8862 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
8863 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
8864 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
8865 {
8866 t_index++;
8867 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
8868 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
8869 *name = string[sindex];
8870 if (string[sindex] == '!')
8871 {
8872 /* indirect reference of $#, $?, $@, or $* */
8873 name[1] = string[sindex + 1];
8874 strcpy (name + 2, temp1);
8875 }
8876 else
8877 strcpy (name + 1, temp1);
8878 free (temp1);
8879 }
8880 sindex = t_index;
8881
8882 /* Find out what character ended the variable name. Then
8883 do the appropriate thing. */
8884 if (c = string[sindex])
8885 sindex++;
8886
8887 /* If c is followed by one of the valid parameter expansion
8888 characters, move past it as normal. If not, assume that
8889 a substring specification is being given, and do not move
8890 past it. */
8891 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
8892 {
8893 check_nullness++;
8894 if (c = string[sindex])
8895 sindex++;
8896 }
8897 else if (c == ':' && string[sindex] != RBRACE)
8898 want_substring = 1;
8899 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
8900 want_patsub = 1;
8901 #if defined (CASEMOD_EXPANSIONS)
8902 else if (c == '^' || c == ',' || c == '~')
8903 {
8904 modspec = c;
8905 want_casemod = 1;
8906 }
8907 #endif
8908 else if (c == '@' && (string[sindex] == 'a' || string[sindex] == 'A') && string[sindex+1] == RBRACE)
8909 {
8910 /* special case because we do not want to shortcut foo as foo[0] here */
8911 want_attributes = 1;
8912 local_pflags |= PF_ALLINDS;
8913 }
8914
8915 /* Catch the valid and invalid brace expressions that made it through the
8916 tests above. */
8917 /* ${#-} is a valid expansion and means to take the length of $-.
8918 Similarly for ${#?} and ${##}... */
8919 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8920 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
8921 {
8922 name = (char *)xrealloc (name, 3);
8923 name[1] = c;
8924 name[2] = '\0';
8925 c = string[sindex++];
8926 }
8927
8928 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8929 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8930 member (c, "%:=+/") && string[sindex] == RBRACE)
8931 {
8932 temp = (char *)NULL;
8933 goto bad_substitution; /* XXX - substitution error */
8934 }
8935
8936 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8937 either a variable name, one of the positional parameters or a special
8938 variable that expands to one of the positional parameters. */
8939 want_indir = *name == '!' &&
8940 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
8941 || VALID_INDIR_PARAM (name[1]));
8942
8943 /* Determine the value of this variable whose name is NAME. */
8944
8945 /* Check for special variables, directly referenced. */
8946 if (SPECIAL_VAR (name, want_indir))
8947 var_is_special++;
8948
8949 /* Check for special expansion things, like the length of a parameter */
8950 if (*name == '#' && name[1])
8951 {
8952 /* If we are not pointing at the character just after the
8953 closing brace, then we haven't gotten all of the name.
8954 Since it begins with a special character, this is a bad
8955 substitution. Also check NAME for validity before trying
8956 to go on. */
8957 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
8958 {
8959 temp = (char *)NULL;
8960 goto bad_substitution; /* substitution error */
8961 }
8962
8963 number = parameter_brace_expand_length (name);
8964 if (number == INTMAX_MIN && unbound_vars_is_error)
8965 {
8966 set_exit_status (EXECUTION_FAILURE);
8967 err_unboundvar (name+1);
8968 free (name);
8969 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8970 }
8971 free (name);
8972
8973 *indexp = sindex;
8974 if (number < 0)
8975 return (&expand_wdesc_error);
8976 else
8977 {
8978 ret = alloc_word_desc ();
8979 ret->word = itos (number);
8980 return ret;
8981 }
8982 }
8983
8984 /* ${@} is identical to $@. */
8985 if (name[0] == '@' && name[1] == '\0')
8986 {
8987 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8988 *quoted_dollar_atp = 1;
8989
8990 if (contains_dollar_at)
8991 *contains_dollar_at = 1;
8992
8993 tflag |= W_DOLLARAT;
8994 }
8995
8996 /* Process ${!PREFIX*} expansion. */
8997 if (want_indir && string[sindex - 1] == RBRACE &&
8998 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
8999 legal_variable_starter ((unsigned char) name[1]))
9000 {
9001 char **x;
9002 WORD_LIST *xlist;
9003
9004 temp1 = savestring (name + 1);
9005 number = strlen (temp1);
9006 temp1[number - 1] = '\0';
9007 x = all_variables_matching_prefix (temp1);
9008 xlist = strvec_to_word_list (x, 0, 0);
9009 if (string[sindex - 2] == '*')
9010 temp = string_list_dollar_star (xlist, quoted, 0);
9011 else
9012 {
9013 temp = string_list_dollar_at (xlist, quoted, 0);
9014 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9015 *quoted_dollar_atp = 1;
9016 if (contains_dollar_at)
9017 *contains_dollar_at = 1;
9018
9019 tflag |= W_DOLLARAT;
9020 }
9021 free (x);
9022 dispose_words (xlist);
9023 free (temp1);
9024 *indexp = sindex;
9025
9026 free (name);
9027
9028 ret = alloc_word_desc ();
9029 ret->word = temp;
9030 ret->flags = tflag; /* XXX */
9031 return ret;
9032 }
9033
9034 #if defined (ARRAY_VARS)
9035 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */
9036 if (want_indir && string[sindex - 1] == RBRACE &&
9037 string[sindex - 2] == RBRACK && valid_array_reference (name+1, 0))
9038 {
9039 char *x, *x1;
9040
9041 temp1 = savestring (name + 1);
9042 x = array_variable_name (temp1, 0, &x1, (int *)0);
9043 FREE (x);
9044 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == RBRACK)
9045 {
9046 temp = array_keys (temp1, quoted, pflags); /* handles assoc vars too */
9047 if (x1[0] == '@')
9048 {
9049 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9050 *quoted_dollar_atp = 1;
9051 if (contains_dollar_at)
9052 *contains_dollar_at = 1;
9053
9054 tflag |= W_DOLLARAT;
9055 }
9056
9057 free (name);
9058 free (temp1);
9059 *indexp = sindex;
9060
9061 ret = alloc_word_desc ();
9062 ret->word = temp;
9063 ret->flags = tflag; /* XXX */
9064 return ret;
9065 }
9066
9067 free (temp1);
9068 }
9069 #endif /* ARRAY_VARS */
9070
9071 /* Make sure that NAME is valid before trying to go on. */
9072 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
9073 var_is_special) == 0)
9074 {
9075 temp = (char *)NULL;
9076 goto bad_substitution; /* substitution error */
9077 }
9078
9079 if (want_indir)
9080 {
9081 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, pflags|local_pflags, quoted_dollar_atp, contains_dollar_at);
9082 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9083 {
9084 temp = (char *)NULL;
9085 goto bad_substitution;
9086 }
9087
9088 /* Turn off the W_ARRAYIND flag because there is no way for this function
9089 to return the index we're supposed to be using. */
9090 if (tdesc && tdesc->flags)
9091 tdesc->flags &= ~W_ARRAYIND;
9092 }
9093 else
9094 {
9095 local_pflags |= PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS));
9096 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, local_pflags, &ind);
9097 }
9098
9099 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9100 {
9101 tflag = 0;
9102 tdesc = 0;
9103 }
9104
9105 if (tdesc)
9106 {
9107 temp = tdesc->word;
9108 tflag = tdesc->flags;
9109 dispose_word_desc (tdesc);
9110 }
9111 else
9112 temp = (char *)0;
9113
9114 if (temp == &expand_param_error || temp == &expand_param_fatal)
9115 {
9116 FREE (name);
9117 FREE (value);
9118 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9119 }
9120
9121 #if defined (ARRAY_VARS)
9122 if (valid_array_reference (name, 0))
9123 {
9124 int qflags;
9125 char *t;
9126
9127 qflags = quoted;
9128 /* If in a context where word splitting will not take place, treat as
9129 if double-quoted. Has effects with $* and ${array[*]} */
9130
9131 if (pflags & PF_ASSIGNRHS)
9132 qflags |= Q_DOUBLE_QUOTES;
9133 /* We duplicate a little code here */
9134 t = mbschr (name, LBRACK);
9135 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == RBRACK)
9136 {
9137 all_element_arrayref = 1;
9138 if (expand_no_split_dollar_star && t[1] == '*') /* XXX */
9139 qflags |= Q_DOUBLE_QUOTES;
9140 }
9141 chk_atstar (name, qflags, pflags, quoted_dollar_atp, contains_dollar_at);
9142 }
9143 #endif
9144
9145 var_is_set = temp != (char *)0;
9146 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
9147 /* XXX - this may not need to be restricted to special variables */
9148 if (check_nullness)
9149 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
9150 #if defined (ARRAY_VARS)
9151 if (check_nullness)
9152 var_is_null |= var_is_set &&
9153 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) &&
9154 QUOTED_NULL (temp) &&
9155 valid_array_reference (name, 0) &&
9156 chk_atstar (name, 0, 0, (int *)0, (int *)0);
9157 #endif
9158
9159 /* Get the rest of the stuff inside the braces. */
9160 if (c && c != RBRACE)
9161 {
9162 /* Extract the contents of the ${ ... } expansion
9163 according to the Posix.2 rules. */
9164 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
9165 if (string[sindex] == RBRACE)
9166 sindex++;
9167 else
9168 goto bad_substitution; /* substitution error */
9169 }
9170 else
9171 value = (char *)NULL;
9172
9173 *indexp = sindex;
9174
9175 /* All the cases where an expansion can possibly generate an unbound
9176 variable error. */
9177 if (want_substring || want_patsub || want_casemod || c == '@' || c == '#' || c == '%' || c == RBRACE)
9178 {
9179 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
9180 {
9181 set_exit_status (EXECUTION_FAILURE);
9182 err_unboundvar (name);
9183 FREE (value);
9184 FREE (temp);
9185 free (name);
9186 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9187 }
9188 }
9189
9190 /* If this is a substring spec, process it and add the result. */
9191 if (want_substring)
9192 {
9193 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9194 FREE (value);
9195 FREE (temp);
9196
9197 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9198 {
9199 FREE (name);
9200 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9201 }
9202
9203 ret = alloc_word_desc ();
9204 ret->word = temp1;
9205 /* We test quoted_dollar_atp because we want variants with double-quoted
9206 "$@" to take a different code path. In fact, we make sure at the end
9207 of expand_word_internal that we're only looking at these flags if
9208 quoted_dollar_at == 0. */
9209 if (temp1 &&
9210 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9211 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9212 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9213 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 &&
9214 (pflags & PF_ASSIGNRHS))
9215 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9216 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9217 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9218 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9219
9220 FREE (name);
9221 return ret;
9222 }
9223 else if (want_patsub)
9224 {
9225 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9226 FREE (value);
9227 FREE (temp);
9228
9229 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9230 {
9231 FREE (name);
9232 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9233 }
9234
9235 ret = alloc_word_desc ();
9236 ret->word = temp1;
9237 if (temp1 &&
9238 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9239 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9240 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9241 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9242 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9243 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9244
9245 FREE (name);
9246 return ret;
9247 }
9248 #if defined (CASEMOD_EXPANSIONS)
9249 else if (want_casemod)
9250 {
9251 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9252 FREE (value);
9253 FREE (temp);
9254
9255 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9256 {
9257 FREE (name);
9258 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9259 }
9260
9261 ret = alloc_word_desc ();
9262 ret->word = temp1;
9263 if (temp1 &&
9264 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9265 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9266 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9267 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9268 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9269 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9270
9271 FREE (name);
9272 return ret;
9273 }
9274 #endif
9275
9276 /* Do the right thing based on which character ended the variable name. */
9277 switch (c)
9278 {
9279 default:
9280 case '\0':
9281 bad_substitution:
9282 set_exit_status (EXECUTION_FAILURE);
9283 report_error (_("%s: bad substitution"), string ? string : "??");
9284 FREE (value);
9285 FREE (temp);
9286 free (name);
9287 if (shell_compatibility_level <= 43)
9288 return &expand_wdesc_error;
9289 else
9290 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
9291
9292 case RBRACE:
9293 break;
9294
9295 case '@':
9296 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9297 free (temp);
9298 free (value);
9299
9300 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9301 {
9302 free (name);
9303 set_exit_status (EXECUTION_FAILURE);
9304 report_error (_("%s: bad substitution"), string ? string : "??");
9305 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9306 }
9307
9308 ret = alloc_word_desc ();
9309 ret->word = temp1;
9310 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9311 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9312 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9313 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9314 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9315
9316 free (name);
9317 return ret;
9318
9319 case '#': /* ${param#[#]pattern} */
9320 case '%': /* ${param%[%]pattern} */
9321 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
9322 {
9323 FREE (value);
9324 break;
9325 }
9326 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9327 free (temp);
9328 free (value);
9329
9330 ret = alloc_word_desc ();
9331 ret->word = temp1;
9332 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9333 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9334 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9335 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9336 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9337
9338 free (name);
9339 return ret;
9340
9341 case '-':
9342 case '=':
9343 case '?':
9344 case '+':
9345 if (var_is_set && var_is_null == 0)
9346 {
9347 /* If the operator is `+', we don't want the value of the named
9348 variable for anything, just the value of the right hand side. */
9349 if (c == '+')
9350 {
9351 /* XXX -- if we're double-quoted and the named variable is "$@",
9352 we want to turn off any special handling of "$@" --
9353 we're not using it, so whatever is on the rhs applies. */
9354 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9355 *quoted_dollar_atp = 0;
9356 if (contains_dollar_at)
9357 *contains_dollar_at = 0;
9358
9359 FREE (temp);
9360 if (value)
9361 {
9362 /* From Posix discussion on austin-group list. Issue 221
9363 requires that backslashes escaping `}' inside
9364 double-quoted ${...} be removed. */
9365 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9366 quoted |= Q_DOLBRACE;
9367 ret = parameter_brace_expand_rhs (name, value, c,
9368 quoted,
9369 pflags,
9370 quoted_dollar_atp,
9371 contains_dollar_at);
9372 /* XXX - fix up later, esp. noting presence of
9373 W_HASQUOTEDNULL in ret->flags */
9374 free (value);
9375 }
9376 else
9377 temp = (char *)NULL;
9378 }
9379 else
9380 {
9381 FREE (value);
9382 }
9383 /* Otherwise do nothing; just use the value in TEMP. */
9384 }
9385 else /* VAR not set or VAR is NULL. */
9386 {
9387 FREE (temp);
9388 temp = (char *)NULL;
9389 if (c == '=' && var_is_special)
9390 {
9391 set_exit_status (EXECUTION_FAILURE);
9392 report_error (_("$%s: cannot assign in this way"), name);
9393 free (name);
9394 free (value);
9395 return &expand_wdesc_error;
9396 }
9397 else if (c == '?')
9398 {
9399 parameter_brace_expand_error (name, value, check_nullness);
9400 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9401 }
9402 else if (c != '+')
9403 {
9404 /* XXX -- if we're double-quoted and the named variable is "$@",
9405 we want to turn off any special handling of "$@" --
9406 we're not using it, so whatever is on the rhs applies. */
9407 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9408 *quoted_dollar_atp = 0;
9409 if (contains_dollar_at)
9410 *contains_dollar_at = 0;
9411
9412 /* From Posix discussion on austin-group list. Issue 221 requires
9413 that backslashes escaping `}' inside double-quoted ${...} be
9414 removed. */
9415 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9416 quoted |= Q_DOLBRACE;
9417 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
9418 quoted_dollar_atp,
9419 contains_dollar_at);
9420 /* XXX - fix up later, esp. noting presence of
9421 W_HASQUOTEDNULL in tdesc->flags */
9422 }
9423 free (value);
9424 }
9425
9426 break;
9427 }
9428 free (name);
9429
9430 if (ret == 0)
9431 {
9432 ret = alloc_word_desc ();
9433 ret->flags = tflag;
9434 ret->word = temp;
9435 }
9436 return (ret);
9437 }
9438
9439 /* Expand a single ${xxx} expansion. The braces are optional. When
9440 the braces are used, parameter_brace_expand() does the work,
9441 possibly calling param_expand recursively. */
9442 static WORD_DESC *
9443 param_expand (string, sindex, quoted, expanded_something,
9444 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
9445 pflags)
9446 char *string;
9447 int *sindex, quoted, *expanded_something, *contains_dollar_at;
9448 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
9449 {
9450 char *temp, *temp1, uerror[3], *savecmd;
9451 int zindex, t_index, expok;
9452 unsigned char c;
9453 intmax_t number;
9454 SHELL_VAR *var;
9455 WORD_LIST *list, *l;
9456 WORD_DESC *tdesc, *ret;
9457 int tflag, nullarg;
9458
9459 /*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
9460 zindex = *sindex;
9461 c = string[++zindex];
9462
9463 temp = (char *)NULL;
9464 ret = tdesc = (WORD_DESC *)NULL;
9465 tflag = 0;
9466
9467 /* Do simple cases first. Switch on what follows '$'. */
9468 switch (c)
9469 {
9470 /* $0 .. $9? */
9471 case '0':
9472 case '1':
9473 case '2':
9474 case '3':
9475 case '4':
9476 case '5':
9477 case '6':
9478 case '7':
9479 case '8':
9480 case '9':
9481 temp1 = dollar_vars[TODIGIT (c)];
9482 /* This doesn't get called when (pflags&PF_IGNUNBOUND) != 0 */
9483 if (unbound_vars_is_error && temp1 == (char *)NULL)
9484 {
9485 uerror[0] = '$';
9486 uerror[1] = c;
9487 uerror[2] = '\0';
9488 set_exit_status (EXECUTION_FAILURE);
9489 err_unboundvar (uerror);
9490 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9491 }
9492 if (temp1)
9493 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9494 ? quote_string (temp1)
9495 : quote_escapes (temp1);
9496 else
9497 temp = (char *)NULL;
9498
9499 break;
9500
9501 /* $$ -- pid of the invoking shell. */
9502 case '$':
9503 temp = itos (dollar_dollar_pid);
9504 break;
9505
9506 /* $# -- number of positional parameters. */
9507 case '#':
9508 temp = itos (number_of_args ());
9509 break;
9510
9511 /* $? -- return value of the last synchronous command. */
9512 case '?':
9513 temp = itos (last_command_exit_value);
9514 break;
9515
9516 /* $- -- flags supplied to the shell on invocation or by `set'. */
9517 case '-':
9518 temp = which_set_flags ();
9519 break;
9520
9521 /* $! -- Pid of the last asynchronous command. */
9522 case '!':
9523 /* If no asynchronous pids have been created, expand to nothing.
9524 If `set -u' has been executed, and no async processes have
9525 been created, this is an expansion error. */
9526 if (last_asynchronous_pid == NO_PID)
9527 {
9528 if (expanded_something)
9529 *expanded_something = 0;
9530 temp = (char *)NULL;
9531 if (unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
9532 {
9533 uerror[0] = '$';
9534 uerror[1] = c;
9535 uerror[2] = '\0';
9536 set_exit_status (EXECUTION_FAILURE);
9537 err_unboundvar (uerror);
9538 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9539 }
9540 }
9541 else
9542 temp = itos (last_asynchronous_pid);
9543 break;
9544
9545 /* The only difference between this and $@ is when the arg is quoted. */
9546 case '*': /* `$*' */
9547 list = list_rest_of_args ();
9548
9549 #if 0
9550 /* According to austin-group posix proposal by Geoff Clare in
9551 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
9552
9553 "The shell shall write a message to standard error and
9554 immediately exit when it tries to expand an unset parameter
9555 other than the '@' and '*' special parameters."
9556 */
9557
9558 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
9559 {
9560 uerror[0] = '$';
9561 uerror[1] = '*';
9562 uerror[2] = '\0';
9563 set_exit_status (EXECUTION_FAILURE);
9564 err_unboundvar (uerror);
9565 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9566 }
9567 #endif
9568
9569 /* If there are no command-line arguments, this should just
9570 disappear if there are other characters in the expansion,
9571 even if it's quoted. */
9572 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
9573 temp = (char *)NULL;
9574 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
9575 {
9576 /* If we have "$*" we want to make a string of the positional
9577 parameters, separated by the first character of $IFS, and
9578 quote the whole string, including the separators. If IFS
9579 is unset, the parameters are separated by ' '; if $IFS is
9580 null, the parameters are concatenated. */
9581 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list, quoted, 0) : string_list (list);
9582 if (temp)
9583 {
9584 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
9585 if (*temp == 0)
9586 tflag |= W_HASQUOTEDNULL;
9587 if (temp != temp1)
9588 free (temp);
9589 temp = temp1;
9590 }
9591 }
9592 else
9593 {
9594 /* We check whether or not we're eventually going to split $* here,
9595 for example when IFS is empty and we are processing the rhs of
9596 an assignment statement. In that case, we don't separate the
9597 arguments at all. Otherwise, if the $* is not quoted it is
9598 identical to $@ */
9599 if (expand_no_split_dollar_star && quoted == 0 && ifs_is_set == 0 && (pflags & PF_ASSIGNRHS))
9600 {
9601 /* Posix interp 888: RHS of assignment, IFS unset: no splitting,
9602 separate with space */
9603 temp1 = string_list_dollar_star (list, quoted, pflags);
9604 temp = temp1 ? quote_string (temp1) : temp1;
9605 /* XXX - tentative - note that we saw a quoted null here */
9606 if (temp1 && *temp1 == 0 && QUOTED_NULL (temp))
9607 tflag |= W_SAWQUOTEDNULL;
9608 FREE (temp1);
9609 }
9610 else if (expand_no_split_dollar_star && quoted == 0 && ifs_is_null && (pflags & PF_ASSIGNRHS))
9611 {
9612 /* Posix interp 888: RHS of assignment, IFS set to '' */
9613 temp1 = string_list_dollar_star (list, quoted, pflags);
9614 temp = temp1 ? quote_escapes (temp1) : temp1;
9615 FREE (temp1);
9616 }
9617 else if (expand_no_split_dollar_star && quoted == 0 && ifs_is_set && ifs_is_null == 0 && (pflags & PF_ASSIGNRHS))
9618 {
9619 /* Posix interp 888: RHS of assignment, IFS set to non-null value */
9620 temp1 = string_list_dollar_star (list, quoted, pflags);
9621 temp = temp1 ? quote_string (temp1) : temp1;
9622
9623 /* XXX - tentative - note that we saw a quoted null here */
9624 if (temp1 && *temp1 == 0 && QUOTED_NULL (temp))
9625 tflag |= W_SAWQUOTEDNULL;
9626 FREE (temp1);
9627 }
9628 /* XXX - should we check ifs_is_set here as well? */
9629 # if defined (HANDLE_MULTIBYTE)
9630 else if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
9631 # else
9632 else if (expand_no_split_dollar_star && ifs_firstc == 0)
9633 # endif
9634 /* Posix interp 888: not RHS, no splitting, IFS set to '' */
9635 temp = string_list_dollar_star (list, quoted, 0);
9636 else
9637 {
9638 temp = string_list_dollar_at (list, quoted, 0);
9639 /* Set W_SPLITSPACE to make sure the individual positional
9640 parameters are split into separate arguments */
9641 #if 0
9642 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
9643 #else /* change with bash-5.0 */
9644 if (quoted == 0 && ifs_is_null)
9645 #endif
9646 tflag |= W_SPLITSPACE;
9647 /* If we're not quoted but we still don't want word splitting, make
9648 we quote the IFS characters to protect them from splitting (e.g.,
9649 when $@ is in the string as well). */
9650 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
9651 {
9652 temp1 = quote_string (temp);
9653 free (temp);
9654 temp = temp1;
9655 }
9656 }
9657
9658 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
9659 *contains_dollar_at = 1;
9660 }
9661
9662 dispose_words (list);
9663 break;
9664
9665 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
9666 means that we have to turn quoting off after we split into
9667 the individually quoted arguments so that the final split
9668 on the first character of $IFS is still done. */
9669 case '@': /* `$@' */
9670 list = list_rest_of_args ();
9671
9672 #if 0
9673 /* According to austin-group posix proposal by Geoff Clare in
9674 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
9675
9676 "The shell shall write a message to standard error and
9677 immediately exit when it tries to expand an unset parameter
9678 other than the '@' and '*' special parameters."
9679 */
9680
9681 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
9682 {
9683 uerror[0] = '$';
9684 uerror[1] = '@';
9685 uerror[2] = '\0';
9686 set_exit_status (EXECUTION_FAILURE);
9687 err_unboundvar (uerror);
9688 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9689 }
9690 #endif
9691
9692 for (nullarg = 0, l = list; l; l = l->next)
9693 {
9694 if (l->word && (l->word->word == 0 || l->word->word[0] == 0))
9695 nullarg = 1;
9696 }
9697
9698 /* We want to flag the fact that we saw this. We can't turn
9699 off quoting entirely, because other characters in the
9700 string might need it (consider "\"$@\""), but we need some
9701 way to signal that the final split on the first character
9702 of $IFS should be done, even though QUOTED is 1. */
9703 /* XXX - should this test include Q_PATQUOTE? */
9704 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9705 *quoted_dollar_at_p = 1;
9706 if (contains_dollar_at)
9707 *contains_dollar_at = 1;
9708
9709 /* We want to separate the positional parameters with the first
9710 character of $IFS in case $IFS is something other than a space.
9711 We also want to make sure that splitting is done no matter what --
9712 according to POSIX.2, this expands to a list of the positional
9713 parameters no matter what IFS is set to. */
9714 /* XXX - what to do when in a context where word splitting is not
9715 performed? Even when IFS is not the default, posix seems to imply
9716 that we have to expand $@ to all the positional parameters and
9717 separate them with spaces, which are preserved because word splitting
9718 doesn't take place. See below for how we use PF_NOSPLIT2 here. */
9719
9720 /* These are the cases where word splitting will not be performed. */
9721 if (pflags & PF_ASSIGNRHS)
9722 {
9723 temp = string_list_dollar_at (list, (quoted|Q_DOUBLE_QUOTES), pflags);
9724 if (nullarg)
9725 tflag |= W_HASQUOTEDNULL; /* we know quoting produces quoted nulls */
9726 }
9727
9728 /* This needs to match what expand_word_internal does with non-quoted $@
9729 does with separating with spaces. Passing Q_DOUBLE_QUOTES means that
9730 the characters in LIST will be quoted, and PF_ASSIGNRHS ensures that
9731 they will separated by spaces. After doing this, we need the special
9732 handling for PF_NOSPLIT2 in expand_word_internal to remove the CTLESC
9733 quotes. */
9734 else if (pflags & PF_NOSPLIT2)
9735 {
9736 #if defined (HANDLE_MULTIBYTE)
9737 if (quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc[0] != ' ')
9738 #else
9739 if (quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc != ' ')
9740 #endif
9741 /* Posix interp 888 */
9742 temp = string_list_dollar_at (list, Q_DOUBLE_QUOTES, pflags);
9743 else
9744 temp = string_list_dollar_at (list, quoted, pflags);
9745 }
9746 else
9747 temp = string_list_dollar_at (list, quoted, pflags);
9748
9749 tflag |= W_DOLLARAT;
9750 dispose_words (list);
9751 break;
9752
9753 case LBRACE:
9754 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
9755 quoted_dollar_at_p,
9756 contains_dollar_at);
9757
9758 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9759 return (tdesc);
9760 temp = tdesc ? tdesc->word : (char *)0;
9761
9762 /* XXX */
9763 /* Quoted nulls should be removed if there is anything else
9764 in the string. */
9765 /* Note that we saw the quoted null so we can add one back at
9766 the end of this function if there are no other characters
9767 in the string, discard TEMP, and go on. The exception to
9768 this is when we have "${@}" and $1 is '', since $@ needs
9769 special handling. */
9770 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
9771 {
9772 if (had_quoted_null_p)
9773 *had_quoted_null_p = 1;
9774 if (*quoted_dollar_at_p == 0)
9775 {
9776 free (temp);
9777 tdesc->word = temp = (char *)NULL;
9778 }
9779
9780 }
9781
9782 ret = tdesc;
9783 goto return0;
9784
9785 /* Do command or arithmetic substitution. */
9786 case LPAREN:
9787 /* We have to extract the contents of this paren substitution. */
9788 t_index = zindex + 1;
9789 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
9790 as arithmetic substitution immediately. */
9791 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
9792 zindex = t_index;
9793
9794 /* For Posix.2-style `$(( ))' arithmetic substitution,
9795 extract the expression and pass it to the evaluator. */
9796 if (temp && *temp == LPAREN)
9797 {
9798 char *temp2;
9799 temp1 = temp + 1;
9800 temp2 = savestring (temp1);
9801 t_index = strlen (temp2) - 1;
9802
9803 if (temp2[t_index] != RPAREN)
9804 {
9805 free (temp2);
9806 goto comsub;
9807 }
9808
9809 /* Cut off ending `)' */
9810 temp2[t_index] = '\0';
9811
9812 if (chk_arithsub (temp2, t_index) == 0)
9813 {
9814 free (temp2);
9815 #if 0
9816 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
9817 #endif
9818 goto comsub;
9819 }
9820
9821 /* Expand variables found inside the expression. */
9822 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
9823 free (temp2);
9824
9825 arithsub:
9826 /* No error messages. */
9827 savecmd = this_command_name;
9828 this_command_name = (char *)NULL;
9829 number = evalexp (temp1, EXP_EXPANDED, &expok);
9830 this_command_name = savecmd;
9831 free (temp);
9832 free (temp1);
9833 if (expok == 0)
9834 {
9835 if (interactive_shell == 0 && posixly_correct)
9836 {
9837 set_exit_status (EXECUTION_FAILURE);
9838 return (&expand_wdesc_fatal);
9839 }
9840 else
9841 return (&expand_wdesc_error);
9842 }
9843 temp = itos (number);
9844 break;
9845 }
9846
9847 comsub:
9848 if (pflags & PF_NOCOMSUB)
9849 /* we need zindex+1 because string[zindex] == RPAREN */
9850 temp1 = substring (string, *sindex, zindex+1);
9851 else
9852 {
9853 tdesc = command_substitute (temp, quoted, pflags&PF_ASSIGNRHS);
9854 temp1 = tdesc ? tdesc->word : (char *)NULL;
9855 if (tdesc)
9856 dispose_word_desc (tdesc);
9857 }
9858 FREE (temp);
9859 temp = temp1;
9860 break;
9861
9862 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
9863 away in a future bash release. */
9864 case '[': /*]*/
9865 /* Extract the contents of this arithmetic substitution. */
9866 t_index = zindex + 1;
9867 temp = extract_arithmetic_subst (string, &t_index);
9868 zindex = t_index;
9869 if (temp == 0)
9870 {
9871 temp = savestring (string);
9872 if (expanded_something)
9873 *expanded_something = 0;
9874 goto return0;
9875 }
9876
9877 /* Do initial variable expansion. */
9878 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
9879
9880 goto arithsub;
9881
9882 default:
9883 /* Find the variable in VARIABLE_LIST. */
9884 temp = (char *)NULL;
9885
9886 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
9887 ;
9888 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
9889
9890 /* If this isn't a variable name, then just output the `$'. */
9891 if (temp1 == 0 || *temp1 == '\0')
9892 {
9893 FREE (temp1);
9894 temp = (char *)xmalloc (2);
9895 temp[0] = '$';
9896 temp[1] = '\0';
9897 if (expanded_something)
9898 *expanded_something = 0;
9899 goto return0;
9900 }
9901
9902 /* If the variable exists, return its value cell. */
9903 var = find_variable (temp1);
9904
9905 if (var && invisible_p (var) == 0 && var_isset (var))
9906 {
9907 #if defined (ARRAY_VARS)
9908 if (assoc_p (var) || array_p (var))
9909 {
9910 temp = array_p (var) ? array_reference (array_cell (var), 0)
9911 : assoc_reference (assoc_cell (var), "0");
9912 if (temp)
9913 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9914 ? quote_string (temp)
9915 : quote_escapes (temp);
9916 else if (unbound_vars_is_error)
9917 goto unbound_variable;
9918 }
9919 else
9920 #endif
9921 {
9922 temp = value_cell (var);
9923
9924 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9925 ? quote_string (temp)
9926 : ((pflags & PF_ASSIGNRHS) ? quote_rhs (temp)
9927 : quote_escapes (temp));
9928 }
9929
9930 free (temp1);
9931
9932 goto return0;
9933 }
9934 else if (var && (invisible_p (var) || var_isset (var) == 0))
9935 temp = (char *)NULL;
9936 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
9937 {
9938 temp = nameref_cell (var);
9939 #if defined (ARRAY_VARS)
9940 if (temp && *temp && valid_array_reference (temp, 0))
9941 {
9942 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
9943 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9944 return (tdesc);
9945 ret = tdesc;
9946 goto return0;
9947 }
9948 else
9949 #endif
9950 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
9951 if (temp && *temp && legal_identifier (temp) == 0)
9952 {
9953 set_exit_status (EXECUTION_FAILURE);
9954 report_error (_("%s: invalid variable name for name reference"), temp);
9955 return (&expand_wdesc_error); /* XXX */
9956 }
9957 else
9958 temp = (char *)NULL;
9959 }
9960
9961 temp = (char *)NULL;
9962
9963 unbound_variable:
9964 if (unbound_vars_is_error)
9965 {
9966 set_exit_status (EXECUTION_FAILURE);
9967 err_unboundvar (temp1);
9968 }
9969 else
9970 {
9971 free (temp1);
9972 goto return0;
9973 }
9974
9975 free (temp1);
9976 set_exit_status (EXECUTION_FAILURE);
9977 return ((unbound_vars_is_error && interactive_shell == 0)
9978 ? &expand_wdesc_fatal
9979 : &expand_wdesc_error);
9980 }
9981
9982 if (string[zindex])
9983 zindex++;
9984
9985 return0:
9986 *sindex = zindex;
9987
9988 if (ret == 0)
9989 {
9990 ret = alloc_word_desc ();
9991 ret->flags = tflag; /* XXX */
9992 ret->word = temp;
9993 }
9994 return ret;
9995 }
9996
9997 void
9998 invalidate_cached_quoted_dollar_at ()
9999 {
10000 dispose_words (cached_quoted_dollar_at);
10001 cached_quoted_dollar_at = 0;
10002 }
10003
10004 /* Make a word list which is the result of parameter and variable
10005 expansion, command substitution, arithmetic substitution, and
10006 quote removal of WORD. Return a pointer to a WORD_LIST which is
10007 the result of the expansion. If WORD contains a null word, the
10008 word list returned is also null.
10009
10010 QUOTED contains flag values defined in shell.h.
10011
10012 ISEXP is used to tell expand_word_internal that the word should be
10013 treated as the result of an expansion. This has implications for
10014 how IFS characters in the word are treated.
10015
10016 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
10017 they point to an integer value which receives information about expansion.
10018 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
10019 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
10020 else zero.
10021
10022 This only does word splitting in the case of $@ expansion. In that
10023 case, we split on ' '. */
10024
10025 /* Values for the local variable quoted_state. */
10026 #define UNQUOTED 0
10027 #define PARTIALLY_QUOTED 1
10028 #define WHOLLY_QUOTED 2
10029
10030 static WORD_LIST *
10031 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
10032 WORD_DESC *word;
10033 int quoted, isexp;
10034 int *contains_dollar_at;
10035 int *expanded_something;
10036 {
10037 WORD_LIST *list;
10038 WORD_DESC *tword;
10039
10040 /* The intermediate string that we build while expanding. */
10041 char *istring;
10042
10043 /* The current size of the above object. */
10044 size_t istring_size;
10045
10046 /* Index into ISTRING. */
10047 int istring_index;
10048
10049 /* Temporary string storage. */
10050 char *temp, *temp1;
10051
10052 /* The text of WORD. */
10053 register char *string;
10054
10055 /* The size of STRING. */
10056 size_t string_size;
10057
10058 /* The index into STRING. */
10059 int sindex;
10060
10061 /* This gets 1 if we see a $@ while quoted. */
10062 int quoted_dollar_at;
10063
10064 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
10065 whether WORD contains no quoting characters, a partially quoted
10066 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
10067 int quoted_state;
10068
10069 /* State flags */
10070 int had_quoted_null;
10071 int has_quoted_ifs; /* did we add a quoted $IFS character here? */
10072 int has_dollar_at, temp_has_dollar_at;
10073 int split_on_spaces;
10074 int local_expanded;
10075 int tflag;
10076 int pflags; /* flags passed to param_expand */
10077 int mb_cur_max;
10078
10079 int assignoff; /* If assignment, offset of `=' */
10080
10081 register unsigned char c; /* Current character. */
10082 int t_index; /* For calls to string_extract_xxx. */
10083
10084 char twochars[2];
10085
10086 DECLARE_MBSTATE;
10087
10088 /* OK, let's see if we can optimize a common idiom: "$@" */
10089 if (STREQ (word->word, "\"$@\"") &&
10090 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
10091 dollar_vars[1]) /* XXX - check IFS here as well? */
10092 {
10093 if (contains_dollar_at)
10094 *contains_dollar_at = 1;
10095 if (expanded_something)
10096 *expanded_something = 1;
10097 if (cached_quoted_dollar_at)
10098 return (copy_word_list (cached_quoted_dollar_at));
10099 list = list_rest_of_args ();
10100 list = quote_list (list);
10101 cached_quoted_dollar_at = copy_word_list (list);
10102 return (list);
10103 }
10104
10105 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
10106 istring[istring_index = 0] = '\0';
10107 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
10108 has_quoted_ifs = 0;
10109 split_on_spaces = 0;
10110 quoted_state = UNQUOTED;
10111
10112 string = word->word;
10113 if (string == 0)
10114 goto finished_with_string;
10115 mb_cur_max = MB_CUR_MAX;
10116
10117 /* Don't need the string length for the SADD... and COPY_ macros unless
10118 multibyte characters are possible, but do need it for bounds checking. */
10119 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
10120
10121 if (contains_dollar_at)
10122 *contains_dollar_at = 0;
10123
10124 assignoff = -1;
10125
10126 /* Begin the expansion. */
10127
10128 for (sindex = 0; ;)
10129 {
10130 c = string[sindex];
10131
10132 /* Case on top-level character. */
10133 switch (c)
10134 {
10135 case '\0':
10136 goto finished_with_string;
10137
10138 case CTLESC:
10139 sindex++;
10140 #if HANDLE_MULTIBYTE
10141 if (mb_cur_max > 1 && string[sindex])
10142 {
10143 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
10144 }
10145 else
10146 #endif
10147 {
10148 temp = (char *)xmalloc (3);
10149 temp[0] = CTLESC;
10150 temp[1] = c = string[sindex];
10151 temp[2] = '\0';
10152 }
10153
10154 dollar_add_string:
10155 if (string[sindex])
10156 sindex++;
10157
10158 add_string:
10159 if (temp)
10160 {
10161 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
10162 temp = (char *)0;
10163 }
10164
10165 break;
10166
10167 #if defined (PROCESS_SUBSTITUTION)
10168 /* Process substitution. */
10169 case '<':
10170 case '>':
10171 {
10172 /* XXX - technically this should only be expanded at the start
10173 of a word */
10174 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)))
10175 {
10176 sindex--; /* add_character: label increments sindex */
10177 goto add_character;
10178 }
10179 else
10180 t_index = sindex + 1; /* skip past both '<' and LPAREN */
10181
10182 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
10183 sindex = t_index;
10184
10185 /* If the process substitution specification is `<()', we want to
10186 open the pipe for writing in the child and produce output; if
10187 it is `>()', we want to open the pipe for reading in the child
10188 and consume input. */
10189 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
10190
10191 FREE (temp1);
10192
10193 goto dollar_add_string;
10194 }
10195 #endif /* PROCESS_SUBSTITUTION */
10196
10197 case '=':
10198 /* Posix.2 section 3.6.1 says that tildes following `=' in words
10199 which are not assignment statements are not expanded. If the
10200 shell isn't in posix mode, though, we perform tilde expansion
10201 on `likely candidate' unquoted assignment statements (flags
10202 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
10203 contains an unquoted :~ or =~. Something to think about: we
10204 now have a flag that says to perform tilde expansion on arguments
10205 to `assignment builtins' like declare and export that look like
10206 assignment statements. We now do tilde expansion on such words
10207 even in POSIX mode. */
10208 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
10209 {
10210 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10211 goto add_ifs_character;
10212 else
10213 goto add_character;
10214 }
10215 /* If we're not in posix mode or forcing assignment-statement tilde
10216 expansion, note where the first `=' appears in the word and prepare
10217 to do tilde expansion following the first `='. We have to keep
10218 track of the first `=' (using assignoff) to avoid being confused
10219 by an `=' in the rhs of the assignment statement. */
10220 if ((word->flags & W_ASSIGNMENT) &&
10221 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
10222 assignoff == -1 && sindex > 0)
10223 assignoff = sindex;
10224 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
10225 word->flags |= W_ITILDE;
10226
10227 if (word->flags & W_ASSIGNARG)
10228 word->flags |= W_ASSIGNRHS; /* affects $@ */
10229
10230 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10231 {
10232 has_quoted_ifs++;
10233 goto add_ifs_character;
10234 }
10235 else
10236 goto add_character;
10237
10238 case ':':
10239 if (word->flags & (W_NOTILDE|W_NOASSNTILDE))
10240 {
10241 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10242 goto add_ifs_character;
10243 else
10244 goto add_character;
10245 }
10246
10247 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS)) &&
10248 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
10249 string[sindex+1] == '~')
10250 word->flags |= W_ITILDE;
10251
10252 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10253 goto add_ifs_character;
10254 else
10255 goto add_character;
10256
10257 case '~':
10258 /* If the word isn't supposed to be tilde expanded, or we're not
10259 at the start of a word or after an unquoted : or = in an
10260 assignment statement, we don't do tilde expansion. We don't
10261 do tilde expansion if quoted or in an arithmetic context. */
10262
10263 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
10264 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
10265 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
10266 {
10267 word->flags &= ~W_ITILDE;
10268 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
10269 goto add_ifs_character;
10270 else
10271 goto add_character;
10272 }
10273
10274 if (word->flags & W_ASSIGNRHS)
10275 tflag = 2;
10276 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
10277 tflag = 1;
10278 else
10279 tflag = 0;
10280
10281 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
10282
10283 word->flags &= ~W_ITILDE;
10284
10285 if (temp && *temp && t_index > 0)
10286 {
10287 temp1 = bash_tilde_expand (temp, tflag);
10288 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
10289 {
10290 FREE (temp);
10291 FREE (temp1);
10292 goto add_character; /* tilde expansion failed */
10293 }
10294 free (temp);
10295 temp = temp1;
10296 sindex += t_index;
10297 goto add_quoted_string; /* XXX was add_string */
10298 }
10299 else
10300 {
10301 FREE (temp);
10302 goto add_character;
10303 }
10304
10305 case '$':
10306 if (expanded_something)
10307 *expanded_something = 1;
10308 local_expanded = 1;
10309
10310 temp_has_dollar_at = 0;
10311 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
10312 if (word->flags & W_NOSPLIT2)
10313 pflags |= PF_NOSPLIT2;
10314 if (word->flags & W_ASSIGNRHS)
10315 pflags |= PF_ASSIGNRHS;
10316 if (word->flags & W_COMPLETE)
10317 pflags |= PF_COMPLETE;
10318
10319 tword = param_expand (string, &sindex, quoted, expanded_something,
10320 &temp_has_dollar_at, &quoted_dollar_at,
10321 &had_quoted_null, pflags);
10322 has_dollar_at += temp_has_dollar_at;
10323 split_on_spaces += (tword->flags & W_SPLITSPACE);
10324
10325 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
10326 {
10327 free (string);
10328 free (istring);
10329 return ((tword == &expand_wdesc_error) ? &expand_word_error
10330 : &expand_word_fatal);
10331 }
10332 if (contains_dollar_at && has_dollar_at)
10333 *contains_dollar_at = 1;
10334
10335 if (tword && (tword->flags & W_HASQUOTEDNULL))
10336 had_quoted_null = 1; /* note for later */
10337 if (tword && (tword->flags & W_SAWQUOTEDNULL))
10338 had_quoted_null = 1; /* XXX */
10339
10340 temp = tword ? tword->word : (char *)NULL;
10341 dispose_word_desc (tword);
10342
10343 /* Kill quoted nulls; we will add them back at the end of
10344 expand_word_internal if nothing else in the string */
10345 if (had_quoted_null && temp && QUOTED_NULL (temp))
10346 {
10347 FREE (temp);
10348 temp = (char *)NULL;
10349 }
10350
10351 goto add_string;
10352 break;
10353
10354 case '`': /* Backquoted command substitution. */
10355 {
10356 t_index = sindex++;
10357
10358 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
10359 /* The test of sindex against t_index is to allow bare instances of
10360 ` to pass through, for backwards compatibility. */
10361 if (temp == &extract_string_error || temp == &extract_string_fatal)
10362 {
10363 if (sindex - 1 == t_index)
10364 {
10365 sindex = t_index;
10366 goto add_character;
10367 }
10368 set_exit_status (EXECUTION_FAILURE);
10369 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
10370 free (string);
10371 free (istring);
10372 return ((temp == &extract_string_error) ? &expand_word_error
10373 : &expand_word_fatal);
10374 }
10375
10376 if (expanded_something)
10377 *expanded_something = 1;
10378 local_expanded = 1;
10379
10380 if (word->flags & W_NOCOMSUB)
10381 /* sindex + 1 because string[sindex] == '`' */
10382 temp1 = substring (string, t_index, sindex + 1);
10383 else
10384 {
10385 de_backslash (temp);
10386 tword = command_substitute (temp, quoted, 0);
10387 temp1 = tword ? tword->word : (char *)NULL;
10388 if (tword)
10389 dispose_word_desc (tword);
10390 }
10391 FREE (temp);
10392 temp = temp1;
10393 goto dollar_add_string;
10394 }
10395
10396 case '\\':
10397 if (string[sindex + 1] == '\n')
10398 {
10399 sindex += 2;
10400 continue;
10401 }
10402
10403 c = string[++sindex];
10404
10405 /* "However, the double-quote character ( '"' ) shall not be treated
10406 specially within a here-document, except when the double-quote
10407 appears within "$()", "``", or "${}"." */
10408 if ((quoted & Q_HERE_DOCUMENT) && (quoted & Q_DOLBRACE) && c == '"')
10409 tflag = CBSDQUOTE; /* special case */
10410 else if (quoted & Q_HERE_DOCUMENT)
10411 tflag = CBSHDOC;
10412 else if (quoted & Q_DOUBLE_QUOTES)
10413 tflag = CBSDQUOTE;
10414 else
10415 tflag = 0;
10416
10417 /* From Posix discussion on austin-group list: Backslash escaping
10418 a } in ${...} is removed. Issue 0000221 */
10419 if ((quoted & Q_DOLBRACE) && c == RBRACE)
10420 {
10421 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10422 }
10423 /* This is the fix for " $@\ " */
10424 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
10425 {
10426 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10427 DEFAULT_ARRAY_SIZE);
10428 istring[istring_index++] = CTLESC;
10429 istring[istring_index++] = '\\';
10430 istring[istring_index] = '\0';
10431
10432 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10433 }
10434 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && c == 0)
10435 {
10436 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10437 DEFAULT_ARRAY_SIZE);
10438 istring[istring_index++] = CTLESC;
10439 istring[istring_index++] = '\\';
10440 istring[istring_index] = '\0';
10441 break;
10442 }
10443 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
10444 {
10445 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
10446 }
10447 else if (c == 0)
10448 {
10449 c = CTLNUL;
10450 sindex--; /* add_character: label increments sindex */
10451 goto add_character;
10452 }
10453 else
10454 {
10455 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10456 }
10457
10458 sindex++;
10459 add_twochars:
10460 /* BEFORE jumping here, we need to increment sindex if appropriate */
10461 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10462 DEFAULT_ARRAY_SIZE);
10463 istring[istring_index++] = twochars[0];
10464 istring[istring_index++] = twochars[1];
10465 istring[istring_index] = '\0';
10466
10467 break;
10468
10469 case '"':
10470 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
10471 goto add_character;
10472
10473 t_index = ++sindex;
10474 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
10475
10476 /* If the quotes surrounded the entire string, then the
10477 whole word was quoted. */
10478 quoted_state = (t_index == 1 && string[sindex] == '\0')
10479 ? WHOLLY_QUOTED
10480 : PARTIALLY_QUOTED;
10481
10482 if (temp && *temp)
10483 {
10484 tword = alloc_word_desc ();
10485 tword->word = temp;
10486
10487 if (word->flags & W_ASSIGNARG)
10488 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
10489 if (word->flags & W_COMPLETE)
10490 tword->flags |= W_COMPLETE; /* for command substitutions */
10491 if (word->flags & W_NOCOMSUB)
10492 tword->flags |= W_NOCOMSUB;
10493 if (word->flags & W_NOPROCSUB)
10494 tword->flags |= W_NOPROCSUB;
10495
10496 if (word->flags & W_ASSIGNRHS)
10497 tword->flags |= W_ASSIGNRHS;
10498
10499 temp = (char *)NULL;
10500
10501 temp_has_dollar_at = 0; /* does this quoted (sub)string include $@? */
10502 /* Need to get W_HASQUOTEDNULL flag through this function. */
10503 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
10504 has_dollar_at += temp_has_dollar_at;
10505
10506 if (list == &expand_word_error || list == &expand_word_fatal)
10507 {
10508 free (istring);
10509 free (string);
10510 /* expand_word_internal has already freed temp_word->word
10511 for us because of the way it prints error messages. */
10512 tword->word = (char *)NULL;
10513 dispose_word (tword);
10514 return list;
10515 }
10516
10517 dispose_word (tword);
10518
10519 /* "$@" (a double-quoted dollar-at) expands into nothing,
10520 not even a NULL word, when there are no positional
10521 parameters. Posix interp 888 says that other parts of the
10522 word that expand to quoted nulls result in quoted nulls, so
10523 we can't just throw the entire word away if we have "$@"
10524 anywhere in it. We use had_quoted_null to keep track */
10525 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
10526 {
10527 quoted_dollar_at++;
10528 break;
10529 }
10530
10531 /* If this list comes back with a quoted null from expansion,
10532 we have either "$x" or "$@" with $1 == ''. In either case,
10533 we need to make sure we add a quoted null argument and
10534 disable the special handling that "$@" gets. */
10535 if (list && list->word && list->next == 0 && (list->word->flags & W_HASQUOTEDNULL))
10536 {
10537 if (had_quoted_null && temp_has_dollar_at)
10538 quoted_dollar_at++;
10539 had_quoted_null = 1; /* XXX */
10540 }
10541
10542 /* If we get "$@", we know we have expanded something, so we
10543 need to remember it for the final split on $IFS. This is
10544 a special case; it's the only case where a quoted string
10545 can expand into more than one word. It's going to come back
10546 from the above call to expand_word_internal as a list with
10547 multiple words. */
10548 if (list)
10549 dequote_list (list);
10550
10551 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
10552 {
10553 quoted_dollar_at++;
10554 if (contains_dollar_at)
10555 *contains_dollar_at = 1;
10556 if (expanded_something)
10557 *expanded_something = 1;
10558 local_expanded = 1;
10559 }
10560 }
10561 else
10562 {
10563 /* What we have is "". This is a minor optimization. */
10564 FREE (temp);
10565 list = (WORD_LIST *)NULL;
10566 had_quoted_null = 1; /* note for later */
10567 }
10568
10569 /* The code above *might* return a list (consider the case of "$@",
10570 where it returns "$1", "$2", etc.). We can't throw away the
10571 rest of the list, and we have to make sure each word gets added
10572 as quoted. We test on tresult->next: if it is non-NULL, we
10573 quote the whole list, save it to a string with string_list, and
10574 add that string. We don't need to quote the results of this
10575 (and it would be wrong, since that would quote the separators
10576 as well), so we go directly to add_string. */
10577 if (list)
10578 {
10579 if (list->next)
10580 {
10581 /* Testing quoted_dollar_at makes sure that "$@" is
10582 split correctly when $IFS does not contain a space. */
10583 temp = quoted_dollar_at
10584 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
10585 : string_list (quote_list (list));
10586 dispose_words (list);
10587 goto add_string;
10588 }
10589 else
10590 {
10591 temp = savestring (list->word->word);
10592 tflag = list->word->flags;
10593 dispose_words (list);
10594
10595 /* If the string is not a quoted null string, we want
10596 to remove any embedded unquoted CTLNUL characters.
10597 We do not want to turn quoted null strings back into
10598 the empty string, though. We do this because we
10599 want to remove any quoted nulls from expansions that
10600 contain other characters. For example, if we have
10601 x"$*"y or "x$*y" and there are no positional parameters,
10602 the $* should expand into nothing. */
10603 /* We use the W_HASQUOTEDNULL flag to differentiate the
10604 cases: a quoted null character as above and when
10605 CTLNUL is contained in the (non-null) expansion
10606 of some variable. We use the had_quoted_null flag to
10607 pass the value through this function to its caller. */
10608 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
10609 remove_quoted_nulls (temp); /* XXX */
10610 }
10611 }
10612 else
10613 temp = (char *)NULL;
10614
10615 if (temp == 0 && quoted_state == PARTIALLY_QUOTED)
10616 had_quoted_null = 1; /* note for later */
10617
10618 /* We do not want to add quoted nulls to strings that are only
10619 partially quoted; we can throw them away. The exception to
10620 this is when we are going to be performing word splitting,
10621 since we have to preserve a null argument if the next character
10622 will cause word splitting. */
10623 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && quoted == 0 && (word->flags & (W_NOSPLIT|W_EXPANDRHS|W_ASSIGNRHS)) == W_EXPANDRHS)
10624 {
10625 c = CTLNUL;
10626 sindex--;
10627 had_quoted_null = 1;
10628 goto add_character;
10629 }
10630 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
10631 continue;
10632
10633 add_quoted_string:
10634
10635 if (temp)
10636 {
10637 temp1 = temp;
10638 temp = quote_string (temp);
10639 free (temp1);
10640 goto add_string;
10641 }
10642 else
10643 {
10644 /* Add NULL arg. */
10645 c = CTLNUL;
10646 sindex--; /* add_character: label increments sindex */
10647 had_quoted_null = 1; /* note for later */
10648 goto add_character;
10649 }
10650
10651 /* break; */
10652
10653 case '\'':
10654 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
10655 goto add_character;
10656
10657 t_index = ++sindex;
10658 temp = string_extract_single_quoted (string, &sindex);
10659
10660 /* If the entire STRING was surrounded by single quotes,
10661 then the string is wholly quoted. */
10662 quoted_state = (t_index == 1 && string[sindex] == '\0')
10663 ? WHOLLY_QUOTED
10664 : PARTIALLY_QUOTED;
10665
10666 /* If all we had was '', it is a null expansion. */
10667 if (*temp == '\0')
10668 {
10669 free (temp);
10670 temp = (char *)NULL;
10671 }
10672 else
10673 remove_quoted_escapes (temp); /* ??? */
10674
10675 if (temp == 0 && quoted_state == PARTIALLY_QUOTED)
10676 had_quoted_null = 1; /* note for later */
10677
10678 /* We do not want to add quoted nulls to strings that are only
10679 partially quoted; such nulls are discarded. See above for the
10680 exception, which is when the string is going to be split.
10681 Posix interp 888/1129 */
10682 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && quoted == 0 && (word->flags & (W_NOSPLIT|W_EXPANDRHS|W_ASSIGNRHS)) == W_EXPANDRHS)
10683 {
10684 c = CTLNUL;
10685 sindex--;
10686 goto add_character;
10687 }
10688
10689 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
10690 continue;
10691
10692 /* If we have a quoted null expansion, add a quoted NULL to istring. */
10693 if (temp == 0)
10694 {
10695 c = CTLNUL;
10696 sindex--; /* add_character: label increments sindex */
10697 goto add_character;
10698 }
10699 else
10700 goto add_quoted_string;
10701
10702 /* break; */
10703
10704 case ' ':
10705 /* If we are in a context where the word is not going to be split, but
10706 we need to account for $@ and $* producing one word for each
10707 positional parameter, add quoted spaces so the spaces in the
10708 expansion of "$@", if any, behave correctly. We still may need to
10709 split if we are expanding the rhs of a word expansion. */
10710 if (ifs_is_null || split_on_spaces || ((word->flags & (W_NOSPLIT|W_NOSPLIT2|W_ASSIGNRHS)) && (word->flags & W_EXPANDRHS) == 0))
10711 {
10712 if (string[sindex])
10713 sindex++;
10714 twochars[0] = CTLESC;
10715 twochars[1] = c;
10716 goto add_twochars;
10717 }
10718 /* FALLTHROUGH */
10719
10720 default:
10721 /* This is the fix for " $@ " */
10722 add_ifs_character:
10723 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
10724 {
10725 if ((quoted&(Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)
10726 has_quoted_ifs++;
10727 add_quoted_character:
10728 if (string[sindex]) /* from old goto dollar_add_string */
10729 sindex++;
10730 if (c == 0)
10731 {
10732 c = CTLNUL;
10733 goto add_character;
10734 }
10735 else
10736 {
10737 #if HANDLE_MULTIBYTE
10738 /* XXX - should make sure that c is actually multibyte,
10739 otherwise we can use the twochars branch */
10740 if (mb_cur_max > 1)
10741 sindex--;
10742
10743 if (mb_cur_max > 1)
10744 {
10745 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
10746 }
10747 else
10748 #endif
10749 {
10750 twochars[0] = CTLESC;
10751 twochars[1] = c;
10752 goto add_twochars;
10753 }
10754 }
10755 }
10756
10757 SADD_MBCHAR (temp, string, sindex, string_size);
10758
10759 add_character:
10760 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
10761 DEFAULT_ARRAY_SIZE);
10762 istring[istring_index++] = c;
10763 istring[istring_index] = '\0';
10764
10765 /* Next character. */
10766 sindex++;
10767 }
10768 }
10769
10770 finished_with_string:
10771 /* OK, we're ready to return. If we have a quoted string, and
10772 quoted_dollar_at is not set, we do no splitting at all; otherwise
10773 we split on ' '. The routines that call this will handle what to
10774 do if nothing has been expanded. */
10775
10776 /* Partially and wholly quoted strings which expand to the empty
10777 string are retained as an empty arguments. Unquoted strings
10778 which expand to the empty string are discarded. The single
10779 exception is the case of expanding "$@" when there are no
10780 positional parameters. In that case, we discard the expansion. */
10781
10782 /* Because of how the code that handles "" and '' in partially
10783 quoted strings works, we need to make ISTRING into a QUOTED_NULL
10784 if we saw quoting characters, but the expansion was empty.
10785 "" and '' are tossed away before we get to this point when
10786 processing partially quoted strings. This makes "" and $xxx""
10787 equivalent when xxx is unset. We also look to see whether we
10788 saw a quoted null from a ${} expansion and add one back if we
10789 need to. */
10790
10791 /* If we expand to nothing and there were no single or double quotes
10792 in the word, we throw it away. Otherwise, we return a NULL word.
10793 The single exception is for $@ surrounded by double quotes when
10794 there are no positional parameters. In that case, we also throw
10795 the word away. */
10796
10797 if (*istring == '\0')
10798 {
10799 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
10800 {
10801 istring[0] = CTLNUL;
10802 istring[1] = '\0';
10803 tword = alloc_word_desc ();
10804 tword->word = istring;
10805 istring = 0; /* avoid later free() */
10806 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10807 list = make_word_list (tword, (WORD_LIST *)NULL);
10808 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
10809 tword->flags |= W_QUOTED;
10810 }
10811 /* According to sh, ksh, and Posix.2, if a word expands into nothing
10812 and a double-quoted "$@" appears anywhere in it, then the entire
10813 word is removed. */
10814 /* XXX - exception appears to be that quoted null strings result in
10815 null arguments */
10816 else if (quoted_state == UNQUOTED || quoted_dollar_at)
10817 list = (WORD_LIST *)NULL;
10818 else
10819 list = (WORD_LIST *)NULL;
10820 }
10821 else if (word->flags & W_NOSPLIT)
10822 {
10823 tword = alloc_word_desc ();
10824 tword->word = istring;
10825 if (had_quoted_null && QUOTED_NULL (istring))
10826 tword->flags |= W_HASQUOTEDNULL;
10827 istring = 0; /* avoid later free() */
10828 if (word->flags & W_ASSIGNMENT)
10829 tword->flags |= W_ASSIGNMENT; /* XXX */
10830 if (word->flags & W_COMPASSIGN)
10831 tword->flags |= W_COMPASSIGN; /* XXX */
10832 if (word->flags & W_NOGLOB)
10833 tword->flags |= W_NOGLOB; /* XXX */
10834 if (word->flags & W_NOBRACE)
10835 tword->flags |= W_NOBRACE; /* XXX */
10836 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
10837 tword->flags |= W_QUOTED;
10838 list = make_word_list (tword, (WORD_LIST *)NULL);
10839 }
10840 else if (word->flags & W_ASSIGNRHS)
10841 {
10842 list = list_string (istring, "", quoted);
10843 tword = list->word;
10844 if (had_quoted_null && QUOTED_NULL (istring))
10845 tword->flags |= W_HASQUOTEDNULL;
10846 free (list);
10847 free (istring);
10848 istring = 0; /* avoid later free() */
10849 goto set_word_flags;
10850 }
10851 else
10852 {
10853 char *ifs_chars;
10854
10855 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
10856
10857 /* If we have $@, we need to split the results no matter what. If
10858 IFS is unset or NULL, string_list_dollar_at has separated the
10859 positional parameters with a space, so we split on space (we have
10860 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
10861 string_list_dollar_at has separated the positional parameters
10862 with the first character of $IFS, so we split on $IFS. If
10863 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
10864 unset or null, and we want to make sure that we split on spaces
10865 regardless of what else has happened to IFS since the expansion,
10866 or we expanded "$@" with IFS null and we need to split the positional
10867 parameters into separate words. */
10868 if (split_on_spaces)
10869 {
10870 /* If IFS is not set, and the word is not quoted, we want to split
10871 the individual words on $' \t\n'. We rely on previous steps to
10872 quote the portions of the word that should not be split */
10873 if (ifs_is_set == 0)
10874 list = list_string (istring, " \t\n", 1); /* XXX quoted == 1? */
10875 else
10876 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
10877 }
10878
10879 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
10880 don't want to split the result (W_NOSPLIT2), and we are not quoted,
10881 we have already separated the arguments with the first character of
10882 $IFS. In this case, we want to return a list with a single word
10883 with the separator possibly replaced with a space (it's what other
10884 shells seem to do).
10885 quoted_dollar_at is internal to this function and is set if we are
10886 passed an argument that is unquoted (quoted == 0) but we encounter a
10887 double-quoted $@ while expanding it. */
10888 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
10889 {
10890 tword = alloc_word_desc ();
10891 /* Only split and rejoin if we have to */
10892 if (*ifs_chars && *ifs_chars != ' ')
10893 {
10894 /* list_string dequotes CTLESCs in the string it's passed, so we
10895 need it to get the space separation right if space isn't the
10896 first character in IFS (but is present) and to remove the
10897 quoting we added back in param_expand(). */
10898 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
10899 /* This isn't exactly right in the case where we're expanding
10900 the RHS of an expansion like ${var-$@} where IFS=: (for
10901 example). The W_NOSPLIT2 means we do the separation with :;
10902 the list_string removes the quotes and breaks the string into
10903 a list, and the string_list rejoins it on spaces. When we
10904 return, we expect to be able to split the results, but the
10905 space separation means the right split doesn't happen. */
10906 tword->word = string_list (list);
10907 }
10908 else
10909 tword->word = istring;
10910 if (had_quoted_null && QUOTED_NULL (istring))
10911 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10912 if (tword->word != istring)
10913 free (istring);
10914 istring = 0; /* avoid later free() */
10915 goto set_word_flags;
10916 }
10917 else if (has_dollar_at && ifs_chars)
10918 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
10919 else
10920 {
10921 tword = alloc_word_desc ();
10922 if (expanded_something && *expanded_something == 0 && has_quoted_ifs)
10923 tword->word = remove_quoted_ifs (istring);
10924 else
10925 tword->word = istring;
10926 if (had_quoted_null && QUOTED_NULL (istring)) /* should check for more than one */
10927 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10928 else if (had_quoted_null)
10929 tword->flags |= W_SAWQUOTEDNULL; /* XXX */
10930 if (tword->word != istring)
10931 free (istring);
10932 istring = 0; /* avoid later free() */
10933 set_word_flags:
10934 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
10935 tword->flags |= W_QUOTED;
10936 if (word->flags & W_ASSIGNMENT)
10937 tword->flags |= W_ASSIGNMENT;
10938 if (word->flags & W_COMPASSIGN)
10939 tword->flags |= W_COMPASSIGN;
10940 if (word->flags & W_NOGLOB)
10941 tword->flags |= W_NOGLOB;
10942 if (word->flags & W_NOBRACE)
10943 tword->flags |= W_NOBRACE;
10944 list = make_word_list (tword, (WORD_LIST *)NULL);
10945 }
10946 }
10947
10948 free (istring);
10949 return (list);
10950 }
10951
10952 /* **************************************************************** */
10953 /* */
10954 /* Functions for Quote Removal */
10955 /* */
10956 /* **************************************************************** */
10957
10958 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
10959 backslash quoting rules for within double quotes or a here document. */
10960 char *
10961 string_quote_removal (string, quoted)
10962 char *string;
10963 int quoted;
10964 {
10965 size_t slen;
10966 char *r, *result_string, *temp, *send;
10967 int sindex, tindex, dquote;
10968 unsigned char c;
10969 DECLARE_MBSTATE;
10970
10971 /* The result can be no longer than the original string. */
10972 slen = strlen (string);
10973 send = string + slen;
10974
10975 r = result_string = (char *)xmalloc (slen + 1);
10976
10977 for (dquote = sindex = 0; c = string[sindex];)
10978 {
10979 switch (c)
10980 {
10981 case '\\':
10982 c = string[++sindex];
10983 if (c == 0)
10984 {
10985 *r++ = '\\';
10986 break;
10987 }
10988 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
10989 *r++ = '\\';
10990 /* FALLTHROUGH */
10991
10992 default:
10993 SCOPY_CHAR_M (r, string, send, sindex);
10994 break;
10995
10996 case '\'':
10997 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
10998 {
10999 *r++ = c;
11000 sindex++;
11001 break;
11002 }
11003 tindex = sindex + 1;
11004 temp = string_extract_single_quoted (string, &tindex);
11005 if (temp)
11006 {
11007 strcpy (r, temp);
11008 r += strlen (r);
11009 free (temp);
11010 }
11011 sindex = tindex;
11012 break;
11013
11014 case '"':
11015 dquote = 1 - dquote;
11016 sindex++;
11017 break;
11018 }
11019 }
11020 *r = '\0';
11021 return (result_string);
11022 }
11023
11024 #if 0
11025 /* UNUSED */
11026 /* Perform quote removal on word WORD. This allocates and returns a new
11027 WORD_DESC *. */
11028 WORD_DESC *
11029 word_quote_removal (word, quoted)
11030 WORD_DESC *word;
11031 int quoted;
11032 {
11033 WORD_DESC *w;
11034 char *t;
11035
11036 t = string_quote_removal (word->word, quoted);
11037 w = alloc_word_desc ();
11038 w->word = t ? t : savestring ("");
11039 return (w);
11040 }
11041
11042 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
11043 the members of the list are treated as if they are surrounded by
11044 double quotes. Return a new list, or NULL if LIST is NULL. */
11045 WORD_LIST *
11046 word_list_quote_removal (list, quoted)
11047 WORD_LIST *list;
11048 int quoted;
11049 {
11050 WORD_LIST *result, *t, *tresult, *e;
11051
11052 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
11053 {
11054 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
11055 #if 0
11056 result = (WORD_LIST *) list_append (result, tresult);
11057 #else
11058 if (result == 0)
11059 result = e = tresult;
11060 else
11061 {
11062 e->next = tresult;
11063 while (e->next)
11064 e = e->next;
11065 }
11066 #endif
11067 }
11068 return (result);
11069 }
11070 #endif
11071
11072 /*******************************************
11073 * *
11074 * Functions to perform word splitting *
11075 * *
11076 *******************************************/
11077
11078 void
11079 setifs (v)
11080 SHELL_VAR *v;
11081 {
11082 char *t;
11083 unsigned char uc;
11084
11085 ifs_var = v;
11086 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
11087
11088 ifs_is_set = ifs_var != 0;
11089 ifs_is_null = ifs_is_set && (*ifs_value == 0);
11090
11091 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
11092 handle multibyte chars in IFS */
11093 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
11094 for (t = ifs_value ; t && *t; t++)
11095 {
11096 uc = *t;
11097 ifs_cmap[uc] = 1;
11098 }
11099
11100 #if defined (HANDLE_MULTIBYTE)
11101 if (ifs_value == 0)
11102 {
11103 ifs_firstc[0] = '\0'; /* XXX - ? */
11104 ifs_firstc_len = 1;
11105 }
11106 else
11107 {
11108 if (locale_utf8locale && UTF8_SINGLEBYTE (*ifs_value))
11109 ifs_firstc_len = (*ifs_value != 0) ? 1 : 0;
11110 else
11111 {
11112 size_t ifs_len;
11113 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
11114 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
11115 }
11116 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
11117 {
11118 ifs_firstc[0] = ifs_value[0];
11119 ifs_firstc[1] = '\0';
11120 ifs_firstc_len = 1;
11121 }
11122 else
11123 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
11124 }
11125 #else
11126 ifs_firstc = ifs_value ? *ifs_value : 0;
11127 #endif
11128 }
11129
11130 char *
11131 getifs ()
11132 {
11133 return ifs_value;
11134 }
11135
11136 /* This splits a single word into a WORD LIST on $IFS, but only if the word
11137 is not quoted. list_string () performs quote removal for us, even if we
11138 don't do any splitting. */
11139 WORD_LIST *
11140 word_split (w, ifs_chars)
11141 WORD_DESC *w;
11142 char *ifs_chars;
11143 {
11144 WORD_LIST *result;
11145
11146 if (w)
11147 {
11148 char *xifs;
11149
11150 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
11151 result = list_string (w->word, xifs, w->flags & W_QUOTED);
11152 }
11153 else
11154 result = (WORD_LIST *)NULL;
11155
11156 return (result);
11157 }
11158
11159 /* Perform word splitting on LIST and return the RESULT. It is possible
11160 to return (WORD_LIST *)NULL. */
11161 static WORD_LIST *
11162 word_list_split (list)
11163 WORD_LIST *list;
11164 {
11165 WORD_LIST *result, *t, *tresult, *e;
11166 WORD_DESC *w;
11167
11168 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
11169 {
11170 tresult = word_split (t->word, ifs_value);
11171 /* POSIX 2.6: "If the complete expansion appropriate for a word results
11172 in an empty field, that empty field shall be deleted from the list
11173 of fields that form the completely expanded command, unless the
11174 original word contained single-quote or double-quote characters."
11175 This is where we handle these words that contain quoted null strings
11176 and other characters that expand to nothing after word splitting. */
11177 if (tresult == 0 && t->word && (t->word->flags & W_SAWQUOTEDNULL)) /* XXX */
11178 {
11179 w = alloc_word_desc ();
11180 w->word = (char *)xmalloc (1);
11181 w->word[0] = '\0';
11182 tresult = make_word_list (w, (WORD_LIST *)NULL);
11183 }
11184 if (result == 0)
11185 result = e = tresult;
11186 else
11187 {
11188 e->next = tresult;
11189 while (e->next)
11190 e = e->next;
11191 }
11192 }
11193 return (result);
11194 }
11195
11196 /**************************************************
11197 * *
11198 * Functions to expand an entire WORD_LIST *
11199 * *
11200 **************************************************/
11201
11202 /* Do any word-expansion-specific cleanup and jump to top_level */
11203 static void
11204 exp_jump_to_top_level (v)
11205 int v;
11206 {
11207 set_pipestatus_from_exit (last_command_exit_value);
11208
11209 /* Cleanup code goes here. */
11210 expand_no_split_dollar_star = 0; /* XXX */
11211 if (expanding_redir)
11212 undo_partial_redirects ();
11213 expanding_redir = 0;
11214 assigning_in_environment = 0;
11215
11216 if (parse_and_execute_level == 0)
11217 top_level_cleanup (); /* from sig.c */
11218
11219 jump_to_top_level (v);
11220 }
11221
11222 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
11223 ELIST, and set ELIST to the new list. */
11224 #define PREPEND_LIST(nlist, elist) \
11225 do { nlist->next = elist; elist = nlist; } while (0)
11226
11227 /* Separate out any initial variable assignments from TLIST. If set -k has
11228 been executed, remove all assignment statements from TLIST. Initial
11229 variable assignments and other environment assignments are placed
11230 on SUBST_ASSIGN_VARLIST. */
11231 static WORD_LIST *
11232 separate_out_assignments (tlist)
11233 WORD_LIST *tlist;
11234 {
11235 register WORD_LIST *vp, *lp;
11236
11237 if (tlist == 0)
11238 return ((WORD_LIST *)NULL);
11239
11240 if (subst_assign_varlist)
11241 dispose_words (subst_assign_varlist); /* Clean up after previous error */
11242
11243 subst_assign_varlist = (WORD_LIST *)NULL;
11244 vp = lp = tlist;
11245
11246 /* Separate out variable assignments at the start of the command.
11247 Loop invariant: vp->next == lp
11248 Loop postcondition:
11249 lp = list of words left after assignment statements skipped
11250 tlist = original list of words
11251 */
11252 while (lp && (lp->word->flags & W_ASSIGNMENT))
11253 {
11254 vp = lp;
11255 lp = lp->next;
11256 }
11257
11258 /* If lp != tlist, we have some initial assignment statements.
11259 We make SUBST_ASSIGN_VARLIST point to the list of assignment
11260 words and TLIST point to the remaining words. */
11261 if (lp != tlist)
11262 {
11263 subst_assign_varlist = tlist;
11264 /* ASSERT(vp->next == lp); */
11265 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
11266 tlist = lp; /* remainder of word list */
11267 }
11268
11269 /* vp == end of variable list */
11270 /* tlist == remainder of original word list without variable assignments */
11271 if (!tlist)
11272 /* All the words in tlist were assignment statements */
11273 return ((WORD_LIST *)NULL);
11274
11275 /* ASSERT(tlist != NULL); */
11276 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
11277
11278 /* If the -k option is in effect, we need to go through the remaining
11279 words, separate out the assignment words, and place them on
11280 SUBST_ASSIGN_VARLIST. */
11281 if (place_keywords_in_env)
11282 {
11283 WORD_LIST *tp; /* tp == running pointer into tlist */
11284
11285 tp = tlist;
11286 lp = tlist->next;
11287
11288 /* Loop Invariant: tp->next == lp */
11289 /* Loop postcondition: tlist == word list without assignment statements */
11290 while (lp)
11291 {
11292 if (lp->word->flags & W_ASSIGNMENT)
11293 {
11294 /* Found an assignment statement, add this word to end of
11295 subst_assign_varlist (vp). */
11296 if (!subst_assign_varlist)
11297 subst_assign_varlist = vp = lp;
11298 else
11299 {
11300 vp->next = lp;
11301 vp = lp;
11302 }
11303
11304 /* Remove the word pointed to by LP from TLIST. */
11305 tp->next = lp->next;
11306 /* ASSERT(vp == lp); */
11307 lp->next = (WORD_LIST *)NULL;
11308 lp = tp->next;
11309 }
11310 else
11311 {
11312 tp = lp;
11313 lp = lp->next;
11314 }
11315 }
11316 }
11317 return (tlist);
11318 }
11319
11320 #define WEXP_VARASSIGN 0x001
11321 #define WEXP_BRACEEXP 0x002
11322 #define WEXP_TILDEEXP 0x004
11323 #define WEXP_PARAMEXP 0x008
11324 #define WEXP_PATHEXP 0x010
11325
11326 /* All of the expansions, including variable assignments at the start of
11327 the list. */
11328 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
11329
11330 /* All of the expansions except variable assignments at the start of
11331 the list. */
11332 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
11333
11334 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
11335 expansion, command substitution, arithmetic expansion, word splitting, and
11336 quote removal. */
11337 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
11338
11339 /* Take the list of words in LIST and do the various substitutions. Return
11340 a new list of words which is the expanded list, and without things like
11341 variable assignments. */
11342
11343 WORD_LIST *
11344 expand_words (list)
11345 WORD_LIST *list;
11346 {
11347 return (expand_word_list_internal (list, WEXP_ALL));
11348 }
11349
11350 /* Same as expand_words (), but doesn't hack variable or environment
11351 variables. */
11352 WORD_LIST *
11353 expand_words_no_vars (list)
11354 WORD_LIST *list;
11355 {
11356 return (expand_word_list_internal (list, WEXP_NOVARS));
11357 }
11358
11359 WORD_LIST *
11360 expand_words_shellexp (list)
11361 WORD_LIST *list;
11362 {
11363 return (expand_word_list_internal (list, WEXP_SHELLEXP));
11364 }
11365
11366 static WORD_LIST *
11367 glob_expand_word_list (tlist, eflags)
11368 WORD_LIST *tlist;
11369 int eflags;
11370 {
11371 char **glob_array, *temp_string;
11372 register int glob_index;
11373 WORD_LIST *glob_list, *output_list, *disposables, *next;
11374 WORD_DESC *tword;
11375 int x;
11376
11377 output_list = disposables = (WORD_LIST *)NULL;
11378 glob_array = (char **)NULL;
11379 while (tlist)
11380 {
11381 /* For each word, either globbing is attempted or the word is
11382 added to orig_list. If globbing succeeds, the results are
11383 added to orig_list and the word (tlist) is added to the list
11384 of disposable words. If globbing fails and failed glob
11385 expansions are left unchanged (the shell default), the
11386 original word is added to orig_list. If globbing fails and
11387 failed glob expansions are removed, the original word is
11388 added to the list of disposable words. orig_list ends up
11389 in reverse order and requires a call to REVERSE_LIST to
11390 be set right. After all words are examined, the disposable
11391 words are freed. */
11392 next = tlist->next;
11393
11394 /* If the word isn't an assignment and contains an unquoted
11395 pattern matching character, then glob it. */
11396 if ((tlist->word->flags & W_NOGLOB) == 0 &&
11397 unquoted_glob_pattern_p (tlist->word->word))
11398 {
11399 glob_array = shell_glob_filename (tlist->word->word, QGLOB_CTLESC); /* XXX */
11400
11401 /* Handle error cases.
11402 I don't think we should report errors like "No such file
11403 or directory". However, I would like to report errors
11404 like "Read failed". */
11405
11406 if (glob_array == 0 || GLOB_FAILED (glob_array))
11407 {
11408 glob_array = (char **)xmalloc (sizeof (char *));
11409 glob_array[0] = (char *)NULL;
11410 }
11411
11412 /* Dequote the current word in case we have to use it. */
11413 if (glob_array[0] == NULL)
11414 {
11415 temp_string = dequote_string (tlist->word->word);
11416 free (tlist->word->word);
11417 tlist->word->word = temp_string;
11418 }
11419
11420 /* Make the array into a word list. */
11421 glob_list = (WORD_LIST *)NULL;
11422 for (glob_index = 0; glob_array[glob_index]; glob_index++)
11423 {
11424 tword = make_bare_word (glob_array[glob_index]);
11425 glob_list = make_word_list (tword, glob_list);
11426 }
11427
11428 if (glob_list)
11429 {
11430 output_list = (WORD_LIST *)list_append (glob_list, output_list);
11431 PREPEND_LIST (tlist, disposables);
11432 }
11433 else if (fail_glob_expansion != 0)
11434 {
11435 last_command_exit_value = EXECUTION_FAILURE;
11436 report_error (_("no match: %s"), tlist->word->word);
11437 exp_jump_to_top_level (DISCARD);
11438 }
11439 else if (allow_null_glob_expansion == 0)
11440 {
11441 /* Failed glob expressions are left unchanged. */
11442 PREPEND_LIST (tlist, output_list);
11443 }
11444 else
11445 {
11446 /* Failed glob expressions are removed. */
11447 PREPEND_LIST (tlist, disposables);
11448 }
11449 }
11450 else
11451 {
11452 /* Dequote the string. */
11453 temp_string = dequote_string (tlist->word->word);
11454 free (tlist->word->word);
11455 tlist->word->word = temp_string;
11456 PREPEND_LIST (tlist, output_list);
11457 }
11458
11459 strvec_dispose (glob_array);
11460 glob_array = (char **)NULL;
11461
11462 tlist = next;
11463 }
11464
11465 if (disposables)
11466 dispose_words (disposables);
11467
11468 if (output_list)
11469 output_list = REVERSE_LIST (output_list, WORD_LIST *);
11470
11471 return (output_list);
11472 }
11473
11474 #if defined (BRACE_EXPANSION)
11475 static WORD_LIST *
11476 brace_expand_word_list (tlist, eflags)
11477 WORD_LIST *tlist;
11478 int eflags;
11479 {
11480 register char **expansions;
11481 char *temp_string;
11482 WORD_LIST *disposables, *output_list, *next;
11483 WORD_DESC *w;
11484 int eindex;
11485
11486 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
11487 {
11488 next = tlist->next;
11489
11490 if (tlist->word->flags & W_NOBRACE)
11491 {
11492 /*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
11493 PREPEND_LIST (tlist, output_list);
11494 continue;
11495 }
11496
11497 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
11498 {
11499 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
11500 PREPEND_LIST (tlist, output_list);
11501 continue;
11502 }
11503
11504 /* Only do brace expansion if the word has a brace character. If
11505 not, just add the word list element to BRACES and continue. In
11506 the common case, at least when running shell scripts, this will
11507 degenerate to a bunch of calls to `mbschr', and then what is
11508 basically a reversal of TLIST into BRACES, which is corrected
11509 by a call to REVERSE_LIST () on BRACES when the end of TLIST
11510 is reached. */
11511 if (mbschr (tlist->word->word, LBRACE))
11512 {
11513 expansions = brace_expand (tlist->word->word);
11514
11515 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
11516 {
11517 w = alloc_word_desc ();
11518 w->word = temp_string;
11519
11520 /* If brace expansion didn't change the word, preserve
11521 the flags. We may want to preserve the flags
11522 unconditionally someday -- XXX */
11523 if (STREQ (temp_string, tlist->word->word))
11524 w->flags = tlist->word->flags;
11525 else
11526 w = make_word_flags (w, temp_string);
11527
11528 output_list = make_word_list (w, output_list);
11529 }
11530 free (expansions);
11531
11532 /* Add TLIST to the list of words to be freed after brace
11533 expansion has been performed. */
11534 PREPEND_LIST (tlist, disposables);
11535 }
11536 else
11537 PREPEND_LIST (tlist, output_list);
11538 }
11539
11540 if (disposables)
11541 dispose_words (disposables);
11542
11543 if (output_list)
11544 output_list = REVERSE_LIST (output_list, WORD_LIST *);
11545
11546 return (output_list);
11547 }
11548 #endif
11549
11550 #if defined (ARRAY_VARS)
11551 /* Take WORD, a compound array assignment, and internally run (for example),
11552 'declare -A w', where W is the variable name portion of WORD. OPTION is
11553 the list of options to supply to `declare'. CMD is the declaration command
11554 we are expanding right now; it's unused currently. */
11555 static int
11556 make_internal_declare (word, option, cmd)
11557 char *word;
11558 char *option;
11559 char *cmd;
11560 {
11561 int t, r;
11562 WORD_LIST *wl;
11563 WORD_DESC *w;
11564
11565 w = make_word (word);
11566
11567 t = assignment (w->word, 0);
11568 if (w->word[t] == '=')
11569 {
11570 w->word[t] = '\0';
11571 if (w->word[t - 1] == '+') /* cut off any append op */
11572 w->word[t - 1] = '\0';
11573 }
11574
11575 wl = make_word_list (w, (WORD_LIST *)NULL);
11576 wl = make_word_list (make_word (option), wl);
11577
11578 r = declare_builtin (wl);
11579
11580 dispose_words (wl);
11581 return r;
11582 }
11583
11584 /* Expand VALUE in NAME[+]=( VALUE ) to a list of words. FLAGS is 1 if NAME
11585 is an associative array.
11586
11587 If we are processing an indexed array, expand_compound_array_assignment
11588 will expand all the individual words and quote_compound_array_list will
11589 single-quote them. If we are processing an associative array, we use
11590 parse_string_to_word_list to split VALUE into a list of words instead of
11591 faking up a shell variable and calling expand_compound_array_assignment.
11592 expand_and_quote_assoc_word expands and single-quotes each word in VALUE
11593 together so we don't have problems finding the end of the subscript when
11594 quoting it.
11595
11596 Words in VALUE can be individual words, which are expanded and single-quoted,
11597 or words of the form [IND]=VALUE, which end up as explained below, as
11598 ['expanded-ind']='expanded-value'. */
11599
11600 static WORD_LIST *
11601 expand_oneword (value, flags)
11602 char *value;
11603 int flags;
11604 {
11605 WORD_LIST *l, *nl;
11606 char *t;
11607
11608 if (flags == 0)
11609 {
11610 /* Indexed array */
11611 l = expand_compound_array_assignment ((SHELL_VAR *)NULL, value, flags);
11612 /* Now we quote the results of the expansion above to prevent double
11613 expansion. */
11614 quote_compound_array_list (l, flags);
11615 return l;
11616 }
11617 else
11618 {
11619 /* Associative array */
11620 l = parse_string_to_word_list (value, 1, "array assign");
11621 /* For associative arrays, with their arbitrary subscripts, we have to
11622 expand and quote in one step so we don't have to search for the
11623 closing right bracket more than once. */
11624 for (nl = l; nl; nl = nl->next)
11625 {
11626 if ((nl->word->flags & W_ASSIGNMENT) == 0)
11627 t = sh_single_quote (nl->word->word ? nl->word->word : "");
11628 else
11629 t = expand_and_quote_assoc_word (nl->word->word, flags);
11630 free (nl->word->word);
11631 nl->word->word = t;
11632 }
11633 return l;
11634 }
11635 }
11636
11637 /* Expand a single compound assignment argument to a declaration builtin.
11638 This word takes the form NAME[+]=( VALUE ). The NAME[+]= is passed through
11639 unchanged. The VALUE is expanded and each word in the result is single-
11640 quoted. Words of the form [key]=value end up as
11641 ['expanded-key']='expanded-value'. Associative arrays have special
11642 handling, see expand_oneword() above. The return value is
11643 NAME[+]=( expanded-and-quoted-VALUE ). */
11644 static void
11645 expand_compound_assignment_word (tlist, flags)
11646 WORD_LIST *tlist;
11647 int flags;
11648 {
11649 WORD_LIST *l;
11650 int wlen, oind, t;
11651 char *value, *temp;
11652
11653 /*itrace("expand_compound_assignment_word: original word = -%s-", tlist->word->word);*/
11654 t = assignment (tlist->word->word, 0);
11655
11656 /* value doesn't have the open and close parens */
11657 oind = 1;
11658 value = extract_array_assignment_list (tlist->word->word + t + 1, &oind);
11659 /* This performs one round of expansion on the index/key and value and
11660 single-quotes each word in the result. */
11661 l = expand_oneword (value, flags);
11662 free (value);
11663
11664 value = string_list (l);
11665 wlen = STRLEN (value);
11666
11667 /* Now, let's rebuild the string */
11668 temp = xmalloc (t + 3 + wlen + 1); /* name[+]=(value) */
11669 memcpy (temp, tlist->word->word, ++t);
11670 temp[t++] = '(';
11671 if (value)
11672 memcpy (temp + t, value, wlen);
11673 t += wlen;
11674 temp[t++] = ')';
11675 temp[t] = '\0';
11676 /*itrace("expand_compound_assignment_word: reconstructed word = -%s-", temp);*/
11677
11678 free (tlist->word->word);
11679 tlist->word->word = temp;
11680
11681 free (value);
11682 }
11683
11684 /* Expand and process an argument to a declaration command. We have already
11685 set flags in TLIST->word->flags depending on the declaration command
11686 (declare, local, etc.) and the options supplied to it (-a, -A, etc.).
11687 TLIST->word->word is of the form NAME[+]=( VALUE ).
11688
11689 This does several things, all using pieces of other functions to get the
11690 evaluation sequence right. It's called for compound array assignments with
11691 the W_ASSIGNMENT flag set (basically, valid identifier names on the lhs).
11692 It parses out which flags need to be set for declare to create the variable
11693 correctly, then calls declare internally (make_internal_declare) to make
11694 sure the variable exists with the correct attributes. Before the variable
11695 is created, it calls expand_compound_assignment_word to expand VALUE to a
11696 list of words, appropriately quoted for further evaluation. This preserves
11697 the semantics of word-expansion-before-calling-builtins. Finally, it calls
11698 do_word_assignment to perform the expansion and assignment with the same
11699 expansion semantics as a standalone assignment statement (no word splitting,
11700 etc.) even though the word is single-quoted so all that needs to happen is
11701 quote removal. */
11702 static WORD_LIST *
11703 expand_declaration_argument (tlist, wcmd)
11704 WORD_LIST *tlist, *wcmd;
11705 {
11706 char opts[16], omap[128];
11707 int t, opti, oind, skip, inheriting;
11708 WORD_LIST *l;
11709
11710 inheriting = localvar_inherit;
11711 opti = 0;
11712 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_CHKLOCAL|W_ASSIGNARRAY))
11713 opts[opti++] = '-';
11714
11715 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
11716 {
11717 opts[opti++] = 'g';
11718 opts[opti++] = 'A';
11719 }
11720 else if (tlist->word->flags & W_ASSIGNASSOC)
11721 {
11722 opts[opti++] = 'A';
11723 }
11724 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
11725 {
11726 opts[opti++] = 'g';
11727 opts[opti++] = 'a';
11728 }
11729 else if (tlist->word->flags & W_ASSIGNARRAY)
11730 {
11731 opts[opti++] = 'a';
11732 }
11733 else if (tlist->word->flags & W_ASSNGLOBAL)
11734 opts[opti++] = 'g';
11735
11736 if (tlist->word->flags & W_CHKLOCAL)
11737 opts[opti++] = 'G';
11738
11739 /* If we have special handling note the integer attribute and others
11740 that transform the value upon assignment. What we do is take all
11741 of the option arguments and scan through them looking for options
11742 that cause such transformations, and add them to the `opts' array. */
11743
11744 memset (omap, '\0', sizeof (omap));
11745 for (l = wcmd->next; l != tlist; l = l->next)
11746 {
11747 if (l->word->word[0] != '-')
11748 break; /* non-option argument */
11749 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
11750 break; /* -- signals end of options */
11751 for (oind = 1; l->word->word[oind]; oind++)
11752 switch (l->word->word[oind])
11753 {
11754 case 'I':
11755 inheriting = 1;
11756 case 'i':
11757 case 'l':
11758 case 'u':
11759 case 'c':
11760 omap[l->word->word[oind]] = 1;
11761 if (opti == 0)
11762 opts[opti++] = '-';
11763 break;
11764 default:
11765 break;
11766 }
11767 }
11768
11769 for (oind = 0; oind < sizeof (omap); oind++)
11770 if (omap[oind])
11771 opts[opti++] = oind;
11772
11773 /* If there are no -a/-A options, but we have a compound assignment,
11774 we have a choice: we can set opts[0]='-', opts[1]='a', since the
11775 default is to create an indexed array, and call
11776 make_internal_declare with that, or we can just skip the -a and let
11777 declare_builtin deal with it. Once we're here, we're better set
11778 up for the latter, since we don't want to deal with looking up
11779 any existing variable here -- better to let declare_builtin do it.
11780 We need the variable created, though, especially if it's local, so
11781 we get the scoping right before we call do_word_assignment.
11782 To ensure that make_local_declare gets called, we add `--' if there
11783 aren't any options. */
11784 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSIGNARRAY)) == 0)
11785 {
11786 if (opti == 0)
11787 {
11788 opts[opti++] = '-';
11789 opts[opti++] = '-';
11790 }
11791 }
11792 opts[opti] = '\0';
11793
11794 /* This isn't perfect, but it's a start. Improvements later. We expand
11795 tlist->word->word and single-quote the results to avoid multiple
11796 expansions by, say, do_assignment_internal(). We have to weigh the
11797 cost of reconstructing the compound assignment string with its single
11798 quoting and letting the declare builtin handle it. The single quotes
11799 will prevent any unwanted additional expansion or word splitting. */
11800 expand_compound_assignment_word (tlist, (tlist->word->flags & W_ASSIGNASSOC) ? 1 : 0);
11801
11802 skip = 0;
11803 if (opti > 0)
11804 {
11805 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
11806 if (t != EXECUTION_SUCCESS)
11807 {
11808 last_command_exit_value = t;
11809 if (tlist->word->flags & W_FORCELOCAL) /* non-fatal error */
11810 skip = 1;
11811 else
11812 exp_jump_to_top_level (DISCARD);
11813 }
11814 }
11815
11816 if (skip == 0)
11817 {
11818 t = do_word_assignment (tlist->word, 0);
11819 if (t == 0)
11820 {
11821 last_command_exit_value = EXECUTION_FAILURE;
11822 exp_jump_to_top_level (DISCARD);
11823 }
11824 }
11825
11826 /* Now transform the word as ksh93 appears to do and go on */
11827 t = assignment (tlist->word->word, 0);
11828 tlist->word->word[t] = '\0';
11829 if (tlist->word->word[t - 1] == '+')
11830 tlist->word->word[t - 1] = '\0'; /* cut off append op */
11831 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
11832
11833 return (tlist);
11834 }
11835 #endif /* ARRAY_VARS */
11836
11837 static WORD_LIST *
11838 shell_expand_word_list (tlist, eflags)
11839 WORD_LIST *tlist;
11840 int eflags;
11841 {
11842 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
11843 int expanded_something, has_dollar_at;
11844
11845 /* We do tilde expansion all the time. This is what 1003.2 says. */
11846 wcmd = new_list = (WORD_LIST *)NULL;
11847
11848 for (orig_list = tlist; tlist; tlist = next)
11849 {
11850 if (wcmd == 0 && (tlist->word->flags & W_ASSNBLTIN))
11851 wcmd = tlist;
11852
11853 next = tlist->next;
11854
11855 #if defined (ARRAY_VARS)
11856 /* If this is a compound array assignment to a builtin that accepts
11857 such assignments (e.g., `declare'), take the assignment and perform
11858 it separately, handling the semantics of declarations inside shell
11859 functions. This avoids the double-evaluation of such arguments,
11860 because `declare' does some evaluation of compound assignments on
11861 its own. */
11862 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
11863 expand_declaration_argument (tlist, wcmd);
11864 #endif
11865
11866 expanded_something = 0;
11867 expanded = expand_word_internal
11868 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
11869
11870 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
11871 {
11872 /* By convention, each time this error is returned,
11873 tlist->word->word has already been freed. */
11874 tlist->word->word = (char *)NULL;
11875
11876 /* Dispose our copy of the original list. */
11877 dispose_words (orig_list);
11878 /* Dispose the new list we're building. */
11879 dispose_words (new_list);
11880
11881 last_command_exit_value = EXECUTION_FAILURE;
11882 if (expanded == &expand_word_error)
11883 exp_jump_to_top_level (DISCARD);
11884 else
11885 exp_jump_to_top_level (FORCE_EOF);
11886 }
11887
11888 /* Don't split words marked W_NOSPLIT. */
11889 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
11890 {
11891 temp_list = word_list_split (expanded);
11892 dispose_words (expanded);
11893 }
11894 else
11895 {
11896 /* If no parameter expansion, command substitution, process
11897 substitution, or arithmetic substitution took place, then
11898 do not do word splitting. We still have to remove quoted
11899 null characters from the result. */
11900 word_list_remove_quoted_nulls (expanded);
11901 temp_list = expanded;
11902 }
11903
11904 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
11905 new_list = (WORD_LIST *)list_append (expanded, new_list);
11906 }
11907
11908 if (orig_list)
11909 dispose_words (orig_list);
11910
11911 if (new_list)
11912 new_list = REVERSE_LIST (new_list, WORD_LIST *);
11913
11914 return (new_list);
11915 }
11916
11917 /* The workhorse for expand_words () and expand_words_no_vars ().
11918 First arg is LIST, a WORD_LIST of words.
11919 Second arg EFLAGS is a flags word controlling which expansions are
11920 performed.
11921
11922 This does all of the substitutions: brace expansion, tilde expansion,
11923 parameter expansion, command substitution, arithmetic expansion,
11924 process substitution, word splitting, and pathname expansion, according
11925 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
11926 set, or for which no expansion is done, do not undergo word splitting.
11927 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
11928 with W_NOBRACE set do not undergo brace expansion (see
11929 brace_expand_word_list above). */
11930 static WORD_LIST *
11931 expand_word_list_internal (list, eflags)
11932 WORD_LIST *list;
11933 int eflags;
11934 {
11935 WORD_LIST *new_list, *temp_list;
11936 int tint;
11937 char *savecmd;
11938
11939 tempenv_assign_error = 0;
11940 if (list == 0)
11941 return ((WORD_LIST *)NULL);
11942
11943 garglist = new_list = copy_word_list (list);
11944 if (eflags & WEXP_VARASSIGN)
11945 {
11946 garglist = new_list = separate_out_assignments (new_list);
11947 if (new_list == 0)
11948 {
11949 if (subst_assign_varlist)
11950 {
11951 /* All the words were variable assignments, so they are placed
11952 into the shell's environment. */
11953 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
11954 {
11955 savecmd = this_command_name;
11956 this_command_name = (char *)NULL; /* no arithmetic errors */
11957 tint = do_word_assignment (temp_list->word, 0);
11958 this_command_name = savecmd;
11959 /* Variable assignment errors in non-interactive shells
11960 running in Posix.2 mode cause the shell to exit, unless
11961 they are being run by the `command' builtin. */
11962 if (tint == 0)
11963 {
11964 last_command_exit_value = EXECUTION_FAILURE;
11965 if (interactive_shell == 0 && posixly_correct && executing_command_builtin == 0)
11966 exp_jump_to_top_level (FORCE_EOF);
11967 else
11968 exp_jump_to_top_level (DISCARD);
11969 }
11970 }
11971 dispose_words (subst_assign_varlist);
11972 subst_assign_varlist = (WORD_LIST *)NULL;
11973 }
11974 return ((WORD_LIST *)NULL);
11975 }
11976 }
11977
11978 /* Begin expanding the words that remain. The expansions take place on
11979 things that aren't really variable assignments. */
11980
11981 #if defined (BRACE_EXPANSION)
11982 /* Do brace expansion on this word if there are any brace characters
11983 in the string. */
11984 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
11985 new_list = brace_expand_word_list (new_list, eflags);
11986 #endif /* BRACE_EXPANSION */
11987
11988 /* Perform the `normal' shell expansions: tilde expansion, parameter and
11989 variable substitution, command substitution, arithmetic expansion,
11990 and word splitting. */
11991 new_list = shell_expand_word_list (new_list, eflags);
11992
11993 /* Okay, we're almost done. Now let's just do some filename
11994 globbing. */
11995 if (new_list)
11996 {
11997 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
11998 /* Glob expand the word list unless globbing has been disabled. */
11999 new_list = glob_expand_word_list (new_list, eflags);
12000 else
12001 /* Dequote the words, because we're not performing globbing. */
12002 new_list = dequote_list (new_list);
12003 }
12004
12005 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
12006 {
12007 sh_wassign_func_t *assign_func;
12008 int is_special_builtin, is_builtin_or_func;
12009
12010 /* If the remainder of the words expand to nothing, Posix.2 requires
12011 that the variable and environment assignments affect the shell's
12012 environment. */
12013 assign_func = new_list ? assign_in_env : do_word_assignment;
12014 tempenv_assign_error = 0;
12015
12016 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
12017 /* Posix says that special builtins exit if a variable assignment error
12018 occurs in an assignment preceding it. */
12019 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
12020
12021 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
12022 {
12023 savecmd = this_command_name;
12024 this_command_name = (char *)NULL;
12025 assigning_in_environment = (assign_func == assign_in_env);
12026 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
12027 assigning_in_environment = 0;
12028 this_command_name = savecmd;
12029 /* Variable assignment errors in non-interactive shells running
12030 in Posix.2 mode cause the shell to exit. */
12031 if (tint == 0)
12032 {
12033 if (assign_func == do_word_assignment)
12034 {
12035 last_command_exit_value = EXECUTION_FAILURE;
12036 if (interactive_shell == 0 && posixly_correct)
12037 exp_jump_to_top_level (FORCE_EOF);
12038 else
12039 exp_jump_to_top_level (DISCARD);
12040 }
12041 else if (interactive_shell == 0 && is_special_builtin)
12042 {
12043 last_command_exit_value = EXECUTION_FAILURE;
12044 exp_jump_to_top_level (FORCE_EOF);
12045 }
12046 else
12047 tempenv_assign_error++;
12048 }
12049 }
12050
12051 dispose_words (subst_assign_varlist);
12052 subst_assign_varlist = (WORD_LIST *)NULL;
12053 }
12054
12055 return (new_list);
12056 }