]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
Bash-4.3 patch 14
[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) 1987-2013 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 #include "bashansi.h"
41 #include "posixstat.h"
42 #include "bashintl.h"
43
44 #include "shell.h"
45 #include "parser.h"
46 #include "flags.h"
47 #include "jobs.h"
48 #include "execute_cmd.h"
49 #include "filecntl.h"
50 #include "trap.h"
51 #include "pathexp.h"
52 #include "mailcheck.h"
53
54 #include "shmbutil.h"
55 #include "typemax.h"
56
57 #include "builtins/getopt.h"
58 #include "builtins/common.h"
59
60 #include "builtins/builtext.h"
61
62 #include <tilde/tilde.h>
63 #include <glob/strmatch.h>
64
65 #if !defined (errno)
66 extern int errno;
67 #endif /* !errno */
68
69 /* The size that strings change by. */
70 #define DEFAULT_INITIAL_ARRAY_SIZE 112
71 #define DEFAULT_ARRAY_SIZE 128
72
73 /* Variable types. */
74 #define VT_VARIABLE 0
75 #define VT_POSPARMS 1
76 #define VT_ARRAYVAR 2
77 #define VT_ARRAYMEMBER 3
78 #define VT_ASSOCVAR 4
79
80 #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */
81
82 /* Flags for quoted_strchr */
83 #define ST_BACKSL 0x01
84 #define ST_CTLESC 0x02
85 #define ST_SQUOTE 0x04 /* unused yet */
86 #define ST_DQUOTE 0x08 /* unused yet */
87
88 /* Flags for the `pflags' argument to param_expand() */
89 #define PF_NOCOMSUB 0x01 /* Do not perform command substitution */
90 #define PF_IGNUNBOUND 0x02 /* ignore unbound vars even if -u set */
91 #define PF_NOSPLIT2 0x04 /* same as W_NOSPLIT2 */
92 #define PF_ASSIGNRHS 0x08 /* same as W_ASSIGNRHS */
93
94 /* These defs make it easier to use the editor. */
95 #define LBRACE '{'
96 #define RBRACE '}'
97 #define LPAREN '('
98 #define RPAREN ')'
99 #define LBRACK '['
100 #define RBRACK ']'
101
102 #if defined (HANDLE_MULTIBYTE)
103 #define WLPAREN L'('
104 #define WRPAREN L')'
105 #endif
106
107 /* Evaluates to 1 if C is one of the shell's special parameters whose length
108 can be taken, but is also one of the special expansion characters. */
109 #define VALID_SPECIAL_LENGTH_PARAM(c) \
110 ((c) == '-' || (c) == '?' || (c) == '#')
111
112 /* Evaluates to 1 if C is one of the shell's special parameters for which an
113 indirect variable reference may be made. */
114 #define VALID_INDIR_PARAM(c) \
115 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
116
117 /* Evaluates to 1 if C is one of the OP characters that follows the parameter
118 in ${parameter[:]OPword}. */
119 #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
120
121 /* Evaluates to 1 if this is one of the shell's special variables. */
122 #define SPECIAL_VAR(name, wi) \
123 ((DIGIT (*name) && all_digits (name)) || \
124 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
125 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1])))
126
127 /* An expansion function that takes a string and a quoted flag and returns
128 a WORD_LIST *. Used as the type of the third argument to
129 expand_string_if_necessary(). */
130 typedef WORD_LIST *EXPFUNC __P((char *, int));
131
132 /* Process ID of the last command executed within command substitution. */
133 pid_t last_command_subst_pid = NO_PID;
134 pid_t current_command_subst_pid = NO_PID;
135
136 /* Variables used to keep track of the characters in IFS. */
137 SHELL_VAR *ifs_var;
138 char *ifs_value;
139 unsigned char ifs_cmap[UCHAR_MAX + 1];
140 int ifs_is_set, ifs_is_null;
141
142 #if defined (HANDLE_MULTIBYTE)
143 unsigned char ifs_firstc[MB_LEN_MAX];
144 size_t ifs_firstc_len;
145 #else
146 unsigned char ifs_firstc;
147 #endif
148
149 /* Sentinel to tell when we are performing variable assignments preceding a
150 command name and putting them into the environment. Used to make sure
151 we use the temporary environment when looking up variable values. */
152 int assigning_in_environment;
153
154 /* Used to hold a list of variable assignments preceding a command. Global
155 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
156 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
157 WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
158
159 /* Extern functions and variables from different files. */
160 extern int last_command_exit_value, last_command_exit_signal;
161 extern int subshell_environment, line_number;
162 extern int subshell_level, parse_and_execute_level, sourcelevel;
163 extern int eof_encountered;
164 extern int return_catch_flag, return_catch_value;
165 extern pid_t dollar_dollar_pid;
166 extern int posixly_correct;
167 extern char *this_command_name;
168 extern struct fd_bitmap *current_fds_to_close;
169 extern int wordexp_only;
170 extern int expanding_redir;
171 extern int tempenv_assign_error;
172 extern int builtin_ignoring_errexit;
173
174 #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
175 extern wchar_t *wcsdup __P((const wchar_t *));
176 #endif
177
178 /* Non-zero means to allow unmatched globbed filenames to expand to
179 a null file. */
180 int allow_null_glob_expansion;
181
182 /* Non-zero means to throw an error when globbing fails to match anything. */
183 int fail_glob_expansion;
184
185 #if 0
186 /* Variables to keep track of which words in an expanded word list (the
187 output of expand_word_list_internal) are the result of globbing
188 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
189 (CURRENTLY UNUSED). */
190 char *glob_argv_flags;
191 static int glob_argv_flags_size;
192 #endif
193
194 static WORD_LIST expand_word_error, expand_word_fatal;
195 static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
196 static char expand_param_error, expand_param_fatal;
197 static char extract_string_error, extract_string_fatal;
198
199 /* Tell the expansion functions to not longjmp back to top_level on fatal
200 errors. Enabled when doing completion and prompt string expansion. */
201 static int no_longjmp_on_fatal_error = 0;
202
203 /* Set by expand_word_unsplit; used to inhibit splitting and re-joining
204 $* on $IFS, primarily when doing assignment statements. */
205 static int expand_no_split_dollar_star = 0;
206
207 /* A WORD_LIST of words to be expanded by expand_word_list_internal,
208 without any leading variable assignments. */
209 static WORD_LIST *garglist = (WORD_LIST *)NULL;
210
211 static char *quoted_substring __P((char *, int, int));
212 static int quoted_strlen __P((char *));
213 static char *quoted_strchr __P((char *, int, int));
214
215 static char *expand_string_if_necessary __P((char *, int, EXPFUNC *));
216 static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *));
217 static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
218 static WORD_LIST *expand_string_internal __P((char *, int));
219 static WORD_LIST *expand_string_leave_quoted __P((char *, int));
220 static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *));
221
222 static WORD_LIST *list_quote_escapes __P((WORD_LIST *));
223 static WORD_LIST *list_dequote_escapes __P((WORD_LIST *));
224
225 static char *make_quoted_char __P((int));
226 static WORD_LIST *quote_list __P((WORD_LIST *));
227
228 static int unquoted_substring __P((char *, char *));
229 static int unquoted_member __P((int, char *));
230
231 #if defined (ARRAY_VARS)
232 static SHELL_VAR *do_compound_assignment __P((char *, char *, int));
233 #endif
234 static int do_assignment_internal __P((const WORD_DESC *, int));
235
236 static char *string_extract_verbatim __P((char *, size_t, int *, char *, int));
237 static char *string_extract __P((char *, int *, char *, int));
238 static char *string_extract_double_quoted __P((char *, int *, int));
239 static inline char *string_extract_single_quoted __P((char *, int *));
240 static inline int skip_single_quoted __P((const char *, size_t, int));
241 static int skip_double_quoted __P((char *, size_t, int));
242 static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int));
243 static char *extract_dollar_brace_string __P((char *, int *, int, int));
244 static int skip_matched_pair __P((const char *, int, int, int, int));
245
246 static char *pos_params __P((char *, int, int, int));
247
248 static unsigned char *mb_getcharlens __P((char *, int));
249
250 static char *remove_upattern __P((char *, char *, int));
251 #if defined (HANDLE_MULTIBYTE)
252 static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int));
253 #endif
254 static char *remove_pattern __P((char *, char *, int));
255
256 static int match_upattern __P((char *, char *, int, char **, char **));
257 #if defined (HANDLE_MULTIBYTE)
258 static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
259 #endif
260 static int match_pattern __P((char *, char *, int, char **, char **));
261 static int getpatspec __P((int, char *));
262 static char *getpattern __P((char *, int, int));
263 static char *variable_remove_pattern __P((char *, char *, int, int));
264 static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int));
265 static char *parameter_list_remove_pattern __P((int, char *, int, int));
266 #ifdef ARRAY_VARS
267 static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int));
268 #endif
269 static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int));
270
271 static char *process_substitute __P((char *, int));
272
273 static char *read_comsub __P((int, int, int *));
274
275 #ifdef ARRAY_VARS
276 static arrayind_t array_length_reference __P((char *));
277 #endif
278
279 static int valid_brace_expansion_word __P((char *, int));
280 static int chk_atstar __P((char *, int, int *, int *));
281 static int chk_arithsub __P((const char *, int));
282
283 static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *));
284 static char *parameter_brace_find_indir __P((char *, int, int, int));
285 static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *));
286 static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *));
287 static void parameter_brace_expand_error __P((char *, char *));
288
289 static int valid_length_expression __P((char *));
290 static intmax_t parameter_brace_expand_length __P((char *));
291
292 static char *skiparith __P((char *, int));
293 static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
294 static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
295 static char *mb_substring __P((char *, int, int));
296 static char *parameter_brace_substring __P((char *, char *, int, char *, int, int));
297
298 static int shouldexp_replacement __P((char *));
299
300 static char *pos_params_pat_subst __P((char *, char *, char *, int));
301
302 static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int));
303
304 static char *pos_params_casemod __P((char *, char *, int, int));
305 static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int));
306
307 static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *));
308 static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int));
309
310 static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
311
312 static WORD_LIST *word_list_split __P((WORD_LIST *));
313
314 static void exp_jump_to_top_level __P((int));
315
316 static WORD_LIST *separate_out_assignments __P((WORD_LIST *));
317 static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int));
318 #ifdef BRACE_EXPANSION
319 static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int));
320 #endif
321 #if defined (ARRAY_VARS)
322 static int make_internal_declare __P((char *, char *));
323 #endif
324 static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int));
325 static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int));
326
327 /* **************************************************************** */
328 /* */
329 /* Utility Functions */
330 /* */
331 /* **************************************************************** */
332
333 #if defined (DEBUG)
334 void
335 dump_word_flags (flags)
336 int flags;
337 {
338 int f;
339
340 f = flags;
341 fprintf (stderr, "%d -> ", f);
342 if (f & W_ASSIGNASSOC)
343 {
344 f &= ~W_ASSIGNASSOC;
345 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
346 }
347 if (f & W_ASSIGNARRAY)
348 {
349 f &= ~W_ASSIGNARRAY;
350 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
351 }
352 if (f & W_HASCTLESC)
353 {
354 f &= ~W_HASCTLESC;
355 fprintf (stderr, "W_HASCTLESC%s", f ? "|" : "");
356 }
357 if (f & W_NOPROCSUB)
358 {
359 f &= ~W_NOPROCSUB;
360 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
361 }
362 if (f & W_DQUOTE)
363 {
364 f &= ~W_DQUOTE;
365 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
366 }
367 if (f & W_HASQUOTEDNULL)
368 {
369 f &= ~W_HASQUOTEDNULL;
370 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
371 }
372 if (f & W_ASSIGNARG)
373 {
374 f &= ~W_ASSIGNARG;
375 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
376 }
377 if (f & W_ASSNBLTIN)
378 {
379 f &= ~W_ASSNBLTIN;
380 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
381 }
382 if (f & W_ASSNGLOBAL)
383 {
384 f &= ~W_ASSNGLOBAL;
385 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
386 }
387 if (f & W_ASSIGNINT)
388 {
389 f &= ~W_ASSIGNINT;
390 fprintf (stderr, "W_ASSIGNINT%s", f ? "|" : "");
391 }
392 if (f & W_COMPASSIGN)
393 {
394 f &= ~W_COMPASSIGN;
395 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
396 }
397 if (f & W_NOEXPAND)
398 {
399 f &= ~W_NOEXPAND;
400 fprintf (stderr, "W_NOEXPAND%s", f ? "|" : "");
401 }
402 if (f & W_ITILDE)
403 {
404 f &= ~W_ITILDE;
405 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
406 }
407 if (f & W_NOTILDE)
408 {
409 f &= ~W_NOTILDE;
410 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
411 }
412 if (f & W_ASSIGNRHS)
413 {
414 f &= ~W_ASSIGNRHS;
415 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
416 }
417 if (f & W_NOCOMSUB)
418 {
419 f &= ~W_NOCOMSUB;
420 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
421 }
422 if (f & W_DOLLARSTAR)
423 {
424 f &= ~W_DOLLARSTAR;
425 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
426 }
427 if (f & W_DOLLARAT)
428 {
429 f &= ~W_DOLLARAT;
430 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
431 }
432 if (f & W_TILDEEXP)
433 {
434 f &= ~W_TILDEEXP;
435 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
436 }
437 if (f & W_NOSPLIT2)
438 {
439 f &= ~W_NOSPLIT2;
440 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
441 }
442 if (f & W_NOSPLIT)
443 {
444 f &= ~W_NOSPLIT;
445 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
446 }
447 if (f & W_NOBRACE)
448 {
449 f &= ~W_NOBRACE;
450 fprintf (stderr, "W_NOBRACE%s", f ? "|" : "");
451 }
452 if (f & W_NOGLOB)
453 {
454 f &= ~W_NOGLOB;
455 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
456 }
457 if (f & W_SPLITSPACE)
458 {
459 f &= ~W_SPLITSPACE;
460 fprintf (stderr, "W_SPLITSPACE%s", f ? "|" : "");
461 }
462 if (f & W_ASSIGNMENT)
463 {
464 f &= ~W_ASSIGNMENT;
465 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
466 }
467 if (f & W_QUOTED)
468 {
469 f &= ~W_QUOTED;
470 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
471 }
472 if (f & W_HASDOLLAR)
473 {
474 f &= ~W_HASDOLLAR;
475 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
476 }
477 fprintf (stderr, "\n");
478 fflush (stderr);
479 }
480 #endif
481
482 #ifdef INCLUDE_UNUSED
483 static char *
484 quoted_substring (string, start, end)
485 char *string;
486 int start, end;
487 {
488 register int len, l;
489 register char *result, *s, *r;
490
491 len = end - start;
492
493 /* Move to string[start], skipping quoted characters. */
494 for (s = string, l = 0; *s && l < start; )
495 {
496 if (*s == CTLESC)
497 {
498 s++;
499 continue;
500 }
501 l++;
502 if (*s == 0)
503 break;
504 }
505
506 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
507
508 /* Copy LEN characters, including quote characters. */
509 s = string + l;
510 for (l = 0; l < len; s++)
511 {
512 if (*s == CTLESC)
513 *r++ = *s++;
514 *r++ = *s;
515 l++;
516 if (*s == 0)
517 break;
518 }
519 *r = '\0';
520 return result;
521 }
522 #endif
523
524 #ifdef INCLUDE_UNUSED
525 /* Return the length of S, skipping over quoted characters */
526 static int
527 quoted_strlen (s)
528 char *s;
529 {
530 register char *p;
531 int i;
532
533 i = 0;
534 for (p = s; *p; p++)
535 {
536 if (*p == CTLESC)
537 {
538 p++;
539 if (*p == 0)
540 return (i + 1);
541 }
542 i++;
543 }
544
545 return i;
546 }
547 #endif
548
549 /* Find the first occurrence of character C in string S, obeying shell
550 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
551 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
552 escaped with CTLESC are skipped. */
553 static char *
554 quoted_strchr (s, c, flags)
555 char *s;
556 int c, flags;
557 {
558 register char *p;
559
560 for (p = s; *p; p++)
561 {
562 if (((flags & ST_BACKSL) && *p == '\\')
563 || ((flags & ST_CTLESC) && *p == CTLESC))
564 {
565 p++;
566 if (*p == '\0')
567 return ((char *)NULL);
568 continue;
569 }
570 else if (*p == c)
571 return p;
572 }
573 return ((char *)NULL);
574 }
575
576 /* Return 1 if CHARACTER appears in an unquoted portion of
577 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
578 static int
579 unquoted_member (character, string)
580 int character;
581 char *string;
582 {
583 size_t slen;
584 int sindex, c;
585 DECLARE_MBSTATE;
586
587 slen = strlen (string);
588 sindex = 0;
589 while (c = string[sindex])
590 {
591 if (c == character)
592 return (1);
593
594 switch (c)
595 {
596 default:
597 ADVANCE_CHAR (string, slen, sindex);
598 break;
599
600 case '\\':
601 sindex++;
602 if (string[sindex])
603 ADVANCE_CHAR (string, slen, sindex);
604 break;
605
606 case '\'':
607 sindex = skip_single_quoted (string, slen, ++sindex);
608 break;
609
610 case '"':
611 sindex = skip_double_quoted (string, slen, ++sindex);
612 break;
613 }
614 }
615 return (0);
616 }
617
618 /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
619 static int
620 unquoted_substring (substr, string)
621 char *substr, *string;
622 {
623 size_t slen;
624 int sindex, c, sublen;
625 DECLARE_MBSTATE;
626
627 if (substr == 0 || *substr == '\0')
628 return (0);
629
630 slen = strlen (string);
631 sublen = strlen (substr);
632 for (sindex = 0; c = string[sindex]; )
633 {
634 if (STREQN (string + sindex, substr, sublen))
635 return (1);
636
637 switch (c)
638 {
639 case '\\':
640 sindex++;
641 if (string[sindex])
642 ADVANCE_CHAR (string, slen, sindex);
643 break;
644
645 case '\'':
646 sindex = skip_single_quoted (string, slen, ++sindex);
647 break;
648
649 case '"':
650 sindex = skip_double_quoted (string, slen, ++sindex);
651 break;
652
653 default:
654 ADVANCE_CHAR (string, slen, sindex);
655 break;
656 }
657 }
658 return (0);
659 }
660
661 /* Most of the substitutions must be done in parallel. In order
662 to avoid using tons of unclear goto's, I have some functions
663 for manipulating malloc'ed strings. They all take INDX, a
664 pointer to an integer which is the offset into the string
665 where manipulation is taking place. They also take SIZE, a
666 pointer to an integer which is the current length of the
667 character array for this string. */
668
669 /* Append SOURCE to TARGET at INDEX. SIZE is the current amount
670 of space allocated to TARGET. SOURCE can be NULL, in which
671 case nothing happens. Gets rid of SOURCE by freeing it.
672 Returns TARGET in case the location has changed. */
673 INLINE char *
674 sub_append_string (source, target, indx, size)
675 char *source, *target;
676 int *indx, *size;
677 {
678 if (source)
679 {
680 int srclen, n;
681
682 srclen = STRLEN (source);
683 if (srclen >= (int)(*size - *indx))
684 {
685 n = srclen + *indx;
686 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
687 target = (char *)xrealloc (target, (*size = n));
688 }
689
690 FASTCOPY (source, target + *indx, srclen);
691 *indx += srclen;
692 target[*indx] = '\0';
693
694 free (source);
695 }
696 return (target);
697 }
698
699 #if 0
700 /* UNUSED */
701 /* Append the textual representation of NUMBER to TARGET.
702 INDX and SIZE are as in SUB_APPEND_STRING. */
703 char *
704 sub_append_number (number, target, indx, size)
705 intmax_t number;
706 int *indx, *size;
707 char *target;
708 {
709 char *temp;
710
711 temp = itos (number);
712 return (sub_append_string (temp, target, indx, size));
713 }
714 #endif
715
716 /* Extract a substring from STRING, starting at SINDEX and ending with
717 one of the characters in CHARLIST. Don't make the ending character
718 part of the string. Leave SINDEX pointing at the ending character.
719 Understand about backslashes in the string. If (flags & SX_VARNAME)
720 is non-zero, and array variables have been compiled into the shell,
721 everything between a `[' and a corresponding `]' is skipped over.
722 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
723 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
724 contain a closing character from CHARLIST. */
725 static char *
726 string_extract (string, sindex, charlist, flags)
727 char *string;
728 int *sindex;
729 char *charlist;
730 int flags;
731 {
732 register int c, i;
733 int found;
734 size_t slen;
735 char *temp;
736 DECLARE_MBSTATE;
737
738 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
739 i = *sindex;
740 found = 0;
741 while (c = string[i])
742 {
743 if (c == '\\')
744 {
745 if (string[i + 1])
746 i++;
747 else
748 break;
749 }
750 #if defined (ARRAY_VARS)
751 else if ((flags & SX_VARNAME) && c == '[')
752 {
753 int ni;
754 /* If this is an array subscript, skip over it and continue. */
755 ni = skipsubscript (string, i, 0);
756 if (string[ni] == ']')
757 i = ni;
758 }
759 #endif
760 else if (MEMBER (c, charlist))
761 {
762 found = 1;
763 break;
764 }
765
766 ADVANCE_CHAR (string, slen, i);
767 }
768
769 /* If we had to have a matching delimiter and didn't find one, return an
770 error and let the caller deal with it. */
771 if ((flags & SX_REQMATCH) && found == 0)
772 {
773 *sindex = i;
774 return (&extract_string_error);
775 }
776
777 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
778 *sindex = i;
779
780 return (temp);
781 }
782
783 /* Extract the contents of STRING as if it is enclosed in double quotes.
784 SINDEX, when passed in, is the offset of the character immediately
785 following the opening double quote; on exit, SINDEX is left pointing after
786 the closing double quote. If STRIPDQ is non-zero, unquoted double
787 quotes are stripped and the string is terminated by a null byte.
788 Backslashes between the embedded double quotes are processed. If STRIPDQ
789 is zero, an unquoted `"' terminates the string. */
790 static char *
791 string_extract_double_quoted (string, sindex, stripdq)
792 char *string;
793 int *sindex, stripdq;
794 {
795 size_t slen;
796 char *send;
797 int j, i, t;
798 unsigned char c;
799 char *temp, *ret; /* The new string we return. */
800 int pass_next, backquote, si; /* State variables for the machine. */
801 int dquote;
802 DECLARE_MBSTATE;
803
804 slen = strlen (string + *sindex) + *sindex;
805 send = string + slen;
806
807 pass_next = backquote = dquote = 0;
808 temp = (char *)xmalloc (1 + slen - *sindex);
809
810 j = 0;
811 i = *sindex;
812 while (c = string[i])
813 {
814 /* Process a character that was quoted by a backslash. */
815 if (pass_next)
816 {
817 /* XXX - take another look at this in light of Interp 221 */
818 /* Posix.2 sez:
819
820 ``The backslash shall retain its special meaning as an escape
821 character only when followed by one of the characters:
822 $ ` " \ <newline>''.
823
824 If STRIPDQ is zero, we handle the double quotes here and let
825 expand_word_internal handle the rest. If STRIPDQ is non-zero,
826 we have already been through one round of backslash stripping,
827 and want to strip these backslashes only if DQUOTE is non-zero,
828 indicating that we are inside an embedded double-quoted string. */
829
830 /* If we are in an embedded quoted string, then don't strip
831 backslashes before characters for which the backslash
832 retains its special meaning, but remove backslashes in
833 front of other characters. If we are not in an
834 embedded quoted string, don't strip backslashes at all.
835 This mess is necessary because the string was already
836 surrounded by double quotes (and sh has some really weird
837 quoting rules).
838 The returned string will be run through expansion as if
839 it were double-quoted. */
840 if ((stripdq == 0 && c != '"') ||
841 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
842 temp[j++] = '\\';
843 pass_next = 0;
844
845 add_one_character:
846 COPY_CHAR_I (temp, j, string, send, i);
847 continue;
848 }
849
850 /* A backslash protects the next character. The code just above
851 handles preserving the backslash in front of any character but
852 a double quote. */
853 if (c == '\\')
854 {
855 pass_next++;
856 i++;
857 continue;
858 }
859
860 /* Inside backquotes, ``the portion of the quoted string from the
861 initial backquote and the characters up to the next backquote
862 that is not preceded by a backslash, having escape characters
863 removed, defines that command''. */
864 if (backquote)
865 {
866 if (c == '`')
867 backquote = 0;
868 temp[j++] = c;
869 i++;
870 continue;
871 }
872
873 if (c == '`')
874 {
875 temp[j++] = c;
876 backquote++;
877 i++;
878 continue;
879 }
880
881 /* Pass everything between `$(' and the matching `)' or a quoted
882 ${ ... } pair through according to the Posix.2 specification. */
883 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
884 {
885 int free_ret = 1;
886
887 si = i + 2;
888 if (string[i + 1] == LPAREN)
889 ret = extract_command_subst (string, &si, 0);
890 else
891 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
892
893 temp[j++] = '$';
894 temp[j++] = string[i + 1];
895
896 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
897 is set. */
898 if (ret == 0 && no_longjmp_on_fatal_error)
899 {
900 free_ret = 0;
901 ret = string + i + 2;
902 }
903
904 for (t = 0; ret[t]; t++, j++)
905 temp[j] = ret[t];
906 temp[j] = string[si];
907
908 if (string[si])
909 {
910 j++;
911 i = si + 1;
912 }
913 else
914 i = si;
915
916 if (free_ret)
917 free (ret);
918 continue;
919 }
920
921 /* Add any character but a double quote to the quoted string we're
922 accumulating. */
923 if (c != '"')
924 goto add_one_character;
925
926 /* c == '"' */
927 if (stripdq)
928 {
929 dquote ^= 1;
930 i++;
931 continue;
932 }
933
934 break;
935 }
936 temp[j] = '\0';
937
938 /* Point to after the closing quote. */
939 if (c)
940 i++;
941 *sindex = i;
942
943 return (temp);
944 }
945
946 /* This should really be another option to string_extract_double_quoted. */
947 static int
948 skip_double_quoted (string, slen, sind)
949 char *string;
950 size_t slen;
951 int sind;
952 {
953 int c, i;
954 char *ret;
955 int pass_next, backquote, si;
956 DECLARE_MBSTATE;
957
958 pass_next = backquote = 0;
959 i = sind;
960 while (c = string[i])
961 {
962 if (pass_next)
963 {
964 pass_next = 0;
965 ADVANCE_CHAR (string, slen, i);
966 continue;
967 }
968 else if (c == '\\')
969 {
970 pass_next++;
971 i++;
972 continue;
973 }
974 else if (backquote)
975 {
976 if (c == '`')
977 backquote = 0;
978 ADVANCE_CHAR (string, slen, i);
979 continue;
980 }
981 else if (c == '`')
982 {
983 backquote++;
984 i++;
985 continue;
986 }
987 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
988 {
989 si = i + 2;
990 if (string[i + 1] == LPAREN)
991 ret = extract_command_subst (string, &si, SX_NOALLOC);
992 else
993 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
994
995 i = si + 1;
996 continue;
997 }
998 else if (c != '"')
999 {
1000 ADVANCE_CHAR (string, slen, i);
1001 continue;
1002 }
1003 else
1004 break;
1005 }
1006
1007 if (c)
1008 i++;
1009
1010 return (i);
1011 }
1012
1013 /* Extract the contents of STRING as if it is enclosed in single quotes.
1014 SINDEX, when passed in, is the offset of the character immediately
1015 following the opening single quote; on exit, SINDEX is left pointing after
1016 the closing single quote. */
1017 static inline char *
1018 string_extract_single_quoted (string, sindex)
1019 char *string;
1020 int *sindex;
1021 {
1022 register int i;
1023 size_t slen;
1024 char *t;
1025 DECLARE_MBSTATE;
1026
1027 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1028 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
1029 i = *sindex;
1030 while (string[i] && string[i] != '\'')
1031 ADVANCE_CHAR (string, slen, i);
1032
1033 t = substring (string, *sindex, i);
1034
1035 if (string[i])
1036 i++;
1037 *sindex = i;
1038
1039 return (t);
1040 }
1041
1042 static inline int
1043 skip_single_quoted (string, slen, sind)
1044 const char *string;
1045 size_t slen;
1046 int sind;
1047 {
1048 register int c;
1049 DECLARE_MBSTATE;
1050
1051 c = sind;
1052 while (string[c] && string[c] != '\'')
1053 ADVANCE_CHAR (string, slen, c);
1054
1055 if (string[c])
1056 c++;
1057 return c;
1058 }
1059
1060 /* Just like string_extract, but doesn't hack backslashes or any of
1061 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
1062 static char *
1063 string_extract_verbatim (string, slen, sindex, charlist, flags)
1064 char *string;
1065 size_t slen;
1066 int *sindex;
1067 char *charlist;
1068 int flags;
1069 {
1070 register int i;
1071 #if defined (HANDLE_MULTIBYTE)
1072 size_t clen;
1073 wchar_t *wcharlist;
1074 #endif
1075 int c;
1076 char *temp;
1077 DECLARE_MBSTATE;
1078
1079 if (charlist[0] == '\'' && charlist[1] == '\0')
1080 {
1081 temp = string_extract_single_quoted (string, sindex);
1082 --*sindex; /* leave *sindex at separator character */
1083 return temp;
1084 }
1085
1086 i = *sindex;
1087 #if 0
1088 /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need
1089 this only if MB_CUR_MAX > 1. */
1090 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1;
1091 #endif
1092 #if defined (HANDLE_MULTIBYTE)
1093 clen = strlen (charlist);
1094 wcharlist = 0;
1095 #endif
1096 while (c = string[i])
1097 {
1098 #if defined (HANDLE_MULTIBYTE)
1099 size_t mblength;
1100 #endif
1101 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1102 {
1103 i += 2;
1104 continue;
1105 }
1106 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1107 through, to protect the CTLNULs from later calls to
1108 remove_quoted_nulls. */
1109 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
1110 {
1111 i += 2;
1112 continue;
1113 }
1114
1115 #if defined (HANDLE_MULTIBYTE)
1116 mblength = MBLEN (string + i, slen - i);
1117 if (mblength > 1)
1118 {
1119 wchar_t wc;
1120 mblength = mbtowc (&wc, string + i, slen - i);
1121 if (MB_INVALIDCH (mblength))
1122 {
1123 if (MEMBER (c, charlist))
1124 break;
1125 }
1126 else
1127 {
1128 if (wcharlist == 0)
1129 {
1130 size_t len;
1131 len = mbstowcs (wcharlist, charlist, 0);
1132 if (len == -1)
1133 len = 0;
1134 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1135 mbstowcs (wcharlist, charlist, len + 1);
1136 }
1137
1138 if (wcschr (wcharlist, wc))
1139 break;
1140 }
1141 }
1142 else
1143 #endif
1144 if (MEMBER (c, charlist))
1145 break;
1146
1147 ADVANCE_CHAR (string, slen, i);
1148 }
1149
1150 #if defined (HANDLE_MULTIBYTE)
1151 FREE (wcharlist);
1152 #endif
1153
1154 temp = substring (string, *sindex, i);
1155 *sindex = i;
1156
1157 return (temp);
1158 }
1159
1160 /* Extract the $( construct in STRING, and return a new string.
1161 Start extracting at (SINDEX) as if we had just seen "$(".
1162 Make (SINDEX) get the position of the matching ")". )
1163 XFLAGS is additional flags to pass to other extraction functions. */
1164 char *
1165 extract_command_subst (string, sindex, xflags)
1166 char *string;
1167 int *sindex;
1168 int xflags;
1169 {
1170 if (string[*sindex] == LPAREN)
1171 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1172 else
1173 {
1174 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1175 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1176 }
1177 }
1178
1179 /* Extract the $[ construct in STRING, and return a new string. (])
1180 Start extracting at (SINDEX) as if we had just seen "$[".
1181 Make (SINDEX) get the position of the matching "]". */
1182 char *
1183 extract_arithmetic_subst (string, sindex)
1184 char *string;
1185 int *sindex;
1186 {
1187 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
1188 }
1189
1190 #if defined (PROCESS_SUBSTITUTION)
1191 /* Extract the <( or >( construct in STRING, and return a new string.
1192 Start extracting at (SINDEX) as if we had just seen "<(".
1193 Make (SINDEX) get the position of the matching ")". */ /*))*/
1194 char *
1195 extract_process_subst (string, starter, sindex)
1196 char *string;
1197 char *starter;
1198 int *sindex;
1199 {
1200 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
1201 }
1202 #endif /* PROCESS_SUBSTITUTION */
1203
1204 #if defined (ARRAY_VARS)
1205 /* This can be fooled by unquoted right parens in the passed string. If
1206 each caller verifies that the last character in STRING is a right paren,
1207 we don't even need to call extract_delimited_string. */
1208 char *
1209 extract_array_assignment_list (string, sindex)
1210 char *string;
1211 int *sindex;
1212 {
1213 int slen;
1214 char *ret;
1215
1216 slen = strlen (string); /* ( */
1217 if (string[slen - 1] == ')')
1218 {
1219 ret = substring (string, *sindex, slen - 1);
1220 *sindex = slen - 1;
1221 return ret;
1222 }
1223 return 0;
1224 }
1225 #endif
1226
1227 /* Extract and create a new string from the contents of STRING, a
1228 character string delimited with OPENER and CLOSER. SINDEX is
1229 the address of an int describing the current offset in STRING;
1230 it should point to just after the first OPENER found. On exit,
1231 SINDEX gets the position of the last character of the matching CLOSER.
1232 If OPENER is more than a single character, ALT_OPENER, if non-null,
1233 contains a character string that can also match CLOSER and thus
1234 needs to be skipped. */
1235 static char *
1236 extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
1237 char *string;
1238 int *sindex;
1239 char *opener, *alt_opener, *closer;
1240 int flags;
1241 {
1242 int i, c, si;
1243 size_t slen;
1244 char *t, *result;
1245 int pass_character, nesting_level, in_comment;
1246 int len_closer, len_opener, len_alt_opener;
1247 DECLARE_MBSTATE;
1248
1249 slen = strlen (string + *sindex) + *sindex;
1250 len_opener = STRLEN (opener);
1251 len_alt_opener = STRLEN (alt_opener);
1252 len_closer = STRLEN (closer);
1253
1254 pass_character = in_comment = 0;
1255
1256 nesting_level = 1;
1257 i = *sindex;
1258
1259 while (nesting_level)
1260 {
1261 c = string[i];
1262
1263 if (c == 0)
1264 break;
1265
1266 if (in_comment)
1267 {
1268 if (c == '\n')
1269 in_comment = 0;
1270 ADVANCE_CHAR (string, slen, i);
1271 continue;
1272 }
1273
1274 if (pass_character) /* previous char was backslash */
1275 {
1276 pass_character = 0;
1277 ADVANCE_CHAR (string, slen, i);
1278 continue;
1279 }
1280
1281 /* Not exactly right yet; should handle shell metacharacters and
1282 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
1283 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
1284 {
1285 in_comment = 1;
1286 ADVANCE_CHAR (string, slen, i);
1287 continue;
1288 }
1289
1290 if (c == CTLESC || c == '\\')
1291 {
1292 pass_character++;
1293 i++;
1294 continue;
1295 }
1296
1297 /* Process a nested command substitution, but only if we're parsing an
1298 arithmetic substitution. */
1299 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1300 {
1301 si = i + 2;
1302 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1303 i = si + 1;
1304 continue;
1305 }
1306
1307 /* Process a nested OPENER. */
1308 if (STREQN (string + i, opener, len_opener))
1309 {
1310 si = i + len_opener;
1311 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
1312 i = si + 1;
1313 continue;
1314 }
1315
1316 /* Process a nested ALT_OPENER */
1317 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
1318 {
1319 si = i + len_alt_opener;
1320 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
1321 i = si + 1;
1322 continue;
1323 }
1324
1325 /* If the current substring terminates the delimited string, decrement
1326 the nesting level. */
1327 if (STREQN (string + i, closer, len_closer))
1328 {
1329 i += len_closer - 1; /* move to last byte of the closer */
1330 nesting_level--;
1331 if (nesting_level == 0)
1332 break;
1333 }
1334
1335 /* Pass old-style command substitution through verbatim. */
1336 if (c == '`')
1337 {
1338 si = i + 1;
1339 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1340 i = si + 1;
1341 continue;
1342 }
1343
1344 /* Pass single-quoted and double-quoted strings through verbatim. */
1345 if (c == '\'' || c == '"')
1346 {
1347 si = i + 1;
1348 i = (c == '\'') ? skip_single_quoted (string, slen, si)
1349 : skip_double_quoted (string, slen, si);
1350 continue;
1351 }
1352
1353 /* move past this character, which was not special. */
1354 ADVANCE_CHAR (string, slen, i);
1355 }
1356
1357 if (c == 0 && nesting_level)
1358 {
1359 if (no_longjmp_on_fatal_error == 0)
1360 {
1361 last_command_exit_value = EXECUTION_FAILURE;
1362 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
1363 exp_jump_to_top_level (DISCARD);
1364 }
1365 else
1366 {
1367 *sindex = i;
1368 return (char *)NULL;
1369 }
1370 }
1371
1372 si = i - *sindex - len_closer + 1;
1373 if (flags & SX_NOALLOC)
1374 result = (char *)NULL;
1375 else
1376 {
1377 result = (char *)xmalloc (1 + si);
1378 strncpy (result, string + *sindex, si);
1379 result[si] = '\0';
1380 }
1381 *sindex = i;
1382
1383 return (result);
1384 }
1385
1386 /* Extract a parameter expansion expression within ${ and } from STRING.
1387 Obey the Posix.2 rules for finding the ending `}': count braces while
1388 skipping over enclosed quoted strings and command substitutions.
1389 SINDEX is the address of an int describing the current offset in STRING;
1390 it should point to just after the first `{' found. On exit, SINDEX
1391 gets the position of the matching `}'. QUOTED is non-zero if this
1392 occurs inside double quotes. */
1393 /* XXX -- this is very similar to extract_delimited_string -- XXX */
1394 static char *
1395 extract_dollar_brace_string (string, sindex, quoted, flags)
1396 char *string;
1397 int *sindex, quoted, flags;
1398 {
1399 register int i, c;
1400 size_t slen;
1401 int pass_character, nesting_level, si, dolbrace_state;
1402 char *result, *t;
1403 DECLARE_MBSTATE;
1404
1405 pass_character = 0;
1406 nesting_level = 1;
1407 slen = strlen (string + *sindex) + *sindex;
1408
1409 /* The handling of dolbrace_state needs to agree with the code in parse.y:
1410 parse_matched_pair(). The different initial value is to handle the
1411 case where this function is called to parse the word in
1412 ${param op word} (SX_WORD). */
1413 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1414 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1415 dolbrace_state = DOLBRACE_QUOTE;
1416
1417 i = *sindex;
1418 while (c = string[i])
1419 {
1420 if (pass_character)
1421 {
1422 pass_character = 0;
1423 ADVANCE_CHAR (string, slen, i);
1424 continue;
1425 }
1426
1427 /* CTLESCs and backslashes quote the next character. */
1428 if (c == CTLESC || c == '\\')
1429 {
1430 pass_character++;
1431 i++;
1432 continue;
1433 }
1434
1435 if (string[i] == '$' && string[i+1] == LBRACE)
1436 {
1437 nesting_level++;
1438 i += 2;
1439 continue;
1440 }
1441
1442 if (c == RBRACE)
1443 {
1444 nesting_level--;
1445 if (nesting_level == 0)
1446 break;
1447 i++;
1448 continue;
1449 }
1450
1451 /* Pass the contents of old-style command substitutions through
1452 verbatim. */
1453 if (c == '`')
1454 {
1455 si = i + 1;
1456 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1457 i = si + 1;
1458 continue;
1459 }
1460
1461 /* Pass the contents of new-style command substitutions and
1462 arithmetic substitutions through verbatim. */
1463 if (string[i] == '$' && string[i+1] == LPAREN)
1464 {
1465 si = i + 2;
1466 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1467 i = si + 1;
1468 continue;
1469 }
1470
1471 /* Pass the contents of double-quoted strings through verbatim. */
1472 if (c == '"')
1473 {
1474 si = i + 1;
1475 i = skip_double_quoted (string, slen, si);
1476 /* skip_XXX_quoted leaves index one past close quote */
1477 continue;
1478 }
1479
1480 if (c == '\'')
1481 {
1482 /*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
1483 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
1484 ADVANCE_CHAR (string, slen, i);
1485 else
1486 {
1487 si = i + 1;
1488 i = skip_single_quoted (string, slen, si);
1489 }
1490
1491 continue;
1492 }
1493
1494 /* move past this character, which was not special. */
1495 ADVANCE_CHAR (string, slen, i);
1496
1497 /* This logic must agree with parse.y:parse_matched_pair, since they
1498 share the same defines. */
1499 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1500 dolbrace_state = DOLBRACE_QUOTE;
1501 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1502 dolbrace_state = DOLBRACE_QUOTE;
1503 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
1504 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
1505 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1506 dolbrace_state = DOLBRACE_QUOTE;
1507 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1508 dolbrace_state = DOLBRACE_QUOTE;
1509 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1510 dolbrace_state = DOLBRACE_OP;
1511 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1512 dolbrace_state = DOLBRACE_WORD;
1513 }
1514
1515 if (c == 0 && nesting_level)
1516 {
1517 if (no_longjmp_on_fatal_error == 0)
1518 { /* { */
1519 last_command_exit_value = EXECUTION_FAILURE;
1520 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
1521 exp_jump_to_top_level (DISCARD);
1522 }
1523 else
1524 {
1525 *sindex = i;
1526 return ((char *)NULL);
1527 }
1528 }
1529
1530 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
1531 *sindex = i;
1532
1533 return (result);
1534 }
1535
1536 /* Remove backslashes which are quoting backquotes from STRING. Modifies
1537 STRING, and returns a pointer to it. */
1538 char *
1539 de_backslash (string)
1540 char *string;
1541 {
1542 register size_t slen;
1543 register int i, j, prev_i;
1544 DECLARE_MBSTATE;
1545
1546 slen = strlen (string);
1547 i = j = 0;
1548
1549 /* Loop copying string[i] to string[j], i >= j. */
1550 while (i < slen)
1551 {
1552 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
1553 string[i + 1] == '$'))
1554 i++;
1555 prev_i = i;
1556 ADVANCE_CHAR (string, slen, i);
1557 if (j < prev_i)
1558 do string[j++] = string[prev_i++]; while (prev_i < i);
1559 else
1560 j = i;
1561 }
1562 string[j] = '\0';
1563
1564 return (string);
1565 }
1566
1567 #if 0
1568 /*UNUSED*/
1569 /* Replace instances of \! in a string with !. */
1570 void
1571 unquote_bang (string)
1572 char *string;
1573 {
1574 register int i, j;
1575 register char *temp;
1576
1577 temp = (char *)xmalloc (1 + strlen (string));
1578
1579 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1580 {
1581 if (string[i] == '\\' && string[i + 1] == '!')
1582 {
1583 temp[j] = '!';
1584 i++;
1585 }
1586 }
1587 strcpy (string, temp);
1588 free (temp);
1589 }
1590 #endif
1591
1592 #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0)
1593
1594 /* This function assumes s[i] == open; returns with s[ret] == close; used to
1595 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1596 matched pairs of quotes or backquotes, or skip word expansions; it is
1597 intended to be used after expansion has been performed and during final
1598 assignment parsing (see arrayfunc.c:assign_compound_array_list()). */
1599 static int
1600 skip_matched_pair (string, start, open, close, flags)
1601 const char *string;
1602 int start, open, close, flags;
1603 {
1604 int i, pass_next, backq, si, c, count;
1605 size_t slen;
1606 char *temp, *ss;
1607 DECLARE_MBSTATE;
1608
1609 slen = strlen (string + start) + start;
1610 no_longjmp_on_fatal_error = 1;
1611
1612 i = start + 1; /* skip over leading bracket */
1613 count = 1;
1614 pass_next = backq = 0;
1615 ss = (char *)string;
1616 while (c = string[i])
1617 {
1618 if (pass_next)
1619 {
1620 pass_next = 0;
1621 if (c == 0)
1622 CQ_RETURN(i);
1623 ADVANCE_CHAR (string, slen, i);
1624 continue;
1625 }
1626 else if (c == '\\')
1627 {
1628 pass_next = 1;
1629 i++;
1630 continue;
1631 }
1632 else if (backq)
1633 {
1634 if (c == '`')
1635 backq = 0;
1636 ADVANCE_CHAR (string, slen, i);
1637 continue;
1638 }
1639 else if ((flags & 1) == 0 && c == '`')
1640 {
1641 backq = 1;
1642 i++;
1643 continue;
1644 }
1645 else if ((flags & 1) == 0 && c == open)
1646 {
1647 count++;
1648 i++;
1649 continue;
1650 }
1651 else if (c == close)
1652 {
1653 count--;
1654 if (count == 0)
1655 break;
1656 i++;
1657 continue;
1658 }
1659 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
1660 {
1661 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i)
1662 : skip_double_quoted (ss, slen, ++i);
1663 /* no increment, the skip functions increment past the closing quote. */
1664 }
1665 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
1666 {
1667 si = i + 2;
1668 if (string[si] == '\0')
1669 CQ_RETURN(si);
1670
1671 if (string[i+1] == LPAREN)
1672 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1673 else
1674 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
1675 i = si;
1676 if (string[i] == '\0') /* don't increment i past EOS in loop */
1677 break;
1678 i++;
1679 continue;
1680 }
1681 else
1682 ADVANCE_CHAR (string, slen, i);
1683 }
1684
1685 CQ_RETURN(i);
1686 }
1687
1688 #if defined (ARRAY_VARS)
1689 int
1690 skipsubscript (string, start, flags)
1691 const char *string;
1692 int start, flags;
1693 {
1694 return (skip_matched_pair (string, start, '[', ']', flags));
1695 }
1696 #endif
1697
1698 /* Skip characters in STRING until we find a character in DELIMS, and return
1699 the index of that character. START is the index into string at which we
1700 begin. This is similar in spirit to strpbrk, but it returns an index into
1701 STRING and takes a starting index. This little piece of code knows quite
1702 a lot of shell syntax. It's very similar to skip_double_quoted and other
1703 functions of that ilk. */
1704 int
1705 skip_to_delim (string, start, delims, flags)
1706 char *string;
1707 int start;
1708 char *delims;
1709 int flags;
1710 {
1711 int i, pass_next, backq, si, c, invert, skipquote, skipcmd;
1712 size_t slen;
1713 char *temp, open[3];
1714 DECLARE_MBSTATE;
1715
1716 slen = strlen (string + start) + start;
1717 if (flags & SD_NOJMP)
1718 no_longjmp_on_fatal_error = 1;
1719 invert = (flags & SD_INVERT);
1720 skipcmd = (flags & SD_NOSKIPCMD) == 0;
1721
1722 i = start;
1723 pass_next = backq = 0;
1724 while (c = string[i])
1725 {
1726 /* If this is non-zero, we should not let quote characters be delimiters
1727 and the current character is a single or double quote. We should not
1728 test whether or not it's a delimiter until after we skip single- or
1729 double-quoted strings. */
1730 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
1731 if (pass_next)
1732 {
1733 pass_next = 0;
1734 if (c == 0)
1735 CQ_RETURN(i);
1736 ADVANCE_CHAR (string, slen, i);
1737 continue;
1738 }
1739 else if (c == '\\')
1740 {
1741 pass_next = 1;
1742 i++;
1743 continue;
1744 }
1745 else if (backq)
1746 {
1747 if (c == '`')
1748 backq = 0;
1749 ADVANCE_CHAR (string, slen, i);
1750 continue;
1751 }
1752 else if (c == '`')
1753 {
1754 backq = 1;
1755 i++;
1756 continue;
1757 }
1758 else if (skipquote == 0 && invert == 0 && member (c, delims))
1759 break;
1760 else if (c == '\'' || c == '"')
1761 {
1762 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1763 : skip_double_quoted (string, slen, ++i);
1764 /* no increment, the skip functions increment past the closing quote. */
1765 }
1766 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
1767 {
1768 si = i + 2;
1769 if (string[si] == '\0')
1770 CQ_RETURN(si);
1771
1772 if (string[i+1] == LPAREN)
1773 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1774 else
1775 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1776 i = si;
1777 if (string[i] == '\0') /* don't increment i past EOS in loop */
1778 break;
1779 i++;
1780 continue;
1781 }
1782 #if defined (PROCESS_SUBSTITUTION)
1783 else if (skipcmd && (c == '<' || c == '>') && string[i+1] == LPAREN)
1784 {
1785 si = i + 2;
1786 if (string[si] == '\0')
1787 CQ_RETURN(si);
1788 temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si);
1789 free (temp); /* no SX_ALLOC here */
1790 i = si;
1791 if (string[i] == '\0')
1792 break;
1793 i++;
1794 continue;
1795 }
1796 #endif /* PROCESS_SUBSTITUTION */
1797 #if defined (EXTENDED_GLOB)
1798 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1799 {
1800 si = i + 2;
1801 if (string[si] == '\0')
1802 CQ_RETURN(si);
1803
1804 open[0] = c;
1805 open[1] = LPAREN;
1806 open[2] = '\0';
1807 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
1808
1809 i = si;
1810 if (string[i] == '\0') /* don't increment i past EOS in loop */
1811 break;
1812 i++;
1813 continue;
1814 }
1815 #endif
1816 else if ((flags & SD_GLOB) && c == LBRACK)
1817 {
1818 si = i + 1;
1819 if (string[si] == '\0')
1820 CQ_RETURN(si);
1821
1822 temp = extract_delimited_string (string, &si, "[", "[", "]", SX_NOALLOC); /* ] */
1823
1824 i = si;
1825 if (string[i] == '\0') /* don't increment i past EOS in loop */
1826 break;
1827 i++;
1828 continue;
1829 }
1830 else if ((skipquote || invert) && (member (c, delims) == 0))
1831 break;
1832 else
1833 ADVANCE_CHAR (string, slen, i);
1834 }
1835
1836 CQ_RETURN(i);
1837 }
1838
1839 #if defined (READLINE)
1840 /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
1841 an unclosed quoted string), or if the character at EINDEX is quoted
1842 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
1843 single and double-quoted string parsing functions should not return an
1844 error if there are unclosed quotes or braces. The characters that this
1845 recognizes need to be the same as the contents of
1846 rl_completer_quote_characters. */
1847
1848 int
1849 char_is_quoted (string, eindex)
1850 char *string;
1851 int eindex;
1852 {
1853 int i, pass_next, c;
1854 size_t slen;
1855 DECLARE_MBSTATE;
1856
1857 slen = strlen (string);
1858 no_longjmp_on_fatal_error = 1;
1859 i = pass_next = 0;
1860 while (i <= eindex)
1861 {
1862 c = string[i];
1863
1864 if (pass_next)
1865 {
1866 pass_next = 0;
1867 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1868 CQ_RETURN(1);
1869 ADVANCE_CHAR (string, slen, i);
1870 continue;
1871 }
1872 else if (c == '\\')
1873 {
1874 pass_next = 1;
1875 i++;
1876 continue;
1877 }
1878 else if (c == '\'' || c == '"')
1879 {
1880 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1881 : skip_double_quoted (string, slen, ++i);
1882 if (i > eindex)
1883 CQ_RETURN(1);
1884 /* no increment, the skip_xxx functions go one past end */
1885 }
1886 else
1887 ADVANCE_CHAR (string, slen, i);
1888 }
1889
1890 CQ_RETURN(0);
1891 }
1892
1893 int
1894 unclosed_pair (string, eindex, openstr)
1895 char *string;
1896 int eindex;
1897 char *openstr;
1898 {
1899 int i, pass_next, openc, olen;
1900 size_t slen;
1901 DECLARE_MBSTATE;
1902
1903 slen = strlen (string);
1904 olen = strlen (openstr);
1905 i = pass_next = openc = 0;
1906 while (i <= eindex)
1907 {
1908 if (pass_next)
1909 {
1910 pass_next = 0;
1911 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1912 return 0;
1913 ADVANCE_CHAR (string, slen, i);
1914 continue;
1915 }
1916 else if (string[i] == '\\')
1917 {
1918 pass_next = 1;
1919 i++;
1920 continue;
1921 }
1922 else if (STREQN (string + i, openstr, olen))
1923 {
1924 openc = 1 - openc;
1925 i += olen;
1926 }
1927 else if (string[i] == '\'' || string[i] == '"')
1928 {
1929 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i)
1930 : skip_double_quoted (string, slen, i);
1931 if (i > eindex)
1932 return 0;
1933 }
1934 else
1935 ADVANCE_CHAR (string, slen, i);
1936 }
1937 return (openc);
1938 }
1939
1940 /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
1941 individual words. If DELIMS is NULL, the current value of $IFS is used
1942 to split the string, and the function follows the shell field splitting
1943 rules. SENTINEL is an index to look for. NWP, if non-NULL,
1944 gets the number of words in the returned list. CWP, if non-NULL, gets
1945 the index of the word containing SENTINEL. Non-whitespace chars in
1946 DELIMS delimit separate fields. */
1947 WORD_LIST *
1948 split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
1949 char *string;
1950 int slen;
1951 char *delims;
1952 int sentinel, flags;
1953 int *nwp, *cwp;
1954 {
1955 int ts, te, i, nw, cw, ifs_split, dflags;
1956 char *token, *d, *d2;
1957 WORD_LIST *ret, *tl;
1958
1959 if (string == 0 || *string == '\0')
1960 {
1961 if (nwp)
1962 *nwp = 0;
1963 if (cwp)
1964 *cwp = 0;
1965 return ((WORD_LIST *)NULL);
1966 }
1967
1968 d = (delims == 0) ? ifs_value : delims;
1969 ifs_split = delims == 0;
1970
1971 /* Make d2 the non-whitespace characters in delims */
1972 d2 = 0;
1973 if (delims)
1974 {
1975 size_t slength;
1976 #if defined (HANDLE_MULTIBYTE)
1977 size_t mblength = 1;
1978 #endif
1979 DECLARE_MBSTATE;
1980
1981 slength = strlen (delims);
1982 d2 = (char *)xmalloc (slength + 1);
1983 i = ts = 0;
1984 while (delims[i])
1985 {
1986 #if defined (HANDLE_MULTIBYTE)
1987 mbstate_t state_bak;
1988 state_bak = state;
1989 mblength = MBRLEN (delims + i, slength, &state);
1990 if (MB_INVALIDCH (mblength))
1991 state = state_bak;
1992 else if (mblength > 1)
1993 {
1994 memcpy (d2 + ts, delims + i, mblength);
1995 ts += mblength;
1996 i += mblength;
1997 slength -= mblength;
1998 continue;
1999 }
2000 #endif
2001 if (whitespace (delims[i]) == 0)
2002 d2[ts++] = delims[i];
2003
2004 i++;
2005 slength--;
2006 }
2007 d2[ts] = '\0';
2008 }
2009
2010 ret = (WORD_LIST *)NULL;
2011
2012 /* Remove sequences of whitespace characters at the start of the string, as
2013 long as those characters are delimiters. */
2014 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
2015 ;
2016 if (string[i] == '\0')
2017 return (ret);
2018
2019 ts = i;
2020 nw = 0;
2021 cw = -1;
2022 dflags = flags|SD_NOJMP;
2023 while (1)
2024 {
2025 te = skip_to_delim (string, ts, d, dflags);
2026
2027 /* If we have a non-whitespace delimiter character, use it to make a
2028 separate field. This is just about what $IFS splitting does and
2029 is closer to the behavior of the shell parser. */
2030 if (ts == te && d2 && member (string[ts], d2))
2031 {
2032 te = ts + 1;
2033 /* If we're using IFS splitting, the non-whitespace delimiter char
2034 and any additional IFS whitespace delimits a field. */
2035 if (ifs_split)
2036 while (member (string[te], d) && spctabnl (string[te]))
2037 te++;
2038 else
2039 while (member (string[te], d2))
2040 te++;
2041 }
2042
2043 token = substring (string, ts, te);
2044
2045 ret = add_string_to_list (token, ret);
2046 free (token);
2047 nw++;
2048
2049 if (sentinel >= ts && sentinel <= te)
2050 cw = nw;
2051
2052 /* If the cursor is at whitespace just before word start, set the
2053 sentinel word to the current word. */
2054 if (cwp && cw == -1 && sentinel == ts-1)
2055 cw = nw;
2056
2057 /* If the cursor is at whitespace between two words, make a new, empty
2058 word, add it before (well, after, since the list is in reverse order)
2059 the word we just added, and set the current word to that one. */
2060 if (cwp && cw == -1 && sentinel < ts)
2061 {
2062 tl = make_word_list (make_word (""), ret->next);
2063 ret->next = tl;
2064 cw = nw;
2065 nw++;
2066 }
2067
2068 if (string[te] == 0)
2069 break;
2070
2071 i = te;
2072 while (member (string[i], d) && (ifs_split || spctabnl(string[i])))
2073 i++;
2074
2075 if (string[i])
2076 ts = i;
2077 else
2078 break;
2079 }
2080
2081 /* Special case for SENTINEL at the end of STRING. If we haven't found
2082 the word containing SENTINEL yet, and the index we're looking for is at
2083 the end of STRING (or past the end of the previously-found token,
2084 possible if the end of the line is composed solely of IFS whitespace)
2085 add an additional null argument and set the current word pointer to that. */
2086 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
2087 {
2088 if (whitespace (string[sentinel - 1]))
2089 {
2090 token = "";
2091 ret = add_string_to_list (token, ret);
2092 nw++;
2093 }
2094 cw = nw;
2095 }
2096
2097 if (nwp)
2098 *nwp = nw;
2099 if (cwp)
2100 *cwp = cw;
2101
2102 FREE (d2);
2103
2104 return (REVERSE_LIST (ret, WORD_LIST *));
2105 }
2106 #endif /* READLINE */
2107
2108 #if 0
2109 /* UNUSED */
2110 /* Extract the name of the variable to bind to from the assignment string. */
2111 char *
2112 assignment_name (string)
2113 char *string;
2114 {
2115 int offset;
2116 char *temp;
2117
2118 offset = assignment (string, 0);
2119 if (offset == 0)
2120 return (char *)NULL;
2121 temp = substring (string, 0, offset);
2122 return (temp);
2123 }
2124 #endif
2125
2126 /* **************************************************************** */
2127 /* */
2128 /* Functions to convert strings to WORD_LISTs and vice versa */
2129 /* */
2130 /* **************************************************************** */
2131
2132 /* Return a single string of all the words in LIST. SEP is the separator
2133 to put between individual elements of LIST in the output string. */
2134 char *
2135 string_list_internal (list, sep)
2136 WORD_LIST *list;
2137 char *sep;
2138 {
2139 register WORD_LIST *t;
2140 char *result, *r;
2141 int word_len, sep_len, result_size;
2142
2143 if (list == 0)
2144 return ((char *)NULL);
2145
2146 /* Short-circuit quickly if we don't need to separate anything. */
2147 if (list->next == 0)
2148 return (savestring (list->word->word));
2149
2150 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2151 sep_len = STRLEN (sep);
2152 result_size = 0;
2153
2154 for (t = list; t; t = t->next)
2155 {
2156 if (t != list)
2157 result_size += sep_len;
2158 result_size += strlen (t->word->word);
2159 }
2160
2161 r = result = (char *)xmalloc (result_size + 1);
2162
2163 for (t = list; t; t = t->next)
2164 {
2165 if (t != list && sep_len)
2166 {
2167 if (sep_len > 1)
2168 {
2169 FASTCOPY (sep, r, sep_len);
2170 r += sep_len;
2171 }
2172 else
2173 *r++ = sep[0];
2174 }
2175
2176 word_len = strlen (t->word->word);
2177 FASTCOPY (t->word->word, r, word_len);
2178 r += word_len;
2179 }
2180
2181 *r = '\0';
2182 return (result);
2183 }
2184
2185 /* Return a single string of all the words present in LIST, separating
2186 each word with a space. */
2187 char *
2188 string_list (list)
2189 WORD_LIST *list;
2190 {
2191 return (string_list_internal (list, " "));
2192 }
2193
2194 /* An external interface that can be used by the rest of the shell to
2195 obtain a string containing the first character in $IFS. Handles all
2196 the multibyte complications. If LENP is non-null, it is set to the
2197 length of the returned string. */
2198 char *
2199 ifs_firstchar (lenp)
2200 int *lenp;
2201 {
2202 char *ret;
2203 int len;
2204
2205 ret = xmalloc (MB_LEN_MAX + 1);
2206 #if defined (HANDLE_MULTIBYTE)
2207 if (ifs_firstc_len == 1)
2208 {
2209 ret[0] = ifs_firstc[0];
2210 ret[1] = '\0';
2211 len = ret[0] ? 1 : 0;
2212 }
2213 else
2214 {
2215 memcpy (ret, ifs_firstc, ifs_firstc_len);
2216 ret[len = ifs_firstc_len] = '\0';
2217 }
2218 #else
2219 ret[0] = ifs_firstc;
2220 ret[1] = '\0';
2221 len = ret[0] ? 0 : 1;
2222 #endif
2223
2224 if (lenp)
2225 *lenp = len;
2226
2227 return ret;
2228 }
2229
2230 /* Return a single string of all the words present in LIST, obeying the
2231 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2232 expansion [of $*] appears within a double quoted string, it expands
2233 to a single field with the value of each parameter separated by the
2234 first character of the IFS variable, or by a <space> if IFS is unset." */
2235 char *
2236 string_list_dollar_star (list)
2237 WORD_LIST *list;
2238 {
2239 char *ret;
2240 #if defined (HANDLE_MULTIBYTE)
2241 # if defined (__GNUC__)
2242 char sep[MB_CUR_MAX + 1];
2243 # else
2244 char *sep = 0;
2245 # endif
2246 #else
2247 char sep[2];
2248 #endif
2249
2250 #if defined (HANDLE_MULTIBYTE)
2251 # if !defined (__GNUC__)
2252 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2253 # endif /* !__GNUC__ */
2254 if (ifs_firstc_len == 1)
2255 {
2256 sep[0] = ifs_firstc[0];
2257 sep[1] = '\0';
2258 }
2259 else
2260 {
2261 memcpy (sep, ifs_firstc, ifs_firstc_len);
2262 sep[ifs_firstc_len] = '\0';
2263 }
2264 #else
2265 sep[0] = ifs_firstc;
2266 sep[1] = '\0';
2267 #endif
2268
2269 ret = string_list_internal (list, sep);
2270 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2271 free (sep);
2272 #endif
2273 return ret;
2274 }
2275
2276 /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2277 is non-zero, the $@ appears within double quotes, and we should quote
2278 the list before converting it into a string. If IFS is unset, and the
2279 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2280 in the words in the list, because the default value of $IFS is
2281 <space><tab><newline>, IFS characters in the words in the list should
2282 also be split. If IFS is null, and the word is not quoted, we need
2283 to quote the words in the list to preserve the positional parameters
2284 exactly. */
2285 char *
2286 string_list_dollar_at (list, quoted)
2287 WORD_LIST *list;
2288 int quoted;
2289 {
2290 char *ifs, *ret;
2291 #if defined (HANDLE_MULTIBYTE)
2292 # if defined (__GNUC__)
2293 char sep[MB_CUR_MAX + 1];
2294 # else
2295 char *sep = 0;
2296 # endif /* !__GNUC__ */
2297 #else
2298 char sep[2];
2299 #endif
2300 WORD_LIST *tlist;
2301
2302 /* XXX this could just be ifs = ifs_value; */
2303 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
2304
2305 #if defined (HANDLE_MULTIBYTE)
2306 # if !defined (__GNUC__)
2307 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2308 # endif /* !__GNUC__ */
2309 if (ifs && *ifs)
2310 {
2311 if (ifs_firstc_len == 1)
2312 {
2313 sep[0] = ifs_firstc[0];
2314 sep[1] = '\0';
2315 }
2316 else
2317 {
2318 memcpy (sep, ifs_firstc, ifs_firstc_len);
2319 sep[ifs_firstc_len] = '\0';
2320 }
2321 }
2322 else
2323 {
2324 sep[0] = ' ';
2325 sep[1] = '\0';
2326 }
2327 #else
2328 sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs;
2329 sep[1] = '\0';
2330 #endif
2331
2332 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2333 it now that quote_escapes quotes spaces */
2334 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
2335 ? quote_list (list)
2336 : list_quote_escapes (list);
2337
2338 ret = string_list_internal (tlist, sep);
2339 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2340 free (sep);
2341 #endif
2342 return ret;
2343 }
2344
2345 /* Turn the positional parameters into a string, understanding quoting and
2346 the various subtleties of using the first character of $IFS as the
2347 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2348 string_list as appropriate. */
2349 char *
2350 string_list_pos_params (pchar, list, quoted)
2351 int pchar;
2352 WORD_LIST *list;
2353 int quoted;
2354 {
2355 char *ret;
2356 WORD_LIST *tlist;
2357
2358 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2359 {
2360 tlist = quote_list (list);
2361 word_list_remove_quoted_nulls (tlist);
2362 ret = string_list_dollar_star (tlist);
2363 }
2364 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2365 {
2366 tlist = quote_list (list);
2367 word_list_remove_quoted_nulls (tlist);
2368 ret = string_list (tlist);
2369 }
2370 else if (pchar == '*')
2371 {
2372 /* Even when unquoted, string_list_dollar_star does the right thing
2373 making sure that the first character of $IFS is used as the
2374 separator. */
2375 ret = string_list_dollar_star (list);
2376 }
2377 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2378 /* We use string_list_dollar_at, but only if the string is quoted, since
2379 that quotes the escapes if it's not, which we don't want. We could
2380 use string_list (the old code did), but that doesn't do the right
2381 thing if the first character of $IFS is not a space. We use
2382 string_list_dollar_star if the string is unquoted so we make sure that
2383 the elements of $@ are separated by the first character of $IFS for
2384 later splitting. */
2385 ret = string_list_dollar_at (list, quoted);
2386 else if (pchar == '@')
2387 ret = string_list_dollar_star (list);
2388 else
2389 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2390
2391 return ret;
2392 }
2393
2394 /* Return the list of words present in STRING. Separate the string into
2395 words at any of the characters found in SEPARATORS. If QUOTED is
2396 non-zero then word in the list will have its quoted flag set, otherwise
2397 the quoted flag is left as make_word () deemed fit.
2398
2399 This obeys the P1003.2 word splitting semantics. If `separators' is
2400 exactly <space><tab><newline>, then the splitting algorithm is that of
2401 the Bourne shell, which treats any sequence of characters from `separators'
2402 as a delimiter. If IFS is unset, which results in `separators' being set
2403 to "", no splitting occurs. If separators has some other value, the
2404 following rules are applied (`IFS white space' means zero or more
2405 occurrences of <space>, <tab>, or <newline>, as long as those characters
2406 are in `separators'):
2407
2408 1) IFS white space is ignored at the start and the end of the
2409 string.
2410 2) Each occurrence of a character in `separators' that is not
2411 IFS white space, along with any adjacent occurrences of
2412 IFS white space delimits a field.
2413 3) Any nonzero-length sequence of IFS white space delimits a field.
2414 */
2415
2416 /* BEWARE! list_string strips null arguments. Don't call it twice and
2417 expect to have "" preserved! */
2418
2419 /* This performs word splitting and quoted null character removal on
2420 STRING. */
2421 #define issep(c) \
2422 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2423 : (c) == (separators)[0]) \
2424 : 0)
2425
2426 WORD_LIST *
2427 list_string (string, separators, quoted)
2428 register char *string, *separators;
2429 int quoted;
2430 {
2431 WORD_LIST *result;
2432 WORD_DESC *t;
2433 char *current_word, *s;
2434 int sindex, sh_style_split, whitesep, xflags;
2435 size_t slen;
2436
2437 if (!string || !*string)
2438 return ((WORD_LIST *)NULL);
2439
2440 sh_style_split = separators && separators[0] == ' ' &&
2441 separators[1] == '\t' &&
2442 separators[2] == '\n' &&
2443 separators[3] == '\0';
2444 for (xflags = 0, s = ifs_value; s && *s; s++)
2445 {
2446 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2447 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2448 }
2449
2450 slen = 0;
2451 /* Remove sequences of whitespace at the beginning of STRING, as
2452 long as those characters appear in IFS. Do not do this if
2453 STRING is quoted or if there are no separator characters. */
2454 if (!quoted || !separators || !*separators)
2455 {
2456 for (s = string; *s && spctabnl (*s) && issep (*s); s++);
2457
2458 if (!*s)
2459 return ((WORD_LIST *)NULL);
2460
2461 string = s;
2462 }
2463
2464 /* OK, now STRING points to a word that does not begin with white space.
2465 The splitting algorithm is:
2466 extract a word, stopping at a separator
2467 skip sequences of spc, tab, or nl as long as they are separators
2468 This obeys the field splitting rules in Posix.2. */
2469 slen = (MB_CUR_MAX > 1) ? strlen (string) : 1;
2470 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
2471 {
2472 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2473 unless multibyte chars are possible. */
2474 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
2475 if (current_word == 0)
2476 break;
2477
2478 /* If we have a quoted empty string, add a quoted null argument. We
2479 want to preserve the quoted null character iff this is a quoted
2480 empty string; otherwise the quoted null characters are removed
2481 below. */
2482 if (QUOTED_NULL (current_word))
2483 {
2484 t = alloc_word_desc ();
2485 t->word = make_quoted_char ('\0');
2486 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2487 result = make_word_list (t, result);
2488 }
2489 else if (current_word[0] != '\0')
2490 {
2491 /* If we have something, then add it regardless. However,
2492 perform quoted null character removal on the current word. */
2493 remove_quoted_nulls (current_word);
2494 result = add_string_to_list (current_word, result);
2495 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
2496 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2497 result->word->flags |= W_QUOTED;
2498 }
2499
2500 /* If we're not doing sequences of separators in the traditional
2501 Bourne shell style, then add a quoted null argument. */
2502 else if (!sh_style_split && !spctabnl (string[sindex]))
2503 {
2504 t = alloc_word_desc ();
2505 t->word = make_quoted_char ('\0');
2506 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2507 result = make_word_list (t, result);
2508 }
2509
2510 free (current_word);
2511
2512 /* Note whether or not the separator is IFS whitespace, used later. */
2513 whitesep = string[sindex] && spctabnl (string[sindex]);
2514
2515 /* Move past the current separator character. */
2516 if (string[sindex])
2517 {
2518 DECLARE_MBSTATE;
2519 ADVANCE_CHAR (string, slen, sindex);
2520 }
2521
2522 /* Now skip sequences of space, tab, or newline characters if they are
2523 in the list of separators. */
2524 while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex]))
2525 sindex++;
2526
2527 /* If the first separator was IFS whitespace and the current character
2528 is a non-whitespace IFS character, it should be part of the current
2529 field delimiter, not a separate delimiter that would result in an
2530 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
2531 if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex]))
2532 {
2533 sindex++;
2534 /* An IFS character that is not IFS white space, along with any
2535 adjacent IFS white space, shall delimit a field. (SUSv3) */
2536 while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex]))
2537 sindex++;
2538 }
2539 }
2540 return (REVERSE_LIST (result, WORD_LIST *));
2541 }
2542
2543 /* Parse a single word from STRING, using SEPARATORS to separate fields.
2544 ENDPTR is set to the first character after the word. This is used by
2545 the `read' builtin. This is never called with SEPARATORS != $IFS;
2546 it should be simplified.
2547
2548 XXX - this function is very similar to list_string; they should be
2549 combined - XXX */
2550 char *
2551 get_word_from_string (stringp, separators, endptr)
2552 char **stringp, *separators, **endptr;
2553 {
2554 register char *s;
2555 char *current_word;
2556 int sindex, sh_style_split, whitesep, xflags;
2557 size_t slen;
2558
2559 if (!stringp || !*stringp || !**stringp)
2560 return ((char *)NULL);
2561
2562 sh_style_split = separators && separators[0] == ' ' &&
2563 separators[1] == '\t' &&
2564 separators[2] == '\n' &&
2565 separators[3] == '\0';
2566 for (xflags = 0, s = ifs_value; s && *s; s++)
2567 {
2568 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2569 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2570 }
2571
2572 s = *stringp;
2573 slen = 0;
2574
2575 /* Remove sequences of whitespace at the beginning of STRING, as
2576 long as those characters appear in IFS. */
2577 if (sh_style_split || !separators || !*separators)
2578 {
2579 for (; *s && spctabnl (*s) && isifs (*s); s++);
2580
2581 /* If the string is nothing but whitespace, update it and return. */
2582 if (!*s)
2583 {
2584 *stringp = s;
2585 if (endptr)
2586 *endptr = s;
2587 return ((char *)NULL);
2588 }
2589 }
2590
2591 /* OK, S points to a word that does not begin with white space.
2592 Now extract a word, stopping at a separator, save a pointer to
2593 the first character after the word, then skip sequences of spc,
2594 tab, or nl as long as they are separators.
2595
2596 This obeys the field splitting rules in Posix.2. */
2597 sindex = 0;
2598 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2599 unless multibyte chars are possible. */
2600 slen = (MB_CUR_MAX > 1) ? strlen (s) : 1;
2601 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
2602
2603 /* Set ENDPTR to the first character after the end of the word. */
2604 if (endptr)
2605 *endptr = s + sindex;
2606
2607 /* Note whether or not the separator is IFS whitespace, used later. */
2608 whitesep = s[sindex] && spctabnl (s[sindex]);
2609
2610 /* Move past the current separator character. */
2611 if (s[sindex])
2612 {
2613 DECLARE_MBSTATE;
2614 ADVANCE_CHAR (s, slen, sindex);
2615 }
2616
2617 /* Now skip sequences of space, tab, or newline characters if they are
2618 in the list of separators. */
2619 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2620 sindex++;
2621
2622 /* If the first separator was IFS whitespace and the current character is
2623 a non-whitespace IFS character, it should be part of the current field
2624 delimiter, not a separate delimiter that would result in an empty field.
2625 Look at POSIX.2, 3.6.5, (3)(b). */
2626 if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex]))
2627 {
2628 sindex++;
2629 /* An IFS character that is not IFS white space, along with any adjacent
2630 IFS white space, shall delimit a field. */
2631 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2632 sindex++;
2633 }
2634
2635 /* Update STRING to point to the next field. */
2636 *stringp = s + sindex;
2637 return (current_word);
2638 }
2639
2640 /* Remove IFS white space at the end of STRING. Start at the end
2641 of the string and walk backwards until the beginning of the string
2642 or we find a character that's not IFS white space and not CTLESC.
2643 Only let CTLESC escape a white space character if SAW_ESCAPE is
2644 non-zero. */
2645 char *
2646 strip_trailing_ifs_whitespace (string, separators, saw_escape)
2647 char *string, *separators;
2648 int saw_escape;
2649 {
2650 char *s;
2651
2652 s = string + STRLEN (string) - 1;
2653 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
2654 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2655 s--;
2656 *++s = '\0';
2657 return string;
2658 }
2659
2660 #if 0
2661 /* UNUSED */
2662 /* Split STRING into words at whitespace. Obeys shell-style quoting with
2663 backslashes, single and double quotes. */
2664 WORD_LIST *
2665 list_string_with_quotes (string)
2666 char *string;
2667 {
2668 WORD_LIST *list;
2669 char *token, *s;
2670 size_t s_len;
2671 int c, i, tokstart, len;
2672
2673 for (s = string; s && *s && spctabnl (*s); s++)
2674 ;
2675 if (s == 0 || *s == 0)
2676 return ((WORD_LIST *)NULL);
2677
2678 s_len = strlen (s);
2679 tokstart = i = 0;
2680 list = (WORD_LIST *)NULL;
2681 while (1)
2682 {
2683 c = s[i];
2684 if (c == '\\')
2685 {
2686 i++;
2687 if (s[i])
2688 i++;
2689 }
2690 else if (c == '\'')
2691 i = skip_single_quoted (s, s_len, ++i);
2692 else if (c == '"')
2693 i = skip_double_quoted (s, s_len, ++i);
2694 else if (c == 0 || spctabnl (c))
2695 {
2696 /* We have found the end of a token. Make a word out of it and
2697 add it to the word list. */
2698 token = substring (s, tokstart, i);
2699 list = add_string_to_list (token, list);
2700 free (token);
2701 while (spctabnl (s[i]))
2702 i++;
2703 if (s[i])
2704 tokstart = i;
2705 else
2706 break;
2707 }
2708 else
2709 i++; /* normal character */
2710 }
2711 return (REVERSE_LIST (list, WORD_LIST *));
2712 }
2713 #endif
2714
2715 /********************************************************/
2716 /* */
2717 /* Functions to perform assignment statements */
2718 /* */
2719 /********************************************************/
2720
2721 #if defined (ARRAY_VARS)
2722 static SHELL_VAR *
2723 do_compound_assignment (name, value, flags)
2724 char *name, *value;
2725 int flags;
2726 {
2727 SHELL_VAR *v;
2728 int mklocal, mkassoc, mkglobal;
2729 WORD_LIST *list;
2730
2731 mklocal = flags & ASS_MKLOCAL;
2732 mkassoc = flags & ASS_MKASSOC;
2733 mkglobal = flags & ASS_MKGLOBAL;
2734
2735 if (mklocal && variable_context)
2736 {
2737 v = find_variable (name);
2738 list = expand_compound_array_assignment (v, value, flags);
2739 if (mkassoc)
2740 v = make_local_assoc_variable (name);
2741 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
2742 v = make_local_array_variable (name, 0);
2743 if (v)
2744 assign_compound_array_list (v, list, flags);
2745 }
2746 /* In a function but forcing assignment in global context */
2747 else if (mkglobal && variable_context)
2748 {
2749 v = find_global_variable (name);
2750 list = expand_compound_array_assignment (v, value, flags);
2751 if (v == 0 && mkassoc)
2752 v = make_new_assoc_variable (name);
2753 else if (v && mkassoc && assoc_p (v) == 0)
2754 v = convert_var_to_assoc (v);
2755 else if (v == 0)
2756 v = make_new_array_variable (name);
2757 else if (v && mkassoc == 0 && array_p (v) == 0)
2758 v = convert_var_to_array (v);
2759 if (v)
2760 assign_compound_array_list (v, list, flags);
2761 }
2762 else
2763 v = assign_array_from_string (name, value, flags);
2764
2765 return (v);
2766 }
2767 #endif
2768
2769 /* Given STRING, an assignment string, get the value of the right side
2770 of the `=', and bind it to the left side. If EXPAND is true, then
2771 perform parameter expansion, command substitution, and arithmetic
2772 expansion on the right-hand side. Perform tilde expansion in any
2773 case. Do not perform word splitting on the result of expansion. */
2774 static int
2775 do_assignment_internal (word, expand)
2776 const WORD_DESC *word;
2777 int expand;
2778 {
2779 int offset, appendop, assign_list, aflags, retval;
2780 char *name, *value, *temp;
2781 SHELL_VAR *entry;
2782 #if defined (ARRAY_VARS)
2783 char *t;
2784 int ni;
2785 #endif
2786 const char *string;
2787
2788 if (word == 0 || word->word == 0)
2789 return 0;
2790
2791 appendop = assign_list = aflags = 0;
2792 string = word->word;
2793 offset = assignment (string, 0);
2794 name = savestring (string);
2795 value = (char *)NULL;
2796
2797 if (name[offset] == '=')
2798 {
2799 if (name[offset - 1] == '+')
2800 {
2801 appendop = 1;
2802 name[offset - 1] = '\0';
2803 }
2804
2805 name[offset] = 0; /* might need this set later */
2806 temp = name + offset + 1;
2807
2808 #if defined (ARRAY_VARS)
2809 if (expand && (word->flags & W_COMPASSIGN))
2810 {
2811 assign_list = ni = 1;
2812 value = extract_array_assignment_list (temp, &ni);
2813 }
2814 else
2815 #endif
2816 if (expand && temp[0])
2817 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
2818 else
2819 value = savestring (temp);
2820 }
2821
2822 if (value == 0)
2823 {
2824 value = (char *)xmalloc (1);
2825 value[0] = '\0';
2826 }
2827
2828 if (echo_command_at_execute)
2829 {
2830 if (appendop)
2831 name[offset - 1] = '+';
2832 xtrace_print_assignment (name, value, assign_list, 1);
2833 if (appendop)
2834 name[offset - 1] = '\0';
2835 }
2836
2837 #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
2838
2839 if (appendop)
2840 aflags |= ASS_APPEND;
2841
2842 #if defined (ARRAY_VARS)
2843 if (t = mbschr (name, '[')) /*]*/
2844 {
2845 if (assign_list)
2846 {
2847 report_error (_("%s: cannot assign list to array member"), name);
2848 ASSIGN_RETURN (0);
2849 }
2850 entry = assign_array_element (name, value, aflags);
2851 if (entry == 0)
2852 ASSIGN_RETURN (0);
2853 }
2854 else if (assign_list)
2855 {
2856 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
2857 aflags |= ASS_MKLOCAL;
2858 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
2859 aflags |= ASS_MKGLOBAL;
2860 if (word->flags & W_ASSIGNASSOC)
2861 aflags |= ASS_MKASSOC;
2862 entry = do_compound_assignment (name, value, aflags);
2863 }
2864 else
2865 #endif /* ARRAY_VARS */
2866 entry = bind_variable (name, value, aflags);
2867
2868 stupidly_hack_special_variables (name);
2869
2870 /* Return 1 if the assignment seems to have been performed correctly. */
2871 if (entry == 0 || readonly_p (entry))
2872 retval = 0; /* assignment failure */
2873 else if (noassign_p (entry))
2874 {
2875 last_command_exit_value = EXECUTION_FAILURE;
2876 retval = 1; /* error status, but not assignment failure */
2877 }
2878 else
2879 retval = 1;
2880
2881 if (entry && retval != 0 && noassign_p (entry) == 0)
2882 VUNSETATTR (entry, att_invisible);
2883
2884 ASSIGN_RETURN (retval);
2885 }
2886
2887 /* Perform the assignment statement in STRING, and expand the
2888 right side by doing tilde, command and parameter expansion. */
2889 int
2890 do_assignment (string)
2891 char *string;
2892 {
2893 WORD_DESC td;
2894
2895 td.flags = W_ASSIGNMENT;
2896 td.word = string;
2897
2898 return do_assignment_internal (&td, 1);
2899 }
2900
2901 int
2902 do_word_assignment (word, flags)
2903 WORD_DESC *word;
2904 int flags;
2905 {
2906 return do_assignment_internal (word, 1);
2907 }
2908
2909 /* Given STRING, an assignment string, get the value of the right side
2910 of the `=', and bind it to the left side. Do not perform any word
2911 expansions on the right hand side. */
2912 int
2913 do_assignment_no_expand (string)
2914 char *string;
2915 {
2916 WORD_DESC td;
2917
2918 td.flags = W_ASSIGNMENT;
2919 td.word = string;
2920
2921 return (do_assignment_internal (&td, 0));
2922 }
2923
2924 /***************************************************
2925 * *
2926 * Functions to manage the positional parameters *
2927 * *
2928 ***************************************************/
2929
2930 /* Return the word list that corresponds to `$*'. */
2931 WORD_LIST *
2932 list_rest_of_args ()
2933 {
2934 register WORD_LIST *list, *args;
2935 int i;
2936
2937 /* Break out of the loop as soon as one of the dollar variables is null. */
2938 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
2939 list = make_word_list (make_bare_word (dollar_vars[i]), list);
2940
2941 for (args = rest_of_args; args; args = args->next)
2942 list = make_word_list (make_bare_word (args->word->word), list);
2943
2944 return (REVERSE_LIST (list, WORD_LIST *));
2945 }
2946
2947 int
2948 number_of_args ()
2949 {
2950 register WORD_LIST *list;
2951 int n;
2952
2953 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
2954 ;
2955 for (list = rest_of_args; list; list = list->next)
2956 n++;
2957 return n;
2958 }
2959
2960 /* Return the value of a positional parameter. This handles values > 10. */
2961 char *
2962 get_dollar_var_value (ind)
2963 intmax_t ind;
2964 {
2965 char *temp;
2966 WORD_LIST *p;
2967
2968 if (ind < 10)
2969 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
2970 else /* We want something like ${11} */
2971 {
2972 ind -= 10;
2973 for (p = rest_of_args; p && ind--; p = p->next)
2974 ;
2975 temp = p ? savestring (p->word->word) : (char *)NULL;
2976 }
2977 return (temp);
2978 }
2979
2980 /* Make a single large string out of the dollar digit variables,
2981 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
2982 case of "$*" with respect to IFS. */
2983 char *
2984 string_rest_of_args (dollar_star)
2985 int dollar_star;
2986 {
2987 register WORD_LIST *list;
2988 char *string;
2989
2990 list = list_rest_of_args ();
2991 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
2992 dispose_words (list);
2993 return (string);
2994 }
2995
2996 /* Return a string containing the positional parameters from START to
2997 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
2998 which only makes a difference if QUOTED is non-zero. If QUOTED includes
2999 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3000 no quoting chars are added. */
3001 static char *
3002 pos_params (string, start, end, quoted)
3003 char *string;
3004 int start, end, quoted;
3005 {
3006 WORD_LIST *save, *params, *h, *t;
3007 char *ret;
3008 int i;
3009
3010 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3011 if (start == end)
3012 return ((char *)NULL);
3013
3014 save = params = list_rest_of_args ();
3015 if (save == 0)
3016 return ((char *)NULL);
3017
3018 if (start == 0) /* handle ${@:0[:x]} specially */
3019 {
3020 t = make_word_list (make_word (dollar_vars[0]), params);
3021 save = params = t;
3022 }
3023
3024 for (i = start ? 1 : 0; params && i < start; i++)
3025 params = params->next;
3026 if (params == 0)
3027 return ((char *)NULL);
3028 for (h = t = params; params && i < end; i++)
3029 {
3030 t = params;
3031 params = params->next;
3032 }
3033
3034 t->next = (WORD_LIST *)NULL;
3035
3036 ret = string_list_pos_params (string[0], h, quoted);
3037
3038 if (t != params)
3039 t->next = params;
3040
3041 dispose_words (save);
3042 return (ret);
3043 }
3044
3045 /******************************************************************/
3046 /* */
3047 /* Functions to expand strings to strings or WORD_LISTs */
3048 /* */
3049 /******************************************************************/
3050
3051 #if defined (PROCESS_SUBSTITUTION)
3052 #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
3053 #else
3054 #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
3055 #endif
3056
3057 /* If there are any characters in STRING that require full expansion,
3058 then call FUNC to expand STRING; otherwise just perform quote
3059 removal if necessary. This returns a new string. */
3060 static char *
3061 expand_string_if_necessary (string, quoted, func)
3062 char *string;
3063 int quoted;
3064 EXPFUNC *func;
3065 {
3066 WORD_LIST *list;
3067 size_t slen;
3068 int i, saw_quote;
3069 char *ret;
3070 DECLARE_MBSTATE;
3071
3072 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3073 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3074 i = saw_quote = 0;
3075 while (string[i])
3076 {
3077 if (EXP_CHAR (string[i]))
3078 break;
3079 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3080 saw_quote = 1;
3081 ADVANCE_CHAR (string, slen, i);
3082 }
3083
3084 if (string[i])
3085 {
3086 list = (*func) (string, quoted);
3087 if (list)
3088 {
3089 ret = string_list (list);
3090 dispose_words (list);
3091 }
3092 else
3093 ret = (char *)NULL;
3094 }
3095 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3096 ret = string_quote_removal (string, quoted);
3097 else
3098 ret = savestring (string);
3099
3100 return ret;
3101 }
3102
3103 static inline char *
3104 expand_string_to_string_internal (string, quoted, func)
3105 char *string;
3106 int quoted;
3107 EXPFUNC *func;
3108 {
3109 WORD_LIST *list;
3110 char *ret;
3111
3112 if (string == 0 || *string == '\0')
3113 return ((char *)NULL);
3114
3115 list = (*func) (string, quoted);
3116 if (list)
3117 {
3118 ret = string_list (list);
3119 dispose_words (list);
3120 }
3121 else
3122 ret = (char *)NULL;
3123
3124 return (ret);
3125 }
3126
3127 char *
3128 expand_string_to_string (string, quoted)
3129 char *string;
3130 int quoted;
3131 {
3132 return (expand_string_to_string_internal (string, quoted, expand_string));
3133 }
3134
3135 char *
3136 expand_string_unsplit_to_string (string, quoted)
3137 char *string;
3138 int quoted;
3139 {
3140 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3141 }
3142
3143 char *
3144 expand_assignment_string_to_string (string, quoted)
3145 char *string;
3146 int quoted;
3147 {
3148 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3149 }
3150
3151 char *
3152 expand_arith_string (string, quoted)
3153 char *string;
3154 int quoted;
3155 {
3156 WORD_DESC td;
3157 WORD_LIST *list, *tlist;
3158 size_t slen;
3159 int i, saw_quote;
3160 char *ret;
3161 DECLARE_MBSTATE;
3162
3163 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3164 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3165 i = saw_quote = 0;
3166 while (string[i])
3167 {
3168 if (EXP_CHAR (string[i]))
3169 break;
3170 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3171 saw_quote = 1;
3172 ADVANCE_CHAR (string, slen, i);
3173 }
3174
3175 if (string[i])
3176 {
3177 /* This is expanded version of expand_string_internal as it's called by
3178 expand_string_leave_quoted */
3179 td.flags = W_NOPROCSUB; /* don't want process substitution */
3180 td.word = savestring (string);
3181 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3182 /* This takes care of the calls from expand_string_leave_quoted and
3183 expand_string */
3184 if (list)
3185 {
3186 tlist = word_list_split (list);
3187 dispose_words (list);
3188 list = tlist;
3189 if (list)
3190 dequote_list (list);
3191 }
3192 /* This comes from expand_string_if_necessary */
3193 if (list)
3194 {
3195 ret = string_list (list);
3196 dispose_words (list);
3197 }
3198 else
3199 ret = (char *)NULL;
3200 FREE (td.word);
3201 }
3202 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3203 ret = string_quote_removal (string, quoted);
3204 else
3205 ret = savestring (string);
3206
3207 return ret;
3208 }
3209
3210 #if defined (COND_COMMAND)
3211 /* Just remove backslashes in STRING. Returns a new string. */
3212 char *
3213 remove_backslashes (string)
3214 char *string;
3215 {
3216 char *r, *ret, *s;
3217
3218 r = ret = (char *)xmalloc (strlen (string) + 1);
3219 for (s = string; s && *s; )
3220 {
3221 if (*s == '\\')
3222 s++;
3223 if (*s == 0)
3224 break;
3225 *r++ = *s++;
3226 }
3227 *r = '\0';
3228 return ret;
3229 }
3230
3231 /* This needs better error handling. */
3232 /* Expand W for use as an argument to a unary or binary operator in a
3233 [[...]] expression. If SPECIAL is 1, this is the rhs argument
3234 to the != or == operator, and should be treated as a pattern. In
3235 this case, we quote the string specially for the globbing code. If
3236 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3237 be quoted appropriately for regcomp/regexec. The caller is responsible
3238 for removing the backslashes if the unquoted word is needed later. */
3239 char *
3240 cond_expand_word (w, special)
3241 WORD_DESC *w;
3242 int special;
3243 {
3244 char *r, *p;
3245 WORD_LIST *l;
3246 int qflags;
3247
3248 if (w->word == 0 || w->word[0] == '\0')
3249 return ((char *)NULL);
3250
3251 expand_no_split_dollar_star = 1;
3252 w->flags |= W_NOSPLIT2;
3253 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3254 expand_no_split_dollar_star = 0;
3255 if (l)
3256 {
3257 if (special == 0) /* LHS */
3258 {
3259 dequote_list (l);
3260 r = string_list (l);
3261 }
3262 else
3263 {
3264 /* Need to figure out whether or not we should call dequote_escapes
3265 or a new dequote_ctlnul function here, and under what
3266 circumstances. */
3267 qflags = QGLOB_CVTNULL;
3268 if (special == 2)
3269 qflags |= QGLOB_REGEXP;
3270 p = string_list (l);
3271 r = quote_string_for_globbing (p, qflags);
3272 free (p);
3273 }
3274 dispose_words (l);
3275 }
3276 else
3277 r = (char *)NULL;
3278
3279 return r;
3280 }
3281 #endif
3282
3283 /* Call expand_word_internal to expand W and handle error returns.
3284 A convenience function for functions that don't want to handle
3285 any errors or free any memory before aborting. */
3286 static WORD_LIST *
3287 call_expand_word_internal (w, q, i, c, e)
3288 WORD_DESC *w;
3289 int q, i, *c, *e;
3290 {
3291 WORD_LIST *result;
3292
3293 result = expand_word_internal (w, q, i, c, e);
3294 if (result == &expand_word_error || result == &expand_word_fatal)
3295 {
3296 /* By convention, each time this error is returned, w->word has
3297 already been freed (it sometimes may not be in the fatal case,
3298 but that doesn't result in a memory leak because we're going
3299 to exit in most cases). */
3300 w->word = (char *)NULL;
3301 last_command_exit_value = EXECUTION_FAILURE;
3302 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
3303 /* NOTREACHED */
3304 return (NULL);
3305 }
3306 else
3307 return (result);
3308 }
3309
3310 /* Perform parameter expansion, command substitution, and arithmetic
3311 expansion on STRING, as if it were a word. Leave the result quoted.
3312 Since this does not perform word splitting, it leaves quoted nulls
3313 in the result. */
3314 static WORD_LIST *
3315 expand_string_internal (string, quoted)
3316 char *string;
3317 int quoted;
3318 {
3319 WORD_DESC td;
3320 WORD_LIST *tresult;
3321
3322 if (string == 0 || *string == 0)
3323 return ((WORD_LIST *)NULL);
3324
3325 td.flags = 0;
3326 td.word = savestring (string);
3327
3328 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3329
3330 FREE (td.word);
3331 return (tresult);
3332 }
3333
3334 /* Expand STRING by performing parameter expansion, command substitution,
3335 and arithmetic expansion. Dequote the resulting WORD_LIST before
3336 returning it, but do not perform word splitting. The call to
3337 remove_quoted_nulls () is in here because word splitting normally
3338 takes care of quote removal. */
3339 WORD_LIST *
3340 expand_string_unsplit (string, quoted)
3341 char *string;
3342 int quoted;
3343 {
3344 WORD_LIST *value;
3345
3346 if (string == 0 || *string == '\0')
3347 return ((WORD_LIST *)NULL);
3348
3349 expand_no_split_dollar_star = 1;
3350 value = expand_string_internal (string, quoted);
3351 expand_no_split_dollar_star = 0;
3352
3353 if (value)
3354 {
3355 if (value->word)
3356 {
3357 remove_quoted_nulls (value->word->word);
3358 value->word->flags &= ~W_HASQUOTEDNULL;
3359 }
3360 dequote_list (value);
3361 }
3362 return (value);
3363 }
3364
3365 /* Expand the rhs of an assignment statement */
3366 WORD_LIST *
3367 expand_string_assignment (string, quoted)
3368 char *string;
3369 int quoted;
3370 {
3371 WORD_DESC td;
3372 WORD_LIST *value;
3373
3374 if (string == 0 || *string == '\0')
3375 return ((WORD_LIST *)NULL);
3376
3377 expand_no_split_dollar_star = 1;
3378
3379 td.flags = W_ASSIGNRHS;
3380 td.word = savestring (string);
3381 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3382 FREE (td.word);
3383
3384 expand_no_split_dollar_star = 0;
3385
3386 if (value)
3387 {
3388 if (value->word)
3389 {
3390 remove_quoted_nulls (value->word->word);
3391 value->word->flags &= ~W_HASQUOTEDNULL;
3392 }
3393 dequote_list (value);
3394 }
3395 return (value);
3396 }
3397
3398
3399 /* Expand one of the PS? prompt strings. This is a sort of combination of
3400 expand_string_unsplit and expand_string_internal, but returns the
3401 passed string when an error occurs. Might want to trap other calls
3402 to jump_to_top_level here so we don't endlessly loop. */
3403 WORD_LIST *
3404 expand_prompt_string (string, quoted, wflags)
3405 char *string;
3406 int quoted;
3407 int wflags;
3408 {
3409 WORD_LIST *value;
3410 WORD_DESC td;
3411
3412 if (string == 0 || *string == 0)
3413 return ((WORD_LIST *)NULL);
3414
3415 td.flags = wflags;
3416 td.word = savestring (string);
3417
3418 no_longjmp_on_fatal_error = 1;
3419 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3420 no_longjmp_on_fatal_error = 0;
3421
3422 if (value == &expand_word_error || value == &expand_word_fatal)
3423 {
3424 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3425 return value;
3426 }
3427 FREE (td.word);
3428 if (value)
3429 {
3430 if (value->word)
3431 {
3432 remove_quoted_nulls (value->word->word);
3433 value->word->flags &= ~W_HASQUOTEDNULL;
3434 }
3435 dequote_list (value);
3436 }
3437 return (value);
3438 }
3439
3440 /* Expand STRING just as if you were expanding a word, but do not dequote
3441 the resultant WORD_LIST. This is called only from within this file,
3442 and is used to correctly preserve quoted characters when expanding
3443 things like ${1+"$@"}. This does parameter expansion, command
3444 substitution, arithmetic expansion, and word splitting. */
3445 static WORD_LIST *
3446 expand_string_leave_quoted (string, quoted)
3447 char *string;
3448 int quoted;
3449 {
3450 WORD_LIST *tlist;
3451 WORD_LIST *tresult;
3452
3453 if (string == 0 || *string == '\0')
3454 return ((WORD_LIST *)NULL);
3455
3456 tlist = expand_string_internal (string, quoted);
3457
3458 if (tlist)
3459 {
3460 tresult = word_list_split (tlist);
3461 dispose_words (tlist);
3462 return (tresult);
3463 }
3464 return ((WORD_LIST *)NULL);
3465 }
3466
3467 /* This does not perform word splitting or dequote the WORD_LIST
3468 it returns. */
3469 static WORD_LIST *
3470 expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at)
3471 char *string;
3472 int quoted, *dollar_at_p, *has_dollar_at;
3473 {
3474 WORD_DESC td;
3475 WORD_LIST *tresult;
3476
3477 if (string == 0 || *string == '\0')
3478 return (WORD_LIST *)NULL;
3479
3480 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
3481 td.word = string;
3482 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at);
3483 return (tresult);
3484 }
3485
3486 /* Expand STRING just as if you were expanding a word. This also returns
3487 a list of words. Note that filename globbing is *NOT* done for word
3488 or string expansion, just when the shell is expanding a command. This
3489 does parameter expansion, command substitution, arithmetic expansion,
3490 and word splitting. Dequote the resultant WORD_LIST before returning. */
3491 WORD_LIST *
3492 expand_string (string, quoted)
3493 char *string;
3494 int quoted;
3495 {
3496 WORD_LIST *result;
3497
3498 if (string == 0 || *string == '\0')
3499 return ((WORD_LIST *)NULL);
3500
3501 result = expand_string_leave_quoted (string, quoted);
3502 return (result ? dequote_list (result) : result);
3503 }
3504
3505 /***************************************************
3506 * *
3507 * Functions to handle quoting chars *
3508 * *
3509 ***************************************************/
3510
3511 /* Conventions:
3512
3513 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3514 The parser passes CTLNUL as CTLESC CTLNUL. */
3515
3516 /* Quote escape characters in string s, but no other characters. This is
3517 used to protect CTLESC and CTLNUL in variable values from the rest of
3518 the word expansion process after the variable is expanded (word splitting
3519 and filename generation). If IFS is null, we quote spaces as well, just
3520 in case we split on spaces later (in the case of unquoted $@, we will
3521 eventually attempt to split the entire word on spaces). Corresponding
3522 code exists in dequote_escapes. Even if we don't end up splitting on
3523 spaces, quoting spaces is not a problem. This should never be called on
3524 a string that is quoted with single or double quotes or part of a here
3525 document (effectively double-quoted). */
3526 char *
3527 quote_escapes (string)
3528 char *string;
3529 {
3530 register char *s, *t;
3531 size_t slen;
3532 char *result, *send;
3533 int quote_spaces, skip_ctlesc, skip_ctlnul;
3534 DECLARE_MBSTATE;
3535
3536 slen = strlen (string);
3537 send = string + slen;
3538
3539 quote_spaces = (ifs_value && *ifs_value == 0);
3540
3541 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3542 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3543
3544 t = result = (char *)xmalloc ((slen * 2) + 1);
3545 s = string;
3546
3547 while (*s)
3548 {
3549 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
3550 *t++ = CTLESC;
3551 COPY_CHAR_P (t, s, send);
3552 }
3553 *t = '\0';
3554
3555 return (result);
3556 }
3557
3558 static WORD_LIST *
3559 list_quote_escapes (list)
3560 WORD_LIST *list;
3561 {
3562 register WORD_LIST *w;
3563 char *t;
3564
3565 for (w = list; w; w = w->next)
3566 {
3567 t = w->word->word;
3568 w->word->word = quote_escapes (t);
3569 free (t);
3570 }
3571 return list;
3572 }
3573
3574 /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3575
3576 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3577 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3578 data stream pass through properly.
3579
3580 We need to remove doubled CTLESC characters inside quoted strings before
3581 quoting the entire string, so we do not double the number of CTLESC
3582 characters.
3583
3584 Also used by parts of the pattern substitution code. */
3585 char *
3586 dequote_escapes (string)
3587 char *string;
3588 {
3589 register char *s, *t, *s1;
3590 size_t slen;
3591 char *result, *send;
3592 int quote_spaces;
3593 DECLARE_MBSTATE;
3594
3595 if (string == 0)
3596 return string;
3597
3598 slen = strlen (string);
3599 send = string + slen;
3600
3601 t = result = (char *)xmalloc (slen + 1);
3602
3603 if (strchr (string, CTLESC) == 0)
3604 return (strcpy (result, string));
3605
3606 quote_spaces = (ifs_value && *ifs_value == 0);
3607
3608 s = string;
3609 while (*s)
3610 {
3611 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
3612 {
3613 s++;
3614 if (*s == '\0')
3615 break;
3616 }
3617 COPY_CHAR_P (t, s, send);
3618 }
3619 *t = '\0';
3620
3621 return result;
3622 }
3623
3624 static WORD_LIST *
3625 list_dequote_escapes (list)
3626 WORD_LIST *list;
3627 {
3628 register WORD_LIST *w;
3629 char *t;
3630
3631 for (w = list; w; w = w->next)
3632 {
3633 t = w->word->word;
3634 w->word->word = dequote_escapes (t);
3635 free (t);
3636 }
3637 return list;
3638 }
3639
3640 /* Return a new string with the quoted representation of character C.
3641 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3642 set in any resultant WORD_DESC where this value is the word. */
3643 static char *
3644 make_quoted_char (c)
3645 int c;
3646 {
3647 char *temp;
3648
3649 temp = (char *)xmalloc (3);
3650 if (c == 0)
3651 {
3652 temp[0] = CTLNUL;
3653 temp[1] = '\0';
3654 }
3655 else
3656 {
3657 temp[0] = CTLESC;
3658 temp[1] = c;
3659 temp[2] = '\0';
3660 }
3661 return (temp);
3662 }
3663
3664 /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3665 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3666 this value is the word. */
3667 char *
3668 quote_string (string)
3669 char *string;
3670 {
3671 register char *t;
3672 size_t slen;
3673 char *result, *send;
3674
3675 if (*string == 0)
3676 {
3677 result = (char *)xmalloc (2);
3678 result[0] = CTLNUL;
3679 result[1] = '\0';
3680 }
3681 else
3682 {
3683 DECLARE_MBSTATE;
3684
3685 slen = strlen (string);
3686 send = string + slen;
3687
3688 result = (char *)xmalloc ((slen * 2) + 1);
3689
3690 for (t = result; string < send; )
3691 {
3692 *t++ = CTLESC;
3693 COPY_CHAR_P (t, string, send);
3694 }
3695 *t = '\0';
3696 }
3697 return (result);
3698 }
3699
3700 /* De-quote quoted characters in STRING. */
3701 char *
3702 dequote_string (string)
3703 char *string;
3704 {
3705 register char *s, *t;
3706 size_t slen;
3707 char *result, *send;
3708 DECLARE_MBSTATE;
3709
3710 slen = strlen (string);
3711
3712 t = result = (char *)xmalloc (slen + 1);
3713
3714 if (QUOTED_NULL (string))
3715 {
3716 result[0] = '\0';
3717 return (result);
3718 }
3719
3720 /* If no character in the string can be quoted, don't bother examining
3721 each character. Just return a copy of the string passed to us. */
3722 if (strchr (string, CTLESC) == NULL)
3723 return (strcpy (result, string));
3724
3725 send = string + slen;
3726 s = string;
3727 while (*s)
3728 {
3729 if (*s == CTLESC)
3730 {
3731 s++;
3732 if (*s == '\0')
3733 break;
3734 }
3735 COPY_CHAR_P (t, s, send);
3736 }
3737
3738 *t = '\0';
3739 return (result);
3740 }
3741
3742 /* Quote the entire WORD_LIST list. */
3743 static WORD_LIST *
3744 quote_list (list)
3745 WORD_LIST *list;
3746 {
3747 register WORD_LIST *w;
3748 char *t;
3749
3750 for (w = list; w; w = w->next)
3751 {
3752 t = w->word->word;
3753 w->word->word = quote_string (t);
3754 if (*t == 0)
3755 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
3756 w->word->flags |= W_QUOTED;
3757 free (t);
3758 }
3759 return list;
3760 }
3761
3762 /* De-quote quoted characters in each word in LIST. */
3763 WORD_LIST *
3764 dequote_list (list)
3765 WORD_LIST *list;
3766 {
3767 register char *s;
3768 register WORD_LIST *tlist;
3769
3770 for (tlist = list; tlist; tlist = tlist->next)
3771 {
3772 s = dequote_string (tlist->word->word);
3773 if (QUOTED_NULL (tlist->word->word))
3774 tlist->word->flags &= ~W_HASQUOTEDNULL;
3775 free (tlist->word->word);
3776 tlist->word->word = s;
3777 }
3778 return list;
3779 }
3780
3781 /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
3782 string. */
3783 char *
3784 remove_quoted_escapes (string)
3785 char *string;
3786 {
3787 char *t;
3788
3789 if (string)
3790 {
3791 t = dequote_escapes (string);
3792 strcpy (string, t);
3793 free (t);
3794 }
3795
3796 return (string);
3797 }
3798
3799 /* Perform quoted null character removal on STRING. We don't allow any
3800 quoted null characters in the middle or at the ends of strings because
3801 of how expand_word_internal works. remove_quoted_nulls () turns
3802 STRING into an empty string iff it only consists of a quoted null,
3803 and removes all unquoted CTLNUL characters. */
3804 char *
3805 remove_quoted_nulls (string)
3806 char *string;
3807 {
3808 register size_t slen;
3809 register int i, j, prev_i;
3810 DECLARE_MBSTATE;
3811
3812 if (strchr (string, CTLNUL) == 0) /* XXX */
3813 return string; /* XXX */
3814
3815 slen = strlen (string);
3816 i = j = 0;
3817
3818 while (i < slen)
3819 {
3820 if (string[i] == CTLESC)
3821 {
3822 /* Old code had j++, but we cannot assume that i == j at this
3823 point -- what if a CTLNUL has already been removed from the
3824 string? We don't want to drop the CTLESC or recopy characters
3825 that we've already copied down. */
3826 i++; string[j++] = CTLESC;
3827 if (i == slen)
3828 break;
3829 }
3830 else if (string[i] == CTLNUL)
3831 {
3832 i++;
3833 continue;
3834 }
3835
3836 prev_i = i;
3837 ADVANCE_CHAR (string, slen, i);
3838 if (j < prev_i)
3839 {
3840 do string[j++] = string[prev_i++]; while (prev_i < i);
3841 }
3842 else
3843 j = i;
3844 }
3845 string[j] = '\0';
3846
3847 return (string);
3848 }
3849
3850 /* Perform quoted null character removal on each element of LIST.
3851 This modifies LIST. */
3852 void
3853 word_list_remove_quoted_nulls (list)
3854 WORD_LIST *list;
3855 {
3856 register WORD_LIST *t;
3857
3858 for (t = list; t; t = t->next)
3859 {
3860 remove_quoted_nulls (t->word->word);
3861 t->word->flags &= ~W_HASQUOTEDNULL;
3862 }
3863 }
3864
3865 /* **************************************************************** */
3866 /* */
3867 /* Functions for Matching and Removing Patterns */
3868 /* */
3869 /* **************************************************************** */
3870
3871 #if defined (HANDLE_MULTIBYTE)
3872 #if 0 /* Currently unused */
3873 static unsigned char *
3874 mb_getcharlens (string, len)
3875 char *string;
3876 int len;
3877 {
3878 int i, offset, last;
3879 unsigned char *ret;
3880 char *p;
3881 DECLARE_MBSTATE;
3882
3883 i = offset = 0;
3884 last = 0;
3885 ret = (unsigned char *)xmalloc (len);
3886 memset (ret, 0, len);
3887 while (string[last])
3888 {
3889 ADVANCE_CHAR (string, len, offset);
3890 ret[last] = offset - last;
3891 last = offset;
3892 }
3893 return ret;
3894 }
3895 #endif
3896 #endif
3897
3898 /* Remove the portion of PARAM matched by PATTERN according to OP, where OP
3899 can have one of 4 values:
3900 RP_LONG_LEFT remove longest matching portion at start of PARAM
3901 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
3902 RP_LONG_RIGHT remove longest matching portion at end of PARAM
3903 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
3904 */
3905
3906 #define RP_LONG_LEFT 1
3907 #define RP_SHORT_LEFT 2
3908 #define RP_LONG_RIGHT 3
3909 #define RP_SHORT_RIGHT 4
3910
3911 /* Returns its first argument if nothing matched; new memory otherwise */
3912 static char *
3913 remove_upattern (param, pattern, op)
3914 char *param, *pattern;
3915 int op;
3916 {
3917 register int len;
3918 register char *end;
3919 register char *p, *ret, c;
3920
3921 len = STRLEN (param);
3922 end = param + len;
3923
3924 switch (op)
3925 {
3926 case RP_LONG_LEFT: /* remove longest match at start */
3927 for (p = end; p >= param; p--)
3928 {
3929 c = *p; *p = '\0';
3930 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3931 {
3932 *p = c;
3933 return (savestring (p));
3934 }
3935 *p = c;
3936
3937 }
3938 break;
3939
3940 case RP_SHORT_LEFT: /* remove shortest match at start */
3941 for (p = param; p <= end; p++)
3942 {
3943 c = *p; *p = '\0';
3944 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3945 {
3946 *p = c;
3947 return (savestring (p));
3948 }
3949 *p = c;
3950 }
3951 break;
3952
3953 case RP_LONG_RIGHT: /* remove longest match at end */
3954 for (p = param; p <= end; p++)
3955 {
3956 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3957 {
3958 c = *p; *p = '\0';
3959 ret = savestring (param);
3960 *p = c;
3961 return (ret);
3962 }
3963 }
3964 break;
3965
3966 case RP_SHORT_RIGHT: /* remove shortest match at end */
3967 for (p = end; p >= param; p--)
3968 {
3969 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3970 {
3971 c = *p; *p = '\0';
3972 ret = savestring (param);
3973 *p = c;
3974 return (ret);
3975 }
3976 }
3977 break;
3978 }
3979
3980 return (param); /* no match, return original string */
3981 }
3982
3983 #if defined (HANDLE_MULTIBYTE)
3984 /* Returns its first argument if nothing matched; new memory otherwise */
3985 static wchar_t *
3986 remove_wpattern (wparam, wstrlen, wpattern, op)
3987 wchar_t *wparam;
3988 size_t wstrlen;
3989 wchar_t *wpattern;
3990 int op;
3991 {
3992 wchar_t wc, *ret;
3993 int n;
3994
3995 switch (op)
3996 {
3997 case RP_LONG_LEFT: /* remove longest match at start */
3998 for (n = wstrlen; n >= 0; n--)
3999 {
4000 wc = wparam[n]; wparam[n] = L'\0';
4001 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4002 {
4003 wparam[n] = wc;
4004 return (wcsdup (wparam + n));
4005 }
4006 wparam[n] = wc;
4007 }
4008 break;
4009
4010 case RP_SHORT_LEFT: /* remove shortest match at start */
4011 for (n = 0; n <= wstrlen; n++)
4012 {
4013 wc = wparam[n]; wparam[n] = L'\0';
4014 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4015 {
4016 wparam[n] = wc;
4017 return (wcsdup (wparam + n));
4018 }
4019 wparam[n] = wc;
4020 }
4021 break;
4022
4023 case RP_LONG_RIGHT: /* remove longest match at end */
4024 for (n = 0; n <= wstrlen; n++)
4025 {
4026 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4027 {
4028 wc = wparam[n]; wparam[n] = L'\0';
4029 ret = wcsdup (wparam);
4030 wparam[n] = wc;
4031 return (ret);
4032 }
4033 }
4034 break;
4035
4036 case RP_SHORT_RIGHT: /* remove shortest match at end */
4037 for (n = wstrlen; n >= 0; n--)
4038 {
4039 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4040 {
4041 wc = wparam[n]; wparam[n] = L'\0';
4042 ret = wcsdup (wparam);
4043 wparam[n] = wc;
4044 return (ret);
4045 }
4046 }
4047 break;
4048 }
4049
4050 return (wparam); /* no match, return original string */
4051 }
4052 #endif /* HANDLE_MULTIBYTE */
4053
4054 static char *
4055 remove_pattern (param, pattern, op)
4056 char *param, *pattern;
4057 int op;
4058 {
4059 char *xret;
4060
4061 if (param == NULL)
4062 return (param);
4063 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4064 return (savestring (param));
4065
4066 #if defined (HANDLE_MULTIBYTE)
4067 if (MB_CUR_MAX > 1)
4068 {
4069 wchar_t *ret, *oret;
4070 size_t n;
4071 wchar_t *wparam, *wpattern;
4072 mbstate_t ps;
4073
4074 n = xdupmbstowcs (&wpattern, NULL, pattern);
4075 if (n == (size_t)-1)
4076 {
4077 xret = remove_upattern (param, pattern, op);
4078 return ((xret == param) ? savestring (param) : xret);
4079 }
4080 n = xdupmbstowcs (&wparam, NULL, param);
4081
4082 if (n == (size_t)-1)
4083 {
4084 free (wpattern);
4085 xret = remove_upattern (param, pattern, op);
4086 return ((xret == param) ? savestring (param) : xret);
4087 }
4088 oret = ret = remove_wpattern (wparam, n, wpattern, op);
4089 /* Don't bother to convert wparam back to multibyte string if nothing
4090 matched; just return copy of original string */
4091 if (ret == wparam)
4092 {
4093 free (wparam);
4094 free (wpattern);
4095 return (savestring (param));
4096 }
4097
4098 free (wparam);
4099 free (wpattern);
4100
4101 n = strlen (param);
4102 xret = (char *)xmalloc (n + 1);
4103 memset (&ps, '\0', sizeof (mbstate_t));
4104 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4105 xret[n] = '\0'; /* just to make sure */
4106 free (oret);
4107 return xret;
4108 }
4109 else
4110 #endif
4111 {
4112 xret = remove_upattern (param, pattern, op);
4113 return ((xret == param) ? savestring (param) : xret);
4114 }
4115 }
4116
4117 /* Match PAT anywhere in STRING and return the match boundaries.
4118 This returns 1 in case of a successful match, 0 otherwise. SP
4119 and EP are pointers into the string where the match begins and
4120 ends, respectively. MTYPE controls what kind of match is attempted.
4121 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4122 of the string, respectively. The longest match is returned. */
4123 static int
4124 match_upattern (string, pat, mtype, sp, ep)
4125 char *string, *pat;
4126 int mtype;
4127 char **sp, **ep;
4128 {
4129 int c, len, mlen;
4130 register char *p, *p1, *npat;
4131 char *end;
4132 int n1;
4133
4134 /* If the pattern doesn't match anywhere in the string, go ahead and
4135 short-circuit right away. A minor optimization, saves a bunch of
4136 unnecessary calls to strmatch (up to N calls for a string of N
4137 characters) if the match is unsuccessful. To preserve the semantics
4138 of the substring matches below, we make sure that the pattern has
4139 `*' as first and last character, making a new pattern if necessary. */
4140 /* XXX - check this later if I ever implement `**' with special meaning,
4141 since this will potentially result in `**' at the beginning or end */
4142 len = STRLEN (pat);
4143 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
4144 {
4145 p = npat = (char *)xmalloc (len + 3);
4146 p1 = pat;
4147 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
4148 *p++ = '*';
4149 while (*p1)
4150 *p++ = *p1++;
4151 if (p1[-1] != '*' || p[-2] == '\\')
4152 *p++ = '*';
4153 *p = '\0';
4154 }
4155 else
4156 npat = pat;
4157 c = strmatch (npat, string, FNMATCH_EXTFLAG);
4158 if (npat != pat)
4159 free (npat);
4160 if (c == FNM_NOMATCH)
4161 return (0);
4162
4163 len = STRLEN (string);
4164 end = string + len;
4165
4166 mlen = umatchlen (pat, len);
4167
4168 switch (mtype)
4169 {
4170 case MATCH_ANY:
4171 for (p = string; p <= end; p++)
4172 {
4173 if (match_pattern_char (pat, p))
4174 {
4175 p1 = (mlen == -1) ? end : p + mlen;
4176 /* p1 - p = length of portion of string to be considered
4177 p = current position in string
4178 mlen = number of characters consumed by match (-1 for entire string)
4179 end = end of string
4180 we want to break immediately if the potential match len
4181 is greater than the number of characters remaining in the
4182 string
4183 */
4184 if (p1 > end)
4185 break;
4186 for ( ; p1 >= p; p1--)
4187 {
4188 c = *p1; *p1 = '\0';
4189 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4190 {
4191 *p1 = c;
4192 *sp = p;
4193 *ep = p1;
4194 return 1;
4195 }
4196 *p1 = c;
4197 #if 1
4198 /* If MLEN != -1, we have a fixed length pattern. */
4199 if (mlen != -1)
4200 break;
4201 #endif
4202 }
4203 }
4204 }
4205
4206 return (0);
4207
4208 case MATCH_BEG:
4209 if (match_pattern_char (pat, string) == 0)
4210 return (0);
4211
4212 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
4213 {
4214 c = *p; *p = '\0';
4215 if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0)
4216 {
4217 *p = c;
4218 *sp = string;
4219 *ep = p;
4220 return 1;
4221 }
4222 *p = c;
4223 /* If MLEN != -1, we have a fixed length pattern. */
4224 if (mlen != -1)
4225 break;
4226 }
4227
4228 return (0);
4229
4230 case MATCH_END:
4231 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
4232 {
4233 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4234 {
4235 *sp = p;
4236 *ep = end;
4237 return 1;
4238 }
4239 /* If MLEN != -1, we have a fixed length pattern. */
4240 if (mlen != -1)
4241 break;
4242 }
4243
4244 return (0);
4245 }
4246
4247 return (0);
4248 }
4249
4250 #if defined (HANDLE_MULTIBYTE)
4251 /* Match WPAT anywhere in WSTRING and return the match boundaries.
4252 This returns 1 in case of a successful match, 0 otherwise. Wide
4253 character version. */
4254 static int
4255 match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4256 wchar_t *wstring;
4257 char **indices;
4258 size_t wstrlen;
4259 wchar_t *wpat;
4260 int mtype;
4261 char **sp, **ep;
4262 {
4263 wchar_t wc, *wp, *nwpat, *wp1;
4264 size_t len;
4265 int mlen;
4266 int n, n1, n2, simple;
4267
4268 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4269 #if defined (EXTENDED_GLOB)
4270 if (extended_glob)
4271 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
4272 #endif
4273
4274 /* If the pattern doesn't match anywhere in the string, go ahead and
4275 short-circuit right away. A minor optimization, saves a bunch of
4276 unnecessary calls to strmatch (up to N calls for a string of N
4277 characters) if the match is unsuccessful. To preserve the semantics
4278 of the substring matches below, we make sure that the pattern has
4279 `*' as first and last character, making a new pattern if necessary. */
4280 len = wcslen (wpat);
4281 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
4282 {
4283 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
4284 wp1 = wpat;
4285 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
4286 *wp++ = L'*';
4287 while (*wp1 != L'\0')
4288 *wp++ = *wp1++;
4289 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4290 *wp++ = L'*';
4291 *wp = '\0';
4292 }
4293 else
4294 nwpat = wpat;
4295 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG);
4296 if (nwpat != wpat)
4297 free (nwpat);
4298 if (len == FNM_NOMATCH)
4299 return (0);
4300
4301 mlen = wmatchlen (wpat, wstrlen);
4302
4303 /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
4304 switch (mtype)
4305 {
4306 case MATCH_ANY:
4307 for (n = 0; n <= wstrlen; n++)
4308 {
4309 n2 = simple ? (*wpat == wstring[n]) : match_pattern_wchar (wpat, wstring + n);
4310 if (n2)
4311 {
4312 n1 = (mlen == -1) ? wstrlen : n + mlen;
4313 if (n1 > wstrlen)
4314 break;
4315
4316 for ( ; n1 >= n; n1--)
4317 {
4318 wc = wstring[n1]; wstring[n1] = L'\0';
4319 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4320 {
4321 wstring[n1] = wc;
4322 *sp = indices[n];
4323 *ep = indices[n1];
4324 return 1;
4325 }
4326 wstring[n1] = wc;
4327 /* If MLEN != -1, we have a fixed length pattern. */
4328 if (mlen != -1)
4329 break;
4330 }
4331 }
4332 }
4333
4334 return (0);
4335
4336 case MATCH_BEG:
4337 if (match_pattern_wchar (wpat, wstring) == 0)
4338 return (0);
4339
4340 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
4341 {
4342 wc = wstring[n]; wstring[n] = L'\0';
4343 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0)
4344 {
4345 wstring[n] = wc;
4346 *sp = indices[0];
4347 *ep = indices[n];
4348 return 1;
4349 }
4350 wstring[n] = wc;
4351 /* If MLEN != -1, we have a fixed length pattern. */
4352 if (mlen != -1)
4353 break;
4354 }
4355
4356 return (0);
4357
4358 case MATCH_END:
4359 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
4360 {
4361 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4362 {
4363 *sp = indices[n];
4364 *ep = indices[wstrlen];
4365 return 1;
4366 }
4367 /* If MLEN != -1, we have a fixed length pattern. */
4368 if (mlen != -1)
4369 break;
4370 }
4371
4372 return (0);
4373 }
4374
4375 return (0);
4376 }
4377 #endif /* HANDLE_MULTIBYTE */
4378
4379 static int
4380 match_pattern (string, pat, mtype, sp, ep)
4381 char *string, *pat;
4382 int mtype;
4383 char **sp, **ep;
4384 {
4385 #if defined (HANDLE_MULTIBYTE)
4386 int ret;
4387 size_t n;
4388 wchar_t *wstring, *wpat;
4389 char **indices;
4390 size_t slen, plen, mslen, mplen;
4391 #endif
4392
4393 if (string == 0 || *string == 0 || pat == 0 || *pat == 0)
4394 return (0);
4395
4396 #if defined (HANDLE_MULTIBYTE)
4397 if (MB_CUR_MAX > 1)
4398 {
4399 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
4400 return (match_upattern (string, pat, mtype, sp, ep));
4401
4402 n = xdupmbstowcs (&wpat, NULL, pat);
4403 if (n == (size_t)-1)
4404 return (match_upattern (string, pat, mtype, sp, ep));
4405 n = xdupmbstowcs (&wstring, &indices, string);
4406 if (n == (size_t)-1)
4407 {
4408 free (wpat);
4409 return (match_upattern (string, pat, mtype, sp, ep));
4410 }
4411 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4412
4413 free (wpat);
4414 free (wstring);
4415 free (indices);
4416
4417 return (ret);
4418 }
4419 else
4420 #endif
4421 return (match_upattern (string, pat, mtype, sp, ep));
4422 }
4423
4424 static int
4425 getpatspec (c, value)
4426 int c;
4427 char *value;
4428 {
4429 if (c == '#')
4430 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4431 else /* c == '%' */
4432 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4433 }
4434
4435 /* Posix.2 says that the WORD should be run through tilde expansion,
4436 parameter expansion, command substitution and arithmetic expansion.
4437 This leaves the result quoted, so quote_string_for_globbing () has
4438 to be called to fix it up for strmatch (). If QUOTED is non-zero,
4439 it means that the entire expression was enclosed in double quotes.
4440 This means that quoting characters in the pattern do not make any
4441 special pattern characters quoted. For example, the `*' in the
4442 following retains its special meaning: "${foo#'*'}". */
4443 static char *
4444 getpattern (value, quoted, expandpat)
4445 char *value;
4446 int quoted, expandpat;
4447 {
4448 char *pat, *tword;
4449 WORD_LIST *l;
4450 #if 0
4451 int i;
4452 #endif
4453 /* There is a problem here: how to handle single or double quotes in the
4454 pattern string when the whole expression is between double quotes?
4455 POSIX.2 says that enclosing double quotes do not cause the pattern to
4456 be quoted, but does that leave us a problem with @ and array[@] and their
4457 expansions inside a pattern? */
4458 #if 0
4459 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4460 {
4461 i = 0;
4462 pat = string_extract_double_quoted (tword, &i, 1);
4463 free (tword);
4464 tword = pat;
4465 }
4466 #endif
4467
4468 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4469 word splitting. */
4470 l = *value ? expand_string_for_rhs (value,
4471 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
4472 (int *)NULL, (int *)NULL)
4473 : (WORD_LIST *)0;
4474 pat = string_list (l);
4475 dispose_words (l);
4476 if (pat)
4477 {
4478 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4479 free (pat);
4480 pat = tword;
4481 }
4482 return (pat);
4483 }
4484
4485 #if 0
4486 /* Handle removing a pattern from a string as a result of ${name%[%]value}
4487 or ${name#[#]value}. */
4488 static char *
4489 variable_remove_pattern (value, pattern, patspec, quoted)
4490 char *value, *pattern;
4491 int patspec, quoted;
4492 {
4493 char *tword;
4494
4495 tword = remove_pattern (value, pattern, patspec);
4496
4497 return (tword);
4498 }
4499 #endif
4500
4501 static char *
4502 list_remove_pattern (list, pattern, patspec, itype, quoted)
4503 WORD_LIST *list;
4504 char *pattern;
4505 int patspec, itype, quoted;
4506 {
4507 WORD_LIST *new, *l;
4508 WORD_DESC *w;
4509 char *tword;
4510
4511 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4512 {
4513 tword = remove_pattern (l->word->word, pattern, patspec);
4514 w = alloc_word_desc ();
4515 w->word = tword ? tword : savestring ("");
4516 new = make_word_list (w, new);
4517 }
4518
4519 l = REVERSE_LIST (new, WORD_LIST *);
4520 tword = string_list_pos_params (itype, l, quoted);
4521 dispose_words (l);
4522
4523 return (tword);
4524 }
4525
4526 static char *
4527 parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4528 int itype;
4529 char *pattern;
4530 int patspec, quoted;
4531 {
4532 char *ret;
4533 WORD_LIST *list;
4534
4535 list = list_rest_of_args ();
4536 if (list == 0)
4537 return ((char *)NULL);
4538 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4539 dispose_words (list);
4540 return (ret);
4541 }
4542
4543 #if defined (ARRAY_VARS)
4544 static char *
4545 array_remove_pattern (var, pattern, patspec, varname, quoted)
4546 SHELL_VAR *var;
4547 char *pattern;
4548 int patspec;
4549 char *varname; /* so we can figure out how it's indexed */
4550 int quoted;
4551 {
4552 ARRAY *a;
4553 HASH_TABLE *h;
4554 int itype;
4555 char *ret;
4556 WORD_LIST *list;
4557 SHELL_VAR *v;
4558
4559 /* compute itype from varname here */
4560 v = array_variable_part (varname, &ret, 0);
4561
4562 /* XXX */
4563 if (v && invisible_p (var))
4564 return ((char *)NULL);
4565
4566 itype = ret[0];
4567
4568 a = (v && array_p (v)) ? array_cell (v) : 0;
4569 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4570
4571 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4572 if (list == 0)
4573 return ((char *)NULL);
4574 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4575 dispose_words (list);
4576
4577 return ret;
4578 }
4579 #endif /* ARRAY_VARS */
4580
4581 static char *
4582 parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4583 char *varname, *value;
4584 int ind;
4585 char *patstr;
4586 int rtype, quoted, flags;
4587 {
4588 int vtype, patspec, starsub;
4589 char *temp1, *val, *pattern;
4590 SHELL_VAR *v;
4591
4592 if (value == 0)
4593 return ((char *)NULL);
4594
4595 this_command_name = varname;
4596
4597 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4598 if (vtype == -1)
4599 return ((char *)NULL);
4600
4601 starsub = vtype & VT_STARSUB;
4602 vtype &= ~VT_STARSUB;
4603
4604 patspec = getpatspec (rtype, patstr);
4605 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4606 patstr++;
4607
4608 /* Need to pass getpattern newly-allocated memory in case of expansion --
4609 the expansion code will free the passed string on an error. */
4610 temp1 = savestring (patstr);
4611 pattern = getpattern (temp1, quoted, 1);
4612 free (temp1);
4613
4614 temp1 = (char *)NULL; /* shut up gcc */
4615 switch (vtype)
4616 {
4617 case VT_VARIABLE:
4618 case VT_ARRAYMEMBER:
4619 temp1 = remove_pattern (val, pattern, patspec);
4620 if (vtype == VT_VARIABLE)
4621 FREE (val);
4622 if (temp1)
4623 {
4624 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
4625 ? quote_string (temp1)
4626 : quote_escapes (temp1);
4627 free (temp1);
4628 temp1 = val;
4629 }
4630 break;
4631 #if defined (ARRAY_VARS)
4632 case VT_ARRAYVAR:
4633 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
4634 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4635 {
4636 val = quote_escapes (temp1);
4637 free (temp1);
4638 temp1 = val;
4639 }
4640 break;
4641 #endif
4642 case VT_POSPARMS:
4643 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
4644 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4645 {
4646 val = quote_escapes (temp1);
4647 free (temp1);
4648 temp1 = val;
4649 }
4650 break;
4651 }
4652
4653 FREE (pattern);
4654 return temp1;
4655 }
4656
4657 /*******************************************
4658 * *
4659 * Functions to expand WORD_DESCs *
4660 * *
4661 *******************************************/
4662
4663 /* Expand WORD, performing word splitting on the result. This does
4664 parameter expansion, command substitution, arithmetic expansion,
4665 word splitting, and quote removal. */
4666
4667 WORD_LIST *
4668 expand_word (word, quoted)
4669 WORD_DESC *word;
4670 int quoted;
4671 {
4672 WORD_LIST *result, *tresult;
4673
4674 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4675 result = word_list_split (tresult);
4676 dispose_words (tresult);
4677 return (result ? dequote_list (result) : result);
4678 }
4679
4680 /* Expand WORD, but do not perform word splitting on the result. This
4681 does parameter expansion, command substitution, arithmetic expansion,
4682 and quote removal. */
4683 WORD_LIST *
4684 expand_word_unsplit (word, quoted)
4685 WORD_DESC *word;
4686 int quoted;
4687 {
4688 WORD_LIST *result;
4689
4690 expand_no_split_dollar_star = 1;
4691 #if defined (HANDLE_MULTIBYTE)
4692 if (ifs_firstc[0] == 0)
4693 #else
4694 if (ifs_firstc == 0)
4695 #endif
4696 word->flags |= W_NOSPLIT;
4697 word->flags |= W_NOSPLIT2;
4698 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4699 expand_no_split_dollar_star = 0;
4700
4701 return (result ? dequote_list (result) : result);
4702 }
4703
4704 /* Perform shell expansions on WORD, but do not perform word splitting or
4705 quote removal on the result. Virtually identical to expand_word_unsplit;
4706 could be combined if implementations don't diverge. */
4707 WORD_LIST *
4708 expand_word_leave_quoted (word, quoted)
4709 WORD_DESC *word;
4710 int quoted;
4711 {
4712 WORD_LIST *result;
4713
4714 expand_no_split_dollar_star = 1;
4715 #if defined (HANDLE_MULTIBYTE)
4716 if (ifs_firstc[0] == 0)
4717 #else
4718 if (ifs_firstc == 0)
4719 #endif
4720 word->flags |= W_NOSPLIT;
4721 word->flags |= W_NOSPLIT2;
4722 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4723 expand_no_split_dollar_star = 0;
4724
4725 return result;
4726 }
4727
4728 #if defined (PROCESS_SUBSTITUTION)
4729
4730 /*****************************************************************/
4731 /* */
4732 /* Hacking Process Substitution */
4733 /* */
4734 /*****************************************************************/
4735
4736 #if !defined (HAVE_DEV_FD)
4737 /* Named pipes must be removed explicitly with `unlink'. This keeps a list
4738 of FIFOs the shell has open. unlink_fifo_list will walk the list and
4739 unlink all of them. add_fifo_list adds the name of an open FIFO to the
4740 list. NFIFO is a count of the number of FIFOs in the list. */
4741 #define FIFO_INCR 20
4742
4743 struct temp_fifo {
4744 char *file;
4745 pid_t proc;
4746 };
4747
4748 static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
4749 static int nfifo;
4750 static int fifo_list_size;
4751
4752 char *
4753 copy_fifo_list (sizep)
4754 int *sizep;
4755 {
4756 if (sizep)
4757 *sizep = 0;
4758 return (char *)NULL;
4759 }
4760
4761 static void
4762 add_fifo_list (pathname)
4763 char *pathname;
4764 {
4765 if (nfifo >= fifo_list_size - 1)
4766 {
4767 fifo_list_size += FIFO_INCR;
4768 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
4769 fifo_list_size * sizeof (struct temp_fifo));
4770 }
4771
4772 fifo_list[nfifo].file = savestring (pathname);
4773 nfifo++;
4774 }
4775
4776 void
4777 unlink_fifo (i)
4778 int i;
4779 {
4780 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4781 {
4782 unlink (fifo_list[i].file);
4783 free (fifo_list[i].file);
4784 fifo_list[i].file = (char *)NULL;
4785 fifo_list[i].proc = -1;
4786 }
4787 }
4788
4789 void
4790 unlink_fifo_list ()
4791 {
4792 int saved, i, j;
4793
4794 if (nfifo == 0)
4795 return;
4796
4797 for (i = saved = 0; i < nfifo; i++)
4798 {
4799 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4800 {
4801 unlink (fifo_list[i].file);
4802 free (fifo_list[i].file);
4803 fifo_list[i].file = (char *)NULL;
4804 fifo_list[i].proc = -1;
4805 }
4806 else
4807 saved++;
4808 }
4809
4810 /* If we didn't remove some of the FIFOs, compact the list. */
4811 if (saved)
4812 {
4813 for (i = j = 0; i < nfifo; i++)
4814 if (fifo_list[i].file)
4815 {
4816 fifo_list[j].file = fifo_list[i].file;
4817 fifo_list[j].proc = fifo_list[i].proc;
4818 j++;
4819 }
4820 nfifo = j;
4821 }
4822 else
4823 nfifo = 0;
4824 }
4825
4826 /* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
4827 from some point in the past, and close all open FIFOs in fifo_list
4828 that are not marked as active in LIST. If LIST is NULL, close
4829 everything in fifo_list. LSIZE is the number of elements in LIST, in
4830 case it's larger than fifo_list_size (size of fifo_list). */
4831 void
4832 close_new_fifos (list, lsize)
4833 char *list;
4834 int lsize;
4835 {
4836 int i;
4837
4838 if (list == 0)
4839 {
4840 unlink_fifo_list ();
4841 return;
4842 }
4843
4844 for (i = 0; i < lsize; i++)
4845 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
4846 unlink_fifo (i);
4847
4848 for (i = lsize; i < fifo_list_size; i++)
4849 unlink_fifo (i);
4850 }
4851
4852 int
4853 fifos_pending ()
4854 {
4855 return nfifo;
4856 }
4857
4858 int
4859 num_fifos ()
4860 {
4861 return nfifo;
4862 }
4863
4864 static char *
4865 make_named_pipe ()
4866 {
4867 char *tname;
4868
4869 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
4870 if (mkfifo (tname, 0600) < 0)
4871 {
4872 free (tname);
4873 return ((char *)NULL);
4874 }
4875
4876 add_fifo_list (tname);
4877 return (tname);
4878 }
4879
4880 #else /* HAVE_DEV_FD */
4881
4882 /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
4883 has open to children. NFDS is a count of the number of bits currently
4884 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
4885 of open files. */
4886 static char *dev_fd_list = (char *)NULL;
4887 static int nfds;
4888 static int totfds; /* The highest possible number of open files. */
4889
4890 char *
4891 copy_fifo_list (sizep)
4892 int *sizep;
4893 {
4894 char *ret;
4895
4896 if (nfds == 0 || totfds == 0)
4897 {
4898 if (sizep)
4899 *sizep = 0;
4900 return (char *)NULL;
4901 }
4902
4903 if (sizep)
4904 *sizep = totfds;
4905 ret = (char *)xmalloc (totfds);
4906 return (memcpy (ret, dev_fd_list, totfds));
4907 }
4908
4909 static void
4910 add_fifo_list (fd)
4911 int fd;
4912 {
4913 if (dev_fd_list == 0 || fd >= totfds)
4914 {
4915 int ofds;
4916
4917 ofds = totfds;
4918 totfds = getdtablesize ();
4919 if (totfds < 0 || totfds > 256)
4920 totfds = 256;
4921 if (fd >= totfds)
4922 totfds = fd + 2;
4923
4924 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
4925 memset (dev_fd_list + ofds, '\0', totfds - ofds);
4926 }
4927
4928 dev_fd_list[fd] = 1;
4929 nfds++;
4930 }
4931
4932 int
4933 fifos_pending ()
4934 {
4935 return 0; /* used for cleanup; not needed with /dev/fd */
4936 }
4937
4938 int
4939 num_fifos ()
4940 {
4941 return nfds;
4942 }
4943
4944 void
4945 unlink_fifo (fd)
4946 int fd;
4947 {
4948 if (dev_fd_list[fd])
4949 {
4950 close (fd);
4951 dev_fd_list[fd] = 0;
4952 nfds--;
4953 }
4954 }
4955
4956 void
4957 unlink_fifo_list ()
4958 {
4959 register int i;
4960
4961 if (nfds == 0)
4962 return;
4963
4964 for (i = 0; nfds && i < totfds; i++)
4965 unlink_fifo (i);
4966
4967 nfds = 0;
4968 }
4969
4970 /* Take LIST, which is a snapshot copy of dev_fd_list from some point in
4971 the past, and close all open fds in dev_fd_list that are not marked
4972 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
4973 LSIZE is the number of elements in LIST, in case it's larger than
4974 totfds (size of dev_fd_list). */
4975 void
4976 close_new_fifos (list, lsize)
4977 char *list;
4978 int lsize;
4979 {
4980 int i;
4981
4982 if (list == 0)
4983 {
4984 unlink_fifo_list ();
4985 return;
4986 }
4987
4988 for (i = 0; i < lsize; i++)
4989 if (list[i] == 0 && i < totfds && dev_fd_list[i])
4990 unlink_fifo (i);
4991
4992 for (i = lsize; i < totfds; i++)
4993 unlink_fifo (i);
4994 }
4995
4996 #if defined (NOTDEF)
4997 print_dev_fd_list ()
4998 {
4999 register int i;
5000
5001 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
5002 fflush (stderr);
5003
5004 for (i = 0; i < totfds; i++)
5005 {
5006 if (dev_fd_list[i])
5007 fprintf (stderr, " %d", i);
5008 }
5009 fprintf (stderr, "\n");
5010 }
5011 #endif /* NOTDEF */
5012
5013 static char *
5014 make_dev_fd_filename (fd)
5015 int fd;
5016 {
5017 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
5018
5019 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
5020
5021 strcpy (ret, DEV_FD_PREFIX);
5022 p = inttostr (fd, intbuf, sizeof (intbuf));
5023 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5024
5025 add_fifo_list (fd);
5026 return (ret);
5027 }
5028
5029 #endif /* HAVE_DEV_FD */
5030
5031 /* Return a filename that will open a connection to the process defined by
5032 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5033 a filename in /dev/fd corresponding to a descriptor that is one of the
5034 ends of the pipe. If not defined, we use named pipes on systems that have
5035 them. Systems without /dev/fd and named pipes are out of luck.
5036
5037 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5038 use the read end of the pipe and dup that file descriptor to fd 0 in
5039 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5040 writing or use the write end of the pipe in the child, and dup that
5041 file descriptor to fd 1 in the child. The parent does the opposite. */
5042
5043 static char *
5044 process_substitute (string, open_for_read_in_child)
5045 char *string;
5046 int open_for_read_in_child;
5047 {
5048 char *pathname;
5049 int fd, result;
5050 pid_t old_pid, pid;
5051 #if defined (HAVE_DEV_FD)
5052 int parent_pipe_fd, child_pipe_fd;
5053 int fildes[2];
5054 #endif /* HAVE_DEV_FD */
5055 #if defined (JOB_CONTROL)
5056 pid_t old_pipeline_pgrp;
5057 #endif
5058
5059 if (!string || !*string || wordexp_only)
5060 return ((char *)NULL);
5061
5062 #if !defined (HAVE_DEV_FD)
5063 pathname = make_named_pipe ();
5064 #else /* HAVE_DEV_FD */
5065 if (pipe (fildes) < 0)
5066 {
5067 sys_error (_("cannot make pipe for process substitution"));
5068 return ((char *)NULL);
5069 }
5070 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5071 the pipe in the parent, otherwise the read end. */
5072 parent_pipe_fd = fildes[open_for_read_in_child];
5073 child_pipe_fd = fildes[1 - open_for_read_in_child];
5074 /* Move the parent end of the pipe to some high file descriptor, to
5075 avoid clashes with FDs used by the script. */
5076 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5077
5078 pathname = make_dev_fd_filename (parent_pipe_fd);
5079 #endif /* HAVE_DEV_FD */
5080
5081 if (pathname == 0)
5082 {
5083 sys_error (_("cannot make pipe for process substitution"));
5084 return ((char *)NULL);
5085 }
5086
5087 old_pid = last_made_pid;
5088
5089 #if defined (JOB_CONTROL)
5090 old_pipeline_pgrp = pipeline_pgrp;
5091 pipeline_pgrp = shell_pgrp;
5092 save_pipeline (1);
5093 #endif /* JOB_CONTROL */
5094
5095 pid = make_child ((char *)NULL, 1);
5096 if (pid == 0)
5097 {
5098 reset_terminating_signals (); /* XXX */
5099 free_pushed_string_input ();
5100 /* Cancel traps, in trap.c. */
5101 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
5102 setup_async_signals ();
5103 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
5104 }
5105
5106 #if defined (JOB_CONTROL)
5107 set_sigchld_handler ();
5108 stop_making_children ();
5109 /* XXX - should we only do this in the parent? (as in command subst) */
5110 pipeline_pgrp = old_pipeline_pgrp;
5111 #endif /* JOB_CONTROL */
5112
5113 if (pid < 0)
5114 {
5115 sys_error (_("cannot make child for process substitution"));
5116 free (pathname);
5117 #if defined (HAVE_DEV_FD)
5118 close (parent_pipe_fd);
5119 close (child_pipe_fd);
5120 #endif /* HAVE_DEV_FD */
5121 return ((char *)NULL);
5122 }
5123
5124 if (pid > 0)
5125 {
5126 #if defined (JOB_CONTROL)
5127 restore_pipeline (1);
5128 #endif
5129
5130 #if !defined (HAVE_DEV_FD)
5131 fifo_list[nfifo-1].proc = pid;
5132 #endif
5133
5134 last_made_pid = old_pid;
5135
5136 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5137 close_pgrp_pipe ();
5138 #endif /* JOB_CONTROL && PGRP_PIPE */
5139
5140 #if defined (HAVE_DEV_FD)
5141 close (child_pipe_fd);
5142 #endif /* HAVE_DEV_FD */
5143
5144 return (pathname);
5145 }
5146
5147 set_sigint_handler ();
5148
5149 #if defined (JOB_CONTROL)
5150 set_job_control (0);
5151 #endif /* JOB_CONTROL */
5152
5153 #if !defined (HAVE_DEV_FD)
5154 /* Open the named pipe in the child. */
5155 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
5156 if (fd < 0)
5157 {
5158 /* Two separate strings for ease of translation. */
5159 if (open_for_read_in_child)
5160 sys_error (_("cannot open named pipe %s for reading"), pathname);
5161 else
5162 sys_error (_("cannot open named pipe %s for writing"), pathname);
5163
5164 exit (127);
5165 }
5166 if (open_for_read_in_child)
5167 {
5168 if (sh_unset_nodelay_mode (fd) < 0)
5169 {
5170 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
5171 exit (127);
5172 }
5173 }
5174 #else /* HAVE_DEV_FD */
5175 fd = child_pipe_fd;
5176 #endif /* HAVE_DEV_FD */
5177
5178 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5179 {
5180 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
5181 open_for_read_in_child ? 0 : 1);
5182 exit (127);
5183 }
5184
5185 if (fd != (open_for_read_in_child ? 0 : 1))
5186 close (fd);
5187
5188 /* Need to close any files that this process has open to pipes inherited
5189 from its parent. */
5190 if (current_fds_to_close)
5191 {
5192 close_fd_bitmap (current_fds_to_close);
5193 current_fds_to_close = (struct fd_bitmap *)NULL;
5194 }
5195
5196 #if defined (HAVE_DEV_FD)
5197 /* Make sure we close the parent's end of the pipe and clear the slot
5198 in the fd list so it is not closed later, if reallocated by, for
5199 instance, pipe(2). */
5200 close (parent_pipe_fd);
5201 dev_fd_list[parent_pipe_fd] = 0;
5202 #endif /* HAVE_DEV_FD */
5203
5204 /* subshells shouldn't have this flag, which controls using the temporary
5205 environment for variable lookups. */
5206 expanding_redir = 0;
5207
5208 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
5209
5210 #if !defined (HAVE_DEV_FD)
5211 /* Make sure we close the named pipe in the child before we exit. */
5212 close (open_for_read_in_child ? 0 : 1);
5213 #endif /* !HAVE_DEV_FD */
5214
5215 last_command_exit_value = result;
5216 result = run_exit_trap ();
5217 exit (result);
5218 /*NOTREACHED*/
5219 }
5220 #endif /* PROCESS_SUBSTITUTION */
5221
5222 /***********************************/
5223 /* */
5224 /* Command Substitution */
5225 /* */
5226 /***********************************/
5227
5228 static char *
5229 read_comsub (fd, quoted, rflag)
5230 int fd, quoted;
5231 int *rflag;
5232 {
5233 char *istring, buf[128], *bufp, *s;
5234 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
5235 ssize_t bufn;
5236
5237 istring = (char *)NULL;
5238 istring_index = istring_size = bufn = tflag = 0;
5239
5240 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5241 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
5242
5243 /* Read the output of the command through the pipe. This may need to be
5244 changed to understand multibyte characters in the future. */
5245 while (1)
5246 {
5247 if (fd < 0)
5248 break;
5249 if (--bufn <= 0)
5250 {
5251 bufn = zread (fd, buf, sizeof (buf));
5252 if (bufn <= 0)
5253 break;
5254 bufp = buf;
5255 }
5256 c = *bufp++;
5257
5258 if (c == 0)
5259 {
5260 #if 0
5261 internal_warning ("read_comsub: ignored null byte in input");
5262 #endif
5263 continue;
5264 }
5265
5266 /* Add the character to ISTRING, possibly after resizing it. */
5267 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5268
5269 /* This is essentially quote_string inline */
5270 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5271 istring[istring_index++] = CTLESC;
5272 /* Escape CTLESC and CTLNUL in the output to protect those characters
5273 from the rest of the word expansions (word splitting and globbing.)
5274 This is essentially quote_escapes inline. */
5275 else if (skip_ctlesc == 0 && c == CTLESC)
5276 {
5277 tflag |= W_HASCTLESC;
5278 istring[istring_index++] = CTLESC;
5279 }
5280 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
5281 istring[istring_index++] = CTLESC;
5282
5283 istring[istring_index++] = c;
5284
5285 #if 0
5286 #if defined (__CYGWIN__)
5287 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5288 {
5289 istring_index--;
5290 istring[istring_index - 1] = '\n';
5291 }
5292 #endif
5293 #endif
5294 }
5295
5296 if (istring)
5297 istring[istring_index] = '\0';
5298
5299 /* If we read no output, just return now and save ourselves some
5300 trouble. */
5301 if (istring_index == 0)
5302 {
5303 FREE (istring);
5304 if (rflag)
5305 *rflag = tflag;
5306 return (char *)NULL;
5307 }
5308
5309 /* Strip trailing newlines from the output of the command. */
5310 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5311 {
5312 while (istring_index > 0)
5313 {
5314 if (istring[istring_index - 1] == '\n')
5315 {
5316 --istring_index;
5317
5318 /* If the newline was quoted, remove the quoting char. */
5319 if (istring[istring_index - 1] == CTLESC)
5320 --istring_index;
5321 }
5322 else
5323 break;
5324 }
5325 istring[istring_index] = '\0';
5326 }
5327 else
5328 strip_trailing (istring, istring_index - 1, 1);
5329
5330 if (rflag)
5331 *rflag = tflag;
5332 return istring;
5333 }
5334
5335 /* Perform command substitution on STRING. This returns a WORD_DESC * with the
5336 contained string possibly quoted. */
5337 WORD_DESC *
5338 command_substitute (string, quoted)
5339 char *string;
5340 int quoted;
5341 {
5342 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
5343 char *istring;
5344 int result, fildes[2], function_value, pflags, rc, tflag;
5345 WORD_DESC *ret;
5346
5347 istring = (char *)NULL;
5348
5349 /* Don't fork () if there is no need to. In the case of no command to
5350 run, just return NULL. */
5351 if (!string || !*string || (string[0] == '\n' && !string[1]))
5352 return ((WORD_DESC *)NULL);
5353
5354 if (wordexp_only && read_but_dont_execute)
5355 {
5356 last_command_exit_value = EX_WEXPCOMSUB;
5357 jump_to_top_level (EXITPROG);
5358 }
5359
5360 /* We're making the assumption here that the command substitution will
5361 eventually run a command from the file system. Since we'll run
5362 maybe_make_export_env in this subshell before executing that command,
5363 the parent shell and any other shells it starts will have to remake
5364 the environment. If we make it before we fork, other shells won't
5365 have to. Don't bother if we have any temporary variable assignments,
5366 though, because the export environment will be remade after this
5367 command completes anyway, but do it if all the words to be expanded
5368 are variable assignments. */
5369 if (subst_assign_varlist == 0 || garglist == 0)
5370 maybe_make_export_env (); /* XXX */
5371
5372 /* Flags to pass to parse_and_execute() */
5373 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
5374
5375 /* Pipe the output of executing STRING into the current shell. */
5376 if (pipe (fildes) < 0)
5377 {
5378 sys_error (_("cannot make pipe for command substitution"));
5379 goto error_exit;
5380 }
5381
5382 old_pid = last_made_pid;
5383 #if defined (JOB_CONTROL)
5384 old_pipeline_pgrp = pipeline_pgrp;
5385 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
5386 if ((subshell_environment & SUBSHELL_PIPE) == 0)
5387 pipeline_pgrp = shell_pgrp;
5388 cleanup_the_pipeline ();
5389 #endif /* JOB_CONTROL */
5390
5391 old_async_pid = last_asynchronous_pid;
5392 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
5393 last_asynchronous_pid = old_async_pid;
5394
5395 if (pid == 0)
5396 {
5397 /* Reset the signal handlers in the child, but don't free the
5398 trap strings. Set a flag noting that we have to free the
5399 trap strings if we run trap to change a signal disposition. */
5400 reset_signal_handlers ();
5401 subshell_environment |= SUBSHELL_RESETTRAP;
5402 }
5403
5404 #if defined (JOB_CONTROL)
5405 /* XXX DO THIS ONLY IN PARENT ? XXX */
5406 set_sigchld_handler ();
5407 stop_making_children ();
5408 if (pid != 0)
5409 pipeline_pgrp = old_pipeline_pgrp;
5410 #else
5411 stop_making_children ();
5412 #endif /* JOB_CONTROL */
5413
5414 if (pid < 0)
5415 {
5416 sys_error (_("cannot make child for command substitution"));
5417 error_exit:
5418
5419 last_made_pid = old_pid;
5420
5421 FREE (istring);
5422 close (fildes[0]);
5423 close (fildes[1]);
5424 return ((WORD_DESC *)NULL);
5425 }
5426
5427 if (pid == 0)
5428 {
5429 set_sigint_handler (); /* XXX */
5430
5431 free_pushed_string_input ();
5432
5433 if (dup2 (fildes[1], 1) < 0)
5434 {
5435 sys_error (_("command_substitute: cannot duplicate pipe as fd 1"));
5436 exit (EXECUTION_FAILURE);
5437 }
5438
5439 /* If standard output is closed in the parent shell
5440 (such as after `exec >&-'), file descriptor 1 will be
5441 the lowest available file descriptor, and end up in
5442 fildes[0]. This can happen for stdin and stderr as well,
5443 but stdout is more important -- it will cause no output
5444 to be generated from this command. */
5445 if ((fildes[1] != fileno (stdin)) &&
5446 (fildes[1] != fileno (stdout)) &&
5447 (fildes[1] != fileno (stderr)))
5448 close (fildes[1]);
5449
5450 if ((fildes[0] != fileno (stdin)) &&
5451 (fildes[0] != fileno (stdout)) &&
5452 (fildes[0] != fileno (stderr)))
5453 close (fildes[0]);
5454
5455 #ifdef __CYGWIN__
5456 /* Let stdio know the fd may have changed from text to binary mode, and
5457 make sure to preserve stdout line buffering. */
5458 freopen (NULL, "w", stdout);
5459 sh_setlinebuf (stdout);
5460 #endif /* __CYGWIN__ */
5461
5462 /* The currently executing shell is not interactive. */
5463 interactive = 0;
5464
5465 /* This is a subshell environment. */
5466 subshell_environment |= SUBSHELL_COMSUB;
5467
5468 /* When not in POSIX mode, command substitution does not inherit
5469 the -e flag. */
5470 if (posixly_correct == 0)
5471 {
5472 builtin_ignoring_errexit = 0;
5473 change_flag ('e', FLAG_OFF);
5474 set_shellopts ();
5475 }
5476
5477 remove_quoted_escapes (string);
5478
5479 startup_state = 2; /* see if we can avoid a fork */
5480 /* Give command substitution a place to jump back to on failure,
5481 so we don't go back up to main (). */
5482 result = setjmp_nosigs (top_level);
5483
5484 /* If we're running a command substitution inside a shell function,
5485 trap `return' so we don't return from the function in the subshell
5486 and go off to never-never land. */
5487 if (result == 0 && return_catch_flag)
5488 function_value = setjmp_nosigs (return_catch);
5489 else
5490 function_value = 0;
5491
5492 if (result == ERREXIT)
5493 rc = last_command_exit_value;
5494 else if (result == EXITPROG)
5495 rc = last_command_exit_value;
5496 else if (result)
5497 rc = EXECUTION_FAILURE;
5498 else if (function_value)
5499 rc = return_catch_value;
5500 else
5501 {
5502 subshell_level++;
5503 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
5504 subshell_level--;
5505 }
5506
5507 last_command_exit_value = rc;
5508 rc = run_exit_trap ();
5509 #if defined (PROCESS_SUBSTITUTION)
5510 unlink_fifo_list ();
5511 #endif
5512 exit (rc);
5513 }
5514 else
5515 {
5516 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5517 close_pgrp_pipe ();
5518 #endif /* JOB_CONTROL && PGRP_PIPE */
5519
5520 close (fildes[1]);
5521
5522 tflag = 0;
5523 istring = read_comsub (fildes[0], quoted, &tflag);
5524
5525 close (fildes[0]);
5526
5527 current_command_subst_pid = pid;
5528 last_command_exit_value = wait_for (pid);
5529 last_command_subst_pid = pid;
5530 last_made_pid = old_pid;
5531
5532 #if defined (JOB_CONTROL)
5533 /* If last_command_exit_value > 128, then the substituted command
5534 was terminated by a signal. If that signal was SIGINT, then send
5535 SIGINT to ourselves. This will break out of loops, for instance. */
5536 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
5537 kill (getpid (), SIGINT);
5538
5539 /* wait_for gives the terminal back to shell_pgrp. If some other
5540 process group should have it, give it away to that group here.
5541 pipeline_pgrp is non-zero only while we are constructing a
5542 pipeline, so what we are concerned about is whether or not that
5543 pipeline was started in the background. A pipeline started in
5544 the background should never get the tty back here. */
5545 if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0)
5546 give_terminal_to (pipeline_pgrp, 0);
5547 #endif /* JOB_CONTROL */
5548
5549 ret = alloc_word_desc ();
5550 ret->word = istring;
5551 ret->flags = tflag;
5552
5553 return ret;
5554 }
5555 }
5556
5557 /********************************************************
5558 * *
5559 * Utility functions for parameter expansion *
5560 * *
5561 ********************************************************/
5562
5563 #if defined (ARRAY_VARS)
5564
5565 static arrayind_t
5566 array_length_reference (s)
5567 char *s;
5568 {
5569 int len;
5570 arrayind_t ind;
5571 char *akey;
5572 char *t, c;
5573 ARRAY *array;
5574 HASH_TABLE *h;
5575 SHELL_VAR *var;
5576
5577 var = array_variable_part (s, &t, &len);
5578
5579 /* If unbound variables should generate an error, report one and return
5580 failure. */
5581 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
5582 {
5583 c = *--t;
5584 *t = '\0';
5585 last_command_exit_value = EXECUTION_FAILURE;
5586 err_unboundvar (s);
5587 *t = c;
5588 return (-1);
5589 }
5590 else if (var == 0 || invisible_p (var))
5591 return 0;
5592
5593 /* We support a couple of expansions for variables that are not arrays.
5594 We'll return the length of the value for v[0], and 1 for v[@] or
5595 v[*]. Return 0 for everything else. */
5596
5597 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
5598 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
5599
5600 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
5601 {
5602 if (assoc_p (var))
5603 return (h ? assoc_num_elements (h) : 0);
5604 else if (array_p (var))
5605 return (array ? array_num_elements (array) : 0);
5606 else
5607 return (var_isset (var) ? 1 : 0);
5608 }
5609
5610 if (assoc_p (var))
5611 {
5612 t[len - 1] = '\0';
5613 akey = expand_assignment_string_to_string (t, 0); /* [ */
5614 t[len - 1] = ']';
5615 if (akey == 0 || *akey == 0)
5616 {
5617 err_badarraysub (t);
5618 FREE (akey);
5619 return (-1);
5620 }
5621 t = assoc_reference (assoc_cell (var), akey);
5622 free (akey);
5623 }
5624 else
5625 {
5626 ind = array_expand_index (var, t, len);
5627 /* negative subscripts to indexed arrays count back from end */
5628 if (var && array_p (var) && ind < 0)
5629 ind = array_max_index (array_cell (var)) + 1 + ind;
5630 if (ind < 0)
5631 {
5632 err_badarraysub (t);
5633 return (-1);
5634 }
5635 if (array_p (var))
5636 t = array_reference (array, ind);
5637 else
5638 t = (ind == 0) ? value_cell (var) : (char *)NULL;
5639 }
5640
5641 len = MB_STRLEN (t);
5642 return (len);
5643 }
5644 #endif /* ARRAY_VARS */
5645
5646 static int
5647 valid_brace_expansion_word (name, var_is_special)
5648 char *name;
5649 int var_is_special;
5650 {
5651 if (DIGIT (*name) && all_digits (name))
5652 return 1;
5653 else if (var_is_special)
5654 return 1;
5655 #if defined (ARRAY_VARS)
5656 else if (valid_array_reference (name))
5657 return 1;
5658 #endif /* ARRAY_VARS */
5659 else if (legal_identifier (name))
5660 return 1;
5661 else
5662 return 0;
5663 }
5664
5665 static int
5666 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
5667 char *name;
5668 int quoted;
5669 int *quoted_dollar_atp, *contains_dollar_at;
5670 {
5671 char *temp1;
5672
5673 if (name == 0)
5674 {
5675 if (quoted_dollar_atp)
5676 *quoted_dollar_atp = 0;
5677 if (contains_dollar_at)
5678 *contains_dollar_at = 0;
5679 return 0;
5680 }
5681
5682 /* check for $@ and $* */
5683 if (name[0] == '@' && name[1] == 0)
5684 {
5685 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5686 *quoted_dollar_atp = 1;
5687 if (contains_dollar_at)
5688 *contains_dollar_at = 1;
5689 return 1;
5690 }
5691 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
5692 {
5693 if (contains_dollar_at)
5694 *contains_dollar_at = 1;
5695 return 1;
5696 }
5697
5698 /* Now check for ${array[@]} and ${array[*]} */
5699 #if defined (ARRAY_VARS)
5700 else if (valid_array_reference (name))
5701 {
5702 temp1 = mbschr (name, '[');
5703 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
5704 {
5705 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5706 *quoted_dollar_atp = 1;
5707 if (contains_dollar_at)
5708 *contains_dollar_at = 1;
5709 return 1;
5710 } /* [ */
5711 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
5712 which should result in separate words even when IFS is unset. */
5713 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
5714 {
5715 if (contains_dollar_at)
5716 *contains_dollar_at = 1;
5717 return 1;
5718 }
5719 }
5720 #endif
5721 return 0;
5722 }
5723
5724 /* Parameter expand NAME, and return a new string which is the expansion,
5725 or NULL if there was no expansion.
5726 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
5727 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
5728 NAME was found inside of a double-quoted expression. */
5729 static WORD_DESC *
5730 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
5731 char *name;
5732 int var_is_special, quoted, pflags;
5733 arrayind_t *indp;
5734 {
5735 WORD_DESC *ret;
5736 char *temp, *tt;
5737 intmax_t arg_index;
5738 SHELL_VAR *var;
5739 int atype, rflags;
5740 arrayind_t ind;
5741
5742 ret = 0;
5743 temp = 0;
5744 rflags = 0;
5745
5746 if (indp)
5747 *indp = INTMAX_MIN;
5748
5749 /* Handle multiple digit arguments, as in ${11}. */
5750 if (legal_number (name, &arg_index))
5751 {
5752 tt = get_dollar_var_value (arg_index);
5753 if (tt)
5754 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5755 ? quote_string (tt)
5756 : quote_escapes (tt);
5757 else
5758 temp = (char *)NULL;
5759 FREE (tt);
5760 }
5761 else if (var_is_special) /* ${@} */
5762 {
5763 int sindex;
5764 tt = (char *)xmalloc (2 + strlen (name));
5765 tt[sindex = 0] = '$';
5766 strcpy (tt + 1, name);
5767
5768 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
5769 (int *)NULL, (int *)NULL, pflags);
5770 free (tt);
5771 }
5772 #if defined (ARRAY_VARS)
5773 else if (valid_array_reference (name))
5774 {
5775 expand_arrayref:
5776 /* XXX - does this leak if name[@] or name[*]? */
5777 if (pflags & PF_ASSIGNRHS)
5778 {
5779 temp = array_variable_name (name, &tt, (int *)0);
5780 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
5781 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, 0, &atype, &ind);
5782 else
5783 temp = array_value (name, quoted, 0, &atype, &ind);
5784 }
5785 else
5786 temp = array_value (name, quoted, 0, &atype, &ind);
5787 if (atype == 0 && temp)
5788 {
5789 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5790 ? quote_string (temp)
5791 : quote_escapes (temp);
5792 rflags |= W_ARRAYIND;
5793 if (indp)
5794 *indp = ind;
5795 }
5796 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5797 rflags |= W_HASQUOTEDNULL;
5798 }
5799 #endif
5800 else if (var = find_variable (name))
5801 {
5802 if (var_isset (var) && invisible_p (var) == 0)
5803 {
5804 #if defined (ARRAY_VARS)
5805 if (assoc_p (var))
5806 temp = assoc_reference (assoc_cell (var), "0");
5807 else if (array_p (var))
5808 temp = array_reference (array_cell (var), 0);
5809 else
5810 temp = value_cell (var);
5811 #else
5812 temp = value_cell (var);
5813 #endif
5814
5815 if (temp)
5816 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5817 ? quote_string (temp)
5818 : quote_escapes (temp);
5819 }
5820 else
5821 temp = (char *)NULL;
5822 }
5823 else if (var = find_variable_last_nameref (name))
5824 {
5825 temp = nameref_cell (var);
5826 #if defined (ARRAY_VARS)
5827 /* Handle expanding nameref whose value is x[n] */
5828 if (temp && *temp && valid_array_reference (temp))
5829 {
5830 name = temp;
5831 goto expand_arrayref;
5832 }
5833 else
5834 #endif
5835 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
5836 if (temp && *temp && legal_identifier (temp) == 0)
5837 {
5838 last_command_exit_value = EXECUTION_FAILURE;
5839 report_error (_("%s: invalid variable name for name reference"), temp);
5840 temp = &expand_param_error;
5841 }
5842 else
5843 temp = (char *)NULL;
5844 }
5845 else
5846 temp = (char *)NULL;
5847
5848 if (ret == 0)
5849 {
5850 ret = alloc_word_desc ();
5851 ret->word = temp;
5852 ret->flags |= rflags;
5853 }
5854 return ret;
5855 }
5856
5857 static char *
5858 parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
5859 char *name;
5860 int var_is_special, quoted, find_nameref;
5861 {
5862 char *temp, *t;
5863 WORD_DESC *w;
5864 SHELL_VAR *v;
5865
5866 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name)) &&
5867 nameref_p (v) && (t = nameref_cell (v)) && *t)
5868 return (savestring (t));
5869
5870 /* If var_is_special == 0, and name is not an array reference, this does
5871 more expansion than necessary. It should really look up the variable's
5872 value and not try to expand it. */
5873 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
5874 t = w->word;
5875 /* Have to dequote here if necessary */
5876 if (t)
5877 {
5878 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
5879 ? dequote_string (t)
5880 : dequote_escapes (t);
5881 free (t);
5882 t = temp;
5883 }
5884 dispose_word_desc (w);
5885
5886 return t;
5887 }
5888
5889 /* Expand an indirect reference to a variable: ${!NAME} expands to the
5890 value of the variable whose name is the value of NAME. */
5891 static WORD_DESC *
5892 parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
5893 char *name;
5894 int var_is_special, quoted;
5895 int *quoted_dollar_atp, *contains_dollar_at;
5896 {
5897 char *temp, *t;
5898 WORD_DESC *w;
5899 SHELL_VAR *v;
5900
5901 /* See if it's a nameref first, behave in ksh93-compatible fashion.
5902 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
5903 bash performs an indirect lookup on foo[0] and expands the result;
5904 ksh93 expands bar[0]. We could do that here -- there are enough usable
5905 primitives to do that -- but do not at this point. */
5906 if (var_is_special == 0 && (v = find_variable_last_nameref (name)))
5907 {
5908 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
5909 {
5910 w = alloc_word_desc ();
5911 w->word = savestring (t);
5912 w->flags = 0;
5913 return w;
5914 }
5915 }
5916
5917 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
5918
5919 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
5920 if (t == 0)
5921 return (WORD_DESC *)NULL;
5922
5923 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
5924 free (t);
5925
5926 return w;
5927 }
5928
5929 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
5930 depending on the value of C, the separating character. C can be one of
5931 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
5932 between double quotes. */
5933 static WORD_DESC *
5934 parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat)
5935 char *name, *value;
5936 int c, quoted, *qdollaratp, *hasdollarat;
5937 {
5938 WORD_DESC *w;
5939 WORD_LIST *l;
5940 char *t, *t1, *temp;
5941 int hasdol;
5942
5943 /* If the entire expression is between double quotes, we want to treat
5944 the value as a double-quoted string, with the exception that we strip
5945 embedded unescaped double quotes (for sh backwards compatibility). */
5946 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
5947 {
5948 hasdol = 0;
5949 temp = string_extract_double_quoted (value, &hasdol, 1);
5950 }
5951 else
5952 temp = value;
5953
5954 w = alloc_word_desc ();
5955 hasdol = 0;
5956 /* XXX was 0 not quoted */
5957 l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL)
5958 : (WORD_LIST *)0;
5959 if (hasdollarat)
5960 *hasdollarat = hasdol || (l && l->next);
5961 if (temp != value)
5962 free (temp);
5963 if (l)
5964 {
5965 /* The expansion of TEMP returned something. We need to treat things
5966 slightly differently if HASDOL is non-zero. If we have "$@", the
5967 individual words have already been quoted. We need to turn them
5968 into a string with the words separated by the first character of
5969 $IFS without any additional quoting, so string_list_dollar_at won't
5970 do the right thing. We use string_list_dollar_star instead. */
5971 temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l);
5972
5973 /* If l->next is not null, we know that TEMP contained "$@", since that
5974 is the only expansion that creates more than one word. */
5975 if (qdollaratp && ((hasdol && quoted) || l->next))
5976 *qdollaratp = 1;
5977 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
5978 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
5979 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
5980 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5981 (which is more paranoia than anything else), we need to return the
5982 quoted null string and set the flags to indicate it. */
5983 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
5984 {
5985 w->flags |= W_HASQUOTEDNULL;
5986 }
5987 dispose_words (l);
5988 }
5989 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol)
5990 {
5991 /* The brace expansion occurred between double quotes and there was
5992 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
5993 it does not expand to anything. In this case, we want to return
5994 a quoted empty string. */
5995 temp = make_quoted_char ('\0');
5996 w->flags |= W_HASQUOTEDNULL;
5997 }
5998 else
5999 temp = (char *)NULL;
6000
6001 if (c == '-' || c == '+')
6002 {
6003 w->word = temp;
6004 return w;
6005 }
6006
6007 /* c == '=' */
6008 t = temp ? savestring (temp) : savestring ("");
6009 t1 = dequote_string (t);
6010 free (t);
6011 #if defined (ARRAY_VARS)
6012 if (valid_array_reference (name))
6013 assign_array_element (name, t1, 0);
6014 else
6015 #endif /* ARRAY_VARS */
6016 bind_variable (name, t1, 0);
6017 #if 0
6018 if (STREQ (name, "IFS") == 0)
6019 #endif
6020 stupidly_hack_special_variables (name);
6021
6022 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6023 free (temp);
6024
6025 w->word = t1;
6026 return w;
6027 }
6028
6029 /* Deal with the right hand side of a ${name:?value} expansion in the case
6030 that NAME is null or not set. If VALUE is non-null it is expanded and
6031 used as the error message to print, otherwise a standard message is
6032 printed. */
6033 static void
6034 parameter_brace_expand_error (name, value)
6035 char *name, *value;
6036 {
6037 WORD_LIST *l;
6038 char *temp;
6039
6040 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
6041 if (value && *value)
6042 {
6043 l = expand_string (value, 0);
6044 temp = string_list (l);
6045 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6046 FREE (temp);
6047 dispose_words (l);
6048 }
6049 else
6050 report_error (_("%s: parameter null or not set"), name);
6051
6052 /* Free the data we have allocated during this expansion, since we
6053 are about to longjmp out. */
6054 free (name);
6055 FREE (value);
6056 }
6057
6058 /* Return 1 if NAME is something for which parameter_brace_expand_length is
6059 OK to do. */
6060 static int
6061 valid_length_expression (name)
6062 char *name;
6063 {
6064 return (name[1] == '\0' || /* ${#} */
6065 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6066 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
6067 #if defined (ARRAY_VARS)
6068 valid_array_reference (name + 1) || /* ${#a[7]} */
6069 #endif
6070 legal_identifier (name + 1)); /* ${#PS1} */
6071 }
6072
6073 /* Handle the parameter brace expansion that requires us to return the
6074 length of a parameter. */
6075 static intmax_t
6076 parameter_brace_expand_length (name)
6077 char *name;
6078 {
6079 char *t, *newname;
6080 intmax_t number, arg_index;
6081 WORD_LIST *list;
6082 #if defined (ARRAY_VARS)
6083 SHELL_VAR *var;
6084 #endif
6085
6086 if (name[1] == '\0') /* ${#} */
6087 number = number_of_args ();
6088 else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */
6089 number = number_of_args ();
6090 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
6091 {
6092 /* Take the lengths of some of the shell's special parameters. */
6093 switch (name[1])
6094 {
6095 case '-':
6096 t = which_set_flags ();
6097 break;
6098 case '?':
6099 t = itos (last_command_exit_value);
6100 break;
6101 case '$':
6102 t = itos (dollar_dollar_pid);
6103 break;
6104 case '!':
6105 if (last_asynchronous_pid == NO_PID)
6106 t = (char *)NULL; /* XXX - error if set -u set? */
6107 else
6108 t = itos (last_asynchronous_pid);
6109 break;
6110 case '#':
6111 t = itos (number_of_args ());
6112 break;
6113 }
6114 number = STRLEN (t);
6115 FREE (t);
6116 }
6117 #if defined (ARRAY_VARS)
6118 else if (valid_array_reference (name + 1))
6119 number = array_length_reference (name + 1);
6120 #endif /* ARRAY_VARS */
6121 else
6122 {
6123 number = 0;
6124
6125 if (legal_number (name + 1, &arg_index)) /* ${#1} */
6126 {
6127 t = get_dollar_var_value (arg_index);
6128 if (t == 0 && unbound_vars_is_error)
6129 return INTMAX_MIN;
6130 number = MB_STRLEN (t);
6131 FREE (t);
6132 }
6133 #if defined (ARRAY_VARS)
6134 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
6135 {
6136 if (assoc_p (var))
6137 t = assoc_reference (assoc_cell (var), "0");
6138 else
6139 t = array_reference (array_cell (var), 0);
6140 if (t == 0 && unbound_vars_is_error)
6141 return INTMAX_MIN;
6142 number = MB_STRLEN (t);
6143 }
6144 #endif
6145 else /* ${#PS1} */
6146 {
6147 newname = savestring (name);
6148 newname[0] = '$';
6149 list = expand_string (newname, Q_DOUBLE_QUOTES);
6150 t = list ? string_list (list) : (char *)NULL;
6151 free (newname);
6152 if (list)
6153 dispose_words (list);
6154
6155 number = t ? MB_STRLEN (t) : 0;
6156 FREE (t);
6157 }
6158 }
6159
6160 return (number);
6161 }
6162
6163 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6164 so we do some ad-hoc parsing of an arithmetic expression to find
6165 the first DELIM, instead of using strchr(3). Two rules:
6166 1. If the substring contains a `(', read until closing `)'.
6167 2. If the substring contains a `?', read past one `:' for each `?'.
6168 */
6169
6170 static char *
6171 skiparith (substr, delim)
6172 char *substr;
6173 int delim;
6174 {
6175 size_t sublen;
6176 int skipcol, pcount, i;
6177 DECLARE_MBSTATE;
6178
6179 sublen = strlen (substr);
6180 i = skipcol = pcount = 0;
6181 while (substr[i])
6182 {
6183 /* Balance parens */
6184 if (substr[i] == LPAREN)
6185 {
6186 pcount++;
6187 i++;
6188 continue;
6189 }
6190 if (substr[i] == RPAREN && pcount)
6191 {
6192 pcount--;
6193 i++;
6194 continue;
6195 }
6196 if (pcount)
6197 {
6198 ADVANCE_CHAR (substr, sublen, i);
6199 continue;
6200 }
6201
6202 /* Skip one `:' for each `?' */
6203 if (substr[i] == ':' && skipcol)
6204 {
6205 skipcol--;
6206 i++;
6207 continue;
6208 }
6209 if (substr[i] == delim)
6210 break;
6211 if (substr[i] == '?')
6212 {
6213 skipcol++;
6214 i++;
6215 continue;
6216 }
6217 ADVANCE_CHAR (substr, sublen, i);
6218 }
6219
6220 return (substr + i);
6221 }
6222
6223 /* Verify and limit the start and end of the desired substring. If
6224 VTYPE == 0, a regular shell variable is being used; if it is 1,
6225 then the positional parameters are being used; if it is 2, then
6226 VALUE is really a pointer to an array variable that should be used.
6227 Return value is 1 if both values were OK, 0 if there was a problem
6228 with an invalid expression, or -1 if the values were out of range. */
6229 static int
6230 verify_substring_values (v, value, substr, vtype, e1p, e2p)
6231 SHELL_VAR *v;
6232 char *value, *substr;
6233 int vtype;
6234 intmax_t *e1p, *e2p;
6235 {
6236 char *t, *temp1, *temp2;
6237 arrayind_t len;
6238 int expok;
6239 #if defined (ARRAY_VARS)
6240 ARRAY *a;
6241 HASH_TABLE *h;
6242 #endif
6243
6244 /* duplicate behavior of strchr(3) */
6245 t = skiparith (substr, ':');
6246 if (*t && *t == ':')
6247 *t = '\0';
6248 else
6249 t = (char *)0;
6250
6251 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
6252 *e1p = evalexp (temp1, &expok);
6253 free (temp1);
6254 if (expok == 0)
6255 return (0);
6256
6257 len = -1; /* paranoia */
6258 switch (vtype)
6259 {
6260 case VT_VARIABLE:
6261 case VT_ARRAYMEMBER:
6262 len = MB_STRLEN (value);
6263 break;
6264 case VT_POSPARMS:
6265 len = number_of_args () + 1;
6266 if (*e1p == 0)
6267 len++; /* add one arg if counting from $0 */
6268 break;
6269 #if defined (ARRAY_VARS)
6270 case VT_ARRAYVAR:
6271 /* For arrays, the first value deals with array indices. Negative
6272 offsets count from one past the array's maximum index. Associative
6273 arrays treat the number of elements as the maximum index. */
6274 if (assoc_p (v))
6275 {
6276 h = assoc_cell (v);
6277 len = assoc_num_elements (h) + (*e1p < 0);
6278 }
6279 else
6280 {
6281 a = (ARRAY *)value;
6282 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
6283 }
6284 break;
6285 #endif
6286 }
6287
6288 if (len == -1) /* paranoia */
6289 return -1;
6290
6291 if (*e1p < 0) /* negative offsets count from end */
6292 *e1p += len;
6293
6294 if (*e1p > len || *e1p < 0)
6295 return (-1);
6296
6297 #if defined (ARRAY_VARS)
6298 /* For arrays, the second offset deals with the number of elements. */
6299 if (vtype == VT_ARRAYVAR)
6300 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
6301 #endif
6302
6303 if (t)
6304 {
6305 t++;
6306 temp2 = savestring (t);
6307 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
6308 free (temp2);
6309 t[-1] = ':';
6310 *e2p = evalexp (temp1, &expok);
6311 free (temp1);
6312 if (expok == 0)
6313 return (0);
6314 #if 1
6315 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
6316 #else
6317 /* bash-4.3: allow positional parameter length < 0 to count backwards
6318 from end of positional parameters */
6319 if (vtype == VT_ARRAYVAR && *e2p < 0)
6320 #endif
6321 {
6322 internal_error (_("%s: substring expression < 0"), t);
6323 return (0);
6324 }
6325 #if defined (ARRAY_VARS)
6326 /* In order to deal with sparse arrays, push the intelligence about how
6327 to deal with the number of elements desired down to the array-
6328 specific functions. */
6329 if (vtype != VT_ARRAYVAR)
6330 #endif
6331 {
6332 if (*e2p < 0)
6333 {
6334 *e2p += len;
6335 if (*e2p < 0 || *e2p < *e1p)
6336 {
6337 internal_error (_("%s: substring expression < 0"), t);
6338 return (0);
6339 }
6340 }
6341 else
6342 *e2p += *e1p; /* want E2 chars starting at E1 */
6343 if (*e2p > len)
6344 *e2p = len;
6345 }
6346 }
6347 else
6348 *e2p = len;
6349
6350 return (1);
6351 }
6352
6353 /* Return the type of variable specified by VARNAME (simple variable,
6354 positional param, or array variable). Also return the value specified
6355 by VARNAME (value of a variable or a reference to an array element).
6356 QUOTED is the standard description of quoting state, using Q_* defines.
6357 FLAGS is currently a set of flags to pass to array_value. If IND is
6358 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
6359 passed to array_value so the array index is not computed again.
6360 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
6361 characters in the value are quoted with CTLESC and takes appropriate
6362 steps. For convenience, *VALP is set to the dequoted VALUE. */
6363 static int
6364 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
6365 char *varname, *value;
6366 arrayind_t ind;
6367 int quoted, flags;
6368 SHELL_VAR **varp;
6369 char **valp;
6370 {
6371 int vtype, want_indir;
6372 char *temp, *vname;
6373 WORD_DESC *wd;
6374 #if defined (ARRAY_VARS)
6375 SHELL_VAR *v;
6376 #endif
6377 arrayind_t lind;
6378
6379 want_indir = *varname == '!' &&
6380 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
6381 || VALID_INDIR_PARAM (varname[1]));
6382 if (want_indir)
6383 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
6384 else
6385 vname = varname;
6386
6387 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
6388 vtype = (vname[0] == '@' || vname[0] == '*') && vname[1] == '\0';
6389 if (vtype == VT_POSPARMS && vname[0] == '*')
6390 vtype |= VT_STARSUB;
6391 *varp = (SHELL_VAR *)NULL;
6392
6393 #if defined (ARRAY_VARS)
6394 if (valid_array_reference (vname))
6395 {
6396 v = array_variable_part (vname, &temp, (int *)0);
6397 /* If we want to signal array_value to use an already-computed index,
6398 set LIND to that index */
6399 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
6400 if (v && invisible_p (v))
6401 {
6402 vtype = VT_ARRAYMEMBER;
6403 *varp = (SHELL_VAR *)NULL;
6404 *valp = (char *)NULL;
6405 }
6406 if (v && (array_p (v) || assoc_p (v)))
6407 { /* [ */
6408 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
6409 {
6410 /* Callers have to differentiate between indexed and associative */
6411 vtype = VT_ARRAYVAR;
6412 if (temp[0] == '*')
6413 vtype |= VT_STARSUB;
6414 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
6415 }
6416 else
6417 {
6418 vtype = VT_ARRAYMEMBER;
6419 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6420 }
6421 *varp = v;
6422 }
6423 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
6424 {
6425 vtype = VT_VARIABLE;
6426 *varp = v;
6427 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6428 *valp = dequote_string (value);
6429 else
6430 *valp = dequote_escapes (value);
6431 }
6432 else
6433 {
6434 vtype = VT_ARRAYMEMBER;
6435 *varp = v;
6436 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6437 }
6438 }
6439 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
6440 {
6441 vtype = VT_ARRAYMEMBER;
6442 *varp = v;
6443 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
6444 }
6445 else
6446 #endif
6447 {
6448 if (value && vtype == VT_VARIABLE)
6449 {
6450 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6451 *valp = dequote_string (value);
6452 else
6453 *valp = dequote_escapes (value);
6454 }
6455 else
6456 *valp = value;
6457 }
6458
6459 if (want_indir)
6460 free (vname);
6461
6462 return vtype;
6463 }
6464
6465 /******************************************************/
6466 /* */
6467 /* Functions to extract substrings of variable values */
6468 /* */
6469 /******************************************************/
6470
6471 #if defined (HANDLE_MULTIBYTE)
6472 /* Character-oriented rather than strictly byte-oriented substrings. S and
6473 E, rather being strict indices into STRING, indicate character (possibly
6474 multibyte character) positions that require calculation.
6475 Used by the ${param:offset[:length]} expansion. */
6476 static char *
6477 mb_substring (string, s, e)
6478 char *string;
6479 int s, e;
6480 {
6481 char *tt;
6482 int start, stop, i, slen;
6483 DECLARE_MBSTATE;
6484
6485 start = 0;
6486 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
6487 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
6488
6489 i = s;
6490 while (string[start] && i--)
6491 ADVANCE_CHAR (string, slen, start);
6492 stop = start;
6493 i = e - s;
6494 while (string[stop] && i--)
6495 ADVANCE_CHAR (string, slen, stop);
6496 tt = substring (string, start, stop);
6497 return tt;
6498 }
6499 #endif
6500
6501 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
6502 is `@', use the positional parameters; otherwise, use the value of
6503 VARNAME. If VARNAME is an array variable, use the array elements. */
6504
6505 static char *
6506 parameter_brace_substring (varname, value, ind, substr, quoted, flags)
6507 char *varname, *value;
6508 int ind;
6509 char *substr;
6510 int quoted, flags;
6511 {
6512 intmax_t e1, e2;
6513 int vtype, r, starsub;
6514 char *temp, *val, *tt, *oname;
6515 SHELL_VAR *v;
6516
6517 if (value == 0)
6518 return ((char *)NULL);
6519
6520 oname = this_command_name;
6521 this_command_name = varname;
6522
6523 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6524 if (vtype == -1)
6525 {
6526 this_command_name = oname;
6527 return ((char *)NULL);
6528 }
6529
6530 starsub = vtype & VT_STARSUB;
6531 vtype &= ~VT_STARSUB;
6532
6533 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
6534 this_command_name = oname;
6535 if (r <= 0)
6536 {
6537 if (vtype == VT_VARIABLE)
6538 FREE (val);
6539 return ((r == 0) ? &expand_param_error : (char *)NULL);
6540 }
6541
6542 switch (vtype)
6543 {
6544 case VT_VARIABLE:
6545 case VT_ARRAYMEMBER:
6546 #if defined (HANDLE_MULTIBYTE)
6547 if (MB_CUR_MAX > 1)
6548 tt = mb_substring (val, e1, e2);
6549 else
6550 #endif
6551 tt = substring (val, e1, e2);
6552
6553 if (vtype == VT_VARIABLE)
6554 FREE (val);
6555 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6556 temp = quote_string (tt);
6557 else
6558 temp = tt ? quote_escapes (tt) : (char *)NULL;
6559 FREE (tt);
6560 break;
6561 case VT_POSPARMS:
6562 tt = pos_params (varname, e1, e2, quoted);
6563 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
6564 {
6565 temp = tt ? quote_escapes (tt) : (char *)NULL;
6566 FREE (tt);
6567 }
6568 else
6569 temp = tt;
6570 break;
6571 #if defined (ARRAY_VARS)
6572 case VT_ARRAYVAR:
6573 if (assoc_p (v))
6574 /* we convert to list and take first e2 elements starting at e1th
6575 element -- officially undefined for now */
6576 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
6577 else
6578 /* We want E2 to be the number of elements desired (arrays can be sparse,
6579 so verify_substring_values just returns the numbers specified and we
6580 rely on array_subrange to understand how to deal with them). */
6581 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
6582 /* array_subrange now calls array_quote_escapes as appropriate, so the
6583 caller no longer needs to. */
6584 break;
6585 #endif
6586 default:
6587 temp = (char *)NULL;
6588 }
6589
6590 return temp;
6591 }
6592
6593 /****************************************************************/
6594 /* */
6595 /* Functions to perform pattern substitution on variable values */
6596 /* */
6597 /****************************************************************/
6598
6599 static int
6600 shouldexp_replacement (s)
6601 char *s;
6602 {
6603 register char *p;
6604
6605 for (p = s; p && *p; p++)
6606 {
6607 if (*p == '\\')
6608 p++;
6609 else if (*p == '&')
6610 return 1;
6611 }
6612 return 0;
6613 }
6614
6615 char *
6616 pat_subst (string, pat, rep, mflags)
6617 char *string, *pat, *rep;
6618 int mflags;
6619 {
6620 char *ret, *s, *e, *str, *rstr, *mstr;
6621 int rsize, rptr, l, replen, mtype, rxpand, rslen, mlen;
6622
6623 if (string == 0)
6624 return (savestring (""));
6625
6626 mtype = mflags & MATCH_TYPEMASK;
6627
6628 #if 0 /* bash-4.2 ? */
6629 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
6630 #else
6631 rxpand = 0;
6632 #endif
6633
6634 /* Special cases:
6635 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
6636 * with REP and return the result.
6637 * 2. A null pattern with mtype == MATCH_END means to append REP to
6638 * STRING and return the result.
6639 * These don't understand or process `&' in the replacement string.
6640 */
6641 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
6642 {
6643 replen = STRLEN (rep);
6644 l = STRLEN (string);
6645 ret = (char *)xmalloc (replen + l + 2);
6646 if (replen == 0)
6647 strcpy (ret, string);
6648 else if (mtype == MATCH_BEG)
6649 {
6650 strcpy (ret, rep);
6651 strcpy (ret + replen, string);
6652 }
6653 else
6654 {
6655 strcpy (ret, string);
6656 strcpy (ret + l, rep);
6657 }
6658 return (ret);
6659 }
6660
6661 ret = (char *)xmalloc (rsize = 64);
6662 ret[0] = '\0';
6663
6664 for (replen = STRLEN (rep), rptr = 0, str = string;;)
6665 {
6666 if (match_pattern (str, pat, mtype, &s, &e) == 0)
6667 break;
6668 l = s - str;
6669
6670 if (rxpand)
6671 {
6672 int x;
6673 mlen = e - s;
6674 mstr = xmalloc (mlen + 1);
6675 for (x = 0; x < mlen; x++)
6676 mstr[x] = s[x];
6677 mstr[mlen] = '\0';
6678 rstr = strcreplace (rep, '&', mstr, 0);
6679 rslen = strlen (rstr);
6680 }
6681 else
6682 {
6683 rstr = rep;
6684 rslen = replen;
6685 }
6686
6687 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
6688
6689 /* OK, now copy the leading unmatched portion of the string (from
6690 str to s) to ret starting at rptr (the current offset). Then copy
6691 the replacement string at ret + rptr + (s - str). Increment
6692 rptr (if necessary) and str and go on. */
6693 if (l)
6694 {
6695 strncpy (ret + rptr, str, l);
6696 rptr += l;
6697 }
6698 if (replen)
6699 {
6700 strncpy (ret + rptr, rstr, rslen);
6701 rptr += rslen;
6702 }
6703 str = e; /* e == end of match */
6704
6705 if (rstr != rep)
6706 free (rstr);
6707
6708 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
6709 break;
6710
6711 if (s == e)
6712 {
6713 /* On a zero-length match, make sure we copy one character, since
6714 we increment one character to avoid infinite recursion. */
6715 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
6716 ret[rptr++] = *str++;
6717 e++; /* avoid infinite recursion on zero-length match */
6718 }
6719 }
6720
6721 /* Now copy the unmatched portion of the input string */
6722 if (str && *str)
6723 {
6724 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
6725 strcpy (ret + rptr, str);
6726 }
6727 else
6728 ret[rptr] = '\0';
6729
6730 return ret;
6731 }
6732
6733 /* Do pattern match and replacement on the positional parameters. */
6734 static char *
6735 pos_params_pat_subst (string, pat, rep, mflags)
6736 char *string, *pat, *rep;
6737 int mflags;
6738 {
6739 WORD_LIST *save, *params;
6740 WORD_DESC *w;
6741 char *ret;
6742 int pchar, qflags;
6743
6744 save = params = list_rest_of_args ();
6745 if (save == 0)
6746 return ((char *)NULL);
6747
6748 for ( ; params; params = params->next)
6749 {
6750 ret = pat_subst (params->word->word, pat, rep, mflags);
6751 w = alloc_word_desc ();
6752 w->word = ret ? ret : savestring ("");
6753 dispose_word (params->word);
6754 params->word = w;
6755 }
6756
6757 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6758 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6759
6760 ret = string_list_pos_params (pchar, save, qflags);
6761
6762 dispose_words (save);
6763
6764 return (ret);
6765 }
6766
6767 /* Perform pattern substitution on VALUE, which is the expansion of
6768 VARNAME. PATSUB is an expression supplying the pattern to match
6769 and the string to substitute. QUOTED is a flags word containing
6770 the type of quoting currently in effect. */
6771 static char *
6772 parameter_brace_patsub (varname, value, ind, patsub, quoted, flags)
6773 char *varname, *value;
6774 int ind;
6775 char *patsub;
6776 int quoted, flags;
6777 {
6778 int vtype, mflags, starsub, delim;
6779 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
6780 SHELL_VAR *v;
6781
6782 if (value == 0)
6783 return ((char *)NULL);
6784
6785 this_command_name = varname;
6786
6787 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6788 if (vtype == -1)
6789 return ((char *)NULL);
6790
6791 starsub = vtype & VT_STARSUB;
6792 vtype &= ~VT_STARSUB;
6793
6794 mflags = 0;
6795 /* PATSUB is never NULL when this is called. */
6796 if (*patsub == '/')
6797 {
6798 mflags |= MATCH_GLOBREP;
6799 patsub++;
6800 }
6801
6802 /* Malloc this because expand_string_if_necessary or one of the expansion
6803 functions in its call chain may free it on a substitution error. */
6804 lpatsub = savestring (patsub);
6805
6806 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6807 mflags |= MATCH_QUOTED;
6808
6809 if (starsub)
6810 mflags |= MATCH_STARSUB;
6811
6812 /* If the pattern starts with a `/', make sure we skip over it when looking
6813 for the replacement delimiter. */
6814 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
6815 if (lpatsub[delim] == '/')
6816 {
6817 lpatsub[delim] = 0;
6818 rep = lpatsub + delim + 1;
6819 }
6820 else
6821 rep = (char *)NULL;
6822
6823 if (rep && *rep == '\0')
6824 rep = (char *)NULL;
6825
6826 /* Perform the same expansions on the pattern as performed by the
6827 pattern removal expansions. */
6828 pat = getpattern (lpatsub, quoted, 1);
6829
6830 if (rep)
6831 {
6832 /* We want to perform quote removal on the expanded replacement even if
6833 the entire expansion is double-quoted because the parser and string
6834 extraction functions treated quotes in the replacement string as
6835 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
6836 if (shell_compatibility_level > 42)
6837 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
6838 /* This is the bash-4.2 code. */
6839 else if ((mflags & MATCH_QUOTED) == 0)
6840 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
6841 else
6842 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
6843 }
6844
6845 /* ksh93 doesn't allow the match specifier to be a part of the expanded
6846 pattern. This is an extension. Make sure we don't anchor the pattern
6847 at the beginning or end of the string if we're doing global replacement,
6848 though. */
6849 p = pat;
6850 if (mflags & MATCH_GLOBREP)
6851 mflags |= MATCH_ANY;
6852 else if (pat && pat[0] == '#')
6853 {
6854 mflags |= MATCH_BEG;
6855 p++;
6856 }
6857 else if (pat && pat[0] == '%')
6858 {
6859 mflags |= MATCH_END;
6860 p++;
6861 }
6862 else
6863 mflags |= MATCH_ANY;
6864
6865 /* OK, we now want to substitute REP for PAT in VAL. If
6866 flags & MATCH_GLOBREP is non-zero, the substitution is done
6867 everywhere, otherwise only the first occurrence of PAT is
6868 replaced. The pattern matching code doesn't understand
6869 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
6870 values passed in (VT_VARIABLE) so the pattern substitution
6871 code works right. We need to requote special chars after
6872 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
6873 other cases if QUOTED == 0, since the posparams and arrays
6874 indexed by * or @ do special things when QUOTED != 0. */
6875
6876 switch (vtype)
6877 {
6878 case VT_VARIABLE:
6879 case VT_ARRAYMEMBER:
6880 temp = pat_subst (val, p, rep, mflags);
6881 if (vtype == VT_VARIABLE)
6882 FREE (val);
6883 if (temp)
6884 {
6885 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
6886 free (temp);
6887 temp = tt;
6888 }
6889 break;
6890 case VT_POSPARMS:
6891 temp = pos_params_pat_subst (val, p, rep, mflags);
6892 if (temp && (mflags & MATCH_QUOTED) == 0)
6893 {
6894 tt = quote_escapes (temp);
6895 free (temp);
6896 temp = tt;
6897 }
6898 break;
6899 #if defined (ARRAY_VARS)
6900 case VT_ARRAYVAR:
6901 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
6902 : array_patsub (array_cell (v), p, rep, mflags);
6903 /* Don't call quote_escapes anymore; array_patsub calls
6904 array_quote_escapes as appropriate before adding the
6905 space separators; ditto for assoc_patsub. */
6906 break;
6907 #endif
6908 }
6909
6910 FREE (pat);
6911 FREE (rep);
6912 free (lpatsub);
6913
6914 return temp;
6915 }
6916
6917 /****************************************************************/
6918 /* */
6919 /* Functions to perform case modification on variable values */
6920 /* */
6921 /****************************************************************/
6922
6923 /* Do case modification on the positional parameters. */
6924
6925 static char *
6926 pos_params_modcase (string, pat, modop, mflags)
6927 char *string, *pat;
6928 int modop;
6929 int mflags;
6930 {
6931 WORD_LIST *save, *params;
6932 WORD_DESC *w;
6933 char *ret;
6934 int pchar, qflags;
6935
6936 save = params = list_rest_of_args ();
6937 if (save == 0)
6938 return ((char *)NULL);
6939
6940 for ( ; params; params = params->next)
6941 {
6942 ret = sh_modcase (params->word->word, pat, modop);
6943 w = alloc_word_desc ();
6944 w->word = ret ? ret : savestring ("");
6945 dispose_word (params->word);
6946 params->word = w;
6947 }
6948
6949 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6950 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6951
6952 ret = string_list_pos_params (pchar, save, qflags);
6953 dispose_words (save);
6954
6955 return (ret);
6956 }
6957
6958 /* Perform case modification on VALUE, which is the expansion of
6959 VARNAME. MODSPEC is an expression supplying the type of modification
6960 to perform. QUOTED is a flags word containing the type of quoting
6961 currently in effect. */
6962 static char *
6963 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
6964 char *varname, *value;
6965 int ind, modspec;
6966 char *patspec;
6967 int quoted, flags;
6968 {
6969 int vtype, starsub, modop, mflags, x;
6970 char *val, *temp, *pat, *p, *lpat, *tt;
6971 SHELL_VAR *v;
6972
6973 if (value == 0)
6974 return ((char *)NULL);
6975
6976 this_command_name = varname;
6977
6978 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6979 if (vtype == -1)
6980 return ((char *)NULL);
6981
6982 starsub = vtype & VT_STARSUB;
6983 vtype &= ~VT_STARSUB;
6984
6985 modop = 0;
6986 mflags = 0;
6987 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6988 mflags |= MATCH_QUOTED;
6989 if (starsub)
6990 mflags |= MATCH_STARSUB;
6991
6992 p = patspec;
6993 if (modspec == '^')
6994 {
6995 x = p && p[0] == modspec;
6996 modop = x ? CASE_UPPER : CASE_UPFIRST;
6997 p += x;
6998 }
6999 else if (modspec == ',')
7000 {
7001 x = p && p[0] == modspec;
7002 modop = x ? CASE_LOWER : CASE_LOWFIRST;
7003 p += x;
7004 }
7005 else if (modspec == '~')
7006 {
7007 x = p && p[0] == modspec;
7008 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7009 p += x;
7010 }
7011
7012 lpat = p ? savestring (p) : 0;
7013 /* Perform the same expansions on the pattern as performed by the
7014 pattern removal expansions. FOR LATER */
7015 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7016
7017 /* OK, now we do the case modification. */
7018 switch (vtype)
7019 {
7020 case VT_VARIABLE:
7021 case VT_ARRAYMEMBER:
7022 temp = sh_modcase (val, pat, modop);
7023 if (vtype == VT_VARIABLE)
7024 FREE (val);
7025 if (temp)
7026 {
7027 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7028 free (temp);
7029 temp = tt;
7030 }
7031 break;
7032
7033 case VT_POSPARMS:
7034 temp = pos_params_modcase (val, pat, modop, mflags);
7035 if (temp && (mflags & MATCH_QUOTED) == 0)
7036 {
7037 tt = quote_escapes (temp);
7038 free (temp);
7039 temp = tt;
7040 }
7041 break;
7042
7043 #if defined (ARRAY_VARS)
7044 case VT_ARRAYVAR:
7045 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7046 : array_modcase (array_cell (v), pat, modop, mflags);
7047 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7048 as appropriate before adding the space separators; ditto for
7049 assoc_modcase. */
7050 break;
7051 #endif
7052 }
7053
7054 FREE (pat);
7055 free (lpat);
7056
7057 return temp;
7058 }
7059
7060 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7061 any occur, this must be a nested command substitution, so return 0.
7062 Otherwise, return 1. A valid arithmetic expression must always have a
7063 ( before a matching ), so any cases where there are more right parens
7064 means that this must not be an arithmetic expression, though the parser
7065 will not accept it without a balanced total number of parens. */
7066 static int
7067 chk_arithsub (s, len)
7068 const char *s;
7069 int len;
7070 {
7071 int i, count;
7072 DECLARE_MBSTATE;
7073
7074 i = count = 0;
7075 while (i < len)
7076 {
7077 if (s[i] == LPAREN)
7078 count++;
7079 else if (s[i] == RPAREN)
7080 {
7081 count--;
7082 if (count < 0)
7083 return 0;
7084 }
7085
7086 switch (s[i])
7087 {
7088 default:
7089 ADVANCE_CHAR (s, len, i);
7090 break;
7091
7092 case '\\':
7093 i++;
7094 if (s[i])
7095 ADVANCE_CHAR (s, len, i);
7096 break;
7097
7098 case '\'':
7099 i = skip_single_quoted (s, len, ++i);
7100 break;
7101
7102 case '"':
7103 i = skip_double_quoted ((char *)s, len, ++i);
7104 break;
7105 }
7106 }
7107
7108 return (count == 0);
7109 }
7110
7111 /****************************************************************/
7112 /* */
7113 /* Functions to perform parameter expansion on a string */
7114 /* */
7115 /****************************************************************/
7116
7117 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
7118 static WORD_DESC *
7119 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
7120 char *string;
7121 int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at, pflags;
7122 {
7123 int check_nullness, var_is_set, var_is_null, var_is_special;
7124 int want_substring, want_indir, want_patsub, want_casemod;
7125 char *name, *value, *temp, *temp1;
7126 WORD_DESC *tdesc, *ret;
7127 int t_index, sindex, c, tflag, modspec;
7128 intmax_t number;
7129 arrayind_t ind;
7130
7131 temp = temp1 = value = (char *)NULL;
7132 var_is_set = var_is_null = var_is_special = check_nullness = 0;
7133 want_substring = want_indir = want_patsub = want_casemod = 0;
7134
7135 sindex = *indexp;
7136 t_index = ++sindex;
7137 /* ${#var} doesn't have any of the other parameter expansions on it. */
7138 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
7139 name = string_extract (string, &t_index, "}", SX_VARNAME);
7140 else
7141 #if defined (CASEMOD_EXPANSIONS)
7142 /* To enable case-toggling expansions using the `~' operator character
7143 change the 1 to 0. */
7144 # if defined (CASEMOD_CAPCASE)
7145 name = string_extract (string, &t_index, "#%^,~:-=?+/}", SX_VARNAME);
7146 # else
7147 name = string_extract (string, &t_index, "#%^,:-=?+/}", SX_VARNAME);
7148 # endif /* CASEMOD_CAPCASE */
7149 #else
7150 name = string_extract (string, &t_index, "#%:-=?+/}", SX_VARNAME);
7151 #endif /* CASEMOD_EXPANSIONS */
7152
7153 ret = 0;
7154 tflag = 0;
7155
7156 ind = INTMAX_MIN;
7157
7158 /* If the name really consists of a special variable, then make sure
7159 that we have the entire name. We don't allow indirect references
7160 to special variables except `#', `?', `@' and `*'. */
7161 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
7162 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
7163 {
7164 t_index++;
7165 temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0);
7166 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
7167 *name = string[sindex];
7168 if (string[sindex] == '!')
7169 {
7170 /* indirect reference of $#, $?, $@, or $* */
7171 name[1] = string[sindex + 1];
7172 strcpy (name + 2, temp1);
7173 }
7174 else
7175 strcpy (name + 1, temp1);
7176 free (temp1);
7177 }
7178 sindex = t_index;
7179
7180 /* Find out what character ended the variable name. Then
7181 do the appropriate thing. */
7182 if (c = string[sindex])
7183 sindex++;
7184
7185 /* If c is followed by one of the valid parameter expansion
7186 characters, move past it as normal. If not, assume that
7187 a substring specification is being given, and do not move
7188 past it. */
7189 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
7190 {
7191 check_nullness++;
7192 if (c = string[sindex])
7193 sindex++;
7194 }
7195 else if (c == ':' && string[sindex] != RBRACE)
7196 want_substring = 1;
7197 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
7198 want_patsub = 1;
7199 #if defined (CASEMOD_EXPANSIONS)
7200 else if (c == '^' || c == ',' || c == '~')
7201 {
7202 modspec = c;
7203 want_casemod = 1;
7204 }
7205 #endif
7206
7207 /* Catch the valid and invalid brace expressions that made it through the
7208 tests above. */
7209 /* ${#-} is a valid expansion and means to take the length of $-.
7210 Similarly for ${#?} and ${##}... */
7211 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7212 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
7213 {
7214 name = (char *)xrealloc (name, 3);
7215 name[1] = c;
7216 name[2] = '\0';
7217 c = string[sindex++];
7218 }
7219
7220 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
7221 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7222 member (c, "%:=+/") && string[sindex] == RBRACE)
7223 {
7224 temp = (char *)NULL;
7225 goto bad_substitution;
7226 }
7227
7228 /* Indirect expansion begins with a `!'. A valid indirect expansion is
7229 either a variable name, one of the positional parameters or a special
7230 variable that expands to one of the positional parameters. */
7231 want_indir = *name == '!' &&
7232 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7233 || VALID_INDIR_PARAM (name[1]));
7234
7235 /* Determine the value of this variable. */
7236
7237 /* Check for special variables, directly referenced. */
7238 if (SPECIAL_VAR (name, want_indir))
7239 var_is_special++;
7240
7241 /* Check for special expansion things, like the length of a parameter */
7242 if (*name == '#' && name[1])
7243 {
7244 /* If we are not pointing at the character just after the
7245 closing brace, then we haven't gotten all of the name.
7246 Since it begins with a special character, this is a bad
7247 substitution. Also check NAME for validity before trying
7248 to go on. */
7249 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
7250 {
7251 temp = (char *)NULL;
7252 goto bad_substitution;
7253 }
7254
7255 number = parameter_brace_expand_length (name);
7256 if (number == INTMAX_MIN && unbound_vars_is_error)
7257 {
7258 last_command_exit_value = EXECUTION_FAILURE;
7259 err_unboundvar (name+1);
7260 free (name);
7261 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7262 }
7263 free (name);
7264
7265 *indexp = sindex;
7266 if (number < 0)
7267 return (&expand_wdesc_error);
7268 else
7269 {
7270 ret = alloc_word_desc ();
7271 ret->word = itos (number);
7272 return ret;
7273 }
7274 }
7275
7276 /* ${@} is identical to $@. */
7277 if (name[0] == '@' && name[1] == '\0')
7278 {
7279 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7280 *quoted_dollar_atp = 1;
7281
7282 if (contains_dollar_at)
7283 *contains_dollar_at = 1;
7284
7285 tflag |= W_DOLLARAT;
7286 }
7287
7288 /* Process ${!PREFIX*} expansion. */
7289 if (want_indir && string[sindex - 1] == RBRACE &&
7290 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
7291 legal_variable_starter ((unsigned char) name[1]))
7292 {
7293 char **x;
7294 WORD_LIST *xlist;
7295
7296 temp1 = savestring (name + 1);
7297 number = strlen (temp1);
7298 temp1[number - 1] = '\0';
7299 x = all_variables_matching_prefix (temp1);
7300 xlist = strvec_to_word_list (x, 0, 0);
7301 if (string[sindex - 2] == '*')
7302 temp = string_list_dollar_star (xlist);
7303 else
7304 {
7305 temp = string_list_dollar_at (xlist, quoted);
7306 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7307 *quoted_dollar_atp = 1;
7308 if (contains_dollar_at)
7309 *contains_dollar_at = 1;
7310
7311 tflag |= W_DOLLARAT;
7312 }
7313 free (x);
7314 dispose_words (xlist);
7315 free (temp1);
7316 *indexp = sindex;
7317
7318 free (name);
7319
7320 ret = alloc_word_desc ();
7321 ret->word = temp;
7322 ret->flags = tflag; /* XXX */
7323 return ret;
7324 }
7325
7326 #if defined (ARRAY_VARS)
7327 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
7328 if (want_indir && string[sindex - 1] == RBRACE &&
7329 string[sindex - 2] == ']' && valid_array_reference (name+1))
7330 {
7331 char *x, *x1;
7332
7333 temp1 = savestring (name + 1);
7334 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
7335 FREE (x);
7336 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
7337 {
7338 temp = array_keys (temp1, quoted); /* handles assoc vars too */
7339 if (x1[0] == '@')
7340 {
7341 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7342 *quoted_dollar_atp = 1;
7343 if (contains_dollar_at)
7344 *contains_dollar_at = 1;
7345
7346 tflag |= W_DOLLARAT;
7347 }
7348
7349 free (temp1);
7350 *indexp = sindex;
7351
7352 ret = alloc_word_desc ();
7353 ret->word = temp;
7354 ret->flags = tflag; /* XXX */
7355 return ret;
7356 }
7357
7358 free (temp1);
7359 }
7360 #endif /* ARRAY_VARS */
7361
7362 /* Make sure that NAME is valid before trying to go on. */
7363 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
7364 var_is_special) == 0)
7365 {
7366 temp = (char *)NULL;
7367 goto bad_substitution;
7368 }
7369
7370 if (want_indir)
7371 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
7372 else
7373 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
7374
7375 if (tdesc)
7376 {
7377 temp = tdesc->word;
7378 tflag = tdesc->flags;
7379 dispose_word_desc (tdesc);
7380 }
7381 else
7382 temp = (char *)0;
7383
7384 if (temp == &expand_param_error || temp == &expand_param_fatal)
7385 {
7386 FREE (name);
7387 FREE (value);
7388 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
7389 }
7390
7391 #if defined (ARRAY_VARS)
7392 if (valid_array_reference (name))
7393 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at);
7394 #endif
7395
7396 var_is_set = temp != (char *)0;
7397 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
7398 /* XXX - this may not need to be restricted to special variables */
7399 if (check_nullness)
7400 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
7401
7402 /* Get the rest of the stuff inside the braces. */
7403 if (c && c != RBRACE)
7404 {
7405 /* Extract the contents of the ${ ... } expansion
7406 according to the Posix.2 rules. */
7407 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
7408 if (string[sindex] == RBRACE)
7409 sindex++;
7410 else
7411 goto bad_substitution;
7412 }
7413 else
7414 value = (char *)NULL;
7415
7416 *indexp = sindex;
7417
7418 /* All the cases where an expansion can possibly generate an unbound
7419 variable error. */
7420 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
7421 {
7422 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]))
7423 {
7424 last_command_exit_value = EXECUTION_FAILURE;
7425 err_unboundvar (name);
7426 FREE (value);
7427 FREE (temp);
7428 free (name);
7429 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7430 }
7431 }
7432
7433 /* If this is a substring spec, process it and add the result. */
7434 if (want_substring)
7435 {
7436 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7437 FREE (name);
7438 FREE (value);
7439 FREE (temp);
7440
7441 if (temp1 == &expand_param_error)
7442 return (&expand_wdesc_error);
7443 else if (temp1 == &expand_param_fatal)
7444 return (&expand_wdesc_fatal);
7445
7446 ret = alloc_word_desc ();
7447 ret->word = temp1;
7448 /* We test quoted_dollar_atp because we want variants with double-quoted
7449 "$@" to take a different code path. In fact, we make sure at the end
7450 of expand_word_internal that we're only looking at these flags if
7451 quoted_dollar_at == 0. */
7452 if (temp1 &&
7453 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7454 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7455 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7456 return ret;
7457 }
7458 else if (want_patsub)
7459 {
7460 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7461 FREE (name);
7462 FREE (value);
7463 FREE (temp);
7464
7465 if (temp1 == &expand_param_error)
7466 return (&expand_wdesc_error);
7467 else if (temp1 == &expand_param_fatal)
7468 return (&expand_wdesc_fatal);
7469
7470 ret = alloc_word_desc ();
7471 ret->word = temp1;
7472 if (temp1 &&
7473 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7474 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7475 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7476 return ret;
7477 }
7478 #if defined (CASEMOD_EXPANSIONS)
7479 else if (want_casemod)
7480 {
7481 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7482 FREE (name);
7483 FREE (value);
7484 FREE (temp);
7485
7486 if (temp1 == &expand_param_error)
7487 return (&expand_wdesc_error);
7488 else if (temp1 == &expand_param_fatal)
7489 return (&expand_wdesc_fatal);
7490
7491 ret = alloc_word_desc ();
7492 ret->word = temp1;
7493 if (temp1 &&
7494 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7495 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7496 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7497 return ret;
7498 }
7499 #endif
7500
7501 /* Do the right thing based on which character ended the variable name. */
7502 switch (c)
7503 {
7504 default:
7505 case '\0':
7506 bad_substitution:
7507 last_command_exit_value = EXECUTION_FAILURE;
7508 report_error (_("%s: bad substitution"), string ? string : "??");
7509 FREE (value);
7510 FREE (temp);
7511 free (name);
7512 return &expand_wdesc_error;
7513
7514 case RBRACE:
7515 break;
7516
7517 case '#': /* ${param#[#]pattern} */
7518 case '%': /* ${param%[%]pattern} */
7519 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
7520 {
7521 FREE (value);
7522 break;
7523 }
7524 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7525 free (temp);
7526 free (value);
7527 free (name);
7528
7529 ret = alloc_word_desc ();
7530 ret->word = temp1;
7531 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7532 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7533 return ret;
7534
7535 case '-':
7536 case '=':
7537 case '?':
7538 case '+':
7539 if (var_is_set && var_is_null == 0)
7540 {
7541 /* If the operator is `+', we don't want the value of the named
7542 variable for anything, just the value of the right hand side. */
7543 if (c == '+')
7544 {
7545 /* XXX -- if we're double-quoted and the named variable is "$@",
7546 we want to turn off any special handling of "$@" --
7547 we're not using it, so whatever is on the rhs applies. */
7548 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7549 *quoted_dollar_atp = 0;
7550 if (contains_dollar_at)
7551 *contains_dollar_at = 0;
7552
7553 FREE (temp);
7554 if (value)
7555 {
7556 /* From Posix discussion on austin-group list. Issue 221
7557 requires that backslashes escaping `}' inside
7558 double-quoted ${...} be removed. */
7559 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7560 quoted |= Q_DOLBRACE;
7561 ret = parameter_brace_expand_rhs (name, value, c,
7562 quoted,
7563 quoted_dollar_atp,
7564 contains_dollar_at);
7565 /* XXX - fix up later, esp. noting presence of
7566 W_HASQUOTEDNULL in ret->flags */
7567 free (value);
7568 }
7569 else
7570 temp = (char *)NULL;
7571 }
7572 else
7573 {
7574 FREE (value);
7575 }
7576 /* Otherwise do nothing; just use the value in TEMP. */
7577 }
7578 else /* VAR not set or VAR is NULL. */
7579 {
7580 FREE (temp);
7581 temp = (char *)NULL;
7582 if (c == '=' && var_is_special)
7583 {
7584 last_command_exit_value = EXECUTION_FAILURE;
7585 report_error (_("$%s: cannot assign in this way"), name);
7586 free (name);
7587 free (value);
7588 return &expand_wdesc_error;
7589 }
7590 else if (c == '?')
7591 {
7592 parameter_brace_expand_error (name, value);
7593 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7594 }
7595 else if (c != '+')
7596 {
7597 /* XXX -- if we're double-quoted and the named variable is "$@",
7598 we want to turn off any special handling of "$@" --
7599 we're not using it, so whatever is on the rhs applies. */
7600 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7601 *quoted_dollar_atp = 0;
7602 if (contains_dollar_at)
7603 *contains_dollar_at = 0;
7604
7605 /* From Posix discussion on austin-group list. Issue 221 requires
7606 that backslashes escaping `}' inside double-quoted ${...} be
7607 removed. */
7608 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7609 quoted |= Q_DOLBRACE;
7610 ret = parameter_brace_expand_rhs (name, value, c, quoted,
7611 quoted_dollar_atp,
7612 contains_dollar_at);
7613 /* XXX - fix up later, esp. noting presence of
7614 W_HASQUOTEDNULL in tdesc->flags */
7615 }
7616 free (value);
7617 }
7618
7619 break;
7620 }
7621 free (name);
7622
7623 if (ret == 0)
7624 {
7625 ret = alloc_word_desc ();
7626 ret->flags = tflag;
7627 ret->word = temp;
7628 }
7629 return (ret);
7630 }
7631
7632 /* Expand a single ${xxx} expansion. The braces are optional. When
7633 the braces are used, parameter_brace_expand() does the work,
7634 possibly calling param_expand recursively. */
7635 static WORD_DESC *
7636 param_expand (string, sindex, quoted, expanded_something,
7637 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
7638 pflags)
7639 char *string;
7640 int *sindex, quoted, *expanded_something, *contains_dollar_at;
7641 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
7642 {
7643 char *temp, *temp1, uerror[3];
7644 int zindex, t_index, expok;
7645 unsigned char c;
7646 intmax_t number;
7647 SHELL_VAR *var;
7648 WORD_LIST *list;
7649 WORD_DESC *tdesc, *ret;
7650 int tflag;
7651
7652 zindex = *sindex;
7653 c = string[++zindex];
7654
7655 temp = (char *)NULL;
7656 ret = tdesc = (WORD_DESC *)NULL;
7657 tflag = 0;
7658
7659 /* Do simple cases first. Switch on what follows '$'. */
7660 switch (c)
7661 {
7662 /* $0 .. $9? */
7663 case '0':
7664 case '1':
7665 case '2':
7666 case '3':
7667 case '4':
7668 case '5':
7669 case '6':
7670 case '7':
7671 case '8':
7672 case '9':
7673 temp1 = dollar_vars[TODIGIT (c)];
7674 if (unbound_vars_is_error && temp1 == (char *)NULL)
7675 {
7676 uerror[0] = '$';
7677 uerror[1] = c;
7678 uerror[2] = '\0';
7679 last_command_exit_value = EXECUTION_FAILURE;
7680 err_unboundvar (uerror);
7681 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7682 }
7683 if (temp1)
7684 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7685 ? quote_string (temp1)
7686 : quote_escapes (temp1);
7687 else
7688 temp = (char *)NULL;
7689
7690 break;
7691
7692 /* $$ -- pid of the invoking shell. */
7693 case '$':
7694 temp = itos (dollar_dollar_pid);
7695 break;
7696
7697 /* $# -- number of positional parameters. */
7698 case '#':
7699 temp = itos (number_of_args ());
7700 break;
7701
7702 /* $? -- return value of the last synchronous command. */
7703 case '?':
7704 temp = itos (last_command_exit_value);
7705 break;
7706
7707 /* $- -- flags supplied to the shell on invocation or by `set'. */
7708 case '-':
7709 temp = which_set_flags ();
7710 break;
7711
7712 /* $! -- Pid of the last asynchronous command. */
7713 case '!':
7714 /* If no asynchronous pids have been created, expand to nothing.
7715 If `set -u' has been executed, and no async processes have
7716 been created, this is an expansion error. */
7717 if (last_asynchronous_pid == NO_PID)
7718 {
7719 if (expanded_something)
7720 *expanded_something = 0;
7721 temp = (char *)NULL;
7722 if (unbound_vars_is_error)
7723 {
7724 uerror[0] = '$';
7725 uerror[1] = c;
7726 uerror[2] = '\0';
7727 last_command_exit_value = EXECUTION_FAILURE;
7728 err_unboundvar (uerror);
7729 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7730 }
7731 }
7732 else
7733 temp = itos (last_asynchronous_pid);
7734 break;
7735
7736 /* The only difference between this and $@ is when the arg is quoted. */
7737 case '*': /* `$*' */
7738 list = list_rest_of_args ();
7739
7740 #if 0
7741 /* According to austin-group posix proposal by Geoff Clare in
7742 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7743
7744 "The shell shall write a message to standard error and
7745 immediately exit when it tries to expand an unset parameter
7746 other than the '@' and '*' special parameters."
7747 */
7748
7749 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7750 {
7751 uerror[0] = '$';
7752 uerror[1] = '*';
7753 uerror[2] = '\0';
7754 last_command_exit_value = EXECUTION_FAILURE;
7755 err_unboundvar (uerror);
7756 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7757 }
7758 #endif
7759
7760 /* If there are no command-line arguments, this should just
7761 disappear if there are other characters in the expansion,
7762 even if it's quoted. */
7763 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
7764 temp = (char *)NULL;
7765 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
7766 {
7767 /* If we have "$*" we want to make a string of the positional
7768 parameters, separated by the first character of $IFS, and
7769 quote the whole string, including the separators. If IFS
7770 is unset, the parameters are separated by ' '; if $IFS is
7771 null, the parameters are concatenated. */
7772 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
7773 if (temp)
7774 {
7775 temp1 = quote_string (temp);
7776 if (*temp == 0)
7777 tflag |= W_HASQUOTEDNULL;
7778 free (temp);
7779 temp = temp1;
7780 }
7781 }
7782 else
7783 {
7784 /* We check whether or not we're eventually going to split $* here,
7785 for example when IFS is empty and we are processing the rhs of
7786 an assignment statement. In that case, we don't separate the
7787 arguments at all. Otherwise, if the $* is not quoted it is
7788 identical to $@ */
7789 # if defined (HANDLE_MULTIBYTE)
7790 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
7791 # else
7792 if (expand_no_split_dollar_star && ifs_firstc == 0)
7793 # endif
7794 temp = string_list_dollar_star (list);
7795 else
7796 {
7797 temp = string_list_dollar_at (list, quoted);
7798 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
7799 tflag |= W_SPLITSPACE;
7800 }
7801
7802 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
7803 *contains_dollar_at = 1;
7804 }
7805
7806 dispose_words (list);
7807 break;
7808
7809 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
7810 means that we have to turn quoting off after we split into
7811 the individually quoted arguments so that the final split
7812 on the first character of $IFS is still done. */
7813 case '@': /* `$@' */
7814 list = list_rest_of_args ();
7815
7816 #if 0
7817 /* According to austin-group posix proposal by Geoff Clare in
7818 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7819
7820 "The shell shall write a message to standard error and
7821 immediately exit when it tries to expand an unset parameter
7822 other than the '@' and '*' special parameters."
7823 */
7824
7825 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7826 {
7827 uerror[0] = '$';
7828 uerror[1] = '@';
7829 uerror[2] = '\0';
7830 last_command_exit_value = EXECUTION_FAILURE;
7831 err_unboundvar (uerror);
7832 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7833 }
7834 #endif
7835
7836 /* We want to flag the fact that we saw this. We can't turn
7837 off quoting entirely, because other characters in the
7838 string might need it (consider "\"$@\""), but we need some
7839 way to signal that the final split on the first character
7840 of $IFS should be done, even though QUOTED is 1. */
7841 /* XXX - should this test include Q_PATQUOTE? */
7842 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7843 *quoted_dollar_at_p = 1;
7844 if (contains_dollar_at)
7845 *contains_dollar_at = 1;
7846
7847 /* We want to separate the positional parameters with the first
7848 character of $IFS in case $IFS is something other than a space.
7849 We also want to make sure that splitting is done no matter what --
7850 according to POSIX.2, this expands to a list of the positional
7851 parameters no matter what IFS is set to. */
7852 /* XXX - what to do when in a context where word splitting is not
7853 performed? Even when IFS is not the default, posix seems to imply
7854 that we behave like unquoted $* ? Maybe we should use PF_NOSPLIT2
7855 here. */
7856 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted);
7857
7858 tflag |= W_DOLLARAT;
7859 dispose_words (list);
7860 break;
7861
7862 case LBRACE:
7863 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
7864 quoted_dollar_at_p,
7865 contains_dollar_at);
7866
7867 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
7868 return (tdesc);
7869 temp = tdesc ? tdesc->word : (char *)0;
7870
7871 /* XXX */
7872 /* Quoted nulls should be removed if there is anything else
7873 in the string. */
7874 /* Note that we saw the quoted null so we can add one back at
7875 the end of this function if there are no other characters
7876 in the string, discard TEMP, and go on. The exception to
7877 this is when we have "${@}" and $1 is '', since $@ needs
7878 special handling. */
7879 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
7880 {
7881 if (had_quoted_null_p)
7882 *had_quoted_null_p = 1;
7883 if (*quoted_dollar_at_p == 0)
7884 {
7885 free (temp);
7886 tdesc->word = temp = (char *)NULL;
7887 }
7888
7889 }
7890
7891 ret = tdesc;
7892 goto return0;
7893
7894 /* Do command or arithmetic substitution. */
7895 case LPAREN:
7896 /* We have to extract the contents of this paren substitution. */
7897 t_index = zindex + 1;
7898 temp = extract_command_subst (string, &t_index, 0);
7899 zindex = t_index;
7900
7901 /* For Posix.2-style `$(( ))' arithmetic substitution,
7902 extract the expression and pass it to the evaluator. */
7903 if (temp && *temp == LPAREN)
7904 {
7905 char *temp2;
7906 temp1 = temp + 1;
7907 temp2 = savestring (temp1);
7908 t_index = strlen (temp2) - 1;
7909
7910 if (temp2[t_index] != RPAREN)
7911 {
7912 free (temp2);
7913 goto comsub;
7914 }
7915
7916 /* Cut off ending `)' */
7917 temp2[t_index] = '\0';
7918
7919 if (chk_arithsub (temp2, t_index) == 0)
7920 {
7921 free (temp2);
7922 #if 0
7923 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
7924 #endif
7925 goto comsub;
7926 }
7927
7928 /* Expand variables found inside the expression. */
7929 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7930 free (temp2);
7931
7932 arithsub:
7933 /* No error messages. */
7934 this_command_name = (char *)NULL;
7935 number = evalexp (temp1, &expok);
7936 free (temp);
7937 free (temp1);
7938 if (expok == 0)
7939 {
7940 if (interactive_shell == 0 && posixly_correct)
7941 {
7942 last_command_exit_value = EXECUTION_FAILURE;
7943 return (&expand_wdesc_fatal);
7944 }
7945 else
7946 return (&expand_wdesc_error);
7947 }
7948 temp = itos (number);
7949 break;
7950 }
7951
7952 comsub:
7953 if (pflags & PF_NOCOMSUB)
7954 /* we need zindex+1 because string[zindex] == RPAREN */
7955 temp1 = substring (string, *sindex, zindex+1);
7956 else
7957 {
7958 tdesc = command_substitute (temp, quoted);
7959 temp1 = tdesc ? tdesc->word : (char *)NULL;
7960 if (tdesc)
7961 dispose_word_desc (tdesc);
7962 }
7963 FREE (temp);
7964 temp = temp1;
7965 break;
7966
7967 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
7968 away in a future bash release. */
7969 case '[':
7970 /* Extract the contents of this arithmetic substitution. */
7971 t_index = zindex + 1;
7972 temp = extract_arithmetic_subst (string, &t_index);
7973 zindex = t_index;
7974 if (temp == 0)
7975 {
7976 temp = savestring (string);
7977 if (expanded_something)
7978 *expanded_something = 0;
7979 goto return0;
7980 }
7981
7982 /* Do initial variable expansion. */
7983 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES);
7984
7985 goto arithsub;
7986
7987 default:
7988 /* Find the variable in VARIABLE_LIST. */
7989 temp = (char *)NULL;
7990
7991 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
7992 ;
7993 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
7994
7995 /* If this isn't a variable name, then just output the `$'. */
7996 if (temp1 == 0 || *temp1 == '\0')
7997 {
7998 FREE (temp1);
7999 temp = (char *)xmalloc (2);
8000 temp[0] = '$';
8001 temp[1] = '\0';
8002 if (expanded_something)
8003 *expanded_something = 0;
8004 goto return0;
8005 }
8006
8007 /* If the variable exists, return its value cell. */
8008 var = find_variable (temp1);
8009
8010 if (var && invisible_p (var) == 0 && var_isset (var))
8011 {
8012 #if defined (ARRAY_VARS)
8013 if (assoc_p (var) || array_p (var))
8014 {
8015 temp = array_p (var) ? array_reference (array_cell (var), 0)
8016 : assoc_reference (assoc_cell (var), "0");
8017 if (temp)
8018 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8019 ? quote_string (temp)
8020 : quote_escapes (temp);
8021 else if (unbound_vars_is_error)
8022 goto unbound_variable;
8023 }
8024 else
8025 #endif
8026 {
8027 temp = value_cell (var);
8028
8029 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8030 ? quote_string (temp)
8031 : quote_escapes (temp);
8032 }
8033
8034 free (temp1);
8035
8036 goto return0;
8037 }
8038 else if (var = find_variable_last_nameref (temp1))
8039 {
8040 temp = nameref_cell (var);
8041 #if defined (ARRAY_VARS)
8042 if (temp && *temp && valid_array_reference (temp))
8043 {
8044 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8045 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8046 return (tdesc);
8047 ret = tdesc;
8048 goto return0;
8049 }
8050 else
8051 #endif
8052 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8053 if (temp && *temp && legal_identifier (temp) == 0)
8054 {
8055 last_command_exit_value = EXECUTION_FAILURE;
8056 report_error (_("%s: invalid variable name for name reference"), temp);
8057 return (&expand_wdesc_error); /* XXX */
8058 }
8059 else
8060 temp = (char *)NULL;
8061 }
8062
8063 temp = (char *)NULL;
8064
8065 unbound_variable:
8066 if (unbound_vars_is_error)
8067 {
8068 last_command_exit_value = EXECUTION_FAILURE;
8069 err_unboundvar (temp1);
8070 }
8071 else
8072 {
8073 free (temp1);
8074 goto return0;
8075 }
8076
8077 free (temp1);
8078 last_command_exit_value = EXECUTION_FAILURE;
8079 return ((unbound_vars_is_error && interactive_shell == 0)
8080 ? &expand_wdesc_fatal
8081 : &expand_wdesc_error);
8082 }
8083
8084 if (string[zindex])
8085 zindex++;
8086
8087 return0:
8088 *sindex = zindex;
8089
8090 if (ret == 0)
8091 {
8092 ret = alloc_word_desc ();
8093 ret->flags = tflag; /* XXX */
8094 ret->word = temp;
8095 }
8096 return ret;
8097 }
8098
8099 /* Make a word list which is the result of parameter and variable
8100 expansion, command substitution, arithmetic substitution, and
8101 quote removal of WORD. Return a pointer to a WORD_LIST which is
8102 the result of the expansion. If WORD contains a null word, the
8103 word list returned is also null.
8104
8105 QUOTED contains flag values defined in shell.h.
8106
8107 ISEXP is used to tell expand_word_internal that the word should be
8108 treated as the result of an expansion. This has implications for
8109 how IFS characters in the word are treated.
8110
8111 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
8112 they point to an integer value which receives information about expansion.
8113 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
8114 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
8115 else zero.
8116
8117 This only does word splitting in the case of $@ expansion. In that
8118 case, we split on ' '. */
8119
8120 /* Values for the local variable quoted_state. */
8121 #define UNQUOTED 0
8122 #define PARTIALLY_QUOTED 1
8123 #define WHOLLY_QUOTED 2
8124
8125 static WORD_LIST *
8126 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
8127 WORD_DESC *word;
8128 int quoted, isexp;
8129 int *contains_dollar_at;
8130 int *expanded_something;
8131 {
8132 WORD_LIST *list;
8133 WORD_DESC *tword;
8134
8135 /* The intermediate string that we build while expanding. */
8136 char *istring;
8137
8138 /* The current size of the above object. */
8139 int istring_size;
8140
8141 /* Index into ISTRING. */
8142 int istring_index;
8143
8144 /* Temporary string storage. */
8145 char *temp, *temp1;
8146
8147 /* The text of WORD. */
8148 register char *string;
8149
8150 /* The size of STRING. */
8151 size_t string_size;
8152
8153 /* The index into STRING. */
8154 int sindex;
8155
8156 /* This gets 1 if we see a $@ while quoted. */
8157 int quoted_dollar_at;
8158
8159 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
8160 whether WORD contains no quoting characters, a partially quoted
8161 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
8162 int quoted_state;
8163
8164 /* State flags */
8165 int had_quoted_null;
8166 int has_dollar_at, temp_has_dollar_at;
8167 int split_on_spaces;
8168 int tflag;
8169 int pflags; /* flags passed to param_expand */
8170
8171 int assignoff; /* If assignment, offset of `=' */
8172
8173 register unsigned char c; /* Current character. */
8174 int t_index; /* For calls to string_extract_xxx. */
8175
8176 char twochars[2];
8177
8178 DECLARE_MBSTATE;
8179
8180 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
8181 istring[istring_index = 0] = '\0';
8182 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
8183 split_on_spaces = 0;
8184 quoted_state = UNQUOTED;
8185
8186 string = word->word;
8187 if (string == 0)
8188 goto finished_with_string;
8189 /* Don't need the string length for the SADD... and COPY_ macros unless
8190 multibyte characters are possible. */
8191 string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1;
8192
8193 if (contains_dollar_at)
8194 *contains_dollar_at = 0;
8195
8196 assignoff = -1;
8197
8198 /* Begin the expansion. */
8199
8200 for (sindex = 0; ;)
8201 {
8202 c = string[sindex];
8203
8204 /* Case on top-level character. */
8205 switch (c)
8206 {
8207 case '\0':
8208 goto finished_with_string;
8209
8210 case CTLESC:
8211 sindex++;
8212 #if HANDLE_MULTIBYTE
8213 if (MB_CUR_MAX > 1 && string[sindex])
8214 {
8215 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8216 }
8217 else
8218 #endif
8219 {
8220 temp = (char *)xmalloc (3);
8221 temp[0] = CTLESC;
8222 temp[1] = c = string[sindex];
8223 temp[2] = '\0';
8224 }
8225
8226 dollar_add_string:
8227 if (string[sindex])
8228 sindex++;
8229
8230 add_string:
8231 if (temp)
8232 {
8233 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
8234 temp = (char *)0;
8235 }
8236
8237 break;
8238
8239 #if defined (PROCESS_SUBSTITUTION)
8240 /* Process substitution. */
8241 case '<':
8242 case '>':
8243 {
8244 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
8245 {
8246 sindex--; /* add_character: label increments sindex */
8247 goto add_character;
8248 }
8249 else
8250 t_index = sindex + 1; /* skip past both '<' and LPAREN */
8251
8252 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index); /*))*/
8253 sindex = t_index;
8254
8255 /* If the process substitution specification is `<()', we want to
8256 open the pipe for writing in the child and produce output; if
8257 it is `>()', we want to open the pipe for reading in the child
8258 and consume input. */
8259 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
8260
8261 FREE (temp1);
8262
8263 goto dollar_add_string;
8264 }
8265 #endif /* PROCESS_SUBSTITUTION */
8266
8267 case '=':
8268 /* Posix.2 section 3.6.1 says that tildes following `=' in words
8269 which are not assignment statements are not expanded. If the
8270 shell isn't in posix mode, though, we perform tilde expansion
8271 on `likely candidate' unquoted assignment statements (flags
8272 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
8273 contains an unquoted :~ or =~. Something to think about: we
8274 now have a flag that says to perform tilde expansion on arguments
8275 to `assignment builtins' like declare and export that look like
8276 assignment statements. We now do tilde expansion on such words
8277 even in POSIX mode. */
8278 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
8279 {
8280 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8281 goto add_ifs_character;
8282 else
8283 goto add_character;
8284 }
8285 /* If we're not in posix mode or forcing assignment-statement tilde
8286 expansion, note where the `=' appears in the word and prepare to
8287 do tilde expansion following the first `='. */
8288 if ((word->flags & W_ASSIGNMENT) &&
8289 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8290 assignoff == -1 && sindex > 0)
8291 assignoff = sindex;
8292 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
8293 word->flags |= W_ITILDE;
8294 #if 0
8295 else if ((word->flags & W_ASSIGNMENT) &&
8296 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8297 string[sindex+1] == '~')
8298 word->flags |= W_ITILDE;
8299 #endif
8300 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8301 goto add_ifs_character;
8302 else
8303 goto add_character;
8304
8305 case ':':
8306 if (word->flags & W_NOTILDE)
8307 {
8308 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8309 goto add_ifs_character;
8310 else
8311 goto add_character;
8312 }
8313
8314 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
8315 string[sindex+1] == '~')
8316 word->flags |= W_ITILDE;
8317
8318 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8319 goto add_ifs_character;
8320 else
8321 goto add_character;
8322
8323 case '~':
8324 /* If the word isn't supposed to be tilde expanded, or we're not
8325 at the start of a word or after an unquoted : or = in an
8326 assignment statement, we don't do tilde expansion. */
8327 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
8328 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
8329 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8330 {
8331 word->flags &= ~W_ITILDE;
8332 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
8333 goto add_ifs_character;
8334 else
8335 goto add_character;
8336 }
8337
8338 if (word->flags & W_ASSIGNRHS)
8339 tflag = 2;
8340 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
8341 tflag = 1;
8342 else
8343 tflag = 0;
8344
8345 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
8346
8347 word->flags &= ~W_ITILDE;
8348
8349 if (temp && *temp && t_index > 0)
8350 {
8351 temp1 = bash_tilde_expand (temp, tflag);
8352 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
8353 {
8354 FREE (temp);
8355 FREE (temp1);
8356 goto add_character; /* tilde expansion failed */
8357 }
8358 free (temp);
8359 temp = temp1;
8360 sindex += t_index;
8361 goto add_quoted_string; /* XXX was add_string */
8362 }
8363 else
8364 {
8365 FREE (temp);
8366 goto add_character;
8367 }
8368
8369 case '$':
8370 if (expanded_something)
8371 *expanded_something = 1;
8372
8373 temp_has_dollar_at = 0;
8374 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
8375 if (word->flags & W_NOSPLIT2)
8376 pflags |= PF_NOSPLIT2;
8377 if (word->flags & W_ASSIGNRHS)
8378 pflags |= PF_ASSIGNRHS;
8379 tword = param_expand (string, &sindex, quoted, expanded_something,
8380 &temp_has_dollar_at, &quoted_dollar_at,
8381 &had_quoted_null, pflags);
8382 has_dollar_at += temp_has_dollar_at;
8383 split_on_spaces += (tword->flags & W_SPLITSPACE);
8384
8385 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
8386 {
8387 free (string);
8388 free (istring);
8389 return ((tword == &expand_wdesc_error) ? &expand_word_error
8390 : &expand_word_fatal);
8391 }
8392 if (contains_dollar_at && has_dollar_at)
8393 *contains_dollar_at = 1;
8394
8395 if (tword && (tword->flags & W_HASQUOTEDNULL))
8396 had_quoted_null = 1;
8397
8398 temp = tword ? tword->word : (char *)NULL;
8399 dispose_word_desc (tword);
8400
8401 /* Kill quoted nulls; we will add them back at the end of
8402 expand_word_internal if nothing else in the string */
8403 if (had_quoted_null && temp && QUOTED_NULL (temp))
8404 {
8405 FREE (temp);
8406 temp = (char *)NULL;
8407 }
8408
8409 goto add_string;
8410 break;
8411
8412 case '`': /* Backquoted command substitution. */
8413 {
8414 t_index = sindex++;
8415
8416 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
8417 /* The test of sindex against t_index is to allow bare instances of
8418 ` to pass through, for backwards compatibility. */
8419 if (temp == &extract_string_error || temp == &extract_string_fatal)
8420 {
8421 if (sindex - 1 == t_index)
8422 {
8423 sindex = t_index;
8424 goto add_character;
8425 }
8426 last_command_exit_value = EXECUTION_FAILURE;
8427 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
8428 free (string);
8429 free (istring);
8430 return ((temp == &extract_string_error) ? &expand_word_error
8431 : &expand_word_fatal);
8432 }
8433
8434 if (expanded_something)
8435 *expanded_something = 1;
8436
8437 if (word->flags & W_NOCOMSUB)
8438 /* sindex + 1 because string[sindex] == '`' */
8439 temp1 = substring (string, t_index, sindex + 1);
8440 else
8441 {
8442 de_backslash (temp);
8443 tword = command_substitute (temp, quoted);
8444 temp1 = tword ? tword->word : (char *)NULL;
8445 if (tword)
8446 dispose_word_desc (tword);
8447 }
8448 FREE (temp);
8449 temp = temp1;
8450 goto dollar_add_string;
8451 }
8452
8453 case '\\':
8454 if (string[sindex + 1] == '\n')
8455 {
8456 sindex += 2;
8457 continue;
8458 }
8459
8460 c = string[++sindex];
8461
8462 if (quoted & Q_HERE_DOCUMENT)
8463 tflag = CBSHDOC;
8464 else if (quoted & Q_DOUBLE_QUOTES)
8465 tflag = CBSDQUOTE;
8466 else
8467 tflag = 0;
8468
8469 /* From Posix discussion on austin-group list: Backslash escaping
8470 a } in ${...} is removed. Issue 0000221 */
8471 if ((quoted & Q_DOLBRACE) && c == RBRACE)
8472 {
8473 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8474 }
8475 /* This is the fix for " $@\ " */
8476 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
8477 {
8478 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8479 DEFAULT_ARRAY_SIZE);
8480 istring[istring_index++] = CTLESC;
8481 istring[istring_index++] = '\\';
8482 istring[istring_index] = '\0';
8483
8484 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8485 }
8486 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
8487 {
8488 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
8489 }
8490 else if (c == 0)
8491 {
8492 c = CTLNUL;
8493 sindex--; /* add_character: label increments sindex */
8494 goto add_character;
8495 }
8496 else
8497 {
8498 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8499 }
8500
8501 sindex++;
8502 add_twochars:
8503 /* BEFORE jumping here, we need to increment sindex if appropriate */
8504 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8505 DEFAULT_ARRAY_SIZE);
8506 istring[istring_index++] = twochars[0];
8507 istring[istring_index++] = twochars[1];
8508 istring[istring_index] = '\0';
8509
8510 break;
8511
8512 case '"':
8513 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8514 goto add_character;
8515
8516 t_index = ++sindex;
8517 temp = string_extract_double_quoted (string, &sindex, 0);
8518
8519 /* If the quotes surrounded the entire string, then the
8520 whole word was quoted. */
8521 quoted_state = (t_index == 1 && string[sindex] == '\0')
8522 ? WHOLLY_QUOTED
8523 : PARTIALLY_QUOTED;
8524
8525 if (temp && *temp)
8526 {
8527 tword = alloc_word_desc ();
8528 tword->word = temp;
8529
8530 temp = (char *)NULL;
8531
8532 temp_has_dollar_at = 0; /* XXX */
8533 /* Need to get W_HASQUOTEDNULL flag through this function. */
8534 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
8535 has_dollar_at += temp_has_dollar_at;
8536
8537 if (list == &expand_word_error || list == &expand_word_fatal)
8538 {
8539 free (istring);
8540 free (string);
8541 /* expand_word_internal has already freed temp_word->word
8542 for us because of the way it prints error messages. */
8543 tword->word = (char *)NULL;
8544 dispose_word (tword);
8545 return list;
8546 }
8547
8548 dispose_word (tword);
8549
8550 /* "$@" (a double-quoted dollar-at) expands into nothing,
8551 not even a NULL word, when there are no positional
8552 parameters. */
8553 if (list == 0 && has_dollar_at)
8554 {
8555 quoted_dollar_at++;
8556 break;
8557 }
8558
8559 /* If we get "$@", we know we have expanded something, so we
8560 need to remember it for the final split on $IFS. This is
8561 a special case; it's the only case where a quoted string
8562 can expand into more than one word. It's going to come back
8563 from the above call to expand_word_internal as a list with
8564 a single word, in which all characters are quoted and
8565 separated by blanks. What we want to do is to turn it back
8566 into a list for the next piece of code. */
8567 if (list)
8568 dequote_list (list);
8569
8570 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
8571 had_quoted_null = 1; /* XXX */
8572
8573 if (has_dollar_at)
8574 {
8575 quoted_dollar_at++;
8576 if (contains_dollar_at)
8577 *contains_dollar_at = 1;
8578 if (expanded_something)
8579 *expanded_something = 1;
8580 }
8581 }
8582 else
8583 {
8584 /* What we have is "". This is a minor optimization. */
8585 FREE (temp);
8586 list = (WORD_LIST *)NULL;
8587 }
8588
8589 /* The code above *might* return a list (consider the case of "$@",
8590 where it returns "$1", "$2", etc.). We can't throw away the
8591 rest of the list, and we have to make sure each word gets added
8592 as quoted. We test on tresult->next: if it is non-NULL, we
8593 quote the whole list, save it to a string with string_list, and
8594 add that string. We don't need to quote the results of this
8595 (and it would be wrong, since that would quote the separators
8596 as well), so we go directly to add_string. */
8597 if (list)
8598 {
8599 if (list->next)
8600 {
8601 /* Testing quoted_dollar_at makes sure that "$@" is
8602 split correctly when $IFS does not contain a space. */
8603 temp = quoted_dollar_at
8604 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES)
8605 : string_list (quote_list (list));
8606 dispose_words (list);
8607 goto add_string;
8608 }
8609 else
8610 {
8611 temp = savestring (list->word->word);
8612 tflag = list->word->flags;
8613 dispose_words (list);
8614
8615 /* If the string is not a quoted null string, we want
8616 to remove any embedded unquoted CTLNUL characters.
8617 We do not want to turn quoted null strings back into
8618 the empty string, though. We do this because we
8619 want to remove any quoted nulls from expansions that
8620 contain other characters. For example, if we have
8621 x"$*"y or "x$*y" and there are no positional parameters,
8622 the $* should expand into nothing. */
8623 /* We use the W_HASQUOTEDNULL flag to differentiate the
8624 cases: a quoted null character as above and when
8625 CTLNUL is contained in the (non-null) expansion
8626 of some variable. We use the had_quoted_null flag to
8627 pass the value through this function to its caller. */
8628 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
8629 remove_quoted_nulls (temp); /* XXX */
8630 }
8631 }
8632 else
8633 temp = (char *)NULL;
8634
8635 /* We do not want to add quoted nulls to strings that are only
8636 partially quoted; we can throw them away. The exception to
8637 this is when we are going to be performing word splitting,
8638 since we have to preserve a null argument if the next character
8639 will cause word splitting. */
8640 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
8641 continue;
8642
8643 add_quoted_string:
8644
8645 if (temp)
8646 {
8647 temp1 = temp;
8648 temp = quote_string (temp);
8649 free (temp1);
8650 goto add_string;
8651 }
8652 else
8653 {
8654 /* Add NULL arg. */
8655 c = CTLNUL;
8656 sindex--; /* add_character: label increments sindex */
8657 goto add_character;
8658 }
8659
8660 /* break; */
8661
8662 case '\'':
8663 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8664 goto add_character;
8665
8666 t_index = ++sindex;
8667 temp = string_extract_single_quoted (string, &sindex);
8668
8669 /* If the entire STRING was surrounded by single quotes,
8670 then the string is wholly quoted. */
8671 quoted_state = (t_index == 1 && string[sindex] == '\0')
8672 ? WHOLLY_QUOTED
8673 : PARTIALLY_QUOTED;
8674
8675 /* If all we had was '', it is a null expansion. */
8676 if (*temp == '\0')
8677 {
8678 free (temp);
8679 temp = (char *)NULL;
8680 }
8681 else
8682 remove_quoted_escapes (temp); /* ??? */
8683
8684 /* We do not want to add quoted nulls to strings that are only
8685 partially quoted; such nulls are discarded. */
8686 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED))
8687 continue;
8688
8689 /* If we have a quoted null expansion, add a quoted NULL to istring. */
8690 if (temp == 0)
8691 {
8692 c = CTLNUL;
8693 sindex--; /* add_character: label increments sindex */
8694 goto add_character;
8695 }
8696 else
8697 goto add_quoted_string;
8698
8699 /* break; */
8700
8701 default:
8702 /* This is the fix for " $@ " */
8703 add_ifs_character:
8704 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c)))
8705 {
8706 if (string[sindex]) /* from old goto dollar_add_string */
8707 sindex++;
8708 if (c == 0)
8709 {
8710 c = CTLNUL;
8711 goto add_character;
8712 }
8713 else
8714 {
8715 #if HANDLE_MULTIBYTE
8716 if (MB_CUR_MAX > 1)
8717 sindex--;
8718
8719 if (MB_CUR_MAX > 1)
8720 {
8721 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8722 }
8723 else
8724 #endif
8725 {
8726 twochars[0] = CTLESC;
8727 twochars[1] = c;
8728 goto add_twochars;
8729 }
8730 }
8731 }
8732
8733 SADD_MBCHAR (temp, string, sindex, string_size);
8734
8735 add_character:
8736 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
8737 DEFAULT_ARRAY_SIZE);
8738 istring[istring_index++] = c;
8739 istring[istring_index] = '\0';
8740
8741 /* Next character. */
8742 sindex++;
8743 }
8744 }
8745
8746 finished_with_string:
8747 /* OK, we're ready to return. If we have a quoted string, and
8748 quoted_dollar_at is not set, we do no splitting at all; otherwise
8749 we split on ' '. The routines that call this will handle what to
8750 do if nothing has been expanded. */
8751
8752 /* Partially and wholly quoted strings which expand to the empty
8753 string are retained as an empty arguments. Unquoted strings
8754 which expand to the empty string are discarded. The single
8755 exception is the case of expanding "$@" when there are no
8756 positional parameters. In that case, we discard the expansion. */
8757
8758 /* Because of how the code that handles "" and '' in partially
8759 quoted strings works, we need to make ISTRING into a QUOTED_NULL
8760 if we saw quoting characters, but the expansion was empty.
8761 "" and '' are tossed away before we get to this point when
8762 processing partially quoted strings. This makes "" and $xxx""
8763 equivalent when xxx is unset. We also look to see whether we
8764 saw a quoted null from a ${} expansion and add one back if we
8765 need to. */
8766
8767 /* If we expand to nothing and there were no single or double quotes
8768 in the word, we throw it away. Otherwise, we return a NULL word.
8769 The single exception is for $@ surrounded by double quotes when
8770 there are no positional parameters. In that case, we also throw
8771 the word away. */
8772
8773 if (*istring == '\0')
8774 {
8775 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
8776 {
8777 istring[0] = CTLNUL;
8778 istring[1] = '\0';
8779 tword = make_bare_word (istring);
8780 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8781 list = make_word_list (tword, (WORD_LIST *)NULL);
8782 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8783 tword->flags |= W_QUOTED;
8784 }
8785 /* According to sh, ksh, and Posix.2, if a word expands into nothing
8786 and a double-quoted "$@" appears anywhere in it, then the entire
8787 word is removed. */
8788 else if (quoted_state == UNQUOTED || quoted_dollar_at)
8789 list = (WORD_LIST *)NULL;
8790 #if 0
8791 else
8792 {
8793 tword = make_bare_word (istring);
8794 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8795 tword->flags |= W_QUOTED;
8796 list = make_word_list (tword, (WORD_LIST *)NULL);
8797 }
8798 #else
8799 else
8800 list = (WORD_LIST *)NULL;
8801 #endif
8802 }
8803 else if (word->flags & W_NOSPLIT)
8804 {
8805 tword = make_bare_word (istring);
8806 if (word->flags & W_ASSIGNMENT)
8807 tword->flags |= W_ASSIGNMENT; /* XXX */
8808 if (word->flags & W_COMPASSIGN)
8809 tword->flags |= W_COMPASSIGN; /* XXX */
8810 if (word->flags & W_NOGLOB)
8811 tword->flags |= W_NOGLOB; /* XXX */
8812 if (word->flags & W_NOBRACE)
8813 tword->flags |= W_NOBRACE; /* XXX */
8814 if (word->flags & W_NOEXPAND)
8815 tword->flags |= W_NOEXPAND; /* XXX */
8816 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8817 tword->flags |= W_QUOTED;
8818 if (had_quoted_null && QUOTED_NULL (istring))
8819 tword->flags |= W_HASQUOTEDNULL;
8820 list = make_word_list (tword, (WORD_LIST *)NULL);
8821 }
8822 else
8823 {
8824 char *ifs_chars;
8825 char *tstring;
8826
8827 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
8828
8829 /* If we have $@, we need to split the results no matter what. If
8830 IFS is unset or NULL, string_list_dollar_at has separated the
8831 positional parameters with a space, so we split on space (we have
8832 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
8833 string_list_dollar_at has separated the positional parameters
8834 with the first character of $IFS, so we split on $IFS. If
8835 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
8836 unset or null, and we want to make sure that we split on spaces
8837 regardless of what else has happened to IFS since the expansion. */
8838 if (split_on_spaces)
8839 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
8840 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
8841 don't want to split the result (W_NOSPLIT2), and we are not quoted,
8842 we have already separated the arguments with the first character of
8843 $IFS. In this case, we want to return a list with a single word
8844 with the separator possibly replaced with a space (it's what other
8845 shells seem to do).
8846 quoted_dollar_at is internal to this function and is set if we are
8847 passed an argument that is unquoted (quoted == 0) but we encounter a
8848 double-quoted $@ while expanding it. */
8849 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
8850 {
8851 /* Only split and rejoin if we have to */
8852 if (*ifs_chars && *ifs_chars != ' ')
8853 {
8854 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
8855 tstring = string_list (list);
8856 }
8857 else
8858 tstring = istring;
8859 tword = make_bare_word (tstring);
8860 if (tstring != istring)
8861 free (tstring);
8862 goto set_word_flags;
8863 }
8864 else if (has_dollar_at && ifs_chars)
8865 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
8866 else
8867 {
8868 tword = make_bare_word (istring);
8869 set_word_flags:
8870 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
8871 tword->flags |= W_QUOTED;
8872 if (word->flags & W_ASSIGNMENT)
8873 tword->flags |= W_ASSIGNMENT;
8874 if (word->flags & W_COMPASSIGN)
8875 tword->flags |= W_COMPASSIGN;
8876 if (word->flags & W_NOGLOB)
8877 tword->flags |= W_NOGLOB;
8878 if (word->flags & W_NOBRACE)
8879 tword->flags |= W_NOBRACE;
8880 if (word->flags & W_NOEXPAND)
8881 tword->flags |= W_NOEXPAND;
8882 if (had_quoted_null && QUOTED_NULL (istring))
8883 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8884 list = make_word_list (tword, (WORD_LIST *)NULL);
8885 }
8886 }
8887
8888 free (istring);
8889 return (list);
8890 }
8891
8892 /* **************************************************************** */
8893 /* */
8894 /* Functions for Quote Removal */
8895 /* */
8896 /* **************************************************************** */
8897
8898 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
8899 backslash quoting rules for within double quotes or a here document. */
8900 char *
8901 string_quote_removal (string, quoted)
8902 char *string;
8903 int quoted;
8904 {
8905 size_t slen;
8906 char *r, *result_string, *temp, *send;
8907 int sindex, tindex, dquote;
8908 unsigned char c;
8909 DECLARE_MBSTATE;
8910
8911 /* The result can be no longer than the original string. */
8912 slen = strlen (string);
8913 send = string + slen;
8914
8915 r = result_string = (char *)xmalloc (slen + 1);
8916
8917 for (dquote = sindex = 0; c = string[sindex];)
8918 {
8919 switch (c)
8920 {
8921 case '\\':
8922 c = string[++sindex];
8923 if (c == 0)
8924 {
8925 *r++ = '\\';
8926 break;
8927 }
8928 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
8929 *r++ = '\\';
8930 /* FALLTHROUGH */
8931
8932 default:
8933 SCOPY_CHAR_M (r, string, send, sindex);
8934 break;
8935
8936 case '\'':
8937 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
8938 {
8939 *r++ = c;
8940 sindex++;
8941 break;
8942 }
8943 tindex = sindex + 1;
8944 temp = string_extract_single_quoted (string, &tindex);
8945 if (temp)
8946 {
8947 strcpy (r, temp);
8948 r += strlen (r);
8949 free (temp);
8950 }
8951 sindex = tindex;
8952 break;
8953
8954 case '"':
8955 dquote = 1 - dquote;
8956 sindex++;
8957 break;
8958 }
8959 }
8960 *r = '\0';
8961 return (result_string);
8962 }
8963
8964 #if 0
8965 /* UNUSED */
8966 /* Perform quote removal on word WORD. This allocates and returns a new
8967 WORD_DESC *. */
8968 WORD_DESC *
8969 word_quote_removal (word, quoted)
8970 WORD_DESC *word;
8971 int quoted;
8972 {
8973 WORD_DESC *w;
8974 char *t;
8975
8976 t = string_quote_removal (word->word, quoted);
8977 w = alloc_word_desc ();
8978 w->word = t ? t : savestring ("");
8979 return (w);
8980 }
8981
8982 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
8983 the members of the list are treated as if they are surrounded by
8984 double quotes. Return a new list, or NULL if LIST is NULL. */
8985 WORD_LIST *
8986 word_list_quote_removal (list, quoted)
8987 WORD_LIST *list;
8988 int quoted;
8989 {
8990 WORD_LIST *result, *t, *tresult, *e;
8991
8992 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
8993 {
8994 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
8995 #if 0
8996 result = (WORD_LIST *) list_append (result, tresult);
8997 #else
8998 if (result == 0)
8999 result = e = tresult;
9000 else
9001 {
9002 e->next = tresult;
9003 while (e->next)
9004 e = e->next;
9005 }
9006 #endif
9007 }
9008 return (result);
9009 }
9010 #endif
9011
9012 /*******************************************
9013 * *
9014 * Functions to perform word splitting *
9015 * *
9016 *******************************************/
9017
9018 void
9019 setifs (v)
9020 SHELL_VAR *v;
9021 {
9022 char *t;
9023 unsigned char uc;
9024
9025 ifs_var = v;
9026 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
9027
9028 ifs_is_set = ifs_var != 0;
9029 ifs_is_null = ifs_is_set && (*ifs_value == 0);
9030
9031 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9032 handle multibyte chars in IFS */
9033 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9034 for (t = ifs_value ; t && *t; t++)
9035 {
9036 uc = *t;
9037 ifs_cmap[uc] = 1;
9038 }
9039
9040 #if defined (HANDLE_MULTIBYTE)
9041 if (ifs_value == 0)
9042 {
9043 ifs_firstc[0] = '\0';
9044 ifs_firstc_len = 1;
9045 }
9046 else
9047 {
9048 size_t ifs_len;
9049 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
9050 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
9051 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
9052 {
9053 ifs_firstc[0] = ifs_value[0];
9054 ifs_firstc[1] = '\0';
9055 ifs_firstc_len = 1;
9056 }
9057 else
9058 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
9059 }
9060 #else
9061 ifs_firstc = ifs_value ? *ifs_value : 0;
9062 #endif
9063 }
9064
9065 char *
9066 getifs ()
9067 {
9068 return ifs_value;
9069 }
9070
9071 /* This splits a single word into a WORD LIST on $IFS, but only if the word
9072 is not quoted. list_string () performs quote removal for us, even if we
9073 don't do any splitting. */
9074 WORD_LIST *
9075 word_split (w, ifs_chars)
9076 WORD_DESC *w;
9077 char *ifs_chars;
9078 {
9079 WORD_LIST *result;
9080
9081 if (w)
9082 {
9083 char *xifs;
9084
9085 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
9086 result = list_string (w->word, xifs, w->flags & W_QUOTED);
9087 }
9088 else
9089 result = (WORD_LIST *)NULL;
9090
9091 return (result);
9092 }
9093
9094 /* Perform word splitting on LIST and return the RESULT. It is possible
9095 to return (WORD_LIST *)NULL. */
9096 static WORD_LIST *
9097 word_list_split (list)
9098 WORD_LIST *list;
9099 {
9100 WORD_LIST *result, *t, *tresult, *e;
9101
9102 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
9103 {
9104 tresult = word_split (t->word, ifs_value);
9105 if (result == 0)
9106 result = e = tresult;
9107 else
9108 {
9109 e->next = tresult;
9110 while (e->next)
9111 e = e->next;
9112 }
9113 }
9114 return (result);
9115 }
9116
9117 /**************************************************
9118 * *
9119 * Functions to expand an entire WORD_LIST *
9120 * *
9121 **************************************************/
9122
9123 /* Do any word-expansion-specific cleanup and jump to top_level */
9124 static void
9125 exp_jump_to_top_level (v)
9126 int v;
9127 {
9128 set_pipestatus_from_exit (last_command_exit_value);
9129
9130 /* Cleanup code goes here. */
9131 expand_no_split_dollar_star = 0; /* XXX */
9132 expanding_redir = 0;
9133 assigning_in_environment = 0;
9134
9135 if (parse_and_execute_level == 0)
9136 top_level_cleanup (); /* from sig.c */
9137
9138 jump_to_top_level (v);
9139 }
9140
9141 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
9142 ELIST, and set ELIST to the new list. */
9143 #define PREPEND_LIST(nlist, elist) \
9144 do { nlist->next = elist; elist = nlist; } while (0)
9145
9146 /* Separate out any initial variable assignments from TLIST. If set -k has
9147 been executed, remove all assignment statements from TLIST. Initial
9148 variable assignments and other environment assignments are placed
9149 on SUBST_ASSIGN_VARLIST. */
9150 static WORD_LIST *
9151 separate_out_assignments (tlist)
9152 WORD_LIST *tlist;
9153 {
9154 register WORD_LIST *vp, *lp;
9155
9156 if (tlist == 0)
9157 return ((WORD_LIST *)NULL);
9158
9159 if (subst_assign_varlist)
9160 dispose_words (subst_assign_varlist); /* Clean up after previous error */
9161
9162 subst_assign_varlist = (WORD_LIST *)NULL;
9163 vp = lp = tlist;
9164
9165 /* Separate out variable assignments at the start of the command.
9166 Loop invariant: vp->next == lp
9167 Loop postcondition:
9168 lp = list of words left after assignment statements skipped
9169 tlist = original list of words
9170 */
9171 while (lp && (lp->word->flags & W_ASSIGNMENT))
9172 {
9173 vp = lp;
9174 lp = lp->next;
9175 }
9176
9177 /* If lp != tlist, we have some initial assignment statements.
9178 We make SUBST_ASSIGN_VARLIST point to the list of assignment
9179 words and TLIST point to the remaining words. */
9180 if (lp != tlist)
9181 {
9182 subst_assign_varlist = tlist;
9183 /* ASSERT(vp->next == lp); */
9184 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
9185 tlist = lp; /* remainder of word list */
9186 }
9187
9188 /* vp == end of variable list */
9189 /* tlist == remainder of original word list without variable assignments */
9190 if (!tlist)
9191 /* All the words in tlist were assignment statements */
9192 return ((WORD_LIST *)NULL);
9193
9194 /* ASSERT(tlist != NULL); */
9195 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
9196
9197 /* If the -k option is in effect, we need to go through the remaining
9198 words, separate out the assignment words, and place them on
9199 SUBST_ASSIGN_VARLIST. */
9200 if (place_keywords_in_env)
9201 {
9202 WORD_LIST *tp; /* tp == running pointer into tlist */
9203
9204 tp = tlist;
9205 lp = tlist->next;
9206
9207 /* Loop Invariant: tp->next == lp */
9208 /* Loop postcondition: tlist == word list without assignment statements */
9209 while (lp)
9210 {
9211 if (lp->word->flags & W_ASSIGNMENT)
9212 {
9213 /* Found an assignment statement, add this word to end of
9214 subst_assign_varlist (vp). */
9215 if (!subst_assign_varlist)
9216 subst_assign_varlist = vp = lp;
9217 else
9218 {
9219 vp->next = lp;
9220 vp = lp;
9221 }
9222
9223 /* Remove the word pointed to by LP from TLIST. */
9224 tp->next = lp->next;
9225 /* ASSERT(vp == lp); */
9226 lp->next = (WORD_LIST *)NULL;
9227 lp = tp->next;
9228 }
9229 else
9230 {
9231 tp = lp;
9232 lp = lp->next;
9233 }
9234 }
9235 }
9236 return (tlist);
9237 }
9238
9239 #define WEXP_VARASSIGN 0x001
9240 #define WEXP_BRACEEXP 0x002
9241 #define WEXP_TILDEEXP 0x004
9242 #define WEXP_PARAMEXP 0x008
9243 #define WEXP_PATHEXP 0x010
9244
9245 /* All of the expansions, including variable assignments at the start of
9246 the list. */
9247 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
9248
9249 /* All of the expansions except variable assignments at the start of
9250 the list. */
9251 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
9252
9253 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
9254 expansion, command substitution, arithmetic expansion, word splitting, and
9255 quote removal. */
9256 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
9257
9258 /* Take the list of words in LIST and do the various substitutions. Return
9259 a new list of words which is the expanded list, and without things like
9260 variable assignments. */
9261
9262 WORD_LIST *
9263 expand_words (list)
9264 WORD_LIST *list;
9265 {
9266 return (expand_word_list_internal (list, WEXP_ALL));
9267 }
9268
9269 /* Same as expand_words (), but doesn't hack variable or environment
9270 variables. */
9271 WORD_LIST *
9272 expand_words_no_vars (list)
9273 WORD_LIST *list;
9274 {
9275 return (expand_word_list_internal (list, WEXP_NOVARS));
9276 }
9277
9278 WORD_LIST *
9279 expand_words_shellexp (list)
9280 WORD_LIST *list;
9281 {
9282 return (expand_word_list_internal (list, WEXP_SHELLEXP));
9283 }
9284
9285 static WORD_LIST *
9286 glob_expand_word_list (tlist, eflags)
9287 WORD_LIST *tlist;
9288 int eflags;
9289 {
9290 char **glob_array, *temp_string;
9291 register int glob_index;
9292 WORD_LIST *glob_list, *output_list, *disposables, *next;
9293 WORD_DESC *tword;
9294
9295 output_list = disposables = (WORD_LIST *)NULL;
9296 glob_array = (char **)NULL;
9297 while (tlist)
9298 {
9299 /* For each word, either globbing is attempted or the word is
9300 added to orig_list. If globbing succeeds, the results are
9301 added to orig_list and the word (tlist) is added to the list
9302 of disposable words. If globbing fails and failed glob
9303 expansions are left unchanged (the shell default), the
9304 original word is added to orig_list. If globbing fails and
9305 failed glob expansions are removed, the original word is
9306 added to the list of disposable words. orig_list ends up
9307 in reverse order and requires a call to REVERSE_LIST to
9308 be set right. After all words are examined, the disposable
9309 words are freed. */
9310 next = tlist->next;
9311
9312 /* If the word isn't an assignment and contains an unquoted
9313 pattern matching character, then glob it. */
9314 if ((tlist->word->flags & W_NOGLOB) == 0 &&
9315 unquoted_glob_pattern_p (tlist->word->word))
9316 {
9317 glob_array = shell_glob_filename (tlist->word->word);
9318
9319 /* Handle error cases.
9320 I don't think we should report errors like "No such file
9321 or directory". However, I would like to report errors
9322 like "Read failed". */
9323
9324 if (glob_array == 0 || GLOB_FAILED (glob_array))
9325 {
9326 glob_array = (char **)xmalloc (sizeof (char *));
9327 glob_array[0] = (char *)NULL;
9328 }
9329
9330 /* Dequote the current word in case we have to use it. */
9331 if (glob_array[0] == NULL)
9332 {
9333 temp_string = dequote_string (tlist->word->word);
9334 free (tlist->word->word);
9335 tlist->word->word = temp_string;
9336 }
9337
9338 /* Make the array into a word list. */
9339 glob_list = (WORD_LIST *)NULL;
9340 for (glob_index = 0; glob_array[glob_index]; glob_index++)
9341 {
9342 tword = make_bare_word (glob_array[glob_index]);
9343 glob_list = make_word_list (tword, glob_list);
9344 }
9345
9346 if (glob_list)
9347 {
9348 output_list = (WORD_LIST *)list_append (glob_list, output_list);
9349 PREPEND_LIST (tlist, disposables);
9350 }
9351 else if (fail_glob_expansion != 0)
9352 {
9353 last_command_exit_value = EXECUTION_FAILURE;
9354 report_error (_("no match: %s"), tlist->word->word);
9355 exp_jump_to_top_level (DISCARD);
9356 }
9357 else if (allow_null_glob_expansion == 0)
9358 {
9359 /* Failed glob expressions are left unchanged. */
9360 PREPEND_LIST (tlist, output_list);
9361 }
9362 else
9363 {
9364 /* Failed glob expressions are removed. */
9365 PREPEND_LIST (tlist, disposables);
9366 }
9367 }
9368 else
9369 {
9370 /* Dequote the string. */
9371 temp_string = dequote_string (tlist->word->word);
9372 free (tlist->word->word);
9373 tlist->word->word = temp_string;
9374 PREPEND_LIST (tlist, output_list);
9375 }
9376
9377 strvec_dispose (glob_array);
9378 glob_array = (char **)NULL;
9379
9380 tlist = next;
9381 }
9382
9383 if (disposables)
9384 dispose_words (disposables);
9385
9386 if (output_list)
9387 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9388
9389 return (output_list);
9390 }
9391
9392 #if defined (BRACE_EXPANSION)
9393 static WORD_LIST *
9394 brace_expand_word_list (tlist, eflags)
9395 WORD_LIST *tlist;
9396 int eflags;
9397 {
9398 register char **expansions;
9399 char *temp_string;
9400 WORD_LIST *disposables, *output_list, *next;
9401 WORD_DESC *w;
9402 int eindex;
9403
9404 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
9405 {
9406 next = tlist->next;
9407
9408 if (tlist->word->flags & W_NOBRACE)
9409 {
9410 /*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
9411 PREPEND_LIST (tlist, output_list);
9412 continue;
9413 }
9414
9415 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9416 {
9417 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
9418 PREPEND_LIST (tlist, output_list);
9419 continue;
9420 }
9421
9422 /* Only do brace expansion if the word has a brace character. If
9423 not, just add the word list element to BRACES and continue. In
9424 the common case, at least when running shell scripts, this will
9425 degenerate to a bunch of calls to `mbschr', and then what is
9426 basically a reversal of TLIST into BRACES, which is corrected
9427 by a call to REVERSE_LIST () on BRACES when the end of TLIST
9428 is reached. */
9429 if (mbschr (tlist->word->word, LBRACE))
9430 {
9431 expansions = brace_expand (tlist->word->word);
9432
9433 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
9434 {
9435 w = alloc_word_desc ();
9436 w->word = temp_string;
9437
9438 /* If brace expansion didn't change the word, preserve
9439 the flags. We may want to preserve the flags
9440 unconditionally someday -- XXX */
9441 if (STREQ (temp_string, tlist->word->word))
9442 w->flags = tlist->word->flags;
9443 else
9444 w = make_word_flags (w, temp_string);
9445
9446 output_list = make_word_list (w, output_list);
9447 }
9448 free (expansions);
9449
9450 /* Add TLIST to the list of words to be freed after brace
9451 expansion has been performed. */
9452 PREPEND_LIST (tlist, disposables);
9453 }
9454 else
9455 PREPEND_LIST (tlist, output_list);
9456 }
9457
9458 if (disposables)
9459 dispose_words (disposables);
9460
9461 if (output_list)
9462 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9463
9464 return (output_list);
9465 }
9466 #endif
9467
9468 #if defined (ARRAY_VARS)
9469 /* Take WORD, a compound associative array assignment, and internally run
9470 'declare -A w', where W is the variable name portion of WORD. */
9471 static int
9472 make_internal_declare (word, option)
9473 char *word;
9474 char *option;
9475 {
9476 int t;
9477 WORD_LIST *wl;
9478 WORD_DESC *w;
9479
9480 w = make_word (word);
9481
9482 t = assignment (w->word, 0);
9483 w->word[t] = '\0';
9484
9485 wl = make_word_list (w, (WORD_LIST *)NULL);
9486 wl = make_word_list (make_word (option), wl);
9487
9488 return (declare_builtin (wl));
9489 }
9490 #endif
9491
9492 static WORD_LIST *
9493 shell_expand_word_list (tlist, eflags)
9494 WORD_LIST *tlist;
9495 int eflags;
9496 {
9497 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list;
9498 int expanded_something, has_dollar_at;
9499 char *temp_string;
9500
9501 /* We do tilde expansion all the time. This is what 1003.2 says. */
9502 new_list = (WORD_LIST *)NULL;
9503 for (orig_list = tlist; tlist; tlist = next)
9504 {
9505 temp_string = tlist->word->word;
9506
9507 next = tlist->next;
9508
9509 #if defined (ARRAY_VARS)
9510 /* If this is a compound array assignment to a builtin that accepts
9511 such assignments (e.g., `declare'), take the assignment and perform
9512 it separately, handling the semantics of declarations inside shell
9513 functions. This avoids the double-evaluation of such arguments,
9514 because `declare' does some evaluation of compound assignments on
9515 its own. */
9516 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9517 {
9518 int t;
9519 char opts[8], opti;
9520
9521 opti = 0;
9522 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
9523 opts[opti++] = '-';
9524
9525 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
9526 {
9527 opts[opti++] = 'g';
9528 opts[opti++] = 'A';
9529 }
9530 else if (tlist->word->flags & W_ASSIGNASSOC)
9531 opts[opti++] = 'A';
9532 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
9533 {
9534 opts[opti++] = 'g';
9535 opts[opti++] = 'a';
9536 }
9537 else if (tlist->word->flags & W_ASSIGNARRAY)
9538 opts[opti++] = 'a';
9539 else if (tlist->word->flags & W_ASSNGLOBAL)
9540 opts[opti++] = 'g';
9541
9542 #if 0
9543 /* If we have special handling note the integer attribute */
9544 if (opti > 0 && (tlist->word->flags & W_ASSIGNINT))
9545 opts[opti++] = 'i';
9546 #endif
9547
9548 opts[opti] = '\0';
9549 if (opti > 0)
9550 make_internal_declare (tlist->word->word, opts);
9551
9552 t = do_word_assignment (tlist->word, 0);
9553 if (t == 0)
9554 {
9555 last_command_exit_value = EXECUTION_FAILURE;
9556 exp_jump_to_top_level (DISCARD);
9557 }
9558
9559 /* Now transform the word as ksh93 appears to do and go on */
9560 t = assignment (tlist->word->word, 0);
9561 tlist->word->word[t] = '\0';
9562 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
9563 }
9564 #endif
9565
9566 expanded_something = 0;
9567 expanded = expand_word_internal
9568 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
9569
9570 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
9571 {
9572 /* By convention, each time this error is returned,
9573 tlist->word->word has already been freed. */
9574 tlist->word->word = (char *)NULL;
9575
9576 /* Dispose our copy of the original list. */
9577 dispose_words (orig_list);
9578 /* Dispose the new list we're building. */
9579 dispose_words (new_list);
9580
9581 last_command_exit_value = EXECUTION_FAILURE;
9582 if (expanded == &expand_word_error)
9583 exp_jump_to_top_level (DISCARD);
9584 else
9585 exp_jump_to_top_level (FORCE_EOF);
9586 }
9587
9588 /* Don't split words marked W_NOSPLIT. */
9589 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
9590 {
9591 temp_list = word_list_split (expanded);
9592 dispose_words (expanded);
9593 }
9594 else
9595 {
9596 /* If no parameter expansion, command substitution, process
9597 substitution, or arithmetic substitution took place, then
9598 do not do word splitting. We still have to remove quoted
9599 null characters from the result. */
9600 word_list_remove_quoted_nulls (expanded);
9601 temp_list = expanded;
9602 }
9603
9604 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
9605 new_list = (WORD_LIST *)list_append (expanded, new_list);
9606 }
9607
9608 if (orig_list)
9609 dispose_words (orig_list);
9610
9611 if (new_list)
9612 new_list = REVERSE_LIST (new_list, WORD_LIST *);
9613
9614 return (new_list);
9615 }
9616
9617 /* The workhorse for expand_words () and expand_words_no_vars ().
9618 First arg is LIST, a WORD_LIST of words.
9619 Second arg EFLAGS is a flags word controlling which expansions are
9620 performed.
9621
9622 This does all of the substitutions: brace expansion, tilde expansion,
9623 parameter expansion, command substitution, arithmetic expansion,
9624 process substitution, word splitting, and pathname expansion, according
9625 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
9626 set, or for which no expansion is done, do not undergo word splitting.
9627 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
9628 with W_NOBRACE set do not undergo brace expansion (see
9629 brace_expand_word_list above). */
9630 static WORD_LIST *
9631 expand_word_list_internal (list, eflags)
9632 WORD_LIST *list;
9633 int eflags;
9634 {
9635 WORD_LIST *new_list, *temp_list;
9636 int tint;
9637
9638 tempenv_assign_error = 0;
9639 if (list == 0)
9640 return ((WORD_LIST *)NULL);
9641
9642 garglist = new_list = copy_word_list (list);
9643 if (eflags & WEXP_VARASSIGN)
9644 {
9645 garglist = new_list = separate_out_assignments (new_list);
9646 if (new_list == 0)
9647 {
9648 if (subst_assign_varlist)
9649 {
9650 /* All the words were variable assignments, so they are placed
9651 into the shell's environment. */
9652 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9653 {
9654 this_command_name = (char *)NULL; /* no arithmetic errors */
9655 tint = do_word_assignment (temp_list->word, 0);
9656 /* Variable assignment errors in non-interactive shells
9657 running in Posix.2 mode cause the shell to exit. */
9658 if (tint == 0)
9659 {
9660 last_command_exit_value = EXECUTION_FAILURE;
9661 if (interactive_shell == 0 && posixly_correct)
9662 exp_jump_to_top_level (FORCE_EOF);
9663 else
9664 exp_jump_to_top_level (DISCARD);
9665 }
9666 }
9667 dispose_words (subst_assign_varlist);
9668 subst_assign_varlist = (WORD_LIST *)NULL;
9669 }
9670 return ((WORD_LIST *)NULL);
9671 }
9672 }
9673
9674 /* Begin expanding the words that remain. The expansions take place on
9675 things that aren't really variable assignments. */
9676
9677 #if defined (BRACE_EXPANSION)
9678 /* Do brace expansion on this word if there are any brace characters
9679 in the string. */
9680 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
9681 new_list = brace_expand_word_list (new_list, eflags);
9682 #endif /* BRACE_EXPANSION */
9683
9684 /* Perform the `normal' shell expansions: tilde expansion, parameter and
9685 variable substitution, command substitution, arithmetic expansion,
9686 and word splitting. */
9687 new_list = shell_expand_word_list (new_list, eflags);
9688
9689 /* Okay, we're almost done. Now let's just do some filename
9690 globbing. */
9691 if (new_list)
9692 {
9693 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
9694 /* Glob expand the word list unless globbing has been disabled. */
9695 new_list = glob_expand_word_list (new_list, eflags);
9696 else
9697 /* Dequote the words, because we're not performing globbing. */
9698 new_list = dequote_list (new_list);
9699 }
9700
9701 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
9702 {
9703 sh_wassign_func_t *assign_func;
9704 int is_special_builtin, is_builtin_or_func;
9705
9706 /* If the remainder of the words expand to nothing, Posix.2 requires
9707 that the variable and environment assignments affect the shell's
9708 environment. */
9709 assign_func = new_list ? assign_in_env : do_word_assignment;
9710 tempenv_assign_error = 0;
9711
9712 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
9713 /* Posix says that special builtins exit if a variable assignment error
9714 occurs in an assignment preceding it. */
9715 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
9716
9717 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9718 {
9719 this_command_name = (char *)NULL;
9720 assigning_in_environment = (assign_func == assign_in_env);
9721 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
9722 assigning_in_environment = 0;
9723 /* Variable assignment errors in non-interactive shells running
9724 in Posix.2 mode cause the shell to exit. */
9725 if (tint == 0)
9726 {
9727 if (assign_func == do_word_assignment)
9728 {
9729 last_command_exit_value = EXECUTION_FAILURE;
9730 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
9731 exp_jump_to_top_level (FORCE_EOF);
9732 else
9733 exp_jump_to_top_level (DISCARD);
9734 }
9735 else
9736 tempenv_assign_error++;
9737 }
9738 }
9739
9740 dispose_words (subst_assign_varlist);
9741 subst_assign_varlist = (WORD_LIST *)NULL;
9742 }
9743
9744 return (new_list);
9745 }