]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
bash-5.1 distribution sources and documentation
[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. */
6360 if ((subshell_environment & SUBSHELL_PIPE) == 0)
6361 pipeline_pgrp = shell_pgrp;
6362 cleanup_the_pipeline ();
6363 #endif /* JOB_CONTROL */
6364
6365 old_async_pid = last_asynchronous_pid;
6366 fork_flags = (subshell_environment&SUBSHELL_ASYNC) ? FORK_ASYNC : 0;
6367 pid = make_child ((char *)NULL, fork_flags|FORK_NOTERM);
6368 last_asynchronous_pid = old_async_pid;
6369
6370 if (pid == 0)
6371 {
6372 /* Reset the signal handlers in the child, but don't free the
6373 trap strings. Set a flag noting that we have to free the
6374 trap strings if we run trap to change a signal disposition. */
6375 reset_signal_handlers ();
6376 if (ISINTERRUPT)
6377 {
6378 kill (getpid (), SIGINT);
6379 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6380 }
6381 QUIT; /* catch any interrupts we got post-fork */
6382 subshell_environment |= SUBSHELL_RESETTRAP;
6383 }
6384
6385 #if defined (JOB_CONTROL)
6386 /* XXX DO THIS ONLY IN PARENT ? XXX */
6387 set_sigchld_handler ();
6388 stop_making_children ();
6389 if (pid != 0)
6390 pipeline_pgrp = old_pipeline_pgrp;
6391 #else
6392 stop_making_children ();
6393 #endif /* JOB_CONTROL */
6394
6395 if (pid < 0)
6396 {
6397 sys_error (_("cannot make child for command substitution"));
6398 error_exit:
6399
6400 last_made_pid = old_pid;
6401
6402 FREE (istring);
6403 close (fildes[0]);
6404 close (fildes[1]);
6405 return ((WORD_DESC *)NULL);
6406 }
6407
6408 if (pid == 0)
6409 {
6410 /* The currently executing shell is not interactive. */
6411 interactive = 0;
6412
6413 set_sigint_handler (); /* XXX */
6414
6415 free_pushed_string_input ();
6416
6417 /* Discard buffered stdio output before replacing the underlying file
6418 descriptor. */
6419 fpurge (stdout);
6420
6421 if (dup2 (fildes[1], 1) < 0)
6422 {
6423 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
6424 exit (EXECUTION_FAILURE);
6425 }
6426
6427 /* If standard output is closed in the parent shell
6428 (such as after `exec >&-'), file descriptor 1 will be
6429 the lowest available file descriptor, and end up in
6430 fildes[0]. This can happen for stdin and stderr as well,
6431 but stdout is more important -- it will cause no output
6432 to be generated from this command. */
6433 if ((fildes[1] != fileno (stdin)) &&
6434 (fildes[1] != fileno (stdout)) &&
6435 (fildes[1] != fileno (stderr)))
6436 close (fildes[1]);
6437
6438 if ((fildes[0] != fileno (stdin)) &&
6439 (fildes[0] != fileno (stdout)) &&
6440 (fildes[0] != fileno (stderr)))
6441 close (fildes[0]);
6442
6443 #ifdef __CYGWIN__
6444 /* Let stdio know the fd may have changed from text to binary mode, and
6445 make sure to preserve stdout line buffering. */
6446 freopen (NULL, "w", stdout);
6447 sh_setlinebuf (stdout);
6448 #endif /* __CYGWIN__ */
6449
6450 /* This is a subshell environment. */
6451 subshell_environment |= SUBSHELL_COMSUB;
6452
6453 /* Many shells do not appear to inherit the -v option for command
6454 substitutions. */
6455 change_flag ('v', FLAG_OFF);
6456
6457 /* When inherit_errexit option is not enabled, command substitution does
6458 not inherit the -e flag. It is enabled when Posix mode is enabled */
6459 if (inherit_errexit == 0)
6460 {
6461 builtin_ignoring_errexit = 0;
6462 change_flag ('e', FLAG_OFF);
6463 }
6464 set_shellopts ();
6465
6466 /* If we are expanding a redirection, we can dispose of any temporary
6467 environment we received, since redirections are not supposed to have
6468 access to the temporary environment. We will have to see whether this
6469 affects temporary environments supplied to `eval', but the temporary
6470 environment gets copied to builtin_env at some point. */
6471 if (expanding_redir)
6472 {
6473 flush_temporary_env ();
6474 expanding_redir = 0;
6475 }
6476
6477 remove_quoted_escapes (string);
6478
6479 startup_state = 2; /* see if we can avoid a fork */
6480 parse_and_execute_level = 0;
6481
6482 /* Give command substitution a place to jump back to on failure,
6483 so we don't go back up to main (). */
6484 result = setjmp_nosigs (top_level);
6485
6486 /* If we're running a command substitution inside a shell function,
6487 trap `return' so we don't return from the function in the subshell
6488 and go off to never-never land. */
6489 if (result == 0 && return_catch_flag)
6490 function_value = setjmp_nosigs (return_catch);
6491 else
6492 function_value = 0;
6493
6494 if (result == ERREXIT)
6495 rc = last_command_exit_value;
6496 else if (result == EXITPROG)
6497 rc = last_command_exit_value;
6498 else if (result)
6499 rc = EXECUTION_FAILURE;
6500 else if (function_value)
6501 rc = return_catch_value;
6502 else
6503 {
6504 subshell_level++;
6505 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
6506 /* leave subshell level intact for any exit trap */
6507 }
6508
6509 last_command_exit_value = rc;
6510 rc = run_exit_trap ();
6511 #if defined (PROCESS_SUBSTITUTION)
6512 unlink_fifo_list ();
6513 #endif
6514 exit (rc);
6515 }
6516 else
6517 {
6518 int dummyfd;
6519
6520 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6521 close_pgrp_pipe ();
6522 #endif /* JOB_CONTROL && PGRP_PIPE */
6523
6524 close (fildes[1]);
6525
6526 begin_unwind_frame ("read-comsub");
6527 dummyfd = fildes[0];
6528 add_unwind_protect (close, dummyfd);
6529
6530 /* Block SIGINT while we're reading from the pipe. If the child
6531 process gets a SIGINT, it will either handle it or die, and the
6532 read will return. */
6533 BLOCK_SIGNAL (SIGINT, set, oset);
6534 tflag = 0;
6535 istring = read_comsub (fildes[0], quoted, flags, &tflag);
6536
6537 close (fildes[0]);
6538 discard_unwind_frame ("read-comsub");
6539 UNBLOCK_SIGNAL (oset);
6540
6541 current_command_subst_pid = pid;
6542 last_command_exit_value = wait_for (pid, JWAIT_NOTERM);
6543 last_command_subst_pid = pid;
6544 last_made_pid = old_pid;
6545
6546 #if defined (JOB_CONTROL)
6547 /* If last_command_exit_value > 128, then the substituted command
6548 was terminated by a signal. If that signal was SIGINT, then send
6549 SIGINT to ourselves. This will break out of loops, for instance. */
6550 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
6551 kill (getpid (), SIGINT);
6552 #endif /* JOB_CONTROL */
6553
6554 ret = alloc_word_desc ();
6555 ret->word = istring;
6556 ret->flags = tflag;
6557
6558 return ret;
6559 }
6560 }
6561
6562 /********************************************************
6563 * *
6564 * Utility functions for parameter expansion *
6565 * *
6566 ********************************************************/
6567
6568 #if defined (ARRAY_VARS)
6569
6570 static arrayind_t
6571 array_length_reference (s)
6572 char *s;
6573 {
6574 int len;
6575 arrayind_t ind;
6576 char *akey;
6577 char *t, c;
6578 ARRAY *array;
6579 HASH_TABLE *h;
6580 SHELL_VAR *var;
6581
6582 var = array_variable_part (s, 0, &t, &len);
6583
6584 /* If unbound variables should generate an error, report one and return
6585 failure. */
6586 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
6587 {
6588 c = *--t;
6589 *t = '\0';
6590 set_exit_status (EXECUTION_FAILURE);
6591 err_unboundvar (s);
6592 *t = c;
6593 return (-1);
6594 }
6595 else if (var == 0 || invisible_p (var))
6596 return 0;
6597
6598 /* We support a couple of expansions for variables that are not arrays.
6599 We'll return the length of the value for v[0], and 1 for v[@] or
6600 v[*]. Return 0 for everything else. */
6601
6602 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
6603 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
6604
6605 if (ALL_ELEMENT_SUB (t[0]) && t[1] == RBRACK)
6606 {
6607 if (assoc_p (var))
6608 return (h ? assoc_num_elements (h) : 0);
6609 else if (array_p (var))
6610 return (array ? array_num_elements (array) : 0);
6611 else
6612 return (var_isset (var) ? 1 : 0);
6613 }
6614
6615 if (assoc_p (var))
6616 {
6617 t[len - 1] = '\0';
6618 akey = expand_assignment_string_to_string (t, 0); /* [ */
6619 t[len - 1] = RBRACK;
6620 if (akey == 0 || *akey == 0)
6621 {
6622 err_badarraysub (t);
6623 FREE (akey);
6624 return (-1);
6625 }
6626 t = assoc_reference (assoc_cell (var), akey);
6627 free (akey);
6628 }
6629 else
6630 {
6631 ind = array_expand_index (var, t, len, 0);
6632 /* negative subscripts to indexed arrays count back from end */
6633 if (var && array_p (var) && ind < 0)
6634 ind = array_max_index (array_cell (var)) + 1 + ind;
6635 if (ind < 0)
6636 {
6637 err_badarraysub (t);
6638 return (-1);
6639 }
6640 if (array_p (var))
6641 t = array_reference (array, ind);
6642 else
6643 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6644 }
6645
6646 len = MB_STRLEN (t);
6647 return (len);
6648 }
6649 #endif /* ARRAY_VARS */
6650
6651 static int
6652 valid_brace_expansion_word (name, var_is_special)
6653 char *name;
6654 int var_is_special;
6655 {
6656 if (DIGIT (*name) && all_digits (name))
6657 return 1;
6658 else if (var_is_special)
6659 return 1;
6660 #if defined (ARRAY_VARS)
6661 else if (valid_array_reference (name, 0))
6662 return 1;
6663 #endif /* ARRAY_VARS */
6664 else if (legal_identifier (name))
6665 return 1;
6666 else
6667 return 0;
6668 }
6669
6670 static int
6671 chk_atstar (name, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
6672 char *name;
6673 int quoted, pflags;
6674 int *quoted_dollar_atp, *contains_dollar_at;
6675 {
6676 char *temp1;
6677
6678 if (name == 0)
6679 {
6680 if (quoted_dollar_atp)
6681 *quoted_dollar_atp = 0;
6682 if (contains_dollar_at)
6683 *contains_dollar_at = 0;
6684 return 0;
6685 }
6686
6687 /* check for $@ and $* */
6688 if (name[0] == '@' && name[1] == 0)
6689 {
6690 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6691 *quoted_dollar_atp = 1;
6692 if (contains_dollar_at)
6693 *contains_dollar_at = 1;
6694 return 1;
6695 }
6696 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6697 {
6698 /* Need more checks here that parallel what string_list_pos_params and
6699 param_expand do. Check expand_no_split_dollar_star and ??? */
6700 if (contains_dollar_at && expand_no_split_dollar_star == 0)
6701 *contains_dollar_at = 1;
6702 return 1;
6703 }
6704
6705 /* Now check for ${array[@]} and ${array[*]} */
6706 #if defined (ARRAY_VARS)
6707 else if (valid_array_reference (name, 0))
6708 {
6709 temp1 = mbschr (name, LBRACK);
6710 if (temp1 && temp1[1] == '@' && temp1[2] == RBRACK)
6711 {
6712 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6713 *quoted_dollar_atp = 1;
6714 if (contains_dollar_at)
6715 *contains_dollar_at = 1;
6716 return 1;
6717 }
6718 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6719 which should result in separate words even when IFS is unset. */
6720 if (temp1 && temp1[1] == '*' && temp1[2] == RBRACK && quoted == 0)
6721 {
6722 if (contains_dollar_at)
6723 *contains_dollar_at = 1;
6724 return 1;
6725 }
6726 }
6727 #endif
6728 return 0;
6729 }
6730
6731 /* Parameter expand NAME, and return a new string which is the expansion,
6732 or NULL if there was no expansion. NAME is as given in ${NAMEcWORD}.
6733 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6734 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6735 NAME was found inside of a double-quoted expression. */
6736 static WORD_DESC *
6737 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
6738 char *name;
6739 int var_is_special, quoted, pflags;
6740 arrayind_t *indp;
6741 {
6742 WORD_DESC *ret;
6743 char *temp, *tt;
6744 intmax_t arg_index;
6745 SHELL_VAR *var;
6746 int atype, rflags;
6747 arrayind_t ind;
6748
6749 ret = 0;
6750 temp = 0;
6751 rflags = 0;
6752
6753 if (indp)
6754 *indp = INTMAX_MIN;
6755
6756 /* Handle multiple digit arguments, as in ${11}. */
6757 if (legal_number (name, &arg_index))
6758 {
6759 tt = get_dollar_var_value (arg_index);
6760 if (tt)
6761 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6762 ? quote_string (tt)
6763 : quote_escapes (tt);
6764 else
6765 temp = (char *)NULL;
6766 FREE (tt);
6767 }
6768 else if (var_is_special) /* ${@} */
6769 {
6770 int sindex;
6771 tt = (char *)xmalloc (2 + strlen (name));
6772 tt[sindex = 0] = '$';
6773 strcpy (tt + 1, name);
6774
6775 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
6776 (int *)NULL, (int *)NULL, pflags);
6777 free (tt);
6778 }
6779 #if defined (ARRAY_VARS)
6780 else if (valid_array_reference (name, 0))
6781 {
6782 expand_arrayref:
6783 var = array_variable_part (name, 0, &tt, (int *)0);
6784 /* These are the cases where word splitting will not be performed */
6785 if (pflags & PF_ASSIGNRHS)
6786 {
6787 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == RBRACK)
6788 {
6789 /* Only treat as double quoted if array variable */
6790 if (var && (array_p (var) || assoc_p (var)))
6791 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6792 else
6793 temp = array_value (name, quoted, 0, &atype, &ind);
6794 }
6795 else
6796 temp = array_value (name, quoted, 0, &atype, &ind);
6797 }
6798 /* Posix interp 888 */
6799 else if (pflags & PF_NOSPLIT2)
6800 {
6801 /* Special cases, then general case, for each of A[@], A[*], A[n] */
6802 #if defined (HANDLE_MULTIBYTE)
6803 if (tt[0] == '@' && tt[1] == RBRACK && var && quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc[0] != ' ')
6804 #else
6805 if (tt[0] == '@' && tt[1] == RBRACK && var && quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc != ' ')
6806 #endif
6807 temp = array_value (name, Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6808 else if (tt[0] == '@' && tt[1] == RBRACK)
6809 temp = array_value (name, quoted, 0, &atype, &ind);
6810 else if (tt[0] == '*' && tt[1] == RBRACK && expand_no_split_dollar_star && ifs_is_null)
6811 temp = array_value (name, Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT, 0, &atype, &ind);
6812 else if (tt[0] == '*' && tt[1] == RBRACK)
6813 temp = array_value (name, quoted, 0, &atype, &ind);
6814 else
6815 temp = array_value (name, quoted, 0, &atype, &ind);
6816 }
6817 else if (tt[0] == '*' && tt[1] == RBRACK && expand_no_split_dollar_star && ifs_is_null)
6818 temp = array_value (name, Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT, 0, &atype, &ind);
6819 else
6820 temp = array_value (name, quoted, 0, &atype, &ind);
6821 if (atype == 0 && temp)
6822 {
6823 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6824 ? quote_string (temp)
6825 : quote_escapes (temp);
6826 rflags |= W_ARRAYIND;
6827 if (indp)
6828 *indp = ind;
6829 }
6830 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6831 rflags |= W_HASQUOTEDNULL;
6832 }
6833 #endif
6834 else if (var = find_variable (name))
6835 {
6836 if (var_isset (var) && invisible_p (var) == 0)
6837 {
6838 #if defined (ARRAY_VARS)
6839 /* We avoid a memory leak by saving TT as the memory allocated by
6840 assoc_to_string or array_to_string and leaving it 0 otherwise,
6841 then freeing TT after quoting temp. */
6842 tt = (char *)NULL;
6843 if ((pflags & PF_ALLINDS) && assoc_p (var))
6844 tt = temp = assoc_empty (assoc_cell (var)) ? (char *)NULL : assoc_to_string (assoc_cell (var), " ", quoted);
6845 else if ((pflags & PF_ALLINDS) && array_p (var))
6846 tt = temp = array_empty (array_cell (var)) ? (char *)NULL : array_to_string (array_cell (var), " ", quoted);
6847 else if (assoc_p (var))
6848 temp = assoc_reference (assoc_cell (var), "0");
6849 else if (array_p (var))
6850 temp = array_reference (array_cell (var), 0);
6851 else
6852 temp = value_cell (var);
6853 #else
6854 temp = value_cell (var);
6855 #endif
6856
6857 if (temp)
6858 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6859 ? quote_string (temp)
6860 : ((pflags & PF_ASSIGNRHS) ? quote_rhs (temp)
6861 : quote_escapes (temp));
6862 FREE (tt);
6863 }
6864 else
6865 temp = (char *)NULL;
6866 }
6867 else if (var = find_variable_last_nameref (name, 0))
6868 {
6869 temp = nameref_cell (var);
6870 #if defined (ARRAY_VARS)
6871 /* Handle expanding nameref whose value is x[n] */
6872 if (temp && *temp && valid_array_reference (temp, 0))
6873 {
6874 name = temp;
6875 goto expand_arrayref;
6876 }
6877 else
6878 #endif
6879 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6880 if (temp && *temp && legal_identifier (temp) == 0)
6881 {
6882 set_exit_status (EXECUTION_FAILURE);
6883 report_error (_("%s: invalid variable name for name reference"), temp);
6884 temp = &expand_param_error;
6885 }
6886 else
6887 temp = (char *)NULL;
6888 }
6889 else
6890 temp = (char *)NULL;
6891
6892 if (ret == 0)
6893 {
6894 ret = alloc_word_desc ();
6895 ret->word = temp;
6896 ret->flags |= rflags;
6897 }
6898 return ret;
6899 }
6900
6901 static char *
6902 parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
6903 char *name;
6904 int var_is_special, quoted, find_nameref;
6905 {
6906 char *temp, *t;
6907 WORD_DESC *w;
6908 SHELL_VAR *v;
6909 int pflags, oldex;
6910
6911 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
6912 nameref_p (v) && (t = nameref_cell (v)) && *t)
6913 return (savestring (t));
6914
6915 /* If var_is_special == 0, and name is not an array reference, this does
6916 more expansion than necessary. It should really look up the variable's
6917 value and not try to expand it. */
6918 pflags = PF_IGNUNBOUND;
6919 /* Note that we're not going to be doing word splitting here */
6920 if (var_is_special)
6921 {
6922 pflags |= PF_ASSIGNRHS; /* suppresses word splitting */
6923 oldex = expand_no_split_dollar_star;
6924 expand_no_split_dollar_star = 1;
6925 }
6926 w = parameter_brace_expand_word (name, var_is_special, quoted, pflags, 0);
6927 if (var_is_special)
6928 expand_no_split_dollar_star = oldex;
6929
6930 t = w->word;
6931 /* Have to dequote here if necessary */
6932 if (t)
6933 {
6934 temp = ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || var_is_special)
6935 ? dequote_string (t)
6936 : dequote_escapes (t);
6937 free (t);
6938 t = temp;
6939 }
6940 dispose_word_desc (w);
6941
6942 return t;
6943 }
6944
6945 /* Expand an indirect reference to a variable: ${!NAME} expands to the
6946 value of the variable whose name is the value of NAME. */
6947 static WORD_DESC *
6948 parameter_brace_expand_indir (name, var_is_special, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
6949 char *name;
6950 int var_is_special, quoted, pflags;
6951 int *quoted_dollar_atp, *contains_dollar_at;
6952 {
6953 char *t;
6954 WORD_DESC *w;
6955 SHELL_VAR *v;
6956
6957 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6958 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6959 bash performs an indirect lookup on foo[0] and expands the result;
6960 ksh93 expands bar[0]. We could do that here -- there are enough usable
6961 primitives to do that -- but do not at this point. */
6962 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
6963 {
6964 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6965 {
6966 w = alloc_word_desc ();
6967 w->word = savestring (t);
6968 w->flags = 0;
6969 return w;
6970 }
6971 }
6972
6973 /* An indirect reference to a positional parameter or a special parameter
6974 is ok. Indirect references to array references, as explained above, are
6975 ok (currently). Only references to unset variables are errors at this
6976 point. */
6977 if (legal_identifier (name) && v == 0)
6978 {
6979 report_error (_("%s: invalid indirect expansion"), name);
6980 w = alloc_word_desc ();
6981 w->word = &expand_param_error;
6982 w->flags = 0;
6983 return (w);
6984 }
6985
6986 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6987
6988 chk_atstar (t, quoted, pflags, quoted_dollar_atp, contains_dollar_at);
6989
6990 #if defined (ARRAY_VARS)
6991 /* Array references to unset variables are also an error */
6992 if (t == 0 && valid_array_reference (name, 0))
6993 {
6994 v = array_variable_part (name, 0, (char **)0, (int *)0);
6995 if (v == 0)
6996 {
6997 report_error (_("%s: invalid indirect expansion"), name);
6998 w = alloc_word_desc ();
6999 w->word = &expand_param_error;
7000 w->flags = 0;
7001 return (w);
7002 }
7003 else
7004 return (WORD_DESC *)NULL;
7005 }
7006 #endif
7007
7008 if (t == 0)
7009 return (WORD_DESC *)NULL;
7010
7011 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
7012 {
7013 report_error (_("%s: invalid variable name"), t);
7014 free (t);
7015 w = alloc_word_desc ();
7016 w->word = &expand_param_error;
7017 w->flags = 0;
7018 return (w);
7019 }
7020
7021 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, pflags, 0);
7022 free (t);
7023
7024 return w;
7025 }
7026
7027 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
7028 depending on the value of C, the separating character. C can be one of
7029 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
7030 between double quotes. */
7031 static WORD_DESC *
7032 parameter_brace_expand_rhs (name, value, op, quoted, pflags, qdollaratp, hasdollarat)
7033 char *name, *value;
7034 int op, quoted, pflags, *qdollaratp, *hasdollarat;
7035 {
7036 WORD_DESC *w;
7037 WORD_LIST *l, *tl;
7038 char *t, *t1, *temp, *vname;
7039 int l_hasdollat, sindex;
7040 SHELL_VAR *v;
7041
7042 /*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
7043 /* If the entire expression is between double quotes, we want to treat
7044 the value as a double-quoted string, with the exception that we strip
7045 embedded unescaped double quotes (for sh backwards compatibility). */
7046 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
7047 {
7048 sindex = 0;
7049 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
7050 }
7051 else
7052 temp = value;
7053
7054 w = alloc_word_desc ();
7055 l_hasdollat = 0;
7056 l = *temp ? expand_string_for_rhs (temp, quoted, op, pflags, &l_hasdollat, (int *)NULL)
7057 : (WORD_LIST *)0;
7058 if (hasdollarat)
7059 *hasdollarat = l_hasdollat || (l && l->next);
7060 if (temp != value)
7061 free (temp);
7062
7063 /* list_string takes multiple CTLNULs and turns them into an empty word
7064 with W_SAWQUOTEDNULL set. Turn it back into a single CTLNUL for the
7065 rest of this function and the caller. */
7066 for (tl = l; tl; tl = tl->next)
7067 {
7068 if (tl->word && (tl->word->word == 0 || tl->word->word[0] == 0) &&
7069 (tl->word->flags | W_SAWQUOTEDNULL))
7070 {
7071 t = make_quoted_char ('\0');
7072 FREE (tl->word->word);
7073 tl->word->word = t;
7074 tl->word->flags |= W_QUOTED|W_HASQUOTEDNULL;
7075 tl->word->flags &= ~W_SAWQUOTEDNULL;
7076 }
7077 }
7078
7079 if (l)
7080 {
7081 /* If l->next is not null, we know that TEMP contained "$@", since that
7082 is the only expansion that creates more than one word. */
7083 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
7084 {
7085 /*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
7086 *qdollaratp = 1;
7087 }
7088
7089 /* The expansion of TEMP returned something. We need to treat things
7090 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
7091 the individual words have already been quoted. We need to turn them
7092 into a string with the words separated by the first character of
7093 $IFS without any additional quoting, so string_list_dollar_at won't
7094 do the right thing. If IFS is null, we want "$@" to split into
7095 separate arguments, not be concatenated, so we use string_list_internal
7096 and mark the word to be split on spaces later. We use
7097 string_list_dollar_star for "$@" otherwise. */
7098 if (l->next && ifs_is_null)
7099 {
7100 temp = string_list_internal (l, " ");
7101 w->flags |= W_SPLITSPACE;
7102 }
7103 else if (l_hasdollat || l->next)
7104 temp = string_list_dollar_star (l, quoted, 0);
7105 else
7106 {
7107 temp = string_list (l);
7108 if (temp && (QUOTED_NULL (temp) == 0) && (l->word->flags & W_SAWQUOTEDNULL))
7109 w->flags |= W_SAWQUOTEDNULL; /* XXX */
7110 }
7111
7112 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
7113 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
7114 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
7115 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7116 (which is more paranoia than anything else), we need to return the
7117 quoted null string and set the flags to indicate it. */
7118 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
7119 {
7120 w->flags |= W_HASQUOTEDNULL;
7121 /*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
7122 /* If we return a quoted null with L_HASDOLLARAT, we either have a
7123 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
7124 parameters or a quoted expansion of "$@" with $1 == ''. In either
7125 case, we don't want to enable special handling of $@. */
7126 if (qdollaratp && l_hasdollat)
7127 *qdollaratp = 0;
7128 }
7129 dispose_words (l);
7130 }
7131 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
7132 {
7133 /* Posix interp 221 changed the rules on this. The idea is that
7134 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
7135 when foo and xxx are unset. The problem is that it's not in any
7136 way backwards compatible and few other shells do it. We're eventually
7137 going to try and split the difference (heh) a little bit here. */
7138 /* l_hasdollat == 1 means we saw a quoted dollar at. */
7139
7140 /* The brace expansion occurred between double quotes and there was
7141 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
7142 it does not expand to anything. In this case, we want to return
7143 a quoted empty string. Posix interp 888 */
7144 temp = make_quoted_char ('\0');
7145 w->flags |= W_HASQUOTEDNULL;
7146 /*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
7147 }
7148 else
7149 temp = (char *)NULL;
7150
7151 if (op == '-' || op == '+')
7152 {
7153 w->word = temp;
7154 return w;
7155 }
7156
7157 /* op == '=' */
7158 t1 = temp ? dequote_string (temp) : savestring ("");
7159 free (temp);
7160
7161 /* bash-4.4/5.0 */
7162 vname = name;
7163 if (*name == '!' &&
7164 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
7165 {
7166 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
7167 if (vname == 0 || *vname == 0)
7168 {
7169 report_error (_("%s: invalid indirect expansion"), name);
7170 free (vname);
7171 free (t1);
7172 dispose_word (w);
7173 return &expand_wdesc_error;
7174 }
7175 if (legal_identifier (vname) == 0)
7176 {
7177 report_error (_("%s: invalid variable name"), vname);
7178 free (vname);
7179 free (t1);
7180 dispose_word (w);
7181 return &expand_wdesc_error;
7182 }
7183 }
7184
7185 #if defined (ARRAY_VARS)
7186 if (valid_array_reference (vname, 0))
7187 v = assign_array_element (vname, t1, 0);
7188 else
7189 #endif /* ARRAY_VARS */
7190 v = bind_variable (vname, t1, 0);
7191
7192 if (v == 0 || readonly_p (v) || noassign_p (v)) /* expansion error */
7193 {
7194 if ((v == 0 || readonly_p (v)) && interactive_shell == 0 && posixly_correct)
7195 {
7196 last_command_exit_value = EXECUTION_FAILURE;
7197 exp_jump_to_top_level (FORCE_EOF);
7198 }
7199 else
7200 {
7201 if (vname != name)
7202 free (vname);
7203 last_command_exit_value = EX_BADUSAGE;
7204 exp_jump_to_top_level (DISCARD);
7205 }
7206 }
7207
7208 stupidly_hack_special_variables (vname);
7209
7210 if (vname != name)
7211 free (vname);
7212
7213 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
7214
7215 /* If we are double-quoted or if we are not going to be performing word
7216 splitting, we want to quote the value we return appropriately, like
7217 the other expansions this function handles. */
7218 w->word = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (t1) : quote_escapes (t1);
7219 /* If we have something that's non-null, that's not a quoted null string,
7220 and we're not going to be performing word splitting (we know we're not
7221 because the operator is `='), we can forget we saw a quoted null. */
7222 if (w->word && w->word[0] && QUOTED_NULL (w->word) == 0)
7223 w->flags &= ~W_SAWQUOTEDNULL;
7224 free (t1);
7225
7226 /* If we convert a null string into a quoted null, make sure the caller
7227 knows it. */
7228 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && QUOTED_NULL (w->word))
7229 w->flags |= W_HASQUOTEDNULL;
7230
7231 return w;
7232 }
7233
7234 /* Deal with the right hand side of a ${name:?value} expansion in the case
7235 that NAME is null or not set. If VALUE is non-null it is expanded and
7236 used as the error message to print, otherwise a standard message is
7237 printed. */
7238 static void
7239 parameter_brace_expand_error (name, value, check_null)
7240 char *name, *value;
7241 int check_null;
7242 {
7243 WORD_LIST *l;
7244 char *temp;
7245
7246 set_exit_status (EXECUTION_FAILURE); /* ensure it's non-zero */
7247 if (value && *value)
7248 {
7249 l = expand_string (value, 0);
7250 temp = string_list (l);
7251 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
7252 FREE (temp);
7253 dispose_words (l);
7254 }
7255 else if (check_null == 0)
7256 report_error (_("%s: parameter not set"), name);
7257 else
7258 report_error (_("%s: parameter null or not set"), name);
7259
7260 /* Free the data we have allocated during this expansion, since we
7261 are about to longjmp out. */
7262 free (name);
7263 FREE (value);
7264 }
7265
7266 /* Return 1 if NAME is something for which parameter_brace_expand_length is
7267 OK to do. */
7268 static int
7269 valid_length_expression (name)
7270 char *name;
7271 {
7272 return (name[1] == '\0' || /* ${#} */
7273 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
7274 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
7275 #if defined (ARRAY_VARS)
7276 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
7277 #endif
7278 legal_identifier (name + 1)); /* ${#PS1} */
7279 }
7280
7281 /* Handle the parameter brace expansion that requires us to return the
7282 length of a parameter. */
7283 static intmax_t
7284 parameter_brace_expand_length (name)
7285 char *name;
7286 {
7287 char *t, *newname;
7288 intmax_t number, arg_index;
7289 WORD_LIST *list;
7290 SHELL_VAR *var;
7291
7292 var = (SHELL_VAR *)NULL;
7293
7294 if (name[1] == '\0') /* ${#} */
7295 number = number_of_args ();
7296 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
7297 number = number_of_args ();
7298 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
7299 {
7300 /* Take the lengths of some of the shell's special parameters. */
7301 switch (name[1])
7302 {
7303 case '-':
7304 t = which_set_flags ();
7305 break;
7306 case '?':
7307 t = itos (last_command_exit_value);
7308 break;
7309 case '$':
7310 t = itos (dollar_dollar_pid);
7311 break;
7312 case '!':
7313 if (last_asynchronous_pid == NO_PID)
7314 t = (char *)NULL; /* XXX - error if set -u set? */
7315 else
7316 t = itos (last_asynchronous_pid);
7317 break;
7318 case '#':
7319 t = itos (number_of_args ());
7320 break;
7321 }
7322 number = STRLEN (t);
7323 FREE (t);
7324 }
7325 #if defined (ARRAY_VARS)
7326 else if (valid_array_reference (name + 1, 0))
7327 number = array_length_reference (name + 1);
7328 #endif /* ARRAY_VARS */
7329 else
7330 {
7331 number = 0;
7332
7333 if (legal_number (name + 1, &arg_index)) /* ${#1} */
7334 {
7335 t = get_dollar_var_value (arg_index);
7336 if (t == 0 && unbound_vars_is_error)
7337 return INTMAX_MIN;
7338 number = MB_STRLEN (t);
7339 FREE (t);
7340 }
7341 #if defined (ARRAY_VARS)
7342 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
7343 {
7344 if (assoc_p (var))
7345 t = assoc_reference (assoc_cell (var), "0");
7346 else
7347 t = array_reference (array_cell (var), 0);
7348 if (t == 0 && unbound_vars_is_error)
7349 return INTMAX_MIN;
7350 number = MB_STRLEN (t);
7351 }
7352 #endif
7353 /* Fast path for the common case of taking the length of a non-dynamic
7354 scalar variable value. */
7355 else if ((var || (var = find_variable (name + 1))) &&
7356 invisible_p (var) == 0 &&
7357 array_p (var) == 0 && assoc_p (var) == 0 &&
7358 var->dynamic_value == 0)
7359 number = value_cell (var) ? MB_STRLEN (value_cell (var)) : 0;
7360 else if (var == 0 && unbound_vars_is_error == 0)
7361 number = 0;
7362 else /* ${#PS1} */
7363 {
7364 newname = savestring (name);
7365 newname[0] = '$';
7366 list = expand_string (newname, Q_DOUBLE_QUOTES);
7367 t = list ? string_list (list) : (char *)NULL;
7368 free (newname);
7369 if (list)
7370 dispose_words (list);
7371
7372 number = t ? MB_STRLEN (t) : 0;
7373 FREE (t);
7374 }
7375 }
7376
7377 return (number);
7378 }
7379
7380 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
7381 so we do some ad-hoc parsing of an arithmetic expression to find
7382 the first DELIM, instead of using strchr(3). Two rules:
7383 1. If the substring contains a `(', read until closing `)'.
7384 2. If the substring contains a `?', read past one `:' for each `?'.
7385 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
7386 */
7387
7388 static char *
7389 skiparith (substr, delim)
7390 char *substr;
7391 int delim;
7392 {
7393 int i;
7394 char delims[2];
7395
7396 delims[0] = delim;
7397 delims[1] = '\0';
7398
7399 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
7400 return (substr + i);
7401 }
7402
7403 /* Verify and limit the start and end of the desired substring. If
7404 VTYPE == 0, a regular shell variable is being used; if it is 1,
7405 then the positional parameters are being used; if it is 2, then
7406 VALUE is really a pointer to an array variable that should be used.
7407 Return value is 1 if both values were OK, 0 if there was a problem
7408 with an invalid expression, or -1 if the values were out of range. */
7409 static int
7410 verify_substring_values (v, value, substr, vtype, e1p, e2p)
7411 SHELL_VAR *v;
7412 char *value, *substr;
7413 int vtype;
7414 intmax_t *e1p, *e2p;
7415 {
7416 char *t, *temp1, *temp2;
7417 arrayind_t len;
7418 int expok;
7419 #if defined (ARRAY_VARS)
7420 ARRAY *a;
7421 HASH_TABLE *h;
7422 #endif
7423
7424 /* duplicate behavior of strchr(3) */
7425 t = skiparith (substr, ':');
7426 if (*t && *t == ':')
7427 *t = '\0';
7428 else
7429 t = (char *)0;
7430
7431 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
7432 *e1p = evalexp (temp1, 0, &expok); /* XXX - EXP_EXPANDED? */
7433 free (temp1);
7434 if (expok == 0)
7435 return (0);
7436
7437 len = -1; /* paranoia */
7438 switch (vtype)
7439 {
7440 case VT_VARIABLE:
7441 case VT_ARRAYMEMBER:
7442 len = MB_STRLEN (value);
7443 break;
7444 case VT_POSPARMS:
7445 len = number_of_args () + 1;
7446 if (*e1p == 0)
7447 len++; /* add one arg if counting from $0 */
7448 break;
7449 #if defined (ARRAY_VARS)
7450 case VT_ARRAYVAR:
7451 /* For arrays, the first value deals with array indices. Negative
7452 offsets count from one past the array's maximum index. Associative
7453 arrays treat the number of elements as the maximum index. */
7454 if (assoc_p (v))
7455 {
7456 h = assoc_cell (v);
7457 len = assoc_num_elements (h) + (*e1p < 0);
7458 }
7459 else
7460 {
7461 a = (ARRAY *)value;
7462 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7463 }
7464 break;
7465 #endif
7466 }
7467
7468 if (len == -1) /* paranoia */
7469 return -1;
7470
7471 if (*e1p < 0) /* negative offsets count from end */
7472 *e1p += len;
7473
7474 if (*e1p > len || *e1p < 0)
7475 return (-1);
7476
7477 #if defined (ARRAY_VARS)
7478 /* For arrays, the second offset deals with the number of elements. */
7479 if (vtype == VT_ARRAYVAR)
7480 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
7481 #endif
7482
7483 if (t)
7484 {
7485 t++;
7486 temp2 = savestring (t);
7487 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7488 free (temp2);
7489 t[-1] = ':';
7490 *e2p = evalexp (temp1, 0, &expok); /* XXX - EXP_EXPANDED? */
7491 free (temp1);
7492 if (expok == 0)
7493 return (0);
7494
7495 /* Should we allow positional parameter length < 0 to count backwards
7496 from end of positional parameters? */
7497 #if 1
7498 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
7499 #else /* TAG: bash-5.2 */
7500 if (vtype == VT_ARRAYVAR && *e2p < 0)
7501 #endif
7502 {
7503 internal_error (_("%s: substring expression < 0"), t);
7504 return (0);
7505 }
7506 #if defined (ARRAY_VARS)
7507 /* In order to deal with sparse arrays, push the intelligence about how
7508 to deal with the number of elements desired down to the array-
7509 specific functions. */
7510 if (vtype != VT_ARRAYVAR)
7511 #endif
7512 {
7513 if (*e2p < 0)
7514 {
7515 *e2p += len;
7516 if (*e2p < 0 || *e2p < *e1p)
7517 {
7518 internal_error (_("%s: substring expression < 0"), t);
7519 return (0);
7520 }
7521 }
7522 else
7523 *e2p += *e1p; /* want E2 chars starting at E1 */
7524 if (*e2p > len)
7525 *e2p = len;
7526 }
7527 }
7528 else
7529 *e2p = len;
7530
7531 return (1);
7532 }
7533
7534 /* Return the type of variable specified by VARNAME (simple variable,
7535 positional param, or array variable). Also return the value specified
7536 by VARNAME (value of a variable or a reference to an array element).
7537 QUOTED is the standard description of quoting state, using Q_* defines.
7538 FLAGS is currently a set of flags to pass to array_value. If IND is
7539 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7540 passed to array_value so the array index is not computed again.
7541 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7542 characters in the value are quoted with CTLESC and takes appropriate
7543 steps. For convenience, *VALP is set to the dequoted VALUE. */
7544 static int
7545 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
7546 char *varname, *value;
7547 arrayind_t ind;
7548 int quoted, flags;
7549 SHELL_VAR **varp;
7550 char **valp;
7551 {
7552 int vtype, want_indir;
7553 char *temp, *vname;
7554 SHELL_VAR *v;
7555 arrayind_t lind;
7556
7557 want_indir = *varname == '!' &&
7558 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7559 || VALID_INDIR_PARAM (varname[1]));
7560 if (want_indir)
7561 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
7562 /* XXX - what if vname == 0 || *vname == 0 ? */
7563 else
7564 vname = varname;
7565
7566 if (vname == 0)
7567 {
7568 vtype = VT_VARIABLE;
7569 *varp = (SHELL_VAR *)NULL;
7570 *valp = (char *)NULL;
7571 return (vtype);
7572 }
7573
7574 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
7575 vtype = STR_DOLLAR_AT_STAR (vname);
7576 if (vtype == VT_POSPARMS && vname[0] == '*')
7577 vtype |= VT_STARSUB;
7578 *varp = (SHELL_VAR *)NULL;
7579
7580 #if defined (ARRAY_VARS)
7581 if (valid_array_reference (vname, 0))
7582 {
7583 v = array_variable_part (vname, 0, &temp, (int *)0);
7584 /* If we want to signal array_value to use an already-computed index,
7585 set LIND to that index */
7586 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
7587 if (v && invisible_p (v))
7588 {
7589 vtype = VT_ARRAYMEMBER;
7590 *varp = (SHELL_VAR *)NULL;
7591 *valp = (char *)NULL;
7592 }
7593 if (v && (array_p (v) || assoc_p (v)))
7594 {
7595 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == RBRACK)
7596 {
7597 /* Callers have to differentiate between indexed and associative */
7598 vtype = VT_ARRAYVAR;
7599 if (temp[0] == '*')
7600 vtype |= VT_STARSUB;
7601 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
7602 }
7603 else
7604 {
7605 vtype = VT_ARRAYMEMBER;
7606 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7607 }
7608 *varp = v;
7609 }
7610 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == RBRACK))
7611 {
7612 vtype = VT_VARIABLE;
7613 *varp = v;
7614 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7615 *valp = value ? dequote_string (value) : (char *)NULL;
7616 else
7617 *valp = value ? dequote_escapes (value) : (char *)NULL;
7618 }
7619 else
7620 {
7621 vtype = VT_ARRAYMEMBER;
7622 *varp = v;
7623 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7624 }
7625 }
7626 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
7627 {
7628 vtype = VT_ARRAYMEMBER;
7629 *varp = v;
7630 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
7631 }
7632 else
7633 #endif
7634 {
7635 if (value && vtype == VT_VARIABLE)
7636 {
7637 *varp = find_variable (vname);
7638 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7639 *valp = dequote_string (value);
7640 else
7641 *valp = dequote_escapes (value);
7642 }
7643 else
7644 *valp = value;
7645 }
7646
7647 if (want_indir)
7648 free (vname);
7649
7650 return vtype;
7651 }
7652
7653 /***********************************************************/
7654 /* */
7655 /* Functions to perform transformations on variable values */
7656 /* */
7657 /***********************************************************/
7658
7659 static char *
7660 string_var_assignment (v, s)
7661 SHELL_VAR *v;
7662 char *s;
7663 {
7664 char flags[MAX_ATTRIBUTES], *ret, *val;
7665 int i;
7666
7667 val = (v && (invisible_p (v) || var_isset (v) == 0)) ? (char *)NULL : sh_quote_reusable (s, 0);
7668 i = var_attribute_string (v, 0, flags);
7669 if (i == 0 && val == 0)
7670 return (char *)NULL;
7671
7672 ret = (char *)xmalloc (i + STRLEN (val) + strlen (v->name) + 16 + MAX_ATTRIBUTES);
7673 if (i > 0 && val == 0)
7674 sprintf (ret, "declare -%s %s", flags, v->name);
7675 else if (i > 0)
7676 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
7677 else
7678 sprintf (ret, "%s=%s", v->name, val);
7679 free (val);
7680 return ret;
7681 }
7682
7683 #if defined (ARRAY_VARS)
7684 static char *
7685 array_var_assignment (v, itype, quoted, atype)
7686 SHELL_VAR *v;
7687 int itype, quoted, atype;
7688 {
7689 char *ret, *val, flags[MAX_ATTRIBUTES];
7690 int i;
7691
7692 if (v == 0)
7693 return (char *)NULL;
7694 if (atype == 2)
7695 val = array_p (v) ? array_to_kvpair (array_cell (v), 0)
7696 : assoc_to_kvpair (assoc_cell (v), 0);
7697 else
7698 val = array_p (v) ? array_to_assign (array_cell (v), 0)
7699 : assoc_to_assign (assoc_cell (v), 0);
7700
7701 if (val == 0 && (invisible_p (v) || var_isset (v) == 0))
7702 ; /* placeholder */
7703 else if (val == 0)
7704 {
7705 val = (char *)xmalloc (3);
7706 val[0] = LPAREN;
7707 val[1] = RPAREN;
7708 val[2] = 0;
7709 }
7710 else
7711 {
7712 ret = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (val) : quote_escapes (val);
7713 free (val);
7714 val = ret;
7715 }
7716
7717 if (atype == 2)
7718 return val;
7719
7720 i = var_attribute_string (v, 0, flags);
7721 ret = (char *)xmalloc (i + STRLEN (val) + strlen (v->name) + 16);
7722 if (val)
7723 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
7724 else
7725 sprintf (ret, "declare -%s %s", flags, v->name);
7726 free (val);
7727 return ret;
7728 }
7729 #endif
7730
7731 static char *
7732 pos_params_assignment (list, itype, quoted)
7733 WORD_LIST *list;
7734 int itype;
7735 int quoted;
7736 {
7737 char *temp, *ret;
7738
7739 /* first, we transform the list to quote each word. */
7740 temp = list_transform ('Q', (SHELL_VAR *)0, list, itype, quoted);
7741 ret = (char *)xmalloc (strlen (temp) + 8);
7742 strcpy (ret, "set -- ");
7743 strcpy (ret + 7, temp);
7744 free (temp);
7745 return ret;
7746 }
7747
7748 static char *
7749 string_transform (xc, v, s)
7750 int xc;
7751 SHELL_VAR *v;
7752 char *s;
7753 {
7754 char *ret, flags[MAX_ATTRIBUTES], *t;
7755 int i;
7756
7757 if (((xc == 'A' || xc == 'a') && v == 0))
7758 return (char *)NULL;
7759 else if (xc != 'a' && xc != 'A' && s == 0)
7760 return (char *)NULL;
7761
7762 switch (xc)
7763 {
7764 /* Transformations that interrogate the variable */
7765 case 'a':
7766 i = var_attribute_string (v, 0, flags);
7767 ret = (i > 0) ? savestring (flags) : (char *)NULL;
7768 break;
7769 case 'A':
7770 ret = string_var_assignment (v, s);
7771 break;
7772 case 'K':
7773 ret = sh_quote_reusable (s, 0);
7774 break;
7775 /* Transformations that modify the variable's value */
7776 case 'E':
7777 t = ansiexpand (s, 0, strlen (s), (int *)0);
7778 ret = dequote_escapes (t);
7779 free (t);
7780 break;
7781 case 'P':
7782 ret = decode_prompt_string (s);
7783 break;
7784 case 'Q':
7785 ret = sh_quote_reusable (s, 0);
7786 break;
7787 case 'U':
7788 ret = sh_modcase (s, 0, CASE_UPPER);
7789 break;
7790 case 'u':
7791 ret = sh_modcase (s, 0, CASE_UPFIRST); /* capitalize */
7792 break;
7793 case 'L':
7794 ret = sh_modcase (s, 0, CASE_LOWER);
7795 break;
7796 default:
7797 ret = (char *)NULL;
7798 break;
7799 }
7800 return ret;
7801 }
7802
7803 static char *
7804 list_transform (xc, v, list, itype, quoted)
7805 int xc;
7806 SHELL_VAR *v;
7807 WORD_LIST *list;
7808 int itype, quoted;
7809 {
7810 WORD_LIST *new, *l;
7811 WORD_DESC *w;
7812 char *tword;
7813 int qflags;
7814
7815 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
7816 {
7817 tword = string_transform (xc, v, l->word->word);
7818 w = alloc_word_desc ();
7819 w->word = tword ? tword : savestring (""); /* XXX */
7820 new = make_word_list (w, new);
7821 }
7822 l = REVERSE_LIST (new, WORD_LIST *);
7823
7824 qflags = quoted;
7825 /* If we are expanding in a context where word splitting will not be
7826 performed, treat as quoted. This changes how $* will be expanded. */
7827 if (itype == '*' && expand_no_split_dollar_star && ifs_is_null)
7828 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
7829
7830 tword = string_list_pos_params (itype, l, qflags, 0);
7831 dispose_words (l);
7832
7833 return (tword);
7834 }
7835
7836 static char *
7837 parameter_list_transform (xc, itype, quoted)
7838 int xc;
7839 int itype;
7840 int quoted;
7841 {
7842 char *ret;
7843 WORD_LIST *list;
7844
7845 list = list_rest_of_args ();
7846 if (list == 0)
7847 return ((char *)NULL);
7848 if (xc == 'A')
7849 ret = pos_params_assignment (list, itype, quoted);
7850 else
7851 ret = list_transform (xc, (SHELL_VAR *)0, list, itype, quoted);
7852 dispose_words (list);
7853 return (ret);
7854 }
7855
7856 #if defined (ARRAY_VARS)
7857 static char *
7858 array_transform (xc, var, starsub, quoted)
7859 int xc;
7860 SHELL_VAR *var;
7861 int starsub; /* so we can figure out how it's indexed */
7862 int quoted;
7863 {
7864 ARRAY *a;
7865 HASH_TABLE *h;
7866 int itype;
7867 char *ret;
7868 WORD_LIST *list;
7869 SHELL_VAR *v;
7870
7871 v = var; /* XXX - for now */
7872
7873 itype = starsub ? '*' : '@';
7874
7875 if (xc == 'A')
7876 return (array_var_assignment (v, itype, quoted, 1));
7877 else if (xc == 'K')
7878 return (array_var_assignment (v, itype, quoted, 2));
7879
7880 /* special case for unset arrays and attributes */
7881 if (xc == 'a' && (invisible_p (v) || var_isset (v) == 0))
7882 {
7883 char flags[MAX_ATTRIBUTES];
7884 int i;
7885
7886 i = var_attribute_string (v, 0, flags);
7887 return ((i > 0) ? savestring (flags) : (char *)NULL);
7888 }
7889
7890 a = (v && array_p (v)) ? array_cell (v) : 0;
7891 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
7892
7893 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
7894 if (list == 0)
7895 return ((char *)NULL);
7896 ret = list_transform (xc, v, list, itype, quoted);
7897 dispose_words (list);
7898
7899 return ret;
7900 }
7901 #endif /* ARRAY_VARS */
7902
7903 static int
7904 valid_parameter_transform (xform)
7905 char *xform;
7906 {
7907 if (xform[1])
7908 return 0;
7909
7910 /* check for valid values of xform[0] */
7911 switch (xform[0])
7912 {
7913 case 'a': /* expand to a string with just attributes */
7914 case 'A': /* expand as an assignment statement with attributes */
7915 case 'K': /* expand assoc array to list of key/value pairs */
7916 case 'E': /* expand like $'...' */
7917 case 'P': /* expand like prompt string */
7918 case 'Q': /* quote reusably */
7919 case 'U': /* transform to uppercase */
7920 case 'u': /* tranform by capitalizing */
7921 case 'L': /* transform to lowercase */
7922 return 1;
7923 default:
7924 return 0;
7925 }
7926 }
7927
7928 static char *
7929 parameter_brace_transform (varname, value, ind, xform, rtype, quoted, pflags, flags)
7930 char *varname, *value;
7931 int ind;
7932 char *xform;
7933 int rtype, quoted, pflags, flags;
7934 {
7935 int vtype, xc, starsub;
7936 char *temp1, *val, *oname;
7937 SHELL_VAR *v;
7938
7939 xc = xform[0];
7940 if (value == 0 && xc != 'A' && xc != 'a')
7941 return ((char *)NULL);
7942
7943 oname = this_command_name;
7944 this_command_name = varname;
7945
7946 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7947 if (vtype == -1)
7948 {
7949 this_command_name = oname;
7950 return ((char *)NULL);
7951 }
7952
7953 if (valid_parameter_transform (xform) == 0)
7954 {
7955 this_command_name = oname;
7956 #if 0 /* TAG: bash-5.2 Martin Schulte <gnu@schrader-schulte.de> 10/2020 */
7957 return (interactive_shell ? &expand_param_error : &expand_param_fatal);
7958 #else
7959 return &expand_param_error;
7960 #endif
7961 }
7962
7963 starsub = vtype & VT_STARSUB;
7964 vtype &= ~VT_STARSUB;
7965
7966 /* If we are asked to display the attributes of an unset variable, V will
7967 be NULL after the call to get_var_and_type. Double-check here. */
7968 if ((xc == 'a' || xc == 'A') && vtype == VT_VARIABLE && varname && v == 0)
7969 v = find_variable (varname);
7970
7971 temp1 = (char *)NULL; /* shut up gcc */
7972 switch (vtype)
7973 {
7974 case VT_VARIABLE:
7975 case VT_ARRAYMEMBER:
7976 temp1 = string_transform (xc, v, val);
7977 if (vtype == VT_VARIABLE)
7978 FREE (val);
7979 if (temp1)
7980 {
7981 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7982 ? quote_string (temp1)
7983 : quote_escapes (temp1);
7984 free (temp1);
7985 temp1 = val;
7986 }
7987 break;
7988 #if defined (ARRAY_VARS)
7989 case VT_ARRAYVAR:
7990 temp1 = array_transform (xc, v, starsub, quoted);
7991 if (temp1 && quoted == 0 && ifs_is_null)
7992 {
7993 /* Posix interp 888 */
7994 }
7995 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
7996 {
7997 val = quote_escapes (temp1);
7998 free (temp1);
7999 temp1 = val;
8000 }
8001 break;
8002 #endif
8003 case VT_POSPARMS:
8004 temp1 = parameter_list_transform (xc, varname[0], quoted);
8005 if (temp1 && quoted == 0 && ifs_is_null)
8006 {
8007 /* Posix interp 888 */
8008 }
8009 else if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
8010 {
8011 val = quote_escapes (temp1);
8012 free (temp1);
8013 temp1 = val;
8014 }
8015 break;
8016 }
8017
8018 this_command_name = oname;
8019 return temp1;
8020 }
8021
8022 /******************************************************/
8023 /* */
8024 /* Functions to extract substrings of variable values */
8025 /* */
8026 /******************************************************/
8027
8028 #if defined (HANDLE_MULTIBYTE)
8029 /* Character-oriented rather than strictly byte-oriented substrings. S and
8030 E, rather being strict indices into STRING, indicate character (possibly
8031 multibyte character) positions that require calculation.
8032 Used by the ${param:offset[:length]} expansion. */
8033 static char *
8034 mb_substring (string, s, e)
8035 char *string;
8036 int s, e;
8037 {
8038 char *tt;
8039 int start, stop, i;
8040 size_t slen;
8041 DECLARE_MBSTATE;
8042
8043 start = 0;
8044 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
8045 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
8046
8047 i = s;
8048 while (string[start] && i--)
8049 ADVANCE_CHAR (string, slen, start);
8050 stop = start;
8051 i = e - s;
8052 while (string[stop] && i--)
8053 ADVANCE_CHAR (string, slen, stop);
8054 tt = substring (string, start, stop);
8055 return tt;
8056 }
8057 #endif
8058
8059 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
8060 is `@', use the positional parameters; otherwise, use the value of
8061 VARNAME. If VARNAME is an array variable, use the array elements. */
8062
8063 static char *
8064 parameter_brace_substring (varname, value, ind, substr, quoted, pflags, flags)
8065 char *varname, *value;
8066 int ind;
8067 char *substr;
8068 int quoted, pflags, flags;
8069 {
8070 intmax_t e1, e2;
8071 int vtype, r, starsub;
8072 char *temp, *val, *tt, *oname;
8073 SHELL_VAR *v;
8074
8075 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
8076 return ((char *)NULL);
8077
8078 oname = this_command_name;
8079 this_command_name = varname;
8080
8081 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
8082 if (vtype == -1)
8083 {
8084 this_command_name = oname;
8085 return ((char *)NULL);
8086 }
8087
8088 starsub = vtype & VT_STARSUB;
8089 vtype &= ~VT_STARSUB;
8090
8091 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
8092 this_command_name = oname;
8093 if (r <= 0)
8094 {
8095 if (vtype == VT_VARIABLE)
8096 FREE (val);
8097 return ((r == 0) ? &expand_param_error : (char *)NULL);
8098 }
8099
8100 switch (vtype)
8101 {
8102 case VT_VARIABLE:
8103 case VT_ARRAYMEMBER:
8104 #if defined (HANDLE_MULTIBYTE)
8105 if (MB_CUR_MAX > 1)
8106 tt = mb_substring (val, e1, e2);
8107 else
8108 #endif
8109 tt = substring (val, e1, e2);
8110
8111 if (vtype == VT_VARIABLE)
8112 FREE (val);
8113 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
8114 temp = quote_string (tt);
8115 else
8116 temp = tt ? quote_escapes (tt) : (char *)NULL;
8117 FREE (tt);
8118 break;
8119 case VT_POSPARMS:
8120 case VT_ARRAYVAR:
8121 if (vtype == VT_POSPARMS)
8122 tt = pos_params (varname, e1, e2, quoted, pflags);
8123 #if defined (ARRAY_VARS)
8124 /* assoc_subrange and array_subrange both call string_list_pos_params,
8125 so we can treat this case just like VT_POSPARAMS. */
8126 else if (assoc_p (v))
8127 /* we convert to list and take first e2 elements starting at e1th
8128 element -- officially undefined for now */
8129 tt = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted, pflags);
8130 else
8131 /* We want E2 to be the number of elements desired (arrays can be
8132 sparse, so verify_substring_values just returns the numbers
8133 specified and we rely on array_subrange to understand how to
8134 deal with them). */
8135 tt = array_subrange (array_cell (v), e1, e2, starsub, quoted, pflags);
8136 #endif
8137 /* We want to leave this alone in every case where pos_params/
8138 string_list_pos_params quotes the list members */
8139 if (tt && quoted == 0 && ifs_is_null)
8140 {
8141 temp = tt; /* Posix interp 888 */
8142 }
8143 else if (tt && quoted == 0 && (pflags & PF_ASSIGNRHS))
8144 {
8145 temp = tt; /* Posix interp 888 */
8146 }
8147 else if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
8148 {
8149 temp = tt ? quote_escapes (tt) : (char *)NULL;
8150 FREE (tt);
8151 }
8152 else
8153 temp = tt;
8154 break;
8155
8156 default:
8157 temp = (char *)NULL;
8158 }
8159
8160 return temp;
8161 }
8162
8163 /****************************************************************/
8164 /* */
8165 /* Functions to perform pattern substitution on variable values */
8166 /* */
8167 /****************************************************************/
8168
8169 #ifdef INCLUDE_UNUSED
8170 static int
8171 shouldexp_replacement (s)
8172 char *s;
8173 {
8174 register char *p;
8175
8176 for (p = s; p && *p; p++)
8177 {
8178 if (*p == '\\')
8179 p++;
8180 else if (*p == '&')
8181 return 1;
8182 }
8183 return 0;
8184 }
8185 #endif
8186
8187 char *
8188 pat_subst (string, pat, rep, mflags)
8189 char *string, *pat, *rep;
8190 int mflags;
8191 {
8192 char *ret, *s, *e, *str, *rstr, *mstr, *send;
8193 int rptr, mtype, rxpand, mlen;
8194 size_t rsize, l, replen, rslen;
8195 DECLARE_MBSTATE;
8196
8197 if (string == 0)
8198 return (savestring (""));
8199
8200 mtype = mflags & MATCH_TYPEMASK;
8201
8202 #if 0 /* TAG: bash-5.2? */
8203 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
8204 #else
8205 rxpand = 0;
8206 #endif
8207
8208 /* Special cases:
8209 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
8210 * with REP and return the result.
8211 * 2. A null pattern with mtype == MATCH_END means to append REP to
8212 * STRING and return the result.
8213 * 3. A null STRING with a matching pattern means to append REP to
8214 * STRING and return the result.
8215 * These don't understand or process `&' in the replacement string.
8216 */
8217 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
8218 {
8219 replen = STRLEN (rep);
8220 l = STRLEN (string);
8221 ret = (char *)xmalloc (replen + l + 2);
8222 if (replen == 0)
8223 strcpy (ret, string);
8224 else if (mtype == MATCH_BEG)
8225 {
8226 strcpy (ret, rep);
8227 strcpy (ret + replen, string);
8228 }
8229 else
8230 {
8231 strcpy (ret, string);
8232 strcpy (ret + l, rep);
8233 }
8234 return (ret);
8235 }
8236 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
8237 {
8238 replen = STRLEN (rep);
8239 ret = (char *)xmalloc (replen + 1);
8240 if (replen == 0)
8241 ret[0] = '\0';
8242 else
8243 strcpy (ret, rep);
8244 return (ret);
8245 }
8246
8247 ret = (char *)xmalloc (rsize = 64);
8248 ret[0] = '\0';
8249 send = string + strlen (string);
8250
8251 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
8252 {
8253 if (match_pattern (str, pat, mtype, &s, &e) == 0)
8254 break;
8255 l = s - str;
8256
8257 if (rep && rxpand)
8258 {
8259 int x;
8260 mlen = e - s;
8261 mstr = xmalloc (mlen + 1);
8262 for (x = 0; x < mlen; x++)
8263 mstr[x] = s[x];
8264 mstr[mlen] = '\0';
8265 rstr = strcreplace (rep, '&', mstr, 0);
8266 free (mstr);
8267 rslen = strlen (rstr);
8268 }
8269 else
8270 {
8271 rstr = rep;
8272 rslen = replen;
8273 }
8274
8275 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
8276
8277 /* OK, now copy the leading unmatched portion of the string (from
8278 str to s) to ret starting at rptr (the current offset). Then copy
8279 the replacement string at ret + rptr + (s - str). Increment
8280 rptr (if necessary) and str and go on. */
8281 if (l)
8282 {
8283 strncpy (ret + rptr, str, l);
8284 rptr += l;
8285 }
8286 if (replen)
8287 {
8288 strncpy (ret + rptr, rstr, rslen);
8289 rptr += rslen;
8290 }
8291 str = e; /* e == end of match */
8292
8293 if (rstr != rep)
8294 free (rstr);
8295
8296 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
8297 break;
8298
8299 if (s == e)
8300 {
8301 /* On a zero-length match, make sure we copy one character, since
8302 we increment one character to avoid infinite recursion. */
8303 char *p, *origp, *origs;
8304 size_t clen;
8305
8306 RESIZE_MALLOCED_BUFFER (ret, rptr, locale_mb_cur_max, rsize, 64);
8307 #if defined (HANDLE_MULTIBYTE)
8308 p = origp = ret + rptr;
8309 origs = str;
8310 COPY_CHAR_P (p, str, send);
8311 rptr += p - origp;
8312 e += str - origs;
8313 #else
8314 ret[rptr++] = *str++;
8315 e++; /* avoid infinite recursion on zero-length match */
8316 #endif
8317 }
8318 }
8319
8320 /* Now copy the unmatched portion of the input string */
8321 if (str && *str)
8322 {
8323 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
8324 strcpy (ret + rptr, str);
8325 }
8326 else
8327 ret[rptr] = '\0';
8328
8329 return ret;
8330 }
8331
8332 /* Do pattern match and replacement on the positional parameters. */
8333 static char *
8334 pos_params_pat_subst (string, pat, rep, mflags)
8335 char *string, *pat, *rep;
8336 int mflags;
8337 {
8338 WORD_LIST *save, *params;
8339 WORD_DESC *w;
8340 char *ret;
8341 int pchar, qflags, pflags;
8342
8343 save = params = list_rest_of_args ();
8344 if (save == 0)
8345 return ((char *)NULL);
8346
8347 for ( ; params; params = params->next)
8348 {
8349 ret = pat_subst (params->word->word, pat, rep, mflags);
8350 w = alloc_word_desc ();
8351 w->word = ret ? ret : savestring ("");
8352 dispose_word (params->word);
8353 params->word = w;
8354 }
8355
8356 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
8357 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
8358 pflags = (mflags & MATCH_ASSIGNRHS) == MATCH_ASSIGNRHS ? PF_ASSIGNRHS : 0;
8359
8360 /* If we are expanding in a context where word splitting will not be
8361 performed, treat as quoted. This changes how $* will be expanded. */
8362 if (pchar == '*' && (mflags & MATCH_ASSIGNRHS) && expand_no_split_dollar_star && ifs_is_null)
8363 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
8364
8365 ret = string_list_pos_params (pchar, save, qflags, pflags);
8366 dispose_words (save);
8367
8368 return (ret);
8369 }
8370
8371 /* Perform pattern substitution on VALUE, which is the expansion of
8372 VARNAME. PATSUB is an expression supplying the pattern to match
8373 and the string to substitute. QUOTED is a flags word containing
8374 the type of quoting currently in effect. */
8375 static char *
8376 parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
8377 char *varname, *value;
8378 int ind;
8379 char *patsub;
8380 int quoted, pflags, flags;
8381 {
8382 int vtype, mflags, starsub, delim;
8383 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt, *oname;
8384 SHELL_VAR *v;
8385
8386 if (value == 0)
8387 return ((char *)NULL);
8388
8389 oname = this_command_name;
8390 this_command_name = varname; /* error messages */
8391
8392 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
8393 if (vtype == -1)
8394 {
8395 this_command_name = oname;
8396 return ((char *)NULL);
8397 }
8398
8399 starsub = vtype & VT_STARSUB;
8400 vtype &= ~VT_STARSUB;
8401
8402 mflags = 0;
8403 /* PATSUB is never NULL when this is called. */
8404 if (*patsub == '/')
8405 {
8406 mflags |= MATCH_GLOBREP;
8407 patsub++;
8408 }
8409
8410 /* Malloc this because expand_string_if_necessary or one of the expansion
8411 functions in its call chain may free it on a substitution error. */
8412 lpatsub = savestring (patsub);
8413
8414 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8415 mflags |= MATCH_QUOTED;
8416
8417 if (starsub)
8418 mflags |= MATCH_STARSUB;
8419
8420 if (pflags & PF_ASSIGNRHS)
8421 mflags |= MATCH_ASSIGNRHS;
8422
8423 /* If the pattern starts with a `/', make sure we skip over it when looking
8424 for the replacement delimiter. */
8425 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
8426 if (lpatsub[delim] == '/')
8427 {
8428 lpatsub[delim] = 0;
8429 rep = lpatsub + delim + 1;
8430 }
8431 else
8432 rep = (char *)NULL;
8433
8434 if (rep && *rep == '\0')
8435 rep = (char *)NULL;
8436
8437 /* Perform the same expansions on the pattern as performed by the
8438 pattern removal expansions. */
8439 pat = getpattern (lpatsub, quoted, 1);
8440
8441 if (rep)
8442 {
8443 /* We want to perform quote removal on the expanded replacement even if
8444 the entire expansion is double-quoted because the parser and string
8445 extraction functions treated quotes in the replacement string as
8446 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
8447 if (shell_compatibility_level > 42)
8448 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
8449 /* This is the bash-4.2 code. */
8450 else if ((mflags & MATCH_QUOTED) == 0)
8451 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
8452 else
8453 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
8454 }
8455
8456 /* ksh93 doesn't allow the match specifier to be a part of the expanded
8457 pattern. This is an extension. Make sure we don't anchor the pattern
8458 at the beginning or end of the string if we're doing global replacement,
8459 though. */
8460 p = pat;
8461 if (mflags & MATCH_GLOBREP)
8462 mflags |= MATCH_ANY;
8463 else if (pat && pat[0] == '#')
8464 {
8465 mflags |= MATCH_BEG;
8466 p++;
8467 }
8468 else if (pat && pat[0] == '%')
8469 {
8470 mflags |= MATCH_END;
8471 p++;
8472 }
8473 else
8474 mflags |= MATCH_ANY;
8475
8476 /* OK, we now want to substitute REP for PAT in VAL. If
8477 flags & MATCH_GLOBREP is non-zero, the substitution is done
8478 everywhere, otherwise only the first occurrence of PAT is
8479 replaced. The pattern matching code doesn't understand
8480 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
8481 values passed in (VT_VARIABLE) so the pattern substitution
8482 code works right. We need to requote special chars after
8483 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
8484 other cases if QUOTED == 0, since the posparams and arrays
8485 indexed by * or @ do special things when QUOTED != 0. */
8486
8487 switch (vtype)
8488 {
8489 case VT_VARIABLE:
8490 case VT_ARRAYMEMBER:
8491 temp = pat_subst (val, p, rep, mflags);
8492 if (vtype == VT_VARIABLE)
8493 FREE (val);
8494 if (temp)
8495 {
8496 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
8497 free (temp);
8498 temp = tt;
8499 }
8500 break;
8501 case VT_POSPARMS:
8502 /* This does the right thing for the case where we are not performing
8503 word splitting. MATCH_STARSUB restricts it to ${* /foo/bar}, and
8504 pos_params_pat_subst/string_list_pos_params will do the right thing
8505 in turn for the case where ifs_is_null. Posix interp 888 */
8506 if ((pflags & PF_NOSPLIT2) && (mflags & MATCH_STARSUB))
8507 mflags |= MATCH_ASSIGNRHS;
8508 temp = pos_params_pat_subst (val, p, rep, mflags);
8509 if (temp && quoted == 0 && ifs_is_null)
8510 {
8511 /* Posix interp 888 */
8512 }
8513 else if (temp && quoted == 0 && (pflags & PF_ASSIGNRHS))
8514 {
8515 /* Posix interp 888 */
8516 }
8517 else if (temp && (mflags & MATCH_QUOTED) == 0)
8518 {
8519 tt = quote_escapes (temp);
8520 free (temp);
8521 temp = tt;
8522 }
8523 break;
8524 #if defined (ARRAY_VARS)
8525 case VT_ARRAYVAR:
8526 /* If we are expanding in a context where word splitting will not be
8527 performed, treat as quoted. This changes how ${A[*]} will be
8528 expanded to make it identical to $*. */
8529 if ((mflags & MATCH_STARSUB) && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8530 mflags |= MATCH_QUOTED; /* Posix interp 888 */
8531
8532 /* these eventually call string_list_pos_params */
8533 if (assoc_p (v))
8534 temp = assoc_patsub (assoc_cell (v), p, rep, mflags);
8535 else
8536 temp = array_patsub (array_cell (v), p, rep, mflags);
8537
8538 if (temp && quoted == 0 && ifs_is_null)
8539 {
8540 /* Posix interp 888 */
8541 }
8542 else if (temp && (mflags & MATCH_QUOTED) == 0)
8543 {
8544 tt = quote_escapes (temp);
8545 free (temp);
8546 temp = tt;
8547 }
8548 break;
8549 #endif
8550 }
8551
8552 FREE (pat);
8553 FREE (rep);
8554 free (lpatsub);
8555
8556 this_command_name = oname;
8557
8558 return temp;
8559 }
8560
8561 /****************************************************************/
8562 /* */
8563 /* Functions to perform case modification on variable values */
8564 /* */
8565 /****************************************************************/
8566
8567 /* Do case modification on the positional parameters. */
8568
8569 static char *
8570 pos_params_modcase (string, pat, modop, mflags)
8571 char *string, *pat;
8572 int modop;
8573 int mflags;
8574 {
8575 WORD_LIST *save, *params;
8576 WORD_DESC *w;
8577 char *ret;
8578 int pchar, qflags, pflags;
8579
8580 save = params = list_rest_of_args ();
8581 if (save == 0)
8582 return ((char *)NULL);
8583
8584 for ( ; params; params = params->next)
8585 {
8586 ret = sh_modcase (params->word->word, pat, modop);
8587 w = alloc_word_desc ();
8588 w->word = ret ? ret : savestring ("");
8589 dispose_word (params->word);
8590 params->word = w;
8591 }
8592
8593 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
8594 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
8595 pflags = (mflags & MATCH_ASSIGNRHS) == MATCH_ASSIGNRHS ? PF_ASSIGNRHS : 0;
8596
8597 /* If we are expanding in a context where word splitting will not be
8598 performed, treat as quoted. This changes how $* will be expanded. */
8599 if (pchar == '*' && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8600 qflags |= Q_DOUBLE_QUOTES; /* Posix interp 888 */
8601
8602 ret = string_list_pos_params (pchar, save, qflags, pflags);
8603 dispose_words (save);
8604
8605 return (ret);
8606 }
8607
8608 /* Perform case modification on VALUE, which is the expansion of
8609 VARNAME. MODSPEC is an expression supplying the type of modification
8610 to perform. QUOTED is a flags word containing the type of quoting
8611 currently in effect. */
8612 static char *
8613 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, pflags, flags)
8614 char *varname, *value;
8615 int ind, modspec;
8616 char *patspec;
8617 int quoted, pflags, flags;
8618 {
8619 int vtype, starsub, modop, mflags, x;
8620 char *val, *temp, *pat, *p, *lpat, *tt, *oname;
8621 SHELL_VAR *v;
8622
8623 if (value == 0)
8624 return ((char *)NULL);
8625
8626 oname = this_command_name;
8627 this_command_name = varname;
8628
8629 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
8630 if (vtype == -1)
8631 {
8632 this_command_name = oname;
8633 return ((char *)NULL);
8634 }
8635
8636 starsub = vtype & VT_STARSUB;
8637 vtype &= ~VT_STARSUB;
8638
8639 modop = 0;
8640 mflags = 0;
8641 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8642 mflags |= MATCH_QUOTED;
8643 if (starsub)
8644 mflags |= MATCH_STARSUB;
8645 if (pflags & PF_ASSIGNRHS)
8646 mflags |= MATCH_ASSIGNRHS;
8647
8648 p = patspec;
8649 if (modspec == '^')
8650 {
8651 x = p && p[0] == modspec;
8652 modop = x ? CASE_UPPER : CASE_UPFIRST;
8653 p += x;
8654 }
8655 else if (modspec == ',')
8656 {
8657 x = p && p[0] == modspec;
8658 modop = x ? CASE_LOWER : CASE_LOWFIRST;
8659 p += x;
8660 }
8661 else if (modspec == '~')
8662 {
8663 x = p && p[0] == modspec;
8664 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
8665 p += x;
8666 }
8667
8668 lpat = p ? savestring (p) : 0;
8669 /* Perform the same expansions on the pattern as performed by the
8670 pattern removal expansions. */
8671 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
8672
8673 /* OK, now we do the case modification. */
8674 switch (vtype)
8675 {
8676 case VT_VARIABLE:
8677 case VT_ARRAYMEMBER:
8678 temp = sh_modcase (val, pat, modop);
8679 if (vtype == VT_VARIABLE)
8680 FREE (val);
8681 if (temp)
8682 {
8683 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
8684 free (temp);
8685 temp = tt;
8686 }
8687 break;
8688
8689 case VT_POSPARMS:
8690 temp = pos_params_modcase (val, pat, modop, mflags);
8691 if (temp && quoted == 0 && ifs_is_null)
8692 {
8693 /* Posix interp 888 */
8694 }
8695 else if (temp && (mflags & MATCH_QUOTED) == 0)
8696 {
8697 tt = quote_escapes (temp);
8698 free (temp);
8699 temp = tt;
8700 }
8701 break;
8702
8703 #if defined (ARRAY_VARS)
8704 case VT_ARRAYVAR:
8705 /* If we are expanding in a context where word splitting will not be
8706 performed, treat as quoted. This changes how ${A[*]} will be
8707 expanded to make it identical to $*. */
8708 if ((mflags & MATCH_STARSUB) && (mflags & MATCH_ASSIGNRHS) && ifs_is_null)
8709 mflags |= MATCH_QUOTED; /* Posix interp 888 */
8710
8711 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
8712 : array_modcase (array_cell (v), pat, modop, mflags);
8713
8714 if (temp && quoted == 0 && ifs_is_null)
8715 {
8716 /* Posix interp 888 */
8717 }
8718 else if (temp && (mflags & MATCH_QUOTED) == 0)
8719 {
8720 tt = quote_escapes (temp);
8721 free (temp);
8722 temp = tt;
8723 }
8724
8725 break;
8726 #endif
8727 }
8728
8729 FREE (pat);
8730 free (lpat);
8731
8732 this_command_name = oname;
8733
8734 return temp;
8735 }
8736
8737 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
8738 any occur, this must be a nested command substitution, so return 0.
8739 Otherwise, return 1. A valid arithmetic expression must always have a
8740 ( before a matching ), so any cases where there are more right parens
8741 means that this must not be an arithmetic expression, though the parser
8742 will not accept it without a balanced total number of parens. */
8743 static int
8744 chk_arithsub (s, len)
8745 const char *s;
8746 int len;
8747 {
8748 int i, count;
8749 DECLARE_MBSTATE;
8750
8751 i = count = 0;
8752 while (i < len)
8753 {
8754 if (s[i] == LPAREN)
8755 count++;
8756 else if (s[i] == RPAREN)
8757 {
8758 count--;
8759 if (count < 0)
8760 return 0;
8761 }
8762
8763 switch (s[i])
8764 {
8765 default:
8766 ADVANCE_CHAR (s, len, i);
8767 break;
8768
8769 case '\\':
8770 i++;
8771 if (s[i])
8772 ADVANCE_CHAR (s, len, i);
8773 break;
8774
8775 case '\'':
8776 i = skip_single_quoted (s, len, ++i, 0);
8777 break;
8778
8779 case '"':
8780 i = skip_double_quoted ((char *)s, len, ++i, 0);
8781 break;
8782 }
8783 }
8784
8785 return (count == 0);
8786 }
8787
8788 /****************************************************************/
8789 /* */
8790 /* Functions to perform parameter expansion on a string */
8791 /* */
8792 /****************************************************************/
8793
8794 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
8795 static WORD_DESC *
8796 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
8797 char *string;
8798 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
8799 {
8800 int check_nullness, var_is_set, var_is_null, var_is_special;
8801 int want_substring, want_indir, want_patsub, want_casemod, want_attributes;
8802 char *name, *value, *temp, *temp1;
8803 WORD_DESC *tdesc, *ret;
8804 int t_index, sindex, c, tflag, modspec, local_pflags, all_element_arrayref;
8805 intmax_t number;
8806 arrayind_t ind;
8807
8808 temp = temp1 = value = (char *)NULL;
8809 var_is_set = var_is_null = var_is_special = check_nullness = 0;
8810 want_substring = want_indir = want_patsub = want_casemod = want_attributes = 0;
8811
8812 local_pflags = 0;
8813 all_element_arrayref = 0;
8814
8815 sindex = *indexp;
8816 t_index = ++sindex;
8817 /* ${#var} doesn't have any of the other parameter expansions on it. */
8818 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
8819 name = string_extract (string, &t_index, "}", SX_VARNAME);
8820 else
8821 #if defined (CASEMOD_EXPANSIONS)
8822 /* To enable case-toggling expansions using the `~' operator character
8823 define CASEMOD_TOGGLECASE in config-top.h */
8824 # if defined (CASEMOD_TOGGLECASE)
8825 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
8826 # else
8827 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
8828 # endif /* CASEMOD_TOGGLECASE */
8829 #else
8830 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
8831 #endif /* CASEMOD_EXPANSIONS */
8832
8833 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
8834 the cleanest code ever. */
8835 if (*name == 0 && sindex == t_index && string[sindex] == '@')
8836 {
8837 name = (char *)xrealloc (name, 2);
8838 name[0] = '@';
8839 name[1] = '\0';
8840 t_index++;
8841 }
8842 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == RBRACE)
8843 {
8844 name = (char *)xrealloc (name, t_index - sindex + 2);
8845 name[t_index - sindex] = '@';
8846 name[t_index - sindex + 1] = '\0';
8847 t_index++;
8848 }
8849
8850 ret = 0;
8851 tflag = 0;
8852
8853 ind = INTMAX_MIN;
8854
8855 /* If the name really consists of a special variable, then make sure
8856 that we have the entire name. We don't allow indirect references
8857 to special variables except `#', `?', `@' and `*'. This clause is
8858 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
8859 general. */
8860 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
8861 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
8862 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
8863 {
8864 t_index++;
8865 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
8866 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
8867 *name = string[sindex];
8868 if (string[sindex] == '!')
8869 {
8870 /* indirect reference of $#, $?, $@, or $* */
8871 name[1] = string[sindex + 1];
8872 strcpy (name + 2, temp1);
8873 }
8874 else
8875 strcpy (name + 1, temp1);
8876 free (temp1);
8877 }
8878 sindex = t_index;
8879
8880 /* Find out what character ended the variable name. Then
8881 do the appropriate thing. */
8882 if (c = string[sindex])
8883 sindex++;
8884
8885 /* If c is followed by one of the valid parameter expansion
8886 characters, move past it as normal. If not, assume that
8887 a substring specification is being given, and do not move
8888 past it. */
8889 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
8890 {
8891 check_nullness++;
8892 if (c = string[sindex])
8893 sindex++;
8894 }
8895 else if (c == ':' && string[sindex] != RBRACE)
8896 want_substring = 1;
8897 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
8898 want_patsub = 1;
8899 #if defined (CASEMOD_EXPANSIONS)
8900 else if (c == '^' || c == ',' || c == '~')
8901 {
8902 modspec = c;
8903 want_casemod = 1;
8904 }
8905 #endif
8906 else if (c == '@' && (string[sindex] == 'a' || string[sindex] == 'A') && string[sindex+1] == RBRACE)
8907 {
8908 /* special case because we do not want to shortcut foo as foo[0] here */
8909 want_attributes = 1;
8910 local_pflags |= PF_ALLINDS;
8911 }
8912
8913 /* Catch the valid and invalid brace expressions that made it through the
8914 tests above. */
8915 /* ${#-} is a valid expansion and means to take the length of $-.
8916 Similarly for ${#?} and ${##}... */
8917 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8918 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
8919 {
8920 name = (char *)xrealloc (name, 3);
8921 name[1] = c;
8922 name[2] = '\0';
8923 c = string[sindex++];
8924 }
8925
8926 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8927 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8928 member (c, "%:=+/") && string[sindex] == RBRACE)
8929 {
8930 temp = (char *)NULL;
8931 goto bad_substitution; /* XXX - substitution error */
8932 }
8933
8934 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8935 either a variable name, one of the positional parameters or a special
8936 variable that expands to one of the positional parameters. */
8937 want_indir = *name == '!' &&
8938 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
8939 || VALID_INDIR_PARAM (name[1]));
8940
8941 /* Determine the value of this variable whose name is NAME. */
8942
8943 /* Check for special variables, directly referenced. */
8944 if (SPECIAL_VAR (name, want_indir))
8945 var_is_special++;
8946
8947 /* Check for special expansion things, like the length of a parameter */
8948 if (*name == '#' && name[1])
8949 {
8950 /* If we are not pointing at the character just after the
8951 closing brace, then we haven't gotten all of the name.
8952 Since it begins with a special character, this is a bad
8953 substitution. Also check NAME for validity before trying
8954 to go on. */
8955 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
8956 {
8957 temp = (char *)NULL;
8958 goto bad_substitution; /* substitution error */
8959 }
8960
8961 number = parameter_brace_expand_length (name);
8962 if (number == INTMAX_MIN && unbound_vars_is_error)
8963 {
8964 set_exit_status (EXECUTION_FAILURE);
8965 err_unboundvar (name+1);
8966 free (name);
8967 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8968 }
8969 free (name);
8970
8971 *indexp = sindex;
8972 if (number < 0)
8973 return (&expand_wdesc_error);
8974 else
8975 {
8976 ret = alloc_word_desc ();
8977 ret->word = itos (number);
8978 return ret;
8979 }
8980 }
8981
8982 /* ${@} is identical to $@. */
8983 if (name[0] == '@' && name[1] == '\0')
8984 {
8985 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8986 *quoted_dollar_atp = 1;
8987
8988 if (contains_dollar_at)
8989 *contains_dollar_at = 1;
8990
8991 tflag |= W_DOLLARAT;
8992 }
8993
8994 /* Process ${!PREFIX*} expansion. */
8995 if (want_indir && string[sindex - 1] == RBRACE &&
8996 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
8997 legal_variable_starter ((unsigned char) name[1]))
8998 {
8999 char **x;
9000 WORD_LIST *xlist;
9001
9002 temp1 = savestring (name + 1);
9003 number = strlen (temp1);
9004 temp1[number - 1] = '\0';
9005 x = all_variables_matching_prefix (temp1);
9006 xlist = strvec_to_word_list (x, 0, 0);
9007 if (string[sindex - 2] == '*')
9008 temp = string_list_dollar_star (xlist, quoted, 0);
9009 else
9010 {
9011 temp = string_list_dollar_at (xlist, quoted, 0);
9012 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9013 *quoted_dollar_atp = 1;
9014 if (contains_dollar_at)
9015 *contains_dollar_at = 1;
9016
9017 tflag |= W_DOLLARAT;
9018 }
9019 free (x);
9020 dispose_words (xlist);
9021 free (temp1);
9022 *indexp = sindex;
9023
9024 free (name);
9025
9026 ret = alloc_word_desc ();
9027 ret->word = temp;
9028 ret->flags = tflag; /* XXX */
9029 return ret;
9030 }
9031
9032 #if defined (ARRAY_VARS)
9033 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */
9034 if (want_indir && string[sindex - 1] == RBRACE &&
9035 string[sindex - 2] == RBRACK && valid_array_reference (name+1, 0))
9036 {
9037 char *x, *x1;
9038
9039 temp1 = savestring (name + 1);
9040 x = array_variable_name (temp1, 0, &x1, (int *)0);
9041 FREE (x);
9042 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == RBRACK)
9043 {
9044 temp = array_keys (temp1, quoted, pflags); /* handles assoc vars too */
9045 if (x1[0] == '@')
9046 {
9047 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9048 *quoted_dollar_atp = 1;
9049 if (contains_dollar_at)
9050 *contains_dollar_at = 1;
9051
9052 tflag |= W_DOLLARAT;
9053 }
9054
9055 free (name);
9056 free (temp1);
9057 *indexp = sindex;
9058
9059 ret = alloc_word_desc ();
9060 ret->word = temp;
9061 ret->flags = tflag; /* XXX */
9062 return ret;
9063 }
9064
9065 free (temp1);
9066 }
9067 #endif /* ARRAY_VARS */
9068
9069 /* Make sure that NAME is valid before trying to go on. */
9070 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
9071 var_is_special) == 0)
9072 {
9073 temp = (char *)NULL;
9074 goto bad_substitution; /* substitution error */
9075 }
9076
9077 if (want_indir)
9078 {
9079 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, pflags|local_pflags, quoted_dollar_atp, contains_dollar_at);
9080 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9081 {
9082 temp = (char *)NULL;
9083 goto bad_substitution;
9084 }
9085
9086 /* Turn off the W_ARRAYIND flag because there is no way for this function
9087 to return the index we're supposed to be using. */
9088 if (tdesc && tdesc->flags)
9089 tdesc->flags &= ~W_ARRAYIND;
9090 }
9091 else
9092 {
9093 local_pflags |= PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS));
9094 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, local_pflags, &ind);
9095 }
9096
9097 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9098 {
9099 tflag = 0;
9100 tdesc = 0;
9101 }
9102
9103 if (tdesc)
9104 {
9105 temp = tdesc->word;
9106 tflag = tdesc->flags;
9107 dispose_word_desc (tdesc);
9108 }
9109 else
9110 temp = (char *)0;
9111
9112 if (temp == &expand_param_error || temp == &expand_param_fatal)
9113 {
9114 FREE (name);
9115 FREE (value);
9116 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9117 }
9118
9119 #if defined (ARRAY_VARS)
9120 if (valid_array_reference (name, 0))
9121 {
9122 int qflags;
9123 char *t;
9124
9125 qflags = quoted;
9126 /* If in a context where word splitting will not take place, treat as
9127 if double-quoted. Has effects with $* and ${array[*]} */
9128
9129 if (pflags & PF_ASSIGNRHS)
9130 qflags |= Q_DOUBLE_QUOTES;
9131 /* We duplicate a little code here */
9132 t = mbschr (name, LBRACK);
9133 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == RBRACK)
9134 {
9135 all_element_arrayref = 1;
9136 if (expand_no_split_dollar_star && t[1] == '*') /* XXX */
9137 qflags |= Q_DOUBLE_QUOTES;
9138 }
9139 chk_atstar (name, qflags, pflags, quoted_dollar_atp, contains_dollar_at);
9140 }
9141 #endif
9142
9143 var_is_set = temp != (char *)0;
9144 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
9145 /* XXX - this may not need to be restricted to special variables */
9146 if (check_nullness)
9147 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
9148 #if defined (ARRAY_VARS)
9149 if (check_nullness)
9150 var_is_null |= var_is_set &&
9151 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) &&
9152 QUOTED_NULL (temp) &&
9153 valid_array_reference (name, 0) &&
9154 chk_atstar (name, 0, 0, (int *)0, (int *)0);
9155 #endif
9156
9157 /* Get the rest of the stuff inside the braces. */
9158 if (c && c != RBRACE)
9159 {
9160 /* Extract the contents of the ${ ... } expansion
9161 according to the Posix.2 rules. */
9162 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
9163 if (string[sindex] == RBRACE)
9164 sindex++;
9165 else
9166 goto bad_substitution; /* substitution error */
9167 }
9168 else
9169 value = (char *)NULL;
9170
9171 *indexp = sindex;
9172
9173 /* All the cases where an expansion can possibly generate an unbound
9174 variable error. */
9175 if (want_substring || want_patsub || want_casemod || c == '@' || c == '#' || c == '%' || c == RBRACE)
9176 {
9177 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
9178 {
9179 set_exit_status (EXECUTION_FAILURE);
9180 err_unboundvar (name);
9181 FREE (value);
9182 FREE (temp);
9183 free (name);
9184 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9185 }
9186 }
9187
9188 /* If this is a substring spec, process it and add the result. */
9189 if (want_substring)
9190 {
9191 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9192 FREE (value);
9193 FREE (temp);
9194
9195 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9196 {
9197 FREE (name);
9198 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9199 }
9200
9201 ret = alloc_word_desc ();
9202 ret->word = temp1;
9203 /* We test quoted_dollar_atp because we want variants with double-quoted
9204 "$@" to take a different code path. In fact, we make sure at the end
9205 of expand_word_internal that we're only looking at these flags if
9206 quoted_dollar_at == 0. */
9207 if (temp1 &&
9208 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9209 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9210 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9211 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 &&
9212 (pflags & PF_ASSIGNRHS))
9213 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9214 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9215 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9216 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9217
9218 FREE (name);
9219 return ret;
9220 }
9221 else if (want_patsub)
9222 {
9223 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9224 FREE (value);
9225 FREE (temp);
9226
9227 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9228 {
9229 FREE (name);
9230 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9231 }
9232
9233 ret = alloc_word_desc ();
9234 ret->word = temp1;
9235 if (temp1 &&
9236 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9237 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9238 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9239 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9240 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9241 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9242
9243 FREE (name);
9244 return ret;
9245 }
9246 #if defined (CASEMOD_EXPANSIONS)
9247 else if (want_casemod)
9248 {
9249 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9250 FREE (value);
9251 FREE (temp);
9252
9253 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9254 {
9255 FREE (name);
9256 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9257 }
9258
9259 ret = alloc_word_desc ();
9260 ret->word = temp1;
9261 if (temp1 &&
9262 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
9263 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9264 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9265 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9266 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9267 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9268
9269 FREE (name);
9270 return ret;
9271 }
9272 #endif
9273
9274 /* Do the right thing based on which character ended the variable name. */
9275 switch (c)
9276 {
9277 default:
9278 case '\0':
9279 bad_substitution:
9280 set_exit_status (EXECUTION_FAILURE);
9281 report_error (_("%s: bad substitution"), string ? string : "??");
9282 FREE (value);
9283 FREE (temp);
9284 free (name);
9285 if (shell_compatibility_level <= 43)
9286 return &expand_wdesc_error;
9287 else
9288 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
9289
9290 case RBRACE:
9291 break;
9292
9293 case '@':
9294 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9295 free (temp);
9296 free (value);
9297
9298 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
9299 {
9300 free (name);
9301 set_exit_status (EXECUTION_FAILURE);
9302 report_error (_("%s: bad substitution"), string ? string : "??");
9303 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
9304 }
9305
9306 ret = alloc_word_desc ();
9307 ret->word = temp1;
9308 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9309 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9310 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9311 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9312 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9313
9314 free (name);
9315 return ret;
9316
9317 case '#': /* ${param#[#]pattern} */
9318 case '%': /* ${param%[%]pattern} */
9319 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
9320 {
9321 FREE (value);
9322 break;
9323 }
9324 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
9325 free (temp);
9326 free (value);
9327
9328 ret = alloc_word_desc ();
9329 ret->word = temp1;
9330 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9331 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
9332 /* Special handling for $* when unquoted and $IFS is null. Posix interp 888 */
9333 else if (temp1 && (name[0] == '*' && name[1] == 0) && quoted == 0 && ifs_is_null)
9334 ret->flags |= W_SPLITSPACE; /* Posix interp 888 */
9335
9336 free (name);
9337 return ret;
9338
9339 case '-':
9340 case '=':
9341 case '?':
9342 case '+':
9343 if (var_is_set && var_is_null == 0)
9344 {
9345 /* If the operator is `+', we don't want the value of the named
9346 variable for anything, just the value of the right hand side. */
9347 if (c == '+')
9348 {
9349 /* XXX -- if we're double-quoted and the named variable is "$@",
9350 we want to turn off any special handling of "$@" --
9351 we're not using it, so whatever is on the rhs applies. */
9352 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9353 *quoted_dollar_atp = 0;
9354 if (contains_dollar_at)
9355 *contains_dollar_at = 0;
9356
9357 FREE (temp);
9358 if (value)
9359 {
9360 /* From Posix discussion on austin-group list. Issue 221
9361 requires that backslashes escaping `}' inside
9362 double-quoted ${...} be removed. */
9363 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9364 quoted |= Q_DOLBRACE;
9365 ret = parameter_brace_expand_rhs (name, value, c,
9366 quoted,
9367 pflags,
9368 quoted_dollar_atp,
9369 contains_dollar_at);
9370 /* XXX - fix up later, esp. noting presence of
9371 W_HASQUOTEDNULL in ret->flags */
9372 free (value);
9373 }
9374 else
9375 temp = (char *)NULL;
9376 }
9377 else
9378 {
9379 FREE (value);
9380 }
9381 /* Otherwise do nothing; just use the value in TEMP. */
9382 }
9383 else /* VAR not set or VAR is NULL. */
9384 {
9385 FREE (temp);
9386 temp = (char *)NULL;
9387 if (c == '=' && var_is_special)
9388 {
9389 set_exit_status (EXECUTION_FAILURE);
9390 report_error (_("$%s: cannot assign in this way"), name);
9391 free (name);
9392 free (value);
9393 return &expand_wdesc_error;
9394 }
9395 else if (c == '?')
9396 {
9397 parameter_brace_expand_error (name, value, check_nullness);
9398 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9399 }
9400 else if (c != '+')
9401 {
9402 /* XXX -- if we're double-quoted and the named variable is "$@",
9403 we want to turn off any special handling of "$@" --
9404 we're not using it, so whatever is on the rhs applies. */
9405 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
9406 *quoted_dollar_atp = 0;
9407 if (contains_dollar_at)
9408 *contains_dollar_at = 0;
9409
9410 /* From Posix discussion on austin-group list. Issue 221 requires
9411 that backslashes escaping `}' inside double-quoted ${...} be
9412 removed. */
9413 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9414 quoted |= Q_DOLBRACE;
9415 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
9416 quoted_dollar_atp,
9417 contains_dollar_at);
9418 /* XXX - fix up later, esp. noting presence of
9419 W_HASQUOTEDNULL in tdesc->flags */
9420 }
9421 free (value);
9422 }
9423
9424 break;
9425 }
9426 free (name);
9427
9428 if (ret == 0)
9429 {
9430 ret = alloc_word_desc ();
9431 ret->flags = tflag;
9432 ret->word = temp;
9433 }
9434 return (ret);
9435 }
9436
9437 /* Expand a single ${xxx} expansion. The braces are optional. When
9438 the braces are used, parameter_brace_expand() does the work,
9439 possibly calling param_expand recursively. */
9440 static WORD_DESC *
9441 param_expand (string, sindex, quoted, expanded_something,
9442 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
9443 pflags)
9444 char *string;
9445 int *sindex, quoted, *expanded_something, *contains_dollar_at;
9446 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
9447 {
9448 char *temp, *temp1, uerror[3], *savecmd;
9449 int zindex, t_index, expok;
9450 unsigned char c;
9451 intmax_t number;
9452 SHELL_VAR *var;
9453 WORD_LIST *list, *l;
9454 WORD_DESC *tdesc, *ret;
9455 int tflag, nullarg;
9456
9457 /*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
9458 zindex = *sindex;
9459 c = string[++zindex];
9460
9461 temp = (char *)NULL;
9462 ret = tdesc = (WORD_DESC *)NULL;
9463 tflag = 0;
9464
9465 /* Do simple cases first. Switch on what follows '$'. */
9466 switch (c)
9467 {
9468 /* $0 .. $9? */
9469 case '0':
9470 case '1':
9471 case '2':
9472 case '3':
9473 case '4':
9474 case '5':
9475 case '6':
9476 case '7':
9477 case '8':
9478 case '9':
9479 temp1 = dollar_vars[TODIGIT (c)];
9480 /* This doesn't get called when (pflags&PF_IGNUNBOUND) != 0 */
9481 if (unbound_vars_is_error && temp1 == (char *)NULL)
9482 {
9483 uerror[0] = '$';
9484 uerror[1] = c;
9485 uerror[2] = '\0';
9486 set_exit_status (EXECUTION_FAILURE);
9487 err_unboundvar (uerror);
9488 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9489 }
9490 if (temp1)
9491 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9492 ? quote_string (temp1)
9493 : quote_escapes (temp1);
9494 else
9495 temp = (char *)NULL;
9496
9497 break;
9498
9499 /* $$ -- pid of the invoking shell. */
9500 case '$':
9501 temp = itos (dollar_dollar_pid);
9502 break;
9503
9504 /* $# -- number of positional parameters. */
9505 case '#':
9506 temp = itos (number_of_args ());
9507 break;
9508
9509 /* $? -- return value of the last synchronous command. */
9510 case '?':
9511 temp = itos (last_command_exit_value);
9512 break;
9513
9514 /* $- -- flags supplied to the shell on invocation or by `set'. */
9515 case '-':
9516 temp = which_set_flags ();
9517 break;
9518
9519 /* $! -- Pid of the last asynchronous command. */
9520 case '!':
9521 /* If no asynchronous pids have been created, expand to nothing.
9522 If `set -u' has been executed, and no async processes have
9523 been created, this is an expansion error. */
9524 if (last_asynchronous_pid == NO_PID)
9525 {
9526 if (expanded_something)
9527 *expanded_something = 0;
9528 temp = (char *)NULL;
9529 if (unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
9530 {
9531 uerror[0] = '$';
9532 uerror[1] = c;
9533 uerror[2] = '\0';
9534 set_exit_status (EXECUTION_FAILURE);
9535 err_unboundvar (uerror);
9536 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9537 }
9538 }
9539 else
9540 temp = itos (last_asynchronous_pid);
9541 break;
9542
9543 /* The only difference between this and $@ is when the arg is quoted. */
9544 case '*': /* `$*' */
9545 list = list_rest_of_args ();
9546
9547 #if 0
9548 /* According to austin-group posix proposal by Geoff Clare in
9549 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
9550
9551 "The shell shall write a message to standard error and
9552 immediately exit when it tries to expand an unset parameter
9553 other than the '@' and '*' special parameters."
9554 */
9555
9556 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
9557 {
9558 uerror[0] = '$';
9559 uerror[1] = '*';
9560 uerror[2] = '\0';
9561 set_exit_status (EXECUTION_FAILURE);
9562 err_unboundvar (uerror);
9563 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9564 }
9565 #endif
9566
9567 /* If there are no command-line arguments, this should just
9568 disappear if there are other characters in the expansion,
9569 even if it's quoted. */
9570 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
9571 temp = (char *)NULL;
9572 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
9573 {
9574 /* If we have "$*" we want to make a string of the positional
9575 parameters, separated by the first character of $IFS, and
9576 quote the whole string, including the separators. If IFS
9577 is unset, the parameters are separated by ' '; if $IFS is
9578 null, the parameters are concatenated. */
9579 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list, quoted, 0) : string_list (list);
9580 if (temp)
9581 {
9582 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
9583 if (*temp == 0)
9584 tflag |= W_HASQUOTEDNULL;
9585 if (temp != temp1)
9586 free (temp);
9587 temp = temp1;
9588 }
9589 }
9590 else
9591 {
9592 /* We check whether or not we're eventually going to split $* here,
9593 for example when IFS is empty and we are processing the rhs of
9594 an assignment statement. In that case, we don't separate the
9595 arguments at all. Otherwise, if the $* is not quoted it is
9596 identical to $@ */
9597 if (expand_no_split_dollar_star && quoted == 0 && ifs_is_set == 0 && (pflags & PF_ASSIGNRHS))
9598 {
9599 /* Posix interp 888: RHS of assignment, IFS unset: no splitting,
9600 separate with space */
9601 temp1 = string_list_dollar_star (list, quoted, pflags);
9602 temp = temp1 ? quote_string (temp1) : temp1;
9603 /* XXX - tentative - note that we saw a quoted null here */
9604 if (temp1 && *temp1 == 0 && QUOTED_NULL (temp))
9605 tflag |= W_SAWQUOTEDNULL;
9606 FREE (temp1);
9607 }
9608 else if (expand_no_split_dollar_star && quoted == 0 && ifs_is_null && (pflags & PF_ASSIGNRHS))
9609 {
9610 /* Posix interp 888: RHS of assignment, IFS set to '' */
9611 temp1 = string_list_dollar_star (list, quoted, pflags);
9612 temp = temp1 ? quote_escapes (temp1) : temp1;
9613 FREE (temp1);
9614 }
9615 else if (expand_no_split_dollar_star && quoted == 0 && ifs_is_set && ifs_is_null == 0 && (pflags & PF_ASSIGNRHS))
9616 {
9617 /* Posix interp 888: RHS of assignment, IFS set to non-null value */
9618 temp1 = string_list_dollar_star (list, quoted, pflags);
9619 temp = temp1 ? quote_string (temp1) : temp1;
9620
9621 /* XXX - tentative - note that we saw a quoted null here */
9622 if (temp1 && *temp1 == 0 && QUOTED_NULL (temp))
9623 tflag |= W_SAWQUOTEDNULL;
9624 FREE (temp1);
9625 }
9626 /* XXX - should we check ifs_is_set here as well? */
9627 # if defined (HANDLE_MULTIBYTE)
9628 else if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
9629 # else
9630 else if (expand_no_split_dollar_star && ifs_firstc == 0)
9631 # endif
9632 /* Posix interp 888: not RHS, no splitting, IFS set to '' */
9633 temp = string_list_dollar_star (list, quoted, 0);
9634 else
9635 {
9636 temp = string_list_dollar_at (list, quoted, 0);
9637 /* Set W_SPLITSPACE to make sure the individual positional
9638 parameters are split into separate arguments */
9639 #if 0
9640 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
9641 #else /* change with bash-5.0 */
9642 if (quoted == 0 && ifs_is_null)
9643 #endif
9644 tflag |= W_SPLITSPACE;
9645 /* If we're not quoted but we still don't want word splitting, make
9646 we quote the IFS characters to protect them from splitting (e.g.,
9647 when $@ is in the string as well). */
9648 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
9649 {
9650 temp1 = quote_string (temp);
9651 free (temp);
9652 temp = temp1;
9653 }
9654 }
9655
9656 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
9657 *contains_dollar_at = 1;
9658 }
9659
9660 dispose_words (list);
9661 break;
9662
9663 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
9664 means that we have to turn quoting off after we split into
9665 the individually quoted arguments so that the final split
9666 on the first character of $IFS is still done. */
9667 case '@': /* `$@' */
9668 list = list_rest_of_args ();
9669
9670 #if 0
9671 /* According to austin-group posix proposal by Geoff Clare in
9672 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
9673
9674 "The shell shall write a message to standard error and
9675 immediately exit when it tries to expand an unset parameter
9676 other than the '@' and '*' special parameters."
9677 */
9678
9679 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
9680 {
9681 uerror[0] = '$';
9682 uerror[1] = '@';
9683 uerror[2] = '\0';
9684 set_exit_status (EXECUTION_FAILURE);
9685 err_unboundvar (uerror);
9686 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
9687 }
9688 #endif
9689
9690 for (nullarg = 0, l = list; l; l = l->next)
9691 {
9692 if (l->word && (l->word->word == 0 || l->word->word[0] == 0))
9693 nullarg = 1;
9694 }
9695
9696 /* We want to flag the fact that we saw this. We can't turn
9697 off quoting entirely, because other characters in the
9698 string might need it (consider "\"$@\""), but we need some
9699 way to signal that the final split on the first character
9700 of $IFS should be done, even though QUOTED is 1. */
9701 /* XXX - should this test include Q_PATQUOTE? */
9702 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9703 *quoted_dollar_at_p = 1;
9704 if (contains_dollar_at)
9705 *contains_dollar_at = 1;
9706
9707 /* We want to separate the positional parameters with the first
9708 character of $IFS in case $IFS is something other than a space.
9709 We also want to make sure that splitting is done no matter what --
9710 according to POSIX.2, this expands to a list of the positional
9711 parameters no matter what IFS is set to. */
9712 /* XXX - what to do when in a context where word splitting is not
9713 performed? Even when IFS is not the default, posix seems to imply
9714 that we have to expand $@ to all the positional parameters and
9715 separate them with spaces, which are preserved because word splitting
9716 doesn't take place. See below for how we use PF_NOSPLIT2 here. */
9717
9718 /* These are the cases where word splitting will not be performed. */
9719 if (pflags & PF_ASSIGNRHS)
9720 {
9721 temp = string_list_dollar_at (list, (quoted|Q_DOUBLE_QUOTES), pflags);
9722 if (nullarg)
9723 tflag |= W_HASQUOTEDNULL; /* we know quoting produces quoted nulls */
9724 }
9725
9726 /* This needs to match what expand_word_internal does with non-quoted $@
9727 does with separating with spaces. Passing Q_DOUBLE_QUOTES means that
9728 the characters in LIST will be quoted, and PF_ASSIGNRHS ensures that
9729 they will separated by spaces. After doing this, we need the special
9730 handling for PF_NOSPLIT2 in expand_word_internal to remove the CTLESC
9731 quotes. */
9732 else if (pflags & PF_NOSPLIT2)
9733 {
9734 #if defined (HANDLE_MULTIBYTE)
9735 if (quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc[0] != ' ')
9736 #else
9737 if (quoted == 0 && ifs_is_set && ifs_is_null == 0 && ifs_firstc != ' ')
9738 #endif
9739 /* Posix interp 888 */
9740 temp = string_list_dollar_at (list, Q_DOUBLE_QUOTES, pflags);
9741 else
9742 temp = string_list_dollar_at (list, quoted, pflags);
9743 }
9744 else
9745 temp = string_list_dollar_at (list, quoted, pflags);
9746
9747 tflag |= W_DOLLARAT;
9748 dispose_words (list);
9749 break;
9750
9751 case LBRACE:
9752 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
9753 quoted_dollar_at_p,
9754 contains_dollar_at);
9755
9756 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9757 return (tdesc);
9758 temp = tdesc ? tdesc->word : (char *)0;
9759
9760 /* XXX */
9761 /* Quoted nulls should be removed if there is anything else
9762 in the string. */
9763 /* Note that we saw the quoted null so we can add one back at
9764 the end of this function if there are no other characters
9765 in the string, discard TEMP, and go on. The exception to
9766 this is when we have "${@}" and $1 is '', since $@ needs
9767 special handling. */
9768 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
9769 {
9770 if (had_quoted_null_p)
9771 *had_quoted_null_p = 1;
9772 if (*quoted_dollar_at_p == 0)
9773 {
9774 free (temp);
9775 tdesc->word = temp = (char *)NULL;
9776 }
9777
9778 }
9779
9780 ret = tdesc;
9781 goto return0;
9782
9783 /* Do command or arithmetic substitution. */
9784 case LPAREN:
9785 /* We have to extract the contents of this paren substitution. */
9786 t_index = zindex + 1;
9787 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
9788 as arithmetic substitution immediately. */
9789 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
9790 zindex = t_index;
9791
9792 /* For Posix.2-style `$(( ))' arithmetic substitution,
9793 extract the expression and pass it to the evaluator. */
9794 if (temp && *temp == LPAREN)
9795 {
9796 char *temp2;
9797 temp1 = temp + 1;
9798 temp2 = savestring (temp1);
9799 t_index = strlen (temp2) - 1;
9800
9801 if (temp2[t_index] != RPAREN)
9802 {
9803 free (temp2);
9804 goto comsub;
9805 }
9806
9807 /* Cut off ending `)' */
9808 temp2[t_index] = '\0';
9809
9810 if (chk_arithsub (temp2, t_index) == 0)
9811 {
9812 free (temp2);
9813 #if 0
9814 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
9815 #endif
9816 goto comsub;
9817 }
9818
9819 /* Expand variables found inside the expression. */
9820 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
9821 free (temp2);
9822
9823 arithsub:
9824 /* No error messages. */
9825 savecmd = this_command_name;
9826 this_command_name = (char *)NULL;
9827 number = evalexp (temp1, EXP_EXPANDED, &expok);
9828 this_command_name = savecmd;
9829 free (temp);
9830 free (temp1);
9831 if (expok == 0)
9832 {
9833 if (interactive_shell == 0 && posixly_correct)
9834 {
9835 set_exit_status (EXECUTION_FAILURE);
9836 return (&expand_wdesc_fatal);
9837 }
9838 else
9839 return (&expand_wdesc_error);
9840 }
9841 temp = itos (number);
9842 break;
9843 }
9844
9845 comsub:
9846 if (pflags & PF_NOCOMSUB)
9847 /* we need zindex+1 because string[zindex] == RPAREN */
9848 temp1 = substring (string, *sindex, zindex+1);
9849 else
9850 {
9851 tdesc = command_substitute (temp, quoted, pflags&PF_ASSIGNRHS);
9852 temp1 = tdesc ? tdesc->word : (char *)NULL;
9853 if (tdesc)
9854 dispose_word_desc (tdesc);
9855 }
9856 FREE (temp);
9857 temp = temp1;
9858 break;
9859
9860 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
9861 away in a future bash release. */
9862 case '[': /*]*/
9863 /* Extract the contents of this arithmetic substitution. */
9864 t_index = zindex + 1;
9865 temp = extract_arithmetic_subst (string, &t_index);
9866 zindex = t_index;
9867 if (temp == 0)
9868 {
9869 temp = savestring (string);
9870 if (expanded_something)
9871 *expanded_something = 0;
9872 goto return0;
9873 }
9874
9875 /* Do initial variable expansion. */
9876 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
9877
9878 goto arithsub;
9879
9880 default:
9881 /* Find the variable in VARIABLE_LIST. */
9882 temp = (char *)NULL;
9883
9884 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
9885 ;
9886 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
9887
9888 /* If this isn't a variable name, then just output the `$'. */
9889 if (temp1 == 0 || *temp1 == '\0')
9890 {
9891 FREE (temp1);
9892 temp = (char *)xmalloc (2);
9893 temp[0] = '$';
9894 temp[1] = '\0';
9895 if (expanded_something)
9896 *expanded_something = 0;
9897 goto return0;
9898 }
9899
9900 /* If the variable exists, return its value cell. */
9901 var = find_variable (temp1);
9902
9903 if (var && invisible_p (var) == 0 && var_isset (var))
9904 {
9905 #if defined (ARRAY_VARS)
9906 if (assoc_p (var) || array_p (var))
9907 {
9908 temp = array_p (var) ? array_reference (array_cell (var), 0)
9909 : assoc_reference (assoc_cell (var), "0");
9910 if (temp)
9911 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9912 ? quote_string (temp)
9913 : quote_escapes (temp);
9914 else if (unbound_vars_is_error)
9915 goto unbound_variable;
9916 }
9917 else
9918 #endif
9919 {
9920 temp = value_cell (var);
9921
9922 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
9923 ? quote_string (temp)
9924 : ((pflags & PF_ASSIGNRHS) ? quote_rhs (temp)
9925 : quote_escapes (temp));
9926 }
9927
9928 free (temp1);
9929
9930 goto return0;
9931 }
9932 else if (var && (invisible_p (var) || var_isset (var) == 0))
9933 temp = (char *)NULL;
9934 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
9935 {
9936 temp = nameref_cell (var);
9937 #if defined (ARRAY_VARS)
9938 if (temp && *temp && valid_array_reference (temp, 0))
9939 {
9940 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
9941 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
9942 return (tdesc);
9943 ret = tdesc;
9944 goto return0;
9945 }
9946 else
9947 #endif
9948 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
9949 if (temp && *temp && legal_identifier (temp) == 0)
9950 {
9951 set_exit_status (EXECUTION_FAILURE);
9952 report_error (_("%s: invalid variable name for name reference"), temp);
9953 return (&expand_wdesc_error); /* XXX */
9954 }
9955 else
9956 temp = (char *)NULL;
9957 }
9958
9959 temp = (char *)NULL;
9960
9961 unbound_variable:
9962 if (unbound_vars_is_error)
9963 {
9964 set_exit_status (EXECUTION_FAILURE);
9965 err_unboundvar (temp1);
9966 }
9967 else
9968 {
9969 free (temp1);
9970 goto return0;
9971 }
9972
9973 free (temp1);
9974 set_exit_status (EXECUTION_FAILURE);
9975 return ((unbound_vars_is_error && interactive_shell == 0)
9976 ? &expand_wdesc_fatal
9977 : &expand_wdesc_error);
9978 }
9979
9980 if (string[zindex])
9981 zindex++;
9982
9983 return0:
9984 *sindex = zindex;
9985
9986 if (ret == 0)
9987 {
9988 ret = alloc_word_desc ();
9989 ret->flags = tflag; /* XXX */
9990 ret->word = temp;
9991 }
9992 return ret;
9993 }
9994
9995 void
9996 invalidate_cached_quoted_dollar_at ()
9997 {
9998 dispose_words (cached_quoted_dollar_at);
9999 cached_quoted_dollar_at = 0;
10000 }
10001
10002 /* Make a word list which is the result of parameter and variable
10003 expansion, command substitution, arithmetic substitution, and
10004 quote removal of WORD. Return a pointer to a WORD_LIST which is
10005 the result of the expansion. If WORD contains a null word, the
10006 word list returned is also null.
10007
10008 QUOTED contains flag values defined in shell.h.
10009
10010 ISEXP is used to tell expand_word_internal that the word should be
10011 treated as the result of an expansion. This has implications for
10012 how IFS characters in the word are treated.
10013
10014 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
10015 they point to an integer value which receives information about expansion.
10016 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
10017 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
10018 else zero.
10019
10020 This only does word splitting in the case of $@ expansion. In that
10021 case, we split on ' '. */
10022
10023 /* Values for the local variable quoted_state. */
10024 #define UNQUOTED 0
10025 #define PARTIALLY_QUOTED 1
10026 #define WHOLLY_QUOTED 2
10027
10028 static WORD_LIST *
10029 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
10030 WORD_DESC *word;
10031 int quoted, isexp;
10032 int *contains_dollar_at;
10033 int *expanded_something;
10034 {
10035 WORD_LIST *list;
10036 WORD_DESC *tword;
10037
10038 /* The intermediate string that we build while expanding. */
10039 char *istring;
10040
10041 /* The current size of the above object. */
10042 size_t istring_size;
10043
10044 /* Index into ISTRING. */
10045 int istring_index;
10046
10047 /* Temporary string storage. */
10048 char *temp, *temp1;
10049
10050 /* The text of WORD. */
10051 register char *string;
10052
10053 /* The size of STRING. */
10054 size_t string_size;
10055
10056 /* The index into STRING. */
10057 int sindex;
10058
10059 /* This gets 1 if we see a $@ while quoted. */
10060 int quoted_dollar_at;
10061
10062 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
10063 whether WORD contains no quoting characters, a partially quoted
10064 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
10065 int quoted_state;
10066
10067 /* State flags */
10068 int had_quoted_null;
10069 int has_quoted_ifs; /* did we add a quoted $IFS character here? */
10070 int has_dollar_at, temp_has_dollar_at;
10071 int split_on_spaces;
10072 int local_expanded;
10073 int tflag;
10074 int pflags; /* flags passed to param_expand */
10075 int mb_cur_max;
10076
10077 int assignoff; /* If assignment, offset of `=' */
10078
10079 register unsigned char c; /* Current character. */
10080 int t_index; /* For calls to string_extract_xxx. */
10081
10082 char twochars[2];
10083
10084 DECLARE_MBSTATE;
10085
10086 /* OK, let's see if we can optimize a common idiom: "$@" */
10087 if (STREQ (word->word, "\"$@\"") &&
10088 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
10089 dollar_vars[1]) /* XXX - check IFS here as well? */
10090 {
10091 if (contains_dollar_at)
10092 *contains_dollar_at = 1;
10093 if (expanded_something)
10094 *expanded_something = 1;
10095 if (cached_quoted_dollar_at)
10096 return (copy_word_list (cached_quoted_dollar_at));
10097 list = list_rest_of_args ();
10098 list = quote_list (list);
10099 cached_quoted_dollar_at = copy_word_list (list);
10100 return (list);
10101 }
10102
10103 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
10104 istring[istring_index = 0] = '\0';
10105 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
10106 has_quoted_ifs = 0;
10107 split_on_spaces = 0;
10108 quoted_state = UNQUOTED;
10109
10110 string = word->word;
10111 if (string == 0)
10112 goto finished_with_string;
10113 mb_cur_max = MB_CUR_MAX;
10114
10115 /* Don't need the string length for the SADD... and COPY_ macros unless
10116 multibyte characters are possible, but do need it for bounds checking. */
10117 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
10118
10119 if (contains_dollar_at)
10120 *contains_dollar_at = 0;
10121
10122 assignoff = -1;
10123
10124 /* Begin the expansion. */
10125
10126 for (sindex = 0; ;)
10127 {
10128 c = string[sindex];
10129
10130 /* Case on top-level character. */
10131 switch (c)
10132 {
10133 case '\0':
10134 goto finished_with_string;
10135
10136 case CTLESC:
10137 sindex++;
10138 #if HANDLE_MULTIBYTE
10139 if (mb_cur_max > 1 && string[sindex])
10140 {
10141 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
10142 }
10143 else
10144 #endif
10145 {
10146 temp = (char *)xmalloc (3);
10147 temp[0] = CTLESC;
10148 temp[1] = c = string[sindex];
10149 temp[2] = '\0';
10150 }
10151
10152 dollar_add_string:
10153 if (string[sindex])
10154 sindex++;
10155
10156 add_string:
10157 if (temp)
10158 {
10159 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
10160 temp = (char *)0;
10161 }
10162
10163 break;
10164
10165 #if defined (PROCESS_SUBSTITUTION)
10166 /* Process substitution. */
10167 case '<':
10168 case '>':
10169 {
10170 /* XXX - technically this should only be expanded at the start
10171 of a word */
10172 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)))
10173 {
10174 sindex--; /* add_character: label increments sindex */
10175 goto add_character;
10176 }
10177 else
10178 t_index = sindex + 1; /* skip past both '<' and LPAREN */
10179
10180 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
10181 sindex = t_index;
10182
10183 /* If the process substitution specification is `<()', we want to
10184 open the pipe for writing in the child and produce output; if
10185 it is `>()', we want to open the pipe for reading in the child
10186 and consume input. */
10187 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
10188
10189 FREE (temp1);
10190
10191 goto dollar_add_string;
10192 }
10193 #endif /* PROCESS_SUBSTITUTION */
10194
10195 case '=':
10196 /* Posix.2 section 3.6.1 says that tildes following `=' in words
10197 which are not assignment statements are not expanded. If the
10198 shell isn't in posix mode, though, we perform tilde expansion
10199 on `likely candidate' unquoted assignment statements (flags
10200 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
10201 contains an unquoted :~ or =~. Something to think about: we
10202 now have a flag that says to perform tilde expansion on arguments
10203 to `assignment builtins' like declare and export that look like
10204 assignment statements. We now do tilde expansion on such words
10205 even in POSIX mode. */
10206 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
10207 {
10208 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10209 goto add_ifs_character;
10210 else
10211 goto add_character;
10212 }
10213 /* If we're not in posix mode or forcing assignment-statement tilde
10214 expansion, note where the first `=' appears in the word and prepare
10215 to do tilde expansion following the first `='. We have to keep
10216 track of the first `=' (using assignoff) to avoid being confused
10217 by an `=' in the rhs of the assignment statement. */
10218 if ((word->flags & W_ASSIGNMENT) &&
10219 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
10220 assignoff == -1 && sindex > 0)
10221 assignoff = sindex;
10222 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
10223 word->flags |= W_ITILDE;
10224
10225 if (word->flags & W_ASSIGNARG)
10226 word->flags |= W_ASSIGNRHS; /* affects $@ */
10227
10228 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10229 {
10230 has_quoted_ifs++;
10231 goto add_ifs_character;
10232 }
10233 else
10234 goto add_character;
10235
10236 case ':':
10237 if (word->flags & (W_NOTILDE|W_NOASSNTILDE))
10238 {
10239 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10240 goto add_ifs_character;
10241 else
10242 goto add_character;
10243 }
10244
10245 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS)) &&
10246 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
10247 string[sindex+1] == '~')
10248 word->flags |= W_ITILDE;
10249
10250 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
10251 goto add_ifs_character;
10252 else
10253 goto add_character;
10254
10255 case '~':
10256 /* If the word isn't supposed to be tilde expanded, or we're not
10257 at the start of a word or after an unquoted : or = in an
10258 assignment statement, we don't do tilde expansion. We don't
10259 do tilde expansion if quoted or in an arithmetic context. */
10260
10261 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
10262 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
10263 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
10264 {
10265 word->flags &= ~W_ITILDE;
10266 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
10267 goto add_ifs_character;
10268 else
10269 goto add_character;
10270 }
10271
10272 if (word->flags & W_ASSIGNRHS)
10273 tflag = 2;
10274 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
10275 tflag = 1;
10276 else
10277 tflag = 0;
10278
10279 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
10280
10281 word->flags &= ~W_ITILDE;
10282
10283 if (temp && *temp && t_index > 0)
10284 {
10285 temp1 = bash_tilde_expand (temp, tflag);
10286 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
10287 {
10288 FREE (temp);
10289 FREE (temp1);
10290 goto add_character; /* tilde expansion failed */
10291 }
10292 free (temp);
10293 temp = temp1;
10294 sindex += t_index;
10295 goto add_quoted_string; /* XXX was add_string */
10296 }
10297 else
10298 {
10299 FREE (temp);
10300 goto add_character;
10301 }
10302
10303 case '$':
10304 if (expanded_something)
10305 *expanded_something = 1;
10306 local_expanded = 1;
10307
10308 temp_has_dollar_at = 0;
10309 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
10310 if (word->flags & W_NOSPLIT2)
10311 pflags |= PF_NOSPLIT2;
10312 if (word->flags & W_ASSIGNRHS)
10313 pflags |= PF_ASSIGNRHS;
10314 if (word->flags & W_COMPLETE)
10315 pflags |= PF_COMPLETE;
10316
10317 tword = param_expand (string, &sindex, quoted, expanded_something,
10318 &temp_has_dollar_at, &quoted_dollar_at,
10319 &had_quoted_null, pflags);
10320 has_dollar_at += temp_has_dollar_at;
10321 split_on_spaces += (tword->flags & W_SPLITSPACE);
10322
10323 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
10324 {
10325 free (string);
10326 free (istring);
10327 return ((tword == &expand_wdesc_error) ? &expand_word_error
10328 : &expand_word_fatal);
10329 }
10330 if (contains_dollar_at && has_dollar_at)
10331 *contains_dollar_at = 1;
10332
10333 if (tword && (tword->flags & W_HASQUOTEDNULL))
10334 had_quoted_null = 1; /* note for later */
10335 if (tword && (tword->flags & W_SAWQUOTEDNULL))
10336 had_quoted_null = 1; /* XXX */
10337
10338 temp = tword ? tword->word : (char *)NULL;
10339 dispose_word_desc (tword);
10340
10341 /* Kill quoted nulls; we will add them back at the end of
10342 expand_word_internal if nothing else in the string */
10343 if (had_quoted_null && temp && QUOTED_NULL (temp))
10344 {
10345 FREE (temp);
10346 temp = (char *)NULL;
10347 }
10348
10349 goto add_string;
10350 break;
10351
10352 case '`': /* Backquoted command substitution. */
10353 {
10354 t_index = sindex++;
10355
10356 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
10357 /* The test of sindex against t_index is to allow bare instances of
10358 ` to pass through, for backwards compatibility. */
10359 if (temp == &extract_string_error || temp == &extract_string_fatal)
10360 {
10361 if (sindex - 1 == t_index)
10362 {
10363 sindex = t_index;
10364 goto add_character;
10365 }
10366 set_exit_status (EXECUTION_FAILURE);
10367 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
10368 free (string);
10369 free (istring);
10370 return ((temp == &extract_string_error) ? &expand_word_error
10371 : &expand_word_fatal);
10372 }
10373
10374 if (expanded_something)
10375 *expanded_something = 1;
10376 local_expanded = 1;
10377
10378 if (word->flags & W_NOCOMSUB)
10379 /* sindex + 1 because string[sindex] == '`' */
10380 temp1 = substring (string, t_index, sindex + 1);
10381 else
10382 {
10383 de_backslash (temp);
10384 tword = command_substitute (temp, quoted, 0);
10385 temp1 = tword ? tword->word : (char *)NULL;
10386 if (tword)
10387 dispose_word_desc (tword);
10388 }
10389 FREE (temp);
10390 temp = temp1;
10391 goto dollar_add_string;
10392 }
10393
10394 case '\\':
10395 if (string[sindex + 1] == '\n')
10396 {
10397 sindex += 2;
10398 continue;
10399 }
10400
10401 c = string[++sindex];
10402
10403 /* "However, the double-quote character ( '"' ) shall not be treated
10404 specially within a here-document, except when the double-quote
10405 appears within "$()", "``", or "${}"." */
10406 if ((quoted & Q_HERE_DOCUMENT) && (quoted & Q_DOLBRACE) && c == '"')
10407 tflag = CBSDQUOTE; /* special case */
10408 else if (quoted & Q_HERE_DOCUMENT)
10409 tflag = CBSHDOC;
10410 else if (quoted & Q_DOUBLE_QUOTES)
10411 tflag = CBSDQUOTE;
10412 else
10413 tflag = 0;
10414
10415 /* From Posix discussion on austin-group list: Backslash escaping
10416 a } in ${...} is removed. Issue 0000221 */
10417 if ((quoted & Q_DOLBRACE) && c == RBRACE)
10418 {
10419 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10420 }
10421 /* This is the fix for " $@\ " */
10422 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
10423 {
10424 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10425 DEFAULT_ARRAY_SIZE);
10426 istring[istring_index++] = CTLESC;
10427 istring[istring_index++] = '\\';
10428 istring[istring_index] = '\0';
10429
10430 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10431 }
10432 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && c == 0)
10433 {
10434 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10435 DEFAULT_ARRAY_SIZE);
10436 istring[istring_index++] = CTLESC;
10437 istring[istring_index++] = '\\';
10438 istring[istring_index] = '\0';
10439 break;
10440 }
10441 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
10442 {
10443 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
10444 }
10445 else if (c == 0)
10446 {
10447 c = CTLNUL;
10448 sindex--; /* add_character: label increments sindex */
10449 goto add_character;
10450 }
10451 else
10452 {
10453 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
10454 }
10455
10456 sindex++;
10457 add_twochars:
10458 /* BEFORE jumping here, we need to increment sindex if appropriate */
10459 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
10460 DEFAULT_ARRAY_SIZE);
10461 istring[istring_index++] = twochars[0];
10462 istring[istring_index++] = twochars[1];
10463 istring[istring_index] = '\0';
10464
10465 break;
10466
10467 case '"':
10468 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
10469 goto add_character;
10470
10471 t_index = ++sindex;
10472 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
10473
10474 /* If the quotes surrounded the entire string, then the
10475 whole word was quoted. */
10476 quoted_state = (t_index == 1 && string[sindex] == '\0')
10477 ? WHOLLY_QUOTED
10478 : PARTIALLY_QUOTED;
10479
10480 if (temp && *temp)
10481 {
10482 tword = alloc_word_desc ();
10483 tword->word = temp;
10484
10485 if (word->flags & W_ASSIGNARG)
10486 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
10487 if (word->flags & W_COMPLETE)
10488 tword->flags |= W_COMPLETE; /* for command substitutions */
10489 if (word->flags & W_NOCOMSUB)
10490 tword->flags |= W_NOCOMSUB;
10491 if (word->flags & W_NOPROCSUB)
10492 tword->flags |= W_NOPROCSUB;
10493
10494 if (word->flags & W_ASSIGNRHS)
10495 tword->flags |= W_ASSIGNRHS;
10496
10497 temp = (char *)NULL;
10498
10499 temp_has_dollar_at = 0; /* does this quoted (sub)string include $@? */
10500 /* Need to get W_HASQUOTEDNULL flag through this function. */
10501 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
10502 has_dollar_at += temp_has_dollar_at;
10503
10504 if (list == &expand_word_error || list == &expand_word_fatal)
10505 {
10506 free (istring);
10507 free (string);
10508 /* expand_word_internal has already freed temp_word->word
10509 for us because of the way it prints error messages. */
10510 tword->word = (char *)NULL;
10511 dispose_word (tword);
10512 return list;
10513 }
10514
10515 dispose_word (tword);
10516
10517 /* "$@" (a double-quoted dollar-at) expands into nothing,
10518 not even a NULL word, when there are no positional
10519 parameters. Posix interp 888 says that other parts of the
10520 word that expand to quoted nulls result in quoted nulls, so
10521 we can't just throw the entire word away if we have "$@"
10522 anywhere in it. We use had_quoted_null to keep track */
10523 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
10524 {
10525 quoted_dollar_at++;
10526 break;
10527 }
10528
10529 /* If this list comes back with a quoted null from expansion,
10530 we have either "$x" or "$@" with $1 == ''. In either case,
10531 we need to make sure we add a quoted null argument and
10532 disable the special handling that "$@" gets. */
10533 if (list && list->word && list->next == 0 && (list->word->flags & W_HASQUOTEDNULL))
10534 {
10535 if (had_quoted_null && temp_has_dollar_at)
10536 quoted_dollar_at++;
10537 had_quoted_null = 1; /* XXX */
10538 }
10539
10540 /* If we get "$@", we know we have expanded something, so we
10541 need to remember it for the final split on $IFS. This is
10542 a special case; it's the only case where a quoted string
10543 can expand into more than one word. It's going to come back
10544 from the above call to expand_word_internal as a list with
10545 multiple words. */
10546 if (list)
10547 dequote_list (list);
10548
10549 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
10550 {
10551 quoted_dollar_at++;
10552 if (contains_dollar_at)
10553 *contains_dollar_at = 1;
10554 if (expanded_something)
10555 *expanded_something = 1;
10556 local_expanded = 1;
10557 }
10558 }
10559 else
10560 {
10561 /* What we have is "". This is a minor optimization. */
10562 FREE (temp);
10563 list = (WORD_LIST *)NULL;
10564 had_quoted_null = 1; /* note for later */
10565 }
10566
10567 /* The code above *might* return a list (consider the case of "$@",
10568 where it returns "$1", "$2", etc.). We can't throw away the
10569 rest of the list, and we have to make sure each word gets added
10570 as quoted. We test on tresult->next: if it is non-NULL, we
10571 quote the whole list, save it to a string with string_list, and
10572 add that string. We don't need to quote the results of this
10573 (and it would be wrong, since that would quote the separators
10574 as well), so we go directly to add_string. */
10575 if (list)
10576 {
10577 if (list->next)
10578 {
10579 /* Testing quoted_dollar_at makes sure that "$@" is
10580 split correctly when $IFS does not contain a space. */
10581 temp = quoted_dollar_at
10582 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
10583 : string_list (quote_list (list));
10584 dispose_words (list);
10585 goto add_string;
10586 }
10587 else
10588 {
10589 temp = savestring (list->word->word);
10590 tflag = list->word->flags;
10591 dispose_words (list);
10592
10593 /* If the string is not a quoted null string, we want
10594 to remove any embedded unquoted CTLNUL characters.
10595 We do not want to turn quoted null strings back into
10596 the empty string, though. We do this because we
10597 want to remove any quoted nulls from expansions that
10598 contain other characters. For example, if we have
10599 x"$*"y or "x$*y" and there are no positional parameters,
10600 the $* should expand into nothing. */
10601 /* We use the W_HASQUOTEDNULL flag to differentiate the
10602 cases: a quoted null character as above and when
10603 CTLNUL is contained in the (non-null) expansion
10604 of some variable. We use the had_quoted_null flag to
10605 pass the value through this function to its caller. */
10606 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
10607 remove_quoted_nulls (temp); /* XXX */
10608 }
10609 }
10610 else
10611 temp = (char *)NULL;
10612
10613 if (temp == 0 && quoted_state == PARTIALLY_QUOTED)
10614 had_quoted_null = 1; /* note for later */
10615
10616 /* We do not want to add quoted nulls to strings that are only
10617 partially quoted; we can throw them away. The exception to
10618 this is when we are going to be performing word splitting,
10619 since we have to preserve a null argument if the next character
10620 will cause word splitting. */
10621 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && quoted == 0 && (word->flags & (W_NOSPLIT|W_EXPANDRHS|W_ASSIGNRHS)) == W_EXPANDRHS)
10622 {
10623 c = CTLNUL;
10624 sindex--;
10625 had_quoted_null = 1;
10626 goto add_character;
10627 }
10628 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
10629 continue;
10630
10631 add_quoted_string:
10632
10633 if (temp)
10634 {
10635 temp1 = temp;
10636 temp = quote_string (temp);
10637 free (temp1);
10638 goto add_string;
10639 }
10640 else
10641 {
10642 /* Add NULL arg. */
10643 c = CTLNUL;
10644 sindex--; /* add_character: label increments sindex */
10645 had_quoted_null = 1; /* note for later */
10646 goto add_character;
10647 }
10648
10649 /* break; */
10650
10651 case '\'':
10652 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
10653 goto add_character;
10654
10655 t_index = ++sindex;
10656 temp = string_extract_single_quoted (string, &sindex);
10657
10658 /* If the entire STRING was surrounded by single quotes,
10659 then the string is wholly quoted. */
10660 quoted_state = (t_index == 1 && string[sindex] == '\0')
10661 ? WHOLLY_QUOTED
10662 : PARTIALLY_QUOTED;
10663
10664 /* If all we had was '', it is a null expansion. */
10665 if (*temp == '\0')
10666 {
10667 free (temp);
10668 temp = (char *)NULL;
10669 }
10670 else
10671 remove_quoted_escapes (temp); /* ??? */
10672
10673 if (temp == 0 && quoted_state == PARTIALLY_QUOTED)
10674 had_quoted_null = 1; /* note for later */
10675
10676 /* We do not want to add quoted nulls to strings that are only
10677 partially quoted; such nulls are discarded. See above for the
10678 exception, which is when the string is going to be split.
10679 Posix interp 888/1129 */
10680 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && quoted == 0 && (word->flags & (W_NOSPLIT|W_EXPANDRHS|W_ASSIGNRHS)) == W_EXPANDRHS)
10681 {
10682 c = CTLNUL;
10683 sindex--;
10684 goto add_character;
10685 }
10686
10687 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
10688 continue;
10689
10690 /* If we have a quoted null expansion, add a quoted NULL to istring. */
10691 if (temp == 0)
10692 {
10693 c = CTLNUL;
10694 sindex--; /* add_character: label increments sindex */
10695 goto add_character;
10696 }
10697 else
10698 goto add_quoted_string;
10699
10700 /* break; */
10701
10702 case ' ':
10703 /* If we are in a context where the word is not going to be split, but
10704 we need to account for $@ and $* producing one word for each
10705 positional parameter, add quoted spaces so the spaces in the
10706 expansion of "$@", if any, behave correctly. We still may need to
10707 split if we are expanding the rhs of a word expansion. */
10708 if (ifs_is_null || split_on_spaces || ((word->flags & (W_NOSPLIT|W_NOSPLIT2|W_ASSIGNRHS)) && (word->flags & W_EXPANDRHS) == 0))
10709 {
10710 if (string[sindex])
10711 sindex++;
10712 twochars[0] = CTLESC;
10713 twochars[1] = c;
10714 goto add_twochars;
10715 }
10716 /* FALLTHROUGH */
10717
10718 default:
10719 /* This is the fix for " $@ " */
10720 add_ifs_character:
10721 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
10722 {
10723 if ((quoted&(Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)
10724 has_quoted_ifs++;
10725 add_quoted_character:
10726 if (string[sindex]) /* from old goto dollar_add_string */
10727 sindex++;
10728 if (c == 0)
10729 {
10730 c = CTLNUL;
10731 goto add_character;
10732 }
10733 else
10734 {
10735 #if HANDLE_MULTIBYTE
10736 /* XXX - should make sure that c is actually multibyte,
10737 otherwise we can use the twochars branch */
10738 if (mb_cur_max > 1)
10739 sindex--;
10740
10741 if (mb_cur_max > 1)
10742 {
10743 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
10744 }
10745 else
10746 #endif
10747 {
10748 twochars[0] = CTLESC;
10749 twochars[1] = c;
10750 goto add_twochars;
10751 }
10752 }
10753 }
10754
10755 SADD_MBCHAR (temp, string, sindex, string_size);
10756
10757 add_character:
10758 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
10759 DEFAULT_ARRAY_SIZE);
10760 istring[istring_index++] = c;
10761 istring[istring_index] = '\0';
10762
10763 /* Next character. */
10764 sindex++;
10765 }
10766 }
10767
10768 finished_with_string:
10769 /* OK, we're ready to return. If we have a quoted string, and
10770 quoted_dollar_at is not set, we do no splitting at all; otherwise
10771 we split on ' '. The routines that call this will handle what to
10772 do if nothing has been expanded. */
10773
10774 /* Partially and wholly quoted strings which expand to the empty
10775 string are retained as an empty arguments. Unquoted strings
10776 which expand to the empty string are discarded. The single
10777 exception is the case of expanding "$@" when there are no
10778 positional parameters. In that case, we discard the expansion. */
10779
10780 /* Because of how the code that handles "" and '' in partially
10781 quoted strings works, we need to make ISTRING into a QUOTED_NULL
10782 if we saw quoting characters, but the expansion was empty.
10783 "" and '' are tossed away before we get to this point when
10784 processing partially quoted strings. This makes "" and $xxx""
10785 equivalent when xxx is unset. We also look to see whether we
10786 saw a quoted null from a ${} expansion and add one back if we
10787 need to. */
10788
10789 /* If we expand to nothing and there were no single or double quotes
10790 in the word, we throw it away. Otherwise, we return a NULL word.
10791 The single exception is for $@ surrounded by double quotes when
10792 there are no positional parameters. In that case, we also throw
10793 the word away. */
10794
10795 if (*istring == '\0')
10796 {
10797 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
10798 {
10799 istring[0] = CTLNUL;
10800 istring[1] = '\0';
10801 tword = alloc_word_desc ();
10802 tword->word = istring;
10803 istring = 0; /* avoid later free() */
10804 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10805 list = make_word_list (tword, (WORD_LIST *)NULL);
10806 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
10807 tword->flags |= W_QUOTED;
10808 }
10809 /* According to sh, ksh, and Posix.2, if a word expands into nothing
10810 and a double-quoted "$@" appears anywhere in it, then the entire
10811 word is removed. */
10812 /* XXX - exception appears to be that quoted null strings result in
10813 null arguments */
10814 else if (quoted_state == UNQUOTED || quoted_dollar_at)
10815 list = (WORD_LIST *)NULL;
10816 else
10817 list = (WORD_LIST *)NULL;
10818 }
10819 else if (word->flags & W_NOSPLIT)
10820 {
10821 tword = alloc_word_desc ();
10822 tword->word = istring;
10823 if (had_quoted_null && QUOTED_NULL (istring))
10824 tword->flags |= W_HASQUOTEDNULL;
10825 istring = 0; /* avoid later free() */
10826 if (word->flags & W_ASSIGNMENT)
10827 tword->flags |= W_ASSIGNMENT; /* XXX */
10828 if (word->flags & W_COMPASSIGN)
10829 tword->flags |= W_COMPASSIGN; /* XXX */
10830 if (word->flags & W_NOGLOB)
10831 tword->flags |= W_NOGLOB; /* XXX */
10832 if (word->flags & W_NOBRACE)
10833 tword->flags |= W_NOBRACE; /* XXX */
10834 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
10835 tword->flags |= W_QUOTED;
10836 list = make_word_list (tword, (WORD_LIST *)NULL);
10837 }
10838 else if (word->flags & W_ASSIGNRHS)
10839 {
10840 list = list_string (istring, "", quoted);
10841 tword = list->word;
10842 if (had_quoted_null && QUOTED_NULL (istring))
10843 tword->flags |= W_HASQUOTEDNULL;
10844 free (list);
10845 free (istring);
10846 istring = 0; /* avoid later free() */
10847 goto set_word_flags;
10848 }
10849 else
10850 {
10851 char *ifs_chars;
10852
10853 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
10854
10855 /* If we have $@, we need to split the results no matter what. If
10856 IFS is unset or NULL, string_list_dollar_at has separated the
10857 positional parameters with a space, so we split on space (we have
10858 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
10859 string_list_dollar_at has separated the positional parameters
10860 with the first character of $IFS, so we split on $IFS. If
10861 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
10862 unset or null, and we want to make sure that we split on spaces
10863 regardless of what else has happened to IFS since the expansion,
10864 or we expanded "$@" with IFS null and we need to split the positional
10865 parameters into separate words. */
10866 if (split_on_spaces)
10867 {
10868 /* If IFS is not set, and the word is not quoted, we want to split
10869 the individual words on $' \t\n'. We rely on previous steps to
10870 quote the portions of the word that should not be split */
10871 if (ifs_is_set == 0)
10872 list = list_string (istring, " \t\n", 1); /* XXX quoted == 1? */
10873 else
10874 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
10875 }
10876
10877 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
10878 don't want to split the result (W_NOSPLIT2), and we are not quoted,
10879 we have already separated the arguments with the first character of
10880 $IFS. In this case, we want to return a list with a single word
10881 with the separator possibly replaced with a space (it's what other
10882 shells seem to do).
10883 quoted_dollar_at is internal to this function and is set if we are
10884 passed an argument that is unquoted (quoted == 0) but we encounter a
10885 double-quoted $@ while expanding it. */
10886 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
10887 {
10888 tword = alloc_word_desc ();
10889 /* Only split and rejoin if we have to */
10890 if (*ifs_chars && *ifs_chars != ' ')
10891 {
10892 /* list_string dequotes CTLESCs in the string it's passed, so we
10893 need it to get the space separation right if space isn't the
10894 first character in IFS (but is present) and to remove the
10895 quoting we added back in param_expand(). */
10896 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
10897 /* This isn't exactly right in the case where we're expanding
10898 the RHS of an expansion like ${var-$@} where IFS=: (for
10899 example). The W_NOSPLIT2 means we do the separation with :;
10900 the list_string removes the quotes and breaks the string into
10901 a list, and the string_list rejoins it on spaces. When we
10902 return, we expect to be able to split the results, but the
10903 space separation means the right split doesn't happen. */
10904 tword->word = string_list (list);
10905 }
10906 else
10907 tword->word = istring;
10908 if (had_quoted_null && QUOTED_NULL (istring))
10909 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10910 if (tword->word != istring)
10911 free (istring);
10912 istring = 0; /* avoid later free() */
10913 goto set_word_flags;
10914 }
10915 else if (has_dollar_at && ifs_chars)
10916 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
10917 else
10918 {
10919 tword = alloc_word_desc ();
10920 if (expanded_something && *expanded_something == 0 && has_quoted_ifs)
10921 tword->word = remove_quoted_ifs (istring);
10922 else
10923 tword->word = istring;
10924 if (had_quoted_null && QUOTED_NULL (istring)) /* should check for more than one */
10925 tword->flags |= W_HASQUOTEDNULL; /* XXX */
10926 else if (had_quoted_null)
10927 tword->flags |= W_SAWQUOTEDNULL; /* XXX */
10928 if (tword->word != istring)
10929 free (istring);
10930 istring = 0; /* avoid later free() */
10931 set_word_flags:
10932 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
10933 tword->flags |= W_QUOTED;
10934 if (word->flags & W_ASSIGNMENT)
10935 tword->flags |= W_ASSIGNMENT;
10936 if (word->flags & W_COMPASSIGN)
10937 tword->flags |= W_COMPASSIGN;
10938 if (word->flags & W_NOGLOB)
10939 tword->flags |= W_NOGLOB;
10940 if (word->flags & W_NOBRACE)
10941 tword->flags |= W_NOBRACE;
10942 list = make_word_list (tword, (WORD_LIST *)NULL);
10943 }
10944 }
10945
10946 free (istring);
10947 return (list);
10948 }
10949
10950 /* **************************************************************** */
10951 /* */
10952 /* Functions for Quote Removal */
10953 /* */
10954 /* **************************************************************** */
10955
10956 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
10957 backslash quoting rules for within double quotes or a here document. */
10958 char *
10959 string_quote_removal (string, quoted)
10960 char *string;
10961 int quoted;
10962 {
10963 size_t slen;
10964 char *r, *result_string, *temp, *send;
10965 int sindex, tindex, dquote;
10966 unsigned char c;
10967 DECLARE_MBSTATE;
10968
10969 /* The result can be no longer than the original string. */
10970 slen = strlen (string);
10971 send = string + slen;
10972
10973 r = result_string = (char *)xmalloc (slen + 1);
10974
10975 for (dquote = sindex = 0; c = string[sindex];)
10976 {
10977 switch (c)
10978 {
10979 case '\\':
10980 c = string[++sindex];
10981 if (c == 0)
10982 {
10983 *r++ = '\\';
10984 break;
10985 }
10986 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
10987 *r++ = '\\';
10988 /* FALLTHROUGH */
10989
10990 default:
10991 SCOPY_CHAR_M (r, string, send, sindex);
10992 break;
10993
10994 case '\'':
10995 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
10996 {
10997 *r++ = c;
10998 sindex++;
10999 break;
11000 }
11001 tindex = sindex + 1;
11002 temp = string_extract_single_quoted (string, &tindex);
11003 if (temp)
11004 {
11005 strcpy (r, temp);
11006 r += strlen (r);
11007 free (temp);
11008 }
11009 sindex = tindex;
11010 break;
11011
11012 case '"':
11013 dquote = 1 - dquote;
11014 sindex++;
11015 break;
11016 }
11017 }
11018 *r = '\0';
11019 return (result_string);
11020 }
11021
11022 #if 0
11023 /* UNUSED */
11024 /* Perform quote removal on word WORD. This allocates and returns a new
11025 WORD_DESC *. */
11026 WORD_DESC *
11027 word_quote_removal (word, quoted)
11028 WORD_DESC *word;
11029 int quoted;
11030 {
11031 WORD_DESC *w;
11032 char *t;
11033
11034 t = string_quote_removal (word->word, quoted);
11035 w = alloc_word_desc ();
11036 w->word = t ? t : savestring ("");
11037 return (w);
11038 }
11039
11040 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
11041 the members of the list are treated as if they are surrounded by
11042 double quotes. Return a new list, or NULL if LIST is NULL. */
11043 WORD_LIST *
11044 word_list_quote_removal (list, quoted)
11045 WORD_LIST *list;
11046 int quoted;
11047 {
11048 WORD_LIST *result, *t, *tresult, *e;
11049
11050 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
11051 {
11052 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
11053 #if 0
11054 result = (WORD_LIST *) list_append (result, tresult);
11055 #else
11056 if (result == 0)
11057 result = e = tresult;
11058 else
11059 {
11060 e->next = tresult;
11061 while (e->next)
11062 e = e->next;
11063 }
11064 #endif
11065 }
11066 return (result);
11067 }
11068 #endif
11069
11070 /*******************************************
11071 * *
11072 * Functions to perform word splitting *
11073 * *
11074 *******************************************/
11075
11076 void
11077 setifs (v)
11078 SHELL_VAR *v;
11079 {
11080 char *t;
11081 unsigned char uc;
11082
11083 ifs_var = v;
11084 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
11085
11086 ifs_is_set = ifs_var != 0;
11087 ifs_is_null = ifs_is_set && (*ifs_value == 0);
11088
11089 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
11090 handle multibyte chars in IFS */
11091 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
11092 for (t = ifs_value ; t && *t; t++)
11093 {
11094 uc = *t;
11095 ifs_cmap[uc] = 1;
11096 }
11097
11098 #if defined (HANDLE_MULTIBYTE)
11099 if (ifs_value == 0)
11100 {
11101 ifs_firstc[0] = '\0'; /* XXX - ? */
11102 ifs_firstc_len = 1;
11103 }
11104 else
11105 {
11106 if (locale_utf8locale && UTF8_SINGLEBYTE (*ifs_value))
11107 ifs_firstc_len = (*ifs_value != 0) ? 1 : 0;
11108 else
11109 {
11110 size_t ifs_len;
11111 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
11112 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
11113 }
11114 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
11115 {
11116 ifs_firstc[0] = ifs_value[0];
11117 ifs_firstc[1] = '\0';
11118 ifs_firstc_len = 1;
11119 }
11120 else
11121 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
11122 }
11123 #else
11124 ifs_firstc = ifs_value ? *ifs_value : 0;
11125 #endif
11126 }
11127
11128 char *
11129 getifs ()
11130 {
11131 return ifs_value;
11132 }
11133
11134 /* This splits a single word into a WORD LIST on $IFS, but only if the word
11135 is not quoted. list_string () performs quote removal for us, even if we
11136 don't do any splitting. */
11137 WORD_LIST *
11138 word_split (w, ifs_chars)
11139 WORD_DESC *w;
11140 char *ifs_chars;
11141 {
11142 WORD_LIST *result;
11143
11144 if (w)
11145 {
11146 char *xifs;
11147
11148 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
11149 result = list_string (w->word, xifs, w->flags & W_QUOTED);
11150 }
11151 else
11152 result = (WORD_LIST *)NULL;
11153
11154 return (result);
11155 }
11156
11157 /* Perform word splitting on LIST and return the RESULT. It is possible
11158 to return (WORD_LIST *)NULL. */
11159 static WORD_LIST *
11160 word_list_split (list)
11161 WORD_LIST *list;
11162 {
11163 WORD_LIST *result, *t, *tresult, *e;
11164 WORD_DESC *w;
11165
11166 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
11167 {
11168 tresult = word_split (t->word, ifs_value);
11169 /* POSIX 2.6: "If the complete expansion appropriate for a word results
11170 in an empty field, that empty field shall be deleted from the list
11171 of fields that form the completely expanded command, unless the
11172 original word contained single-quote or double-quote characters."
11173 This is where we handle these words that contain quoted null strings
11174 and other characters that expand to nothing after word splitting. */
11175 if (tresult == 0 && t->word && (t->word->flags & W_SAWQUOTEDNULL)) /* XXX */
11176 {
11177 w = alloc_word_desc ();
11178 w->word = (char *)xmalloc (1);
11179 w->word[0] = '\0';
11180 tresult = make_word_list (w, (WORD_LIST *)NULL);
11181 }
11182 if (result == 0)
11183 result = e = tresult;
11184 else
11185 {
11186 e->next = tresult;
11187 while (e->next)
11188 e = e->next;
11189 }
11190 }
11191 return (result);
11192 }
11193
11194 /**************************************************
11195 * *
11196 * Functions to expand an entire WORD_LIST *
11197 * *
11198 **************************************************/
11199
11200 /* Do any word-expansion-specific cleanup and jump to top_level */
11201 static void
11202 exp_jump_to_top_level (v)
11203 int v;
11204 {
11205 set_pipestatus_from_exit (last_command_exit_value);
11206
11207 /* Cleanup code goes here. */
11208 expand_no_split_dollar_star = 0; /* XXX */
11209 if (expanding_redir)
11210 undo_partial_redirects ();
11211 expanding_redir = 0;
11212 assigning_in_environment = 0;
11213
11214 if (parse_and_execute_level == 0)
11215 top_level_cleanup (); /* from sig.c */
11216
11217 jump_to_top_level (v);
11218 }
11219
11220 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
11221 ELIST, and set ELIST to the new list. */
11222 #define PREPEND_LIST(nlist, elist) \
11223 do { nlist->next = elist; elist = nlist; } while (0)
11224
11225 /* Separate out any initial variable assignments from TLIST. If set -k has
11226 been executed, remove all assignment statements from TLIST. Initial
11227 variable assignments and other environment assignments are placed
11228 on SUBST_ASSIGN_VARLIST. */
11229 static WORD_LIST *
11230 separate_out_assignments (tlist)
11231 WORD_LIST *tlist;
11232 {
11233 register WORD_LIST *vp, *lp;
11234
11235 if (tlist == 0)
11236 return ((WORD_LIST *)NULL);
11237
11238 if (subst_assign_varlist)
11239 dispose_words (subst_assign_varlist); /* Clean up after previous error */
11240
11241 subst_assign_varlist = (WORD_LIST *)NULL;
11242 vp = lp = tlist;
11243
11244 /* Separate out variable assignments at the start of the command.
11245 Loop invariant: vp->next == lp
11246 Loop postcondition:
11247 lp = list of words left after assignment statements skipped
11248 tlist = original list of words
11249 */
11250 while (lp && (lp->word->flags & W_ASSIGNMENT))
11251 {
11252 vp = lp;
11253 lp = lp->next;
11254 }
11255
11256 /* If lp != tlist, we have some initial assignment statements.
11257 We make SUBST_ASSIGN_VARLIST point to the list of assignment
11258 words and TLIST point to the remaining words. */
11259 if (lp != tlist)
11260 {
11261 subst_assign_varlist = tlist;
11262 /* ASSERT(vp->next == lp); */
11263 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
11264 tlist = lp; /* remainder of word list */
11265 }
11266
11267 /* vp == end of variable list */
11268 /* tlist == remainder of original word list without variable assignments */
11269 if (!tlist)
11270 /* All the words in tlist were assignment statements */
11271 return ((WORD_LIST *)NULL);
11272
11273 /* ASSERT(tlist != NULL); */
11274 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
11275
11276 /* If the -k option is in effect, we need to go through the remaining
11277 words, separate out the assignment words, and place them on
11278 SUBST_ASSIGN_VARLIST. */
11279 if (place_keywords_in_env)
11280 {
11281 WORD_LIST *tp; /* tp == running pointer into tlist */
11282
11283 tp = tlist;
11284 lp = tlist->next;
11285
11286 /* Loop Invariant: tp->next == lp */
11287 /* Loop postcondition: tlist == word list without assignment statements */
11288 while (lp)
11289 {
11290 if (lp->word->flags & W_ASSIGNMENT)
11291 {
11292 /* Found an assignment statement, add this word to end of
11293 subst_assign_varlist (vp). */
11294 if (!subst_assign_varlist)
11295 subst_assign_varlist = vp = lp;
11296 else
11297 {
11298 vp->next = lp;
11299 vp = lp;
11300 }
11301
11302 /* Remove the word pointed to by LP from TLIST. */
11303 tp->next = lp->next;
11304 /* ASSERT(vp == lp); */
11305 lp->next = (WORD_LIST *)NULL;
11306 lp = tp->next;
11307 }
11308 else
11309 {
11310 tp = lp;
11311 lp = lp->next;
11312 }
11313 }
11314 }
11315 return (tlist);
11316 }
11317
11318 #define WEXP_VARASSIGN 0x001
11319 #define WEXP_BRACEEXP 0x002
11320 #define WEXP_TILDEEXP 0x004
11321 #define WEXP_PARAMEXP 0x008
11322 #define WEXP_PATHEXP 0x010
11323
11324 /* All of the expansions, including variable assignments at the start of
11325 the list. */
11326 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
11327
11328 /* All of the expansions except variable assignments at the start of
11329 the list. */
11330 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
11331
11332 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
11333 expansion, command substitution, arithmetic expansion, word splitting, and
11334 quote removal. */
11335 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
11336
11337 /* Take the list of words in LIST and do the various substitutions. Return
11338 a new list of words which is the expanded list, and without things like
11339 variable assignments. */
11340
11341 WORD_LIST *
11342 expand_words (list)
11343 WORD_LIST *list;
11344 {
11345 return (expand_word_list_internal (list, WEXP_ALL));
11346 }
11347
11348 /* Same as expand_words (), but doesn't hack variable or environment
11349 variables. */
11350 WORD_LIST *
11351 expand_words_no_vars (list)
11352 WORD_LIST *list;
11353 {
11354 return (expand_word_list_internal (list, WEXP_NOVARS));
11355 }
11356
11357 WORD_LIST *
11358 expand_words_shellexp (list)
11359 WORD_LIST *list;
11360 {
11361 return (expand_word_list_internal (list, WEXP_SHELLEXP));
11362 }
11363
11364 static WORD_LIST *
11365 glob_expand_word_list (tlist, eflags)
11366 WORD_LIST *tlist;
11367 int eflags;
11368 {
11369 char **glob_array, *temp_string;
11370 register int glob_index;
11371 WORD_LIST *glob_list, *output_list, *disposables, *next;
11372 WORD_DESC *tword;
11373 int x;
11374
11375 output_list = disposables = (WORD_LIST *)NULL;
11376 glob_array = (char **)NULL;
11377 while (tlist)
11378 {
11379 /* For each word, either globbing is attempted or the word is
11380 added to orig_list. If globbing succeeds, the results are
11381 added to orig_list and the word (tlist) is added to the list
11382 of disposable words. If globbing fails and failed glob
11383 expansions are left unchanged (the shell default), the
11384 original word is added to orig_list. If globbing fails and
11385 failed glob expansions are removed, the original word is
11386 added to the list of disposable words. orig_list ends up
11387 in reverse order and requires a call to REVERSE_LIST to
11388 be set right. After all words are examined, the disposable
11389 words are freed. */
11390 next = tlist->next;
11391
11392 /* If the word isn't an assignment and contains an unquoted
11393 pattern matching character, then glob it. */
11394 if ((tlist->word->flags & W_NOGLOB) == 0 &&
11395 unquoted_glob_pattern_p (tlist->word->word))
11396 {
11397 glob_array = shell_glob_filename (tlist->word->word, QGLOB_CTLESC); /* XXX */
11398
11399 /* Handle error cases.
11400 I don't think we should report errors like "No such file
11401 or directory". However, I would like to report errors
11402 like "Read failed". */
11403
11404 if (glob_array == 0 || GLOB_FAILED (glob_array))
11405 {
11406 glob_array = (char **)xmalloc (sizeof (char *));
11407 glob_array[0] = (char *)NULL;
11408 }
11409
11410 /* Dequote the current word in case we have to use it. */
11411 if (glob_array[0] == NULL)
11412 {
11413 temp_string = dequote_string (tlist->word->word);
11414 free (tlist->word->word);
11415 tlist->word->word = temp_string;
11416 }
11417
11418 /* Make the array into a word list. */
11419 glob_list = (WORD_LIST *)NULL;
11420 for (glob_index = 0; glob_array[glob_index]; glob_index++)
11421 {
11422 tword = make_bare_word (glob_array[glob_index]);
11423 glob_list = make_word_list (tword, glob_list);
11424 }
11425
11426 if (glob_list)
11427 {
11428 output_list = (WORD_LIST *)list_append (glob_list, output_list);
11429 PREPEND_LIST (tlist, disposables);
11430 }
11431 else if (fail_glob_expansion != 0)
11432 {
11433 last_command_exit_value = EXECUTION_FAILURE;
11434 report_error (_("no match: %s"), tlist->word->word);
11435 exp_jump_to_top_level (DISCARD);
11436 }
11437 else if (allow_null_glob_expansion == 0)
11438 {
11439 /* Failed glob expressions are left unchanged. */
11440 PREPEND_LIST (tlist, output_list);
11441 }
11442 else
11443 {
11444 /* Failed glob expressions are removed. */
11445 PREPEND_LIST (tlist, disposables);
11446 }
11447 }
11448 else
11449 {
11450 /* Dequote the string. */
11451 temp_string = dequote_string (tlist->word->word);
11452 free (tlist->word->word);
11453 tlist->word->word = temp_string;
11454 PREPEND_LIST (tlist, output_list);
11455 }
11456
11457 strvec_dispose (glob_array);
11458 glob_array = (char **)NULL;
11459
11460 tlist = next;
11461 }
11462
11463 if (disposables)
11464 dispose_words (disposables);
11465
11466 if (output_list)
11467 output_list = REVERSE_LIST (output_list, WORD_LIST *);
11468
11469 return (output_list);
11470 }
11471
11472 #if defined (BRACE_EXPANSION)
11473 static WORD_LIST *
11474 brace_expand_word_list (tlist, eflags)
11475 WORD_LIST *tlist;
11476 int eflags;
11477 {
11478 register char **expansions;
11479 char *temp_string;
11480 WORD_LIST *disposables, *output_list, *next;
11481 WORD_DESC *w;
11482 int eindex;
11483
11484 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
11485 {
11486 next = tlist->next;
11487
11488 if (tlist->word->flags & W_NOBRACE)
11489 {
11490 /*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
11491 PREPEND_LIST (tlist, output_list);
11492 continue;
11493 }
11494
11495 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
11496 {
11497 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
11498 PREPEND_LIST (tlist, output_list);
11499 continue;
11500 }
11501
11502 /* Only do brace expansion if the word has a brace character. If
11503 not, just add the word list element to BRACES and continue. In
11504 the common case, at least when running shell scripts, this will
11505 degenerate to a bunch of calls to `mbschr', and then what is
11506 basically a reversal of TLIST into BRACES, which is corrected
11507 by a call to REVERSE_LIST () on BRACES when the end of TLIST
11508 is reached. */
11509 if (mbschr (tlist->word->word, LBRACE))
11510 {
11511 expansions = brace_expand (tlist->word->word);
11512
11513 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
11514 {
11515 w = alloc_word_desc ();
11516 w->word = temp_string;
11517
11518 /* If brace expansion didn't change the word, preserve
11519 the flags. We may want to preserve the flags
11520 unconditionally someday -- XXX */
11521 if (STREQ (temp_string, tlist->word->word))
11522 w->flags = tlist->word->flags;
11523 else
11524 w = make_word_flags (w, temp_string);
11525
11526 output_list = make_word_list (w, output_list);
11527 }
11528 free (expansions);
11529
11530 /* Add TLIST to the list of words to be freed after brace
11531 expansion has been performed. */
11532 PREPEND_LIST (tlist, disposables);
11533 }
11534 else
11535 PREPEND_LIST (tlist, output_list);
11536 }
11537
11538 if (disposables)
11539 dispose_words (disposables);
11540
11541 if (output_list)
11542 output_list = REVERSE_LIST (output_list, WORD_LIST *);
11543
11544 return (output_list);
11545 }
11546 #endif
11547
11548 #if defined (ARRAY_VARS)
11549 /* Take WORD, a compound array assignment, and internally run (for example),
11550 'declare -A w', where W is the variable name portion of WORD. OPTION is
11551 the list of options to supply to `declare'. CMD is the declaration command
11552 we are expanding right now; it's unused currently. */
11553 static int
11554 make_internal_declare (word, option, cmd)
11555 char *word;
11556 char *option;
11557 char *cmd;
11558 {
11559 int t, r;
11560 WORD_LIST *wl;
11561 WORD_DESC *w;
11562
11563 w = make_word (word);
11564
11565 t = assignment (w->word, 0);
11566 if (w->word[t] == '=')
11567 {
11568 w->word[t] = '\0';
11569 if (w->word[t - 1] == '+') /* cut off any append op */
11570 w->word[t - 1] = '\0';
11571 }
11572
11573 wl = make_word_list (w, (WORD_LIST *)NULL);
11574 wl = make_word_list (make_word (option), wl);
11575
11576 r = declare_builtin (wl);
11577
11578 dispose_words (wl);
11579 return r;
11580 }
11581
11582 /* Expand VALUE in NAME[+]=( VALUE ) to a list of words. FLAGS is 1 if NAME
11583 is an associative array.
11584
11585 If we are processing an indexed array, expand_compound_array_assignment
11586 will expand all the individual words and quote_compound_array_list will
11587 single-quote them. If we are processing an associative array, we use
11588 parse_string_to_word_list to split VALUE into a list of words instead of
11589 faking up a shell variable and calling expand_compound_array_assignment.
11590 expand_and_quote_assoc_word expands and single-quotes each word in VALUE
11591 together so we don't have problems finding the end of the subscript when
11592 quoting it.
11593
11594 Words in VALUE can be individual words, which are expanded and single-quoted,
11595 or words of the form [IND]=VALUE, which end up as explained below, as
11596 ['expanded-ind']='expanded-value'. */
11597
11598 static WORD_LIST *
11599 expand_oneword (value, flags)
11600 char *value;
11601 int flags;
11602 {
11603 WORD_LIST *l, *nl;
11604 char *t;
11605
11606 if (flags == 0)
11607 {
11608 /* Indexed array */
11609 l = expand_compound_array_assignment ((SHELL_VAR *)NULL, value, flags);
11610 /* Now we quote the results of the expansion above to prevent double
11611 expansion. */
11612 quote_compound_array_list (l, flags);
11613 return l;
11614 }
11615 else
11616 {
11617 /* Associative array */
11618 l = parse_string_to_word_list (value, 1, "array assign");
11619 /* For associative arrays, with their arbitrary subscripts, we have to
11620 expand and quote in one step so we don't have to search for the
11621 closing right bracket more than once. */
11622 for (nl = l; nl; nl = nl->next)
11623 {
11624 if ((nl->word->flags & W_ASSIGNMENT) == 0)
11625 t = sh_single_quote (nl->word->word ? nl->word->word : "");
11626 else
11627 t = expand_and_quote_assoc_word (nl->word->word, flags);
11628 free (nl->word->word);
11629 nl->word->word = t;
11630 }
11631 return l;
11632 }
11633 }
11634
11635 /* Expand a single compound assignment argument to a declaration builtin.
11636 This word takes the form NAME[+]=( VALUE ). The NAME[+]= is passed through
11637 unchanged. The VALUE is expanded and each word in the result is single-
11638 quoted. Words of the form [key]=value end up as
11639 ['expanded-key']='expanded-value'. Associative arrays have special
11640 handling, see expand_oneword() above. The return value is
11641 NAME[+]=( expanded-and-quoted-VALUE ). */
11642 static void
11643 expand_compound_assignment_word (tlist, flags)
11644 WORD_LIST *tlist;
11645 int flags;
11646 {
11647 WORD_LIST *l;
11648 int wlen, oind, t;
11649 char *value, *temp;
11650
11651 /*itrace("expand_compound_assignment_word: original word = -%s-", tlist->word->word);*/
11652 t = assignment (tlist->word->word, 0);
11653
11654 /* value doesn't have the open and close parens */
11655 oind = 1;
11656 value = extract_array_assignment_list (tlist->word->word + t + 1, &oind);
11657 /* This performs one round of expansion on the index/key and value and
11658 single-quotes each word in the result. */
11659 l = expand_oneword (value, flags);
11660 free (value);
11661
11662 value = string_list (l);
11663 wlen = STRLEN (value);
11664
11665 /* Now, let's rebuild the string */
11666 temp = xmalloc (t + 3 + wlen + 1); /* name[+]=(value) */
11667 memcpy (temp, tlist->word->word, ++t);
11668 temp[t++] = '(';
11669 if (value)
11670 memcpy (temp + t, value, wlen);
11671 t += wlen;
11672 temp[t++] = ')';
11673 temp[t] = '\0';
11674 /*itrace("expand_compound_assignment_word: reconstructed word = -%s-", temp);*/
11675
11676 free (tlist->word->word);
11677 tlist->word->word = temp;
11678
11679 free (value);
11680 }
11681
11682 /* Expand and process an argument to a declaration command. We have already
11683 set flags in TLIST->word->flags depending on the declaration command
11684 (declare, local, etc.) and the options supplied to it (-a, -A, etc.).
11685 TLIST->word->word is of the form NAME[+]=( VALUE ).
11686
11687 This does several things, all using pieces of other functions to get the
11688 evaluation sequence right. It's called for compound array assignments with
11689 the W_ASSIGNMENT flag set (basically, valid identifier names on the lhs).
11690 It parses out which flags need to be set for declare to create the variable
11691 correctly, then calls declare internally (make_internal_declare) to make
11692 sure the variable exists with the correct attributes. Before the variable
11693 is created, it calls expand_compound_assignment_word to expand VALUE to a
11694 list of words, appropriately quoted for further evaluation. This preserves
11695 the semantics of word-expansion-before-calling-builtins. Finally, it calls
11696 do_word_assignment to perform the expansion and assignment with the same
11697 expansion semantics as a standalone assignment statement (no word splitting,
11698 etc.) even though the word is single-quoted so all that needs to happen is
11699 quote removal. */
11700 static WORD_LIST *
11701 expand_declaration_argument (tlist, wcmd)
11702 WORD_LIST *tlist, *wcmd;
11703 {
11704 char opts[16], omap[128];
11705 int t, opti, oind, skip, inheriting;
11706 WORD_LIST *l;
11707
11708 inheriting = localvar_inherit;
11709 opti = 0;
11710 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_CHKLOCAL|W_ASSIGNARRAY))
11711 opts[opti++] = '-';
11712
11713 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
11714 {
11715 opts[opti++] = 'g';
11716 opts[opti++] = 'A';
11717 }
11718 else if (tlist->word->flags & W_ASSIGNASSOC)
11719 {
11720 opts[opti++] = 'A';
11721 }
11722 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
11723 {
11724 opts[opti++] = 'g';
11725 opts[opti++] = 'a';
11726 }
11727 else if (tlist->word->flags & W_ASSIGNARRAY)
11728 {
11729 opts[opti++] = 'a';
11730 }
11731 else if (tlist->word->flags & W_ASSNGLOBAL)
11732 opts[opti++] = 'g';
11733
11734 if (tlist->word->flags & W_CHKLOCAL)
11735 opts[opti++] = 'G';
11736
11737 /* If we have special handling note the integer attribute and others
11738 that transform the value upon assignment. What we do is take all
11739 of the option arguments and scan through them looking for options
11740 that cause such transformations, and add them to the `opts' array. */
11741
11742 memset (omap, '\0', sizeof (omap));
11743 for (l = wcmd->next; l != tlist; l = l->next)
11744 {
11745 if (l->word->word[0] != '-')
11746 break; /* non-option argument */
11747 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
11748 break; /* -- signals end of options */
11749 for (oind = 1; l->word->word[oind]; oind++)
11750 switch (l->word->word[oind])
11751 {
11752 case 'I':
11753 inheriting = 1;
11754 case 'i':
11755 case 'l':
11756 case 'u':
11757 case 'c':
11758 omap[l->word->word[oind]] = 1;
11759 if (opti == 0)
11760 opts[opti++] = '-';
11761 break;
11762 default:
11763 break;
11764 }
11765 }
11766
11767 for (oind = 0; oind < sizeof (omap); oind++)
11768 if (omap[oind])
11769 opts[opti++] = oind;
11770
11771 /* If there are no -a/-A options, but we have a compound assignment,
11772 we have a choice: we can set opts[0]='-', opts[1]='a', since the
11773 default is to create an indexed array, and call
11774 make_internal_declare with that, or we can just skip the -a and let
11775 declare_builtin deal with it. Once we're here, we're better set
11776 up for the latter, since we don't want to deal with looking up
11777 any existing variable here -- better to let declare_builtin do it.
11778 We need the variable created, though, especially if it's local, so
11779 we get the scoping right before we call do_word_assignment.
11780 To ensure that make_local_declare gets called, we add `--' if there
11781 aren't any options. */
11782 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSIGNARRAY)) == 0)
11783 {
11784 if (opti == 0)
11785 {
11786 opts[opti++] = '-';
11787 opts[opti++] = '-';
11788 }
11789 }
11790 opts[opti] = '\0';
11791
11792 /* This isn't perfect, but it's a start. Improvements later. We expand
11793 tlist->word->word and single-quote the results to avoid multiple
11794 expansions by, say, do_assignment_internal(). We have to weigh the
11795 cost of reconstructing the compound assignment string with its single
11796 quoting and letting the declare builtin handle it. The single quotes
11797 will prevent any unwanted additional expansion or word splitting. */
11798 expand_compound_assignment_word (tlist, (tlist->word->flags & W_ASSIGNASSOC) ? 1 : 0);
11799
11800 skip = 0;
11801 if (opti > 0)
11802 {
11803 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
11804 if (t != EXECUTION_SUCCESS)
11805 {
11806 last_command_exit_value = t;
11807 if (tlist->word->flags & W_FORCELOCAL) /* non-fatal error */
11808 skip = 1;
11809 else
11810 exp_jump_to_top_level (DISCARD);
11811 }
11812 }
11813
11814 if (skip == 0)
11815 {
11816 t = do_word_assignment (tlist->word, 0);
11817 if (t == 0)
11818 {
11819 last_command_exit_value = EXECUTION_FAILURE;
11820 exp_jump_to_top_level (DISCARD);
11821 }
11822 }
11823
11824 /* Now transform the word as ksh93 appears to do and go on */
11825 t = assignment (tlist->word->word, 0);
11826 tlist->word->word[t] = '\0';
11827 if (tlist->word->word[t - 1] == '+')
11828 tlist->word->word[t - 1] = '\0'; /* cut off append op */
11829 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
11830
11831 return (tlist);
11832 }
11833 #endif /* ARRAY_VARS */
11834
11835 static WORD_LIST *
11836 shell_expand_word_list (tlist, eflags)
11837 WORD_LIST *tlist;
11838 int eflags;
11839 {
11840 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
11841 int expanded_something, has_dollar_at;
11842
11843 /* We do tilde expansion all the time. This is what 1003.2 says. */
11844 wcmd = new_list = (WORD_LIST *)NULL;
11845
11846 for (orig_list = tlist; tlist; tlist = next)
11847 {
11848 if (wcmd == 0 && (tlist->word->flags & W_ASSNBLTIN))
11849 wcmd = tlist;
11850
11851 next = tlist->next;
11852
11853 #if defined (ARRAY_VARS)
11854 /* If this is a compound array assignment to a builtin that accepts
11855 such assignments (e.g., `declare'), take the assignment and perform
11856 it separately, handling the semantics of declarations inside shell
11857 functions. This avoids the double-evaluation of such arguments,
11858 because `declare' does some evaluation of compound assignments on
11859 its own. */
11860 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
11861 expand_declaration_argument (tlist, wcmd);
11862 #endif
11863
11864 expanded_something = 0;
11865 expanded = expand_word_internal
11866 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
11867
11868 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
11869 {
11870 /* By convention, each time this error is returned,
11871 tlist->word->word has already been freed. */
11872 tlist->word->word = (char *)NULL;
11873
11874 /* Dispose our copy of the original list. */
11875 dispose_words (orig_list);
11876 /* Dispose the new list we're building. */
11877 dispose_words (new_list);
11878
11879 last_command_exit_value = EXECUTION_FAILURE;
11880 if (expanded == &expand_word_error)
11881 exp_jump_to_top_level (DISCARD);
11882 else
11883 exp_jump_to_top_level (FORCE_EOF);
11884 }
11885
11886 /* Don't split words marked W_NOSPLIT. */
11887 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
11888 {
11889 temp_list = word_list_split (expanded);
11890 dispose_words (expanded);
11891 }
11892 else
11893 {
11894 /* If no parameter expansion, command substitution, process
11895 substitution, or arithmetic substitution took place, then
11896 do not do word splitting. We still have to remove quoted
11897 null characters from the result. */
11898 word_list_remove_quoted_nulls (expanded);
11899 temp_list = expanded;
11900 }
11901
11902 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
11903 new_list = (WORD_LIST *)list_append (expanded, new_list);
11904 }
11905
11906 if (orig_list)
11907 dispose_words (orig_list);
11908
11909 if (new_list)
11910 new_list = REVERSE_LIST (new_list, WORD_LIST *);
11911
11912 return (new_list);
11913 }
11914
11915 /* The workhorse for expand_words () and expand_words_no_vars ().
11916 First arg is LIST, a WORD_LIST of words.
11917 Second arg EFLAGS is a flags word controlling which expansions are
11918 performed.
11919
11920 This does all of the substitutions: brace expansion, tilde expansion,
11921 parameter expansion, command substitution, arithmetic expansion,
11922 process substitution, word splitting, and pathname expansion, according
11923 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
11924 set, or for which no expansion is done, do not undergo word splitting.
11925 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
11926 with W_NOBRACE set do not undergo brace expansion (see
11927 brace_expand_word_list above). */
11928 static WORD_LIST *
11929 expand_word_list_internal (list, eflags)
11930 WORD_LIST *list;
11931 int eflags;
11932 {
11933 WORD_LIST *new_list, *temp_list;
11934 int tint;
11935 char *savecmd;
11936
11937 tempenv_assign_error = 0;
11938 if (list == 0)
11939 return ((WORD_LIST *)NULL);
11940
11941 garglist = new_list = copy_word_list (list);
11942 if (eflags & WEXP_VARASSIGN)
11943 {
11944 garglist = new_list = separate_out_assignments (new_list);
11945 if (new_list == 0)
11946 {
11947 if (subst_assign_varlist)
11948 {
11949 /* All the words were variable assignments, so they are placed
11950 into the shell's environment. */
11951 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
11952 {
11953 savecmd = this_command_name;
11954 this_command_name = (char *)NULL; /* no arithmetic errors */
11955 tint = do_word_assignment (temp_list->word, 0);
11956 this_command_name = savecmd;
11957 /* Variable assignment errors in non-interactive shells
11958 running in Posix.2 mode cause the shell to exit, unless
11959 they are being run by the `command' builtin. */
11960 if (tint == 0)
11961 {
11962 last_command_exit_value = EXECUTION_FAILURE;
11963 if (interactive_shell == 0 && posixly_correct && executing_command_builtin == 0)
11964 exp_jump_to_top_level (FORCE_EOF);
11965 else
11966 exp_jump_to_top_level (DISCARD);
11967 }
11968 }
11969 dispose_words (subst_assign_varlist);
11970 subst_assign_varlist = (WORD_LIST *)NULL;
11971 }
11972 return ((WORD_LIST *)NULL);
11973 }
11974 }
11975
11976 /* Begin expanding the words that remain. The expansions take place on
11977 things that aren't really variable assignments. */
11978
11979 #if defined (BRACE_EXPANSION)
11980 /* Do brace expansion on this word if there are any brace characters
11981 in the string. */
11982 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
11983 new_list = brace_expand_word_list (new_list, eflags);
11984 #endif /* BRACE_EXPANSION */
11985
11986 /* Perform the `normal' shell expansions: tilde expansion, parameter and
11987 variable substitution, command substitution, arithmetic expansion,
11988 and word splitting. */
11989 new_list = shell_expand_word_list (new_list, eflags);
11990
11991 /* Okay, we're almost done. Now let's just do some filename
11992 globbing. */
11993 if (new_list)
11994 {
11995 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
11996 /* Glob expand the word list unless globbing has been disabled. */
11997 new_list = glob_expand_word_list (new_list, eflags);
11998 else
11999 /* Dequote the words, because we're not performing globbing. */
12000 new_list = dequote_list (new_list);
12001 }
12002
12003 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
12004 {
12005 sh_wassign_func_t *assign_func;
12006 int is_special_builtin, is_builtin_or_func;
12007
12008 /* If the remainder of the words expand to nothing, Posix.2 requires
12009 that the variable and environment assignments affect the shell's
12010 environment. */
12011 assign_func = new_list ? assign_in_env : do_word_assignment;
12012 tempenv_assign_error = 0;
12013
12014 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
12015 /* Posix says that special builtins exit if a variable assignment error
12016 occurs in an assignment preceding it. */
12017 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
12018
12019 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
12020 {
12021 savecmd = this_command_name;
12022 this_command_name = (char *)NULL;
12023 assigning_in_environment = (assign_func == assign_in_env);
12024 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
12025 assigning_in_environment = 0;
12026 this_command_name = savecmd;
12027 /* Variable assignment errors in non-interactive shells running
12028 in Posix.2 mode cause the shell to exit. */
12029 if (tint == 0)
12030 {
12031 if (assign_func == do_word_assignment)
12032 {
12033 last_command_exit_value = EXECUTION_FAILURE;
12034 if (interactive_shell == 0 && posixly_correct)
12035 exp_jump_to_top_level (FORCE_EOF);
12036 else
12037 exp_jump_to_top_level (DISCARD);
12038 }
12039 else if (interactive_shell == 0 && is_special_builtin)
12040 {
12041 last_command_exit_value = EXECUTION_FAILURE;
12042 exp_jump_to_top_level (FORCE_EOF);
12043 }
12044 else
12045 tempenv_assign_error++;
12046 }
12047 }
12048
12049 dispose_words (subst_assign_varlist);
12050 subst_assign_varlist = (WORD_LIST *)NULL;
12051 }
12052
12053 return (new_list);
12054 }