]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
Bash-4.2 patch 4
[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) 1987-2010 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
93 /* These defs make it easier to use the editor. */
94 #define LBRACE '{'
95 #define RBRACE '}'
96 #define LPAREN '('
97 #define RPAREN ')'
98
99 #if defined (HANDLE_MULTIBYTE)
100 #define WLPAREN L'('
101 #define WRPAREN L')'
102 #endif
103
104 /* Evaluates to 1 if C is one of the shell's special parameters whose length
105 can be taken, but is also one of the special expansion characters. */
106 #define VALID_SPECIAL_LENGTH_PARAM(c) \
107 ((c) == '-' || (c) == '?' || (c) == '#')
108
109 /* Evaluates to 1 if C is one of the shell's special parameters for which an
110 indirect variable reference may be made. */
111 #define VALID_INDIR_PARAM(c) \
112 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
113
114 /* Evaluates to 1 if C is one of the OP characters that follows the parameter
115 in ${parameter[:]OPword}. */
116 #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
117
118 /* Evaluates to 1 if this is one of the shell's special variables. */
119 #define SPECIAL_VAR(name, wi) \
120 ((DIGIT (*name) && all_digits (name)) || \
121 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
122 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1])))
123
124 /* An expansion function that takes a string and a quoted flag and returns
125 a WORD_LIST *. Used as the type of the third argument to
126 expand_string_if_necessary(). */
127 typedef WORD_LIST *EXPFUNC __P((char *, int));
128
129 /* Process ID of the last command executed within command substitution. */
130 pid_t last_command_subst_pid = NO_PID;
131 pid_t current_command_subst_pid = NO_PID;
132
133 /* Variables used to keep track of the characters in IFS. */
134 SHELL_VAR *ifs_var;
135 char *ifs_value;
136 unsigned char ifs_cmap[UCHAR_MAX + 1];
137
138 #if defined (HANDLE_MULTIBYTE)
139 unsigned char ifs_firstc[MB_LEN_MAX];
140 size_t ifs_firstc_len;
141 #else
142 unsigned char ifs_firstc;
143 #endif
144
145 /* Sentinel to tell when we are performing variable assignments preceding a
146 command name and putting them into the environment. Used to make sure
147 we use the temporary environment when looking up variable values. */
148 int assigning_in_environment;
149
150 /* Used to hold a list of variable assignments preceding a command. Global
151 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
152 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
153 WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
154
155 /* Extern functions and variables from different files. */
156 extern int last_command_exit_value, last_command_exit_signal;
157 extern int subshell_environment, line_number;
158 extern int subshell_level, parse_and_execute_level, sourcelevel;
159 extern int eof_encountered;
160 extern int return_catch_flag, return_catch_value;
161 extern pid_t dollar_dollar_pid;
162 extern int posixly_correct;
163 extern char *this_command_name;
164 extern struct fd_bitmap *current_fds_to_close;
165 extern int wordexp_only;
166 extern int expanding_redir;
167 extern int tempenv_assign_error;
168
169 #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
170 extern wchar_t *wcsdup __P((const wchar_t *));
171 #endif
172
173 /* Non-zero means to allow unmatched globbed filenames to expand to
174 a null file. */
175 int allow_null_glob_expansion;
176
177 /* Non-zero means to throw an error when globbing fails to match anything. */
178 int fail_glob_expansion;
179
180 #if 0
181 /* Variables to keep track of which words in an expanded word list (the
182 output of expand_word_list_internal) are the result of globbing
183 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
184 (CURRENTLY UNUSED). */
185 char *glob_argv_flags;
186 static int glob_argv_flags_size;
187 #endif
188
189 static WORD_LIST expand_word_error, expand_word_fatal;
190 static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
191 static char expand_param_error, expand_param_fatal;
192 static char extract_string_error, extract_string_fatal;
193
194 /* Tell the expansion functions to not longjmp back to top_level on fatal
195 errors. Enabled when doing completion and prompt string expansion. */
196 static int no_longjmp_on_fatal_error = 0;
197
198 /* Set by expand_word_unsplit; used to inhibit splitting and re-joining
199 $* on $IFS, primarily when doing assignment statements. */
200 static int expand_no_split_dollar_star = 0;
201
202 /* A WORD_LIST of words to be expanded by expand_word_list_internal,
203 without any leading variable assignments. */
204 static WORD_LIST *garglist = (WORD_LIST *)NULL;
205
206 static char *quoted_substring __P((char *, int, int));
207 static int quoted_strlen __P((char *));
208 static char *quoted_strchr __P((char *, int, int));
209
210 static char *expand_string_if_necessary __P((char *, int, EXPFUNC *));
211 static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *));
212 static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
213 static WORD_LIST *expand_string_internal __P((char *, int));
214 static WORD_LIST *expand_string_leave_quoted __P((char *, int));
215 static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *));
216
217 static WORD_LIST *list_quote_escapes __P((WORD_LIST *));
218 static char *make_quoted_char __P((int));
219 static WORD_LIST *quote_list __P((WORD_LIST *));
220
221 static int unquoted_substring __P((char *, char *));
222 static int unquoted_member __P((int, char *));
223
224 #if defined (ARRAY_VARS)
225 static SHELL_VAR *do_compound_assignment __P((char *, char *, int));
226 #endif
227 static int do_assignment_internal __P((const WORD_DESC *, int));
228
229 static char *string_extract_verbatim __P((char *, size_t, int *, char *, int));
230 static char *string_extract __P((char *, int *, char *, int));
231 static char *string_extract_double_quoted __P((char *, int *, int));
232 static inline char *string_extract_single_quoted __P((char *, int *));
233 static inline int skip_single_quoted __P((const char *, size_t, int));
234 static int skip_double_quoted __P((char *, size_t, int));
235 static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int));
236 static char *extract_dollar_brace_string __P((char *, int *, int, int));
237 static int skip_matched_pair __P((const char *, int, int, int, int));
238
239 static char *pos_params __P((char *, int, int, int));
240
241 static unsigned char *mb_getcharlens __P((char *, int));
242
243 static char *remove_upattern __P((char *, char *, int));
244 #if defined (HANDLE_MULTIBYTE)
245 static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int));
246 #endif
247 static char *remove_pattern __P((char *, char *, int));
248
249 static int match_upattern __P((char *, char *, int, char **, char **));
250 #if defined (HANDLE_MULTIBYTE)
251 static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
252 #endif
253 static int match_pattern __P((char *, char *, int, char **, char **));
254 static int getpatspec __P((int, char *));
255 static char *getpattern __P((char *, int, int));
256 static char *variable_remove_pattern __P((char *, char *, int, int));
257 static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int));
258 static char *parameter_list_remove_pattern __P((int, char *, int, int));
259 #ifdef ARRAY_VARS
260 static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int));
261 #endif
262 static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int));
263
264 static char *process_substitute __P((char *, int));
265
266 static char *read_comsub __P((int, int, int *));
267
268 #ifdef ARRAY_VARS
269 static arrayind_t array_length_reference __P((char *));
270 #endif
271
272 static int valid_brace_expansion_word __P((char *, int));
273 static int chk_atstar __P((char *, int, int *, int *));
274 static int chk_arithsub __P((const char *, int));
275
276 static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *));
277 static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *));
278 static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *));
279 static void parameter_brace_expand_error __P((char *, char *));
280
281 static int valid_length_expression __P((char *));
282 static intmax_t parameter_brace_expand_length __P((char *));
283
284 static char *skiparith __P((char *, int));
285 static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
286 static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
287 static char *mb_substring __P((char *, int, int));
288 static char *parameter_brace_substring __P((char *, char *, int, char *, int, int));
289
290 static int shouldexp_replacement __P((char *));
291
292 static char *pos_params_pat_subst __P((char *, char *, char *, int));
293
294 static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int));
295
296 static char *pos_params_casemod __P((char *, char *, int, int));
297 static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int));
298
299 static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *));
300 static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int));
301
302 static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
303
304 static WORD_LIST *word_list_split __P((WORD_LIST *));
305
306 static void exp_jump_to_top_level __P((int));
307
308 static WORD_LIST *separate_out_assignments __P((WORD_LIST *));
309 static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int));
310 #ifdef BRACE_EXPANSION
311 static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int));
312 #endif
313 #if defined (ARRAY_VARS)
314 static int make_internal_declare __P((char *, char *));
315 #endif
316 static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int));
317 static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int));
318
319 /* **************************************************************** */
320 /* */
321 /* Utility Functions */
322 /* */
323 /* **************************************************************** */
324
325 #if defined (DEBUG)
326 void
327 dump_word_flags (flags)
328 int flags;
329 {
330 int f;
331
332 f = flags;
333 fprintf (stderr, "%d -> ", f);
334 if (f & W_ASSIGNASSOC)
335 {
336 f &= ~W_ASSIGNASSOC;
337 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
338 }
339 if (f & W_HASCTLESC)
340 {
341 f &= ~W_HASCTLESC;
342 fprintf (stderr, "W_HASCTLESC%s", f ? "|" : "");
343 }
344 if (f & W_NOPROCSUB)
345 {
346 f &= ~W_NOPROCSUB;
347 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
348 }
349 if (f & W_DQUOTE)
350 {
351 f &= ~W_DQUOTE;
352 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
353 }
354 if (f & W_HASQUOTEDNULL)
355 {
356 f &= ~W_HASQUOTEDNULL;
357 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
358 }
359 if (f & W_ASSIGNARG)
360 {
361 f &= ~W_ASSIGNARG;
362 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
363 }
364 if (f & W_ASSNBLTIN)
365 {
366 f &= ~W_ASSNBLTIN;
367 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
368 }
369 if (f & W_COMPASSIGN)
370 {
371 f &= ~W_COMPASSIGN;
372 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
373 }
374 if (f & W_NOEXPAND)
375 {
376 f &= ~W_NOEXPAND;
377 fprintf (stderr, "W_NOEXPAND%s", f ? "|" : "");
378 }
379 if (f & W_ITILDE)
380 {
381 f &= ~W_ITILDE;
382 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
383 }
384 if (f & W_NOTILDE)
385 {
386 f &= ~W_NOTILDE;
387 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
388 }
389 if (f & W_ASSIGNRHS)
390 {
391 f &= ~W_ASSIGNRHS;
392 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
393 }
394 if (f & W_NOCOMSUB)
395 {
396 f &= ~W_NOCOMSUB;
397 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
398 }
399 if (f & W_DOLLARSTAR)
400 {
401 f &= ~W_DOLLARSTAR;
402 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
403 }
404 if (f & W_DOLLARAT)
405 {
406 f &= ~W_DOLLARAT;
407 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
408 }
409 if (f & W_TILDEEXP)
410 {
411 f &= ~W_TILDEEXP;
412 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
413 }
414 if (f & W_NOSPLIT2)
415 {
416 f &= ~W_NOSPLIT2;
417 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
418 }
419 if (f & W_NOGLOB)
420 {
421 f &= ~W_NOGLOB;
422 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
423 }
424 if (f & W_NOSPLIT)
425 {
426 f &= ~W_NOSPLIT;
427 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
428 }
429 if (f & W_GLOBEXP)
430 {
431 f &= ~W_GLOBEXP;
432 fprintf (stderr, "W_GLOBEXP%s", f ? "|" : "");
433 }
434 if (f & W_ASSIGNMENT)
435 {
436 f &= ~W_ASSIGNMENT;
437 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
438 }
439 if (f & W_QUOTED)
440 {
441 f &= ~W_QUOTED;
442 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
443 }
444 if (f & W_HASDOLLAR)
445 {
446 f &= ~W_HASDOLLAR;
447 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
448 }
449 fprintf (stderr, "\n");
450 fflush (stderr);
451 }
452 #endif
453
454 #ifdef INCLUDE_UNUSED
455 static char *
456 quoted_substring (string, start, end)
457 char *string;
458 int start, end;
459 {
460 register int len, l;
461 register char *result, *s, *r;
462
463 len = end - start;
464
465 /* Move to string[start], skipping quoted characters. */
466 for (s = string, l = 0; *s && l < start; )
467 {
468 if (*s == CTLESC)
469 {
470 s++;
471 continue;
472 }
473 l++;
474 if (*s == 0)
475 break;
476 }
477
478 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
479
480 /* Copy LEN characters, including quote characters. */
481 s = string + l;
482 for (l = 0; l < len; s++)
483 {
484 if (*s == CTLESC)
485 *r++ = *s++;
486 *r++ = *s;
487 l++;
488 if (*s == 0)
489 break;
490 }
491 *r = '\0';
492 return result;
493 }
494 #endif
495
496 #ifdef INCLUDE_UNUSED
497 /* Return the length of S, skipping over quoted characters */
498 static int
499 quoted_strlen (s)
500 char *s;
501 {
502 register char *p;
503 int i;
504
505 i = 0;
506 for (p = s; *p; p++)
507 {
508 if (*p == CTLESC)
509 {
510 p++;
511 if (*p == 0)
512 return (i + 1);
513 }
514 i++;
515 }
516
517 return i;
518 }
519 #endif
520
521 /* Find the first occurrence of character C in string S, obeying shell
522 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
523 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
524 escaped with CTLESC are skipped. */
525 static char *
526 quoted_strchr (s, c, flags)
527 char *s;
528 int c, flags;
529 {
530 register char *p;
531
532 for (p = s; *p; p++)
533 {
534 if (((flags & ST_BACKSL) && *p == '\\')
535 || ((flags & ST_CTLESC) && *p == CTLESC))
536 {
537 p++;
538 if (*p == '\0')
539 return ((char *)NULL);
540 continue;
541 }
542 else if (*p == c)
543 return p;
544 }
545 return ((char *)NULL);
546 }
547
548 /* Return 1 if CHARACTER appears in an unquoted portion of
549 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
550 static int
551 unquoted_member (character, string)
552 int character;
553 char *string;
554 {
555 size_t slen;
556 int sindex, c;
557 DECLARE_MBSTATE;
558
559 slen = strlen (string);
560 sindex = 0;
561 while (c = string[sindex])
562 {
563 if (c == character)
564 return (1);
565
566 switch (c)
567 {
568 default:
569 ADVANCE_CHAR (string, slen, sindex);
570 break;
571
572 case '\\':
573 sindex++;
574 if (string[sindex])
575 ADVANCE_CHAR (string, slen, sindex);
576 break;
577
578 case '\'':
579 sindex = skip_single_quoted (string, slen, ++sindex);
580 break;
581
582 case '"':
583 sindex = skip_double_quoted (string, slen, ++sindex);
584 break;
585 }
586 }
587 return (0);
588 }
589
590 /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
591 static int
592 unquoted_substring (substr, string)
593 char *substr, *string;
594 {
595 size_t slen;
596 int sindex, c, sublen;
597 DECLARE_MBSTATE;
598
599 if (substr == 0 || *substr == '\0')
600 return (0);
601
602 slen = strlen (string);
603 sublen = strlen (substr);
604 for (sindex = 0; c = string[sindex]; )
605 {
606 if (STREQN (string + sindex, substr, sublen))
607 return (1);
608
609 switch (c)
610 {
611 case '\\':
612 sindex++;
613 if (string[sindex])
614 ADVANCE_CHAR (string, slen, sindex);
615 break;
616
617 case '\'':
618 sindex = skip_single_quoted (string, slen, ++sindex);
619 break;
620
621 case '"':
622 sindex = skip_double_quoted (string, slen, ++sindex);
623 break;
624
625 default:
626 ADVANCE_CHAR (string, slen, sindex);
627 break;
628 }
629 }
630 return (0);
631 }
632
633 /* Most of the substitutions must be done in parallel. In order
634 to avoid using tons of unclear goto's, I have some functions
635 for manipulating malloc'ed strings. They all take INDX, a
636 pointer to an integer which is the offset into the string
637 where manipulation is taking place. They also take SIZE, a
638 pointer to an integer which is the current length of the
639 character array for this string. */
640
641 /* Append SOURCE to TARGET at INDEX. SIZE is the current amount
642 of space allocated to TARGET. SOURCE can be NULL, in which
643 case nothing happens. Gets rid of SOURCE by freeing it.
644 Returns TARGET in case the location has changed. */
645 INLINE char *
646 sub_append_string (source, target, indx, size)
647 char *source, *target;
648 int *indx, *size;
649 {
650 if (source)
651 {
652 int srclen, n;
653
654 srclen = STRLEN (source);
655 if (srclen >= (int)(*size - *indx))
656 {
657 n = srclen + *indx;
658 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
659 target = (char *)xrealloc (target, (*size = n));
660 }
661
662 FASTCOPY (source, target + *indx, srclen);
663 *indx += srclen;
664 target[*indx] = '\0';
665
666 free (source);
667 }
668 return (target);
669 }
670
671 #if 0
672 /* UNUSED */
673 /* Append the textual representation of NUMBER to TARGET.
674 INDX and SIZE are as in SUB_APPEND_STRING. */
675 char *
676 sub_append_number (number, target, indx, size)
677 intmax_t number;
678 int *indx, *size;
679 char *target;
680 {
681 char *temp;
682
683 temp = itos (number);
684 return (sub_append_string (temp, target, indx, size));
685 }
686 #endif
687
688 /* Extract a substring from STRING, starting at SINDEX and ending with
689 one of the characters in CHARLIST. Don't make the ending character
690 part of the string. Leave SINDEX pointing at the ending character.
691 Understand about backslashes in the string. If (flags & SX_VARNAME)
692 is non-zero, and array variables have been compiled into the shell,
693 everything between a `[' and a corresponding `]' is skipped over.
694 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
695 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
696 contain a closing character from CHARLIST. */
697 static char *
698 string_extract (string, sindex, charlist, flags)
699 char *string;
700 int *sindex;
701 char *charlist;
702 int flags;
703 {
704 register int c, i;
705 int found;
706 size_t slen;
707 char *temp;
708 DECLARE_MBSTATE;
709
710 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
711 i = *sindex;
712 found = 0;
713 while (c = string[i])
714 {
715 if (c == '\\')
716 {
717 if (string[i + 1])
718 i++;
719 else
720 break;
721 }
722 #if defined (ARRAY_VARS)
723 else if ((flags & SX_VARNAME) && c == '[')
724 {
725 int ni;
726 /* If this is an array subscript, skip over it and continue. */
727 ni = skipsubscript (string, i, 0);
728 if (string[ni] == ']')
729 i = ni;
730 }
731 #endif
732 else if (MEMBER (c, charlist))
733 {
734 found = 1;
735 break;
736 }
737
738 ADVANCE_CHAR (string, slen, i);
739 }
740
741 /* If we had to have a matching delimiter and didn't find one, return an
742 error and let the caller deal with it. */
743 if ((flags & SX_REQMATCH) && found == 0)
744 {
745 *sindex = i;
746 return (&extract_string_error);
747 }
748
749 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
750 *sindex = i;
751
752 return (temp);
753 }
754
755 /* Extract the contents of STRING as if it is enclosed in double quotes.
756 SINDEX, when passed in, is the offset of the character immediately
757 following the opening double quote; on exit, SINDEX is left pointing after
758 the closing double quote. If STRIPDQ is non-zero, unquoted double
759 quotes are stripped and the string is terminated by a null byte.
760 Backslashes between the embedded double quotes are processed. If STRIPDQ
761 is zero, an unquoted `"' terminates the string. */
762 static char *
763 string_extract_double_quoted (string, sindex, stripdq)
764 char *string;
765 int *sindex, stripdq;
766 {
767 size_t slen;
768 char *send;
769 int j, i, t;
770 unsigned char c;
771 char *temp, *ret; /* The new string we return. */
772 int pass_next, backquote, si; /* State variables for the machine. */
773 int dquote;
774 DECLARE_MBSTATE;
775
776 slen = strlen (string + *sindex) + *sindex;
777 send = string + slen;
778
779 pass_next = backquote = dquote = 0;
780 temp = (char *)xmalloc (1 + slen - *sindex);
781
782 j = 0;
783 i = *sindex;
784 while (c = string[i])
785 {
786 /* Process a character that was quoted by a backslash. */
787 if (pass_next)
788 {
789 /* XXX - take another look at this in light of Interp 221 */
790 /* Posix.2 sez:
791
792 ``The backslash shall retain its special meaning as an escape
793 character only when followed by one of the characters:
794 $ ` " \ <newline>''.
795
796 If STRIPDQ is zero, we handle the double quotes here and let
797 expand_word_internal handle the rest. If STRIPDQ is non-zero,
798 we have already been through one round of backslash stripping,
799 and want to strip these backslashes only if DQUOTE is non-zero,
800 indicating that we are inside an embedded double-quoted string. */
801
802 /* If we are in an embedded quoted string, then don't strip
803 backslashes before characters for which the backslash
804 retains its special meaning, but remove backslashes in
805 front of other characters. If we are not in an
806 embedded quoted string, don't strip backslashes at all.
807 This mess is necessary because the string was already
808 surrounded by double quotes (and sh has some really weird
809 quoting rules).
810 The returned string will be run through expansion as if
811 it were double-quoted. */
812 if ((stripdq == 0 && c != '"') ||
813 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
814 temp[j++] = '\\';
815 pass_next = 0;
816
817 add_one_character:
818 COPY_CHAR_I (temp, j, string, send, i);
819 continue;
820 }
821
822 /* A backslash protects the next character. The code just above
823 handles preserving the backslash in front of any character but
824 a double quote. */
825 if (c == '\\')
826 {
827 pass_next++;
828 i++;
829 continue;
830 }
831
832 /* Inside backquotes, ``the portion of the quoted string from the
833 initial backquote and the characters up to the next backquote
834 that is not preceded by a backslash, having escape characters
835 removed, defines that command''. */
836 if (backquote)
837 {
838 if (c == '`')
839 backquote = 0;
840 temp[j++] = c;
841 i++;
842 continue;
843 }
844
845 if (c == '`')
846 {
847 temp[j++] = c;
848 backquote++;
849 i++;
850 continue;
851 }
852
853 /* Pass everything between `$(' and the matching `)' or a quoted
854 ${ ... } pair through according to the Posix.2 specification. */
855 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
856 {
857 int free_ret = 1;
858
859 si = i + 2;
860 if (string[i + 1] == LPAREN)
861 ret = extract_command_subst (string, &si, 0);
862 else
863 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
864
865 temp[j++] = '$';
866 temp[j++] = string[i + 1];
867
868 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
869 is set. */
870 if (ret == 0 && no_longjmp_on_fatal_error)
871 {
872 free_ret = 0;
873 ret = string + i + 2;
874 }
875
876 for (t = 0; ret[t]; t++, j++)
877 temp[j] = ret[t];
878 temp[j] = string[si];
879
880 if (string[si])
881 {
882 j++;
883 i = si + 1;
884 }
885 else
886 i = si;
887
888 if (free_ret)
889 free (ret);
890 continue;
891 }
892
893 /* Add any character but a double quote to the quoted string we're
894 accumulating. */
895 if (c != '"')
896 goto add_one_character;
897
898 /* c == '"' */
899 if (stripdq)
900 {
901 dquote ^= 1;
902 i++;
903 continue;
904 }
905
906 break;
907 }
908 temp[j] = '\0';
909
910 /* Point to after the closing quote. */
911 if (c)
912 i++;
913 *sindex = i;
914
915 return (temp);
916 }
917
918 /* This should really be another option to string_extract_double_quoted. */
919 static int
920 skip_double_quoted (string, slen, sind)
921 char *string;
922 size_t slen;
923 int sind;
924 {
925 int c, i;
926 char *ret;
927 int pass_next, backquote, si;
928 DECLARE_MBSTATE;
929
930 pass_next = backquote = 0;
931 i = sind;
932 while (c = string[i])
933 {
934 if (pass_next)
935 {
936 pass_next = 0;
937 ADVANCE_CHAR (string, slen, i);
938 continue;
939 }
940 else if (c == '\\')
941 {
942 pass_next++;
943 i++;
944 continue;
945 }
946 else if (backquote)
947 {
948 if (c == '`')
949 backquote = 0;
950 ADVANCE_CHAR (string, slen, i);
951 continue;
952 }
953 else if (c == '`')
954 {
955 backquote++;
956 i++;
957 continue;
958 }
959 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
960 {
961 si = i + 2;
962 if (string[i + 1] == LPAREN)
963 ret = extract_command_subst (string, &si, SX_NOALLOC);
964 else
965 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
966
967 i = si + 1;
968 continue;
969 }
970 else if (c != '"')
971 {
972 ADVANCE_CHAR (string, slen, i);
973 continue;
974 }
975 else
976 break;
977 }
978
979 if (c)
980 i++;
981
982 return (i);
983 }
984
985 /* Extract the contents of STRING as if it is enclosed in single quotes.
986 SINDEX, when passed in, is the offset of the character immediately
987 following the opening single quote; on exit, SINDEX is left pointing after
988 the closing single quote. */
989 static inline char *
990 string_extract_single_quoted (string, sindex)
991 char *string;
992 int *sindex;
993 {
994 register int i;
995 size_t slen;
996 char *t;
997 DECLARE_MBSTATE;
998
999 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1000 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
1001 i = *sindex;
1002 while (string[i] && string[i] != '\'')
1003 ADVANCE_CHAR (string, slen, i);
1004
1005 t = substring (string, *sindex, i);
1006
1007 if (string[i])
1008 i++;
1009 *sindex = i;
1010
1011 return (t);
1012 }
1013
1014 static inline int
1015 skip_single_quoted (string, slen, sind)
1016 const char *string;
1017 size_t slen;
1018 int sind;
1019 {
1020 register int c;
1021 DECLARE_MBSTATE;
1022
1023 c = sind;
1024 while (string[c] && string[c] != '\'')
1025 ADVANCE_CHAR (string, slen, c);
1026
1027 if (string[c])
1028 c++;
1029 return c;
1030 }
1031
1032 /* Just like string_extract, but doesn't hack backslashes or any of
1033 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
1034 static char *
1035 string_extract_verbatim (string, slen, sindex, charlist, flags)
1036 char *string;
1037 size_t slen;
1038 int *sindex;
1039 char *charlist;
1040 int flags;
1041 {
1042 register int i;
1043 #if defined (HANDLE_MULTIBYTE)
1044 size_t clen;
1045 wchar_t *wcharlist;
1046 #endif
1047 int c;
1048 char *temp;
1049 DECLARE_MBSTATE;
1050
1051 if (charlist[0] == '\'' && charlist[1] == '\0')
1052 {
1053 temp = string_extract_single_quoted (string, sindex);
1054 --*sindex; /* leave *sindex at separator character */
1055 return temp;
1056 }
1057
1058 i = *sindex;
1059 #if 0
1060 /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need
1061 this only if MB_CUR_MAX > 1. */
1062 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1;
1063 #endif
1064 #if defined (HANDLE_MULTIBYTE)
1065 clen = strlen (charlist);
1066 wcharlist = 0;
1067 #endif
1068 while (c = string[i])
1069 {
1070 #if defined (HANDLE_MULTIBYTE)
1071 size_t mblength;
1072 #endif
1073 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1074 {
1075 i += 2;
1076 continue;
1077 }
1078 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1079 through, to protect the CTLNULs from later calls to
1080 remove_quoted_nulls. */
1081 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
1082 {
1083 i += 2;
1084 continue;
1085 }
1086
1087 #if defined (HANDLE_MULTIBYTE)
1088 mblength = MBLEN (string + i, slen - i);
1089 if (mblength > 1)
1090 {
1091 wchar_t wc;
1092 mblength = mbtowc (&wc, string + i, slen - i);
1093 if (MB_INVALIDCH (mblength))
1094 {
1095 if (MEMBER (c, charlist))
1096 break;
1097 }
1098 else
1099 {
1100 if (wcharlist == 0)
1101 {
1102 size_t len;
1103 len = mbstowcs (wcharlist, charlist, 0);
1104 if (len == -1)
1105 len = 0;
1106 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1107 mbstowcs (wcharlist, charlist, len + 1);
1108 }
1109
1110 if (wcschr (wcharlist, wc))
1111 break;
1112 }
1113 }
1114 else
1115 #endif
1116 if (MEMBER (c, charlist))
1117 break;
1118
1119 ADVANCE_CHAR (string, slen, i);
1120 }
1121
1122 #if defined (HANDLE_MULTIBYTE)
1123 FREE (wcharlist);
1124 #endif
1125
1126 temp = substring (string, *sindex, i);
1127 *sindex = i;
1128
1129 return (temp);
1130 }
1131
1132 /* Extract the $( construct in STRING, and return a new string.
1133 Start extracting at (SINDEX) as if we had just seen "$(".
1134 Make (SINDEX) get the position of the matching ")". )
1135 XFLAGS is additional flags to pass to other extraction functions. */
1136 char *
1137 extract_command_subst (string, sindex, xflags)
1138 char *string;
1139 int *sindex;
1140 int xflags;
1141 {
1142 if (string[*sindex] == LPAREN)
1143 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1144 else
1145 {
1146 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1147 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1148 }
1149 }
1150
1151 /* Extract the $[ construct in STRING, and return a new string. (])
1152 Start extracting at (SINDEX) as if we had just seen "$[".
1153 Make (SINDEX) get the position of the matching "]". */
1154 char *
1155 extract_arithmetic_subst (string, sindex)
1156 char *string;
1157 int *sindex;
1158 {
1159 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
1160 }
1161
1162 #if defined (PROCESS_SUBSTITUTION)
1163 /* Extract the <( or >( construct in STRING, and return a new string.
1164 Start extracting at (SINDEX) as if we had just seen "<(".
1165 Make (SINDEX) get the position of the matching ")". */ /*))*/
1166 char *
1167 extract_process_subst (string, starter, sindex)
1168 char *string;
1169 char *starter;
1170 int *sindex;
1171 {
1172 return (extract_delimited_string (string, sindex, starter, "(", ")", 0));
1173 }
1174 #endif /* PROCESS_SUBSTITUTION */
1175
1176 #if defined (ARRAY_VARS)
1177 /* This can be fooled by unquoted right parens in the passed string. If
1178 each caller verifies that the last character in STRING is a right paren,
1179 we don't even need to call extract_delimited_string. */
1180 char *
1181 extract_array_assignment_list (string, sindex)
1182 char *string;
1183 int *sindex;
1184 {
1185 int slen;
1186 char *ret;
1187
1188 slen = strlen (string); /* ( */
1189 if (string[slen - 1] == ')')
1190 {
1191 ret = substring (string, *sindex, slen - 1);
1192 *sindex = slen - 1;
1193 return ret;
1194 }
1195 return 0;
1196 }
1197 #endif
1198
1199 /* Extract and create a new string from the contents of STRING, a
1200 character string delimited with OPENER and CLOSER. SINDEX is
1201 the address of an int describing the current offset in STRING;
1202 it should point to just after the first OPENER found. On exit,
1203 SINDEX gets the position of the last character of the matching CLOSER.
1204 If OPENER is more than a single character, ALT_OPENER, if non-null,
1205 contains a character string that can also match CLOSER and thus
1206 needs to be skipped. */
1207 static char *
1208 extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
1209 char *string;
1210 int *sindex;
1211 char *opener, *alt_opener, *closer;
1212 int flags;
1213 {
1214 int i, c, si;
1215 size_t slen;
1216 char *t, *result;
1217 int pass_character, nesting_level, in_comment;
1218 int len_closer, len_opener, len_alt_opener;
1219 DECLARE_MBSTATE;
1220
1221 slen = strlen (string + *sindex) + *sindex;
1222 len_opener = STRLEN (opener);
1223 len_alt_opener = STRLEN (alt_opener);
1224 len_closer = STRLEN (closer);
1225
1226 pass_character = in_comment = 0;
1227
1228 nesting_level = 1;
1229 i = *sindex;
1230
1231 while (nesting_level)
1232 {
1233 c = string[i];
1234
1235 if (c == 0)
1236 break;
1237
1238 if (in_comment)
1239 {
1240 if (c == '\n')
1241 in_comment = 0;
1242 ADVANCE_CHAR (string, slen, i);
1243 continue;
1244 }
1245
1246 if (pass_character) /* previous char was backslash */
1247 {
1248 pass_character = 0;
1249 ADVANCE_CHAR (string, slen, i);
1250 continue;
1251 }
1252
1253 /* Not exactly right yet; should handle shell metacharacters and
1254 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
1255 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
1256 {
1257 in_comment = 1;
1258 ADVANCE_CHAR (string, slen, i);
1259 continue;
1260 }
1261
1262 if (c == CTLESC || c == '\\')
1263 {
1264 pass_character++;
1265 i++;
1266 continue;
1267 }
1268
1269 /* Process a nested command substitution, but only if we're parsing an
1270 arithmetic substitution. */
1271 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1272 {
1273 si = i + 2;
1274 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1275 i = si + 1;
1276 continue;
1277 }
1278
1279 /* Process a nested OPENER. */
1280 if (STREQN (string + i, opener, len_opener))
1281 {
1282 si = i + len_opener;
1283 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
1284 i = si + 1;
1285 continue;
1286 }
1287
1288 /* Process a nested ALT_OPENER */
1289 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
1290 {
1291 si = i + len_alt_opener;
1292 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
1293 i = si + 1;
1294 continue;
1295 }
1296
1297 /* If the current substring terminates the delimited string, decrement
1298 the nesting level. */
1299 if (STREQN (string + i, closer, len_closer))
1300 {
1301 i += len_closer - 1; /* move to last byte of the closer */
1302 nesting_level--;
1303 if (nesting_level == 0)
1304 break;
1305 }
1306
1307 /* Pass old-style command substitution through verbatim. */
1308 if (c == '`')
1309 {
1310 si = i + 1;
1311 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1312 i = si + 1;
1313 continue;
1314 }
1315
1316 /* Pass single-quoted and double-quoted strings through verbatim. */
1317 if (c == '\'' || c == '"')
1318 {
1319 si = i + 1;
1320 i = (c == '\'') ? skip_single_quoted (string, slen, si)
1321 : skip_double_quoted (string, slen, si);
1322 continue;
1323 }
1324
1325 /* move past this character, which was not special. */
1326 ADVANCE_CHAR (string, slen, i);
1327 }
1328
1329 if (c == 0 && nesting_level)
1330 {
1331 if (no_longjmp_on_fatal_error == 0)
1332 {
1333 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
1334 last_command_exit_value = EXECUTION_FAILURE;
1335 exp_jump_to_top_level (DISCARD);
1336 }
1337 else
1338 {
1339 *sindex = i;
1340 return (char *)NULL;
1341 }
1342 }
1343
1344 si = i - *sindex - len_closer + 1;
1345 if (flags & SX_NOALLOC)
1346 result = (char *)NULL;
1347 else
1348 {
1349 result = (char *)xmalloc (1 + si);
1350 strncpy (result, string + *sindex, si);
1351 result[si] = '\0';
1352 }
1353 *sindex = i;
1354
1355 return (result);
1356 }
1357
1358 /* Extract a parameter expansion expression within ${ and } from STRING.
1359 Obey the Posix.2 rules for finding the ending `}': count braces while
1360 skipping over enclosed quoted strings and command substitutions.
1361 SINDEX is the address of an int describing the current offset in STRING;
1362 it should point to just after the first `{' found. On exit, SINDEX
1363 gets the position of the matching `}'. QUOTED is non-zero if this
1364 occurs inside double quotes. */
1365 /* XXX -- this is very similar to extract_delimited_string -- XXX */
1366 static char *
1367 extract_dollar_brace_string (string, sindex, quoted, flags)
1368 char *string;
1369 int *sindex, quoted, flags;
1370 {
1371 register int i, c;
1372 size_t slen;
1373 int pass_character, nesting_level, si, dolbrace_state;
1374 char *result, *t;
1375 DECLARE_MBSTATE;
1376
1377 pass_character = 0;
1378 nesting_level = 1;
1379 slen = strlen (string + *sindex) + *sindex;
1380
1381 /* The handling of dolbrace_state needs to agree with the code in parse.y:
1382 parse_matched_pair(). The different initial value is to handle the
1383 case where this function is called to parse the word in
1384 ${param op word} (SX_WORD). */
1385 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1386 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1387 dolbrace_state = DOLBRACE_QUOTE;
1388
1389 i = *sindex;
1390 while (c = string[i])
1391 {
1392 if (pass_character)
1393 {
1394 pass_character = 0;
1395 ADVANCE_CHAR (string, slen, i);
1396 continue;
1397 }
1398
1399 /* CTLESCs and backslashes quote the next character. */
1400 if (c == CTLESC || c == '\\')
1401 {
1402 pass_character++;
1403 i++;
1404 continue;
1405 }
1406
1407 if (string[i] == '$' && string[i+1] == LBRACE)
1408 {
1409 nesting_level++;
1410 i += 2;
1411 continue;
1412 }
1413
1414 if (c == RBRACE)
1415 {
1416 nesting_level--;
1417 if (nesting_level == 0)
1418 break;
1419 i++;
1420 continue;
1421 }
1422
1423 /* Pass the contents of old-style command substitutions through
1424 verbatim. */
1425 if (c == '`')
1426 {
1427 si = i + 1;
1428 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1429 i = si + 1;
1430 continue;
1431 }
1432
1433 /* Pass the contents of new-style command substitutions and
1434 arithmetic substitutions through verbatim. */
1435 if (string[i] == '$' && string[i+1] == LPAREN)
1436 {
1437 si = i + 2;
1438 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1439 i = si + 1;
1440 continue;
1441 }
1442
1443 #if 0
1444 /* Pass the contents of single-quoted and double-quoted strings
1445 through verbatim. */
1446 if (c == '\'' || c == '"')
1447 {
1448 si = i + 1;
1449 i = (c == '\'') ? skip_single_quoted (string, slen, si)
1450 : skip_double_quoted (string, slen, si);
1451 /* skip_XXX_quoted leaves index one past close quote */
1452 continue;
1453 }
1454 #else /* XXX - bash-4.2 */
1455 /* Pass the contents of double-quoted strings through verbatim. */
1456 if (c == '"')
1457 {
1458 si = i + 1;
1459 i = skip_double_quoted (string, slen, si);
1460 /* skip_XXX_quoted leaves index one past close quote */
1461 continue;
1462 }
1463
1464 if (c == '\'')
1465 {
1466 /*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
1467 if (posixly_correct && shell_compatibility_level > 41 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
1468 ADVANCE_CHAR (string, slen, i);
1469 else
1470 {
1471 si = i + 1;
1472 i = skip_single_quoted (string, slen, si);
1473 }
1474
1475 continue;
1476 }
1477 #endif
1478
1479 /* move past this character, which was not special. */
1480 ADVANCE_CHAR (string, slen, i);
1481
1482 /* This logic must agree with parse.y:parse_matched_pair, since they
1483 share the same defines. */
1484 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1485 dolbrace_state = DOLBRACE_QUOTE;
1486 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1487 dolbrace_state = DOLBRACE_QUOTE;
1488 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
1489 dolbrace_state = DOLBRACE_QUOTE;
1490 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1491 dolbrace_state = DOLBRACE_QUOTE;
1492 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1493 dolbrace_state = DOLBRACE_QUOTE;
1494 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1495 dolbrace_state = DOLBRACE_OP;
1496 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1497 dolbrace_state = DOLBRACE_WORD;
1498 }
1499
1500 if (c == 0 && nesting_level)
1501 {
1502 if (no_longjmp_on_fatal_error == 0)
1503 { /* { */
1504 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
1505 last_command_exit_value = EXECUTION_FAILURE;
1506 exp_jump_to_top_level (DISCARD);
1507 }
1508 else
1509 {
1510 *sindex = i;
1511 return ((char *)NULL);
1512 }
1513 }
1514
1515 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
1516 *sindex = i;
1517
1518 return (result);
1519 }
1520
1521 /* Remove backslashes which are quoting backquotes from STRING. Modifies
1522 STRING, and returns a pointer to it. */
1523 char *
1524 de_backslash (string)
1525 char *string;
1526 {
1527 register size_t slen;
1528 register int i, j, prev_i;
1529 DECLARE_MBSTATE;
1530
1531 slen = strlen (string);
1532 i = j = 0;
1533
1534 /* Loop copying string[i] to string[j], i >= j. */
1535 while (i < slen)
1536 {
1537 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
1538 string[i + 1] == '$'))
1539 i++;
1540 prev_i = i;
1541 ADVANCE_CHAR (string, slen, i);
1542 if (j < prev_i)
1543 do string[j++] = string[prev_i++]; while (prev_i < i);
1544 else
1545 j = i;
1546 }
1547 string[j] = '\0';
1548
1549 return (string);
1550 }
1551
1552 #if 0
1553 /*UNUSED*/
1554 /* Replace instances of \! in a string with !. */
1555 void
1556 unquote_bang (string)
1557 char *string;
1558 {
1559 register int i, j;
1560 register char *temp;
1561
1562 temp = (char *)xmalloc (1 + strlen (string));
1563
1564 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1565 {
1566 if (string[i] == '\\' && string[i + 1] == '!')
1567 {
1568 temp[j] = '!';
1569 i++;
1570 }
1571 }
1572 strcpy (string, temp);
1573 free (temp);
1574 }
1575 #endif
1576
1577 #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0)
1578
1579 /* This function assumes s[i] == open; returns with s[ret] == close; used to
1580 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1581 matched pairs of quotes or backquotes, or skip word expansions; it is
1582 intended to be used after expansion has been performed and during final
1583 assignment parsing (see arrayfunc.c:assign_compound_array_list()). */
1584 static int
1585 skip_matched_pair (string, start, open, close, flags)
1586 const char *string;
1587 int start, open, close, flags;
1588 {
1589 int i, pass_next, backq, si, c, count;
1590 size_t slen;
1591 char *temp, *ss;
1592 DECLARE_MBSTATE;
1593
1594 slen = strlen (string + start) + start;
1595 no_longjmp_on_fatal_error = 1;
1596
1597 i = start + 1; /* skip over leading bracket */
1598 count = 1;
1599 pass_next = backq = 0;
1600 ss = (char *)string;
1601 while (c = string[i])
1602 {
1603 if (pass_next)
1604 {
1605 pass_next = 0;
1606 if (c == 0)
1607 CQ_RETURN(i);
1608 ADVANCE_CHAR (string, slen, i);
1609 continue;
1610 }
1611 else if (c == '\\')
1612 {
1613 pass_next = 1;
1614 i++;
1615 continue;
1616 }
1617 else if (backq)
1618 {
1619 if (c == '`')
1620 backq = 0;
1621 ADVANCE_CHAR (string, slen, i);
1622 continue;
1623 }
1624 else if ((flags & 1) == 0 && c == '`')
1625 {
1626 backq = 1;
1627 i++;
1628 continue;
1629 }
1630 else if ((flags & 1) == 0 && c == open)
1631 {
1632 count++;
1633 i++;
1634 continue;
1635 }
1636 else if (c == close)
1637 {
1638 count--;
1639 if (count == 0)
1640 break;
1641 i++;
1642 continue;
1643 }
1644 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
1645 {
1646 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i)
1647 : skip_double_quoted (ss, slen, ++i);
1648 /* no increment, the skip functions increment past the closing quote. */
1649 }
1650 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
1651 {
1652 si = i + 2;
1653 if (string[si] == '\0')
1654 CQ_RETURN(si);
1655
1656 if (string[i+1] == LPAREN)
1657 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1658 else
1659 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
1660 i = si;
1661 if (string[i] == '\0') /* don't increment i past EOS in loop */
1662 break;
1663 i++;
1664 continue;
1665 }
1666 else
1667 ADVANCE_CHAR (string, slen, i);
1668 }
1669
1670 CQ_RETURN(i);
1671 }
1672
1673 #if defined (ARRAY_VARS)
1674 int
1675 skipsubscript (string, start, flags)
1676 const char *string;
1677 int start, flags;
1678 {
1679 return (skip_matched_pair (string, start, '[', ']', flags));
1680 }
1681 #endif
1682
1683 /* Skip characters in STRING until we find a character in DELIMS, and return
1684 the index of that character. START is the index into string at which we
1685 begin. This is similar in spirit to strpbrk, but it returns an index into
1686 STRING and takes a starting index. This little piece of code knows quite
1687 a lot of shell syntax. It's very similar to skip_double_quoted and other
1688 functions of that ilk. */
1689 int
1690 skip_to_delim (string, start, delims, flags)
1691 char *string;
1692 int start;
1693 char *delims;
1694 int flags;
1695 {
1696 int i, pass_next, backq, si, c, invert, skipquote, skipcmd;
1697 size_t slen;
1698 char *temp, open[3];
1699 DECLARE_MBSTATE;
1700
1701 slen = strlen (string + start) + start;
1702 if (flags & SD_NOJMP)
1703 no_longjmp_on_fatal_error = 1;
1704 invert = (flags & SD_INVERT);
1705 skipcmd = (flags & SD_NOSKIPCMD) == 0;
1706
1707 i = start;
1708 pass_next = backq = 0;
1709 while (c = string[i])
1710 {
1711 /* If this is non-zero, we should not let quote characters be delimiters
1712 and the current character is a single or double quote. We should not
1713 test whether or not it's a delimiter until after we skip single- or
1714 double-quoted strings. */
1715 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
1716 if (pass_next)
1717 {
1718 pass_next = 0;
1719 if (c == 0)
1720 CQ_RETURN(i);
1721 ADVANCE_CHAR (string, slen, i);
1722 continue;
1723 }
1724 else if (c == '\\')
1725 {
1726 pass_next = 1;
1727 i++;
1728 continue;
1729 }
1730 else if (backq)
1731 {
1732 if (c == '`')
1733 backq = 0;
1734 ADVANCE_CHAR (string, slen, i);
1735 continue;
1736 }
1737 else if (c == '`')
1738 {
1739 backq = 1;
1740 i++;
1741 continue;
1742 }
1743 else if (skipquote == 0 && invert == 0 && member (c, delims))
1744 break;
1745 else if (c == '\'' || c == '"')
1746 {
1747 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1748 : skip_double_quoted (string, slen, ++i);
1749 /* no increment, the skip functions increment past the closing quote. */
1750 }
1751 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
1752 {
1753 si = i + 2;
1754 if (string[si] == '\0')
1755 CQ_RETURN(si);
1756
1757 if (string[i+1] == LPAREN)
1758 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1759 else
1760 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1761 i = si;
1762 if (string[i] == '\0') /* don't increment i past EOS in loop */
1763 break;
1764 i++;
1765 continue;
1766 }
1767 #if defined (PROCESS_SUBSTITUTION)
1768 else if (skipcmd && (c == '<' || c == '>') && string[i+1] == LPAREN)
1769 {
1770 si = i + 2;
1771 if (string[si] == '\0')
1772 CQ_RETURN(si);
1773 temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si);
1774 i = si;
1775 if (string[i] == '\0')
1776 break;
1777 i++;
1778 continue;
1779 }
1780 #endif /* PROCESS_SUBSTITUTION */
1781 #if defined (EXTENDED_GLOB)
1782 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1783 {
1784 si = i + 2;
1785 if (string[si] == '\0')
1786 CQ_RETURN(si);
1787
1788 open[0] = c;
1789 open[1] = LPAREN;
1790 open[2] = '\0';
1791 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
1792
1793 i = si;
1794 if (string[i] == '\0') /* don't increment i past EOS in loop */
1795 break;
1796 i++;
1797 continue;
1798 }
1799 #endif
1800 else if ((skipquote || invert) && (member (c, delims) == 0))
1801 break;
1802 else
1803 ADVANCE_CHAR (string, slen, i);
1804 }
1805
1806 CQ_RETURN(i);
1807 }
1808
1809 #if defined (READLINE)
1810 /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
1811 an unclosed quoted string), or if the character at EINDEX is quoted
1812 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
1813 single and double-quoted string parsing functions should not return an
1814 error if there are unclosed quotes or braces. The characters that this
1815 recognizes need to be the same as the contents of
1816 rl_completer_quote_characters. */
1817
1818 int
1819 char_is_quoted (string, eindex)
1820 char *string;
1821 int eindex;
1822 {
1823 int i, pass_next, c;
1824 size_t slen;
1825 DECLARE_MBSTATE;
1826
1827 slen = strlen (string);
1828 no_longjmp_on_fatal_error = 1;
1829 i = pass_next = 0;
1830 while (i <= eindex)
1831 {
1832 c = string[i];
1833
1834 if (pass_next)
1835 {
1836 pass_next = 0;
1837 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1838 CQ_RETURN(1);
1839 ADVANCE_CHAR (string, slen, i);
1840 continue;
1841 }
1842 else if (c == '\\')
1843 {
1844 pass_next = 1;
1845 i++;
1846 continue;
1847 }
1848 else if (c == '\'' || c == '"')
1849 {
1850 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1851 : skip_double_quoted (string, slen, ++i);
1852 if (i > eindex)
1853 CQ_RETURN(1);
1854 /* no increment, the skip_xxx functions go one past end */
1855 }
1856 else
1857 ADVANCE_CHAR (string, slen, i);
1858 }
1859
1860 CQ_RETURN(0);
1861 }
1862
1863 int
1864 unclosed_pair (string, eindex, openstr)
1865 char *string;
1866 int eindex;
1867 char *openstr;
1868 {
1869 int i, pass_next, openc, olen;
1870 size_t slen;
1871 DECLARE_MBSTATE;
1872
1873 slen = strlen (string);
1874 olen = strlen (openstr);
1875 i = pass_next = openc = 0;
1876 while (i <= eindex)
1877 {
1878 if (pass_next)
1879 {
1880 pass_next = 0;
1881 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1882 return 0;
1883 ADVANCE_CHAR (string, slen, i);
1884 continue;
1885 }
1886 else if (string[i] == '\\')
1887 {
1888 pass_next = 1;
1889 i++;
1890 continue;
1891 }
1892 else if (STREQN (string + i, openstr, olen))
1893 {
1894 openc = 1 - openc;
1895 i += olen;
1896 }
1897 else if (string[i] == '\'' || string[i] == '"')
1898 {
1899 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i)
1900 : skip_double_quoted (string, slen, i);
1901 if (i > eindex)
1902 return 0;
1903 }
1904 else
1905 ADVANCE_CHAR (string, slen, i);
1906 }
1907 return (openc);
1908 }
1909
1910 /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
1911 individual words. If DELIMS is NULL, the current value of $IFS is used
1912 to split the string, and the function follows the shell field splitting
1913 rules. SENTINEL is an index to look for. NWP, if non-NULL,
1914 gets the number of words in the returned list. CWP, if non-NULL, gets
1915 the index of the word containing SENTINEL. Non-whitespace chars in
1916 DELIMS delimit separate fields. */
1917 WORD_LIST *
1918 split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
1919 char *string;
1920 int slen;
1921 char *delims;
1922 int sentinel, flags;
1923 int *nwp, *cwp;
1924 {
1925 int ts, te, i, nw, cw, ifs_split, dflags;
1926 char *token, *d, *d2;
1927 WORD_LIST *ret, *tl;
1928
1929 if (string == 0 || *string == '\0')
1930 {
1931 if (nwp)
1932 *nwp = 0;
1933 if (cwp)
1934 *cwp = 0;
1935 return ((WORD_LIST *)NULL);
1936 }
1937
1938 d = (delims == 0) ? ifs_value : delims;
1939 ifs_split = delims == 0;
1940
1941 /* Make d2 the non-whitespace characters in delims */
1942 d2 = 0;
1943 if (delims)
1944 {
1945 size_t slength;
1946 #if defined (HANDLE_MULTIBYTE)
1947 size_t mblength = 1;
1948 #endif
1949 DECLARE_MBSTATE;
1950
1951 slength = strlen (delims);
1952 d2 = (char *)xmalloc (slength + 1);
1953 i = ts = 0;
1954 while (delims[i])
1955 {
1956 #if defined (HANDLE_MULTIBYTE)
1957 mbstate_t state_bak;
1958 state_bak = state;
1959 mblength = MBRLEN (delims + i, slength, &state);
1960 if (MB_INVALIDCH (mblength))
1961 state = state_bak;
1962 else if (mblength > 1)
1963 {
1964 memcpy (d2 + ts, delims + i, mblength);
1965 ts += mblength;
1966 i += mblength;
1967 slength -= mblength;
1968 continue;
1969 }
1970 #endif
1971 if (whitespace (delims[i]) == 0)
1972 d2[ts++] = delims[i];
1973
1974 i++;
1975 slength--;
1976 }
1977 d2[ts] = '\0';
1978 }
1979
1980 ret = (WORD_LIST *)NULL;
1981
1982 /* Remove sequences of whitespace characters at the start of the string, as
1983 long as those characters are delimiters. */
1984 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
1985 ;
1986 if (string[i] == '\0')
1987 return (ret);
1988
1989 ts = i;
1990 nw = 0;
1991 cw = -1;
1992 dflags = flags|SD_NOJMP;
1993 while (1)
1994 {
1995 te = skip_to_delim (string, ts, d, dflags);
1996
1997 /* If we have a non-whitespace delimiter character, use it to make a
1998 separate field. This is just about what $IFS splitting does and
1999 is closer to the behavior of the shell parser. */
2000 if (ts == te && d2 && member (string[ts], d2))
2001 {
2002 te = ts + 1;
2003 /* If we're using IFS splitting, the non-whitespace delimiter char
2004 and any additional IFS whitespace delimits a field. */
2005 if (ifs_split)
2006 while (member (string[te], d) && spctabnl (string[te]))
2007 te++;
2008 else
2009 while (member (string[te], d2))
2010 te++;
2011 }
2012
2013 token = substring (string, ts, te);
2014
2015 ret = add_string_to_list (token, ret);
2016 free (token);
2017 nw++;
2018
2019 if (sentinel >= ts && sentinel <= te)
2020 cw = nw;
2021
2022 /* If the cursor is at whitespace just before word start, set the
2023 sentinel word to the current word. */
2024 if (cwp && cw == -1 && sentinel == ts-1)
2025 cw = nw;
2026
2027 /* If the cursor is at whitespace between two words, make a new, empty
2028 word, add it before (well, after, since the list is in reverse order)
2029 the word we just added, and set the current word to that one. */
2030 if (cwp && cw == -1 && sentinel < ts)
2031 {
2032 tl = make_word_list (make_word (""), ret->next);
2033 ret->next = tl;
2034 cw = nw;
2035 nw++;
2036 }
2037
2038 if (string[te] == 0)
2039 break;
2040
2041 i = te;
2042 while (member (string[i], d) && (ifs_split || spctabnl(string[i])))
2043 i++;
2044
2045 if (string[i])
2046 ts = i;
2047 else
2048 break;
2049 }
2050
2051 /* Special case for SENTINEL at the end of STRING. If we haven't found
2052 the word containing SENTINEL yet, and the index we're looking for is at
2053 the end of STRING (or past the end of the previously-found token,
2054 possible if the end of the line is composed solely of IFS whitespace)
2055 add an additional null argument and set the current word pointer to that. */
2056 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
2057 {
2058 if (whitespace (string[sentinel - 1]))
2059 {
2060 token = "";
2061 ret = add_string_to_list (token, ret);
2062 nw++;
2063 }
2064 cw = nw;
2065 }
2066
2067 if (nwp)
2068 *nwp = nw;
2069 if (cwp)
2070 *cwp = cw;
2071
2072 return (REVERSE_LIST (ret, WORD_LIST *));
2073 }
2074 #endif /* READLINE */
2075
2076 #if 0
2077 /* UNUSED */
2078 /* Extract the name of the variable to bind to from the assignment string. */
2079 char *
2080 assignment_name (string)
2081 char *string;
2082 {
2083 int offset;
2084 char *temp;
2085
2086 offset = assignment (string, 0);
2087 if (offset == 0)
2088 return (char *)NULL;
2089 temp = substring (string, 0, offset);
2090 return (temp);
2091 }
2092 #endif
2093
2094 /* **************************************************************** */
2095 /* */
2096 /* Functions to convert strings to WORD_LISTs and vice versa */
2097 /* */
2098 /* **************************************************************** */
2099
2100 /* Return a single string of all the words in LIST. SEP is the separator
2101 to put between individual elements of LIST in the output string. */
2102 char *
2103 string_list_internal (list, sep)
2104 WORD_LIST *list;
2105 char *sep;
2106 {
2107 register WORD_LIST *t;
2108 char *result, *r;
2109 int word_len, sep_len, result_size;
2110
2111 if (list == 0)
2112 return ((char *)NULL);
2113
2114 /* Short-circuit quickly if we don't need to separate anything. */
2115 if (list->next == 0)
2116 return (savestring (list->word->word));
2117
2118 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2119 sep_len = STRLEN (sep);
2120 result_size = 0;
2121
2122 for (t = list; t; t = t->next)
2123 {
2124 if (t != list)
2125 result_size += sep_len;
2126 result_size += strlen (t->word->word);
2127 }
2128
2129 r = result = (char *)xmalloc (result_size + 1);
2130
2131 for (t = list; t; t = t->next)
2132 {
2133 if (t != list && sep_len)
2134 {
2135 if (sep_len > 1)
2136 {
2137 FASTCOPY (sep, r, sep_len);
2138 r += sep_len;
2139 }
2140 else
2141 *r++ = sep[0];
2142 }
2143
2144 word_len = strlen (t->word->word);
2145 FASTCOPY (t->word->word, r, word_len);
2146 r += word_len;
2147 }
2148
2149 *r = '\0';
2150 return (result);
2151 }
2152
2153 /* Return a single string of all the words present in LIST, separating
2154 each word with a space. */
2155 char *
2156 string_list (list)
2157 WORD_LIST *list;
2158 {
2159 return (string_list_internal (list, " "));
2160 }
2161
2162 /* An external interface that can be used by the rest of the shell to
2163 obtain a string containing the first character in $IFS. Handles all
2164 the multibyte complications. If LENP is non-null, it is set to the
2165 length of the returned string. */
2166 char *
2167 ifs_firstchar (lenp)
2168 int *lenp;
2169 {
2170 char *ret;
2171 int len;
2172
2173 ret = xmalloc (MB_LEN_MAX + 1);
2174 #if defined (HANDLE_MULTIBYTE)
2175 if (ifs_firstc_len == 1)
2176 {
2177 ret[0] = ifs_firstc[0];
2178 ret[1] = '\0';
2179 len = ret[0] ? 1 : 0;
2180 }
2181 else
2182 {
2183 memcpy (ret, ifs_firstc, ifs_firstc_len);
2184 ret[len = ifs_firstc_len] = '\0';
2185 }
2186 #else
2187 ret[0] = ifs_firstc;
2188 ret[1] = '\0';
2189 len = ret[0] ? 0 : 1;
2190 #endif
2191
2192 if (lenp)
2193 *lenp = len;
2194
2195 return ret;
2196 }
2197
2198 /* Return a single string of all the words present in LIST, obeying the
2199 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2200 expansion [of $*] appears within a double quoted string, it expands
2201 to a single field with the value of each parameter separated by the
2202 first character of the IFS variable, or by a <space> if IFS is unset." */
2203 char *
2204 string_list_dollar_star (list)
2205 WORD_LIST *list;
2206 {
2207 char *ret;
2208 #if defined (HANDLE_MULTIBYTE)
2209 # if defined (__GNUC__)
2210 char sep[MB_CUR_MAX + 1];
2211 # else
2212 char *sep = 0;
2213 # endif
2214 #else
2215 char sep[2];
2216 #endif
2217
2218 #if defined (HANDLE_MULTIBYTE)
2219 # if !defined (__GNUC__)
2220 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2221 # endif /* !__GNUC__ */
2222 if (ifs_firstc_len == 1)
2223 {
2224 sep[0] = ifs_firstc[0];
2225 sep[1] = '\0';
2226 }
2227 else
2228 {
2229 memcpy (sep, ifs_firstc, ifs_firstc_len);
2230 sep[ifs_firstc_len] = '\0';
2231 }
2232 #else
2233 sep[0] = ifs_firstc;
2234 sep[1] = '\0';
2235 #endif
2236
2237 ret = string_list_internal (list, sep);
2238 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2239 free (sep);
2240 #endif
2241 return ret;
2242 }
2243
2244 /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2245 is non-zero, the $@ appears within double quotes, and we should quote
2246 the list before converting it into a string. If IFS is unset, and the
2247 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2248 in the words in the list, because the default value of $IFS is
2249 <space><tab><newline>, IFS characters in the words in the list should
2250 also be split. If IFS is null, and the word is not quoted, we need
2251 to quote the words in the list to preserve the positional parameters
2252 exactly. */
2253 char *
2254 string_list_dollar_at (list, quoted)
2255 WORD_LIST *list;
2256 int quoted;
2257 {
2258 char *ifs, *ret;
2259 #if defined (HANDLE_MULTIBYTE)
2260 # if defined (__GNUC__)
2261 char sep[MB_CUR_MAX + 1];
2262 # else
2263 char *sep = 0;
2264 # endif /* !__GNUC__ */
2265 #else
2266 char sep[2];
2267 #endif
2268 WORD_LIST *tlist;
2269
2270 /* XXX this could just be ifs = ifs_value; */
2271 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
2272
2273 #if defined (HANDLE_MULTIBYTE)
2274 # if !defined (__GNUC__)
2275 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2276 # endif /* !__GNUC__ */
2277 if (ifs && *ifs)
2278 {
2279 if (ifs_firstc_len == 1)
2280 {
2281 sep[0] = ifs_firstc[0];
2282 sep[1] = '\0';
2283 }
2284 else
2285 {
2286 memcpy (sep, ifs_firstc, ifs_firstc_len);
2287 sep[ifs_firstc_len] = '\0';
2288 }
2289 }
2290 else
2291 {
2292 sep[0] = ' ';
2293 sep[1] = '\0';
2294 }
2295 #else
2296 sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs;
2297 sep[1] = '\0';
2298 #endif
2299
2300 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2301 it now that quote_escapes quotes spaces */
2302 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
2303 ? quote_list (list)
2304 : list_quote_escapes (list);
2305
2306 ret = string_list_internal (tlist, sep);
2307 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2308 free (sep);
2309 #endif
2310 return ret;
2311 }
2312
2313 /* Turn the positional paramters into a string, understanding quoting and
2314 the various subtleties of using the first character of $IFS as the
2315 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2316 string_list as appropriate. */
2317 char *
2318 string_list_pos_params (pchar, list, quoted)
2319 int pchar;
2320 WORD_LIST *list;
2321 int quoted;
2322 {
2323 char *ret;
2324 WORD_LIST *tlist;
2325
2326 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2327 {
2328 tlist = quote_list (list);
2329 word_list_remove_quoted_nulls (tlist);
2330 ret = string_list_dollar_star (tlist);
2331 }
2332 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2333 {
2334 tlist = quote_list (list);
2335 word_list_remove_quoted_nulls (tlist);
2336 ret = string_list (tlist);
2337 }
2338 else if (pchar == '*')
2339 {
2340 /* Even when unquoted, string_list_dollar_star does the right thing
2341 making sure that the first character of $IFS is used as the
2342 separator. */
2343 ret = string_list_dollar_star (list);
2344 }
2345 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2346 /* We use string_list_dollar_at, but only if the string is quoted, since
2347 that quotes the escapes if it's not, which we don't want. We could
2348 use string_list (the old code did), but that doesn't do the right
2349 thing if the first character of $IFS is not a space. We use
2350 string_list_dollar_star if the string is unquoted so we make sure that
2351 the elements of $@ are separated by the first character of $IFS for
2352 later splitting. */
2353 ret = string_list_dollar_at (list, quoted);
2354 else if (pchar == '@')
2355 ret = string_list_dollar_star (list);
2356 else
2357 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2358
2359 return ret;
2360 }
2361
2362 /* Return the list of words present in STRING. Separate the string into
2363 words at any of the characters found in SEPARATORS. If QUOTED is
2364 non-zero then word in the list will have its quoted flag set, otherwise
2365 the quoted flag is left as make_word () deemed fit.
2366
2367 This obeys the P1003.2 word splitting semantics. If `separators' is
2368 exactly <space><tab><newline>, then the splitting algorithm is that of
2369 the Bourne shell, which treats any sequence of characters from `separators'
2370 as a delimiter. If IFS is unset, which results in `separators' being set
2371 to "", no splitting occurs. If separators has some other value, the
2372 following rules are applied (`IFS white space' means zero or more
2373 occurrences of <space>, <tab>, or <newline>, as long as those characters
2374 are in `separators'):
2375
2376 1) IFS white space is ignored at the start and the end of the
2377 string.
2378 2) Each occurrence of a character in `separators' that is not
2379 IFS white space, along with any adjacent occurrences of
2380 IFS white space delimits a field.
2381 3) Any nonzero-length sequence of IFS white space delimits a field.
2382 */
2383
2384 /* BEWARE! list_string strips null arguments. Don't call it twice and
2385 expect to have "" preserved! */
2386
2387 /* This performs word splitting and quoted null character removal on
2388 STRING. */
2389 #define issep(c) \
2390 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2391 : (c) == (separators)[0]) \
2392 : 0)
2393
2394 WORD_LIST *
2395 list_string (string, separators, quoted)
2396 register char *string, *separators;
2397 int quoted;
2398 {
2399 WORD_LIST *result;
2400 WORD_DESC *t;
2401 char *current_word, *s;
2402 int sindex, sh_style_split, whitesep, xflags;
2403 size_t slen;
2404
2405 if (!string || !*string)
2406 return ((WORD_LIST *)NULL);
2407
2408 sh_style_split = separators && separators[0] == ' ' &&
2409 separators[1] == '\t' &&
2410 separators[2] == '\n' &&
2411 separators[3] == '\0';
2412 for (xflags = 0, s = ifs_value; s && *s; s++)
2413 {
2414 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2415 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2416 }
2417
2418 slen = 0;
2419 /* Remove sequences of whitespace at the beginning of STRING, as
2420 long as those characters appear in IFS. Do not do this if
2421 STRING is quoted or if there are no separator characters. */
2422 if (!quoted || !separators || !*separators)
2423 {
2424 for (s = string; *s && spctabnl (*s) && issep (*s); s++);
2425
2426 if (!*s)
2427 return ((WORD_LIST *)NULL);
2428
2429 string = s;
2430 }
2431
2432 /* OK, now STRING points to a word that does not begin with white space.
2433 The splitting algorithm is:
2434 extract a word, stopping at a separator
2435 skip sequences of spc, tab, or nl as long as they are separators
2436 This obeys the field splitting rules in Posix.2. */
2437 slen = (MB_CUR_MAX > 1) ? strlen (string) : 1;
2438 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
2439 {
2440 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2441 unless multibyte chars are possible. */
2442 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
2443 if (current_word == 0)
2444 break;
2445
2446 /* If we have a quoted empty string, add a quoted null argument. We
2447 want to preserve the quoted null character iff this is a quoted
2448 empty string; otherwise the quoted null characters are removed
2449 below. */
2450 if (QUOTED_NULL (current_word))
2451 {
2452 t = alloc_word_desc ();
2453 t->word = make_quoted_char ('\0');
2454 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2455 result = make_word_list (t, result);
2456 }
2457 else if (current_word[0] != '\0')
2458 {
2459 /* If we have something, then add it regardless. However,
2460 perform quoted null character removal on the current word. */
2461 remove_quoted_nulls (current_word);
2462 result = add_string_to_list (current_word, result);
2463 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
2464 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2465 result->word->flags |= W_QUOTED;
2466 }
2467
2468 /* If we're not doing sequences of separators in the traditional
2469 Bourne shell style, then add a quoted null argument. */
2470 else if (!sh_style_split && !spctabnl (string[sindex]))
2471 {
2472 t = alloc_word_desc ();
2473 t->word = make_quoted_char ('\0');
2474 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2475 result = make_word_list (t, result);
2476 }
2477
2478 free (current_word);
2479
2480 /* Note whether or not the separator is IFS whitespace, used later. */
2481 whitesep = string[sindex] && spctabnl (string[sindex]);
2482
2483 /* Move past the current separator character. */
2484 if (string[sindex])
2485 {
2486 DECLARE_MBSTATE;
2487 ADVANCE_CHAR (string, slen, sindex);
2488 }
2489
2490 /* Now skip sequences of space, tab, or newline characters if they are
2491 in the list of separators. */
2492 while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex]))
2493 sindex++;
2494
2495 /* If the first separator was IFS whitespace and the current character
2496 is a non-whitespace IFS character, it should be part of the current
2497 field delimiter, not a separate delimiter that would result in an
2498 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
2499 if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex]))
2500 {
2501 sindex++;
2502 /* An IFS character that is not IFS white space, along with any
2503 adjacent IFS white space, shall delimit a field. (SUSv3) */
2504 while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex]))
2505 sindex++;
2506 }
2507 }
2508 return (REVERSE_LIST (result, WORD_LIST *));
2509 }
2510
2511 /* Parse a single word from STRING, using SEPARATORS to separate fields.
2512 ENDPTR is set to the first character after the word. This is used by
2513 the `read' builtin. This is never called with SEPARATORS != $IFS;
2514 it should be simplified.
2515
2516 XXX - this function is very similar to list_string; they should be
2517 combined - XXX */
2518 char *
2519 get_word_from_string (stringp, separators, endptr)
2520 char **stringp, *separators, **endptr;
2521 {
2522 register char *s;
2523 char *current_word;
2524 int sindex, sh_style_split, whitesep, xflags;
2525 size_t slen;
2526
2527 if (!stringp || !*stringp || !**stringp)
2528 return ((char *)NULL);
2529
2530 sh_style_split = separators && separators[0] == ' ' &&
2531 separators[1] == '\t' &&
2532 separators[2] == '\n' &&
2533 separators[3] == '\0';
2534 for (xflags = 0, s = ifs_value; s && *s; s++)
2535 {
2536 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2537 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2538 }
2539
2540 s = *stringp;
2541 slen = 0;
2542
2543 /* Remove sequences of whitespace at the beginning of STRING, as
2544 long as those characters appear in IFS. */
2545 if (sh_style_split || !separators || !*separators)
2546 {
2547 for (; *s && spctabnl (*s) && isifs (*s); s++);
2548
2549 /* If the string is nothing but whitespace, update it and return. */
2550 if (!*s)
2551 {
2552 *stringp = s;
2553 if (endptr)
2554 *endptr = s;
2555 return ((char *)NULL);
2556 }
2557 }
2558
2559 /* OK, S points to a word that does not begin with white space.
2560 Now extract a word, stopping at a separator, save a pointer to
2561 the first character after the word, then skip sequences of spc,
2562 tab, or nl as long as they are separators.
2563
2564 This obeys the field splitting rules in Posix.2. */
2565 sindex = 0;
2566 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2567 unless multibyte chars are possible. */
2568 slen = (MB_CUR_MAX > 1) ? strlen (s) : 1;
2569 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
2570
2571 /* Set ENDPTR to the first character after the end of the word. */
2572 if (endptr)
2573 *endptr = s + sindex;
2574
2575 /* Note whether or not the separator is IFS whitespace, used later. */
2576 whitesep = s[sindex] && spctabnl (s[sindex]);
2577
2578 /* Move past the current separator character. */
2579 if (s[sindex])
2580 {
2581 DECLARE_MBSTATE;
2582 ADVANCE_CHAR (s, slen, sindex);
2583 }
2584
2585 /* Now skip sequences of space, tab, or newline characters if they are
2586 in the list of separators. */
2587 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2588 sindex++;
2589
2590 /* If the first separator was IFS whitespace and the current character is
2591 a non-whitespace IFS character, it should be part of the current field
2592 delimiter, not a separate delimiter that would result in an empty field.
2593 Look at POSIX.2, 3.6.5, (3)(b). */
2594 if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex]))
2595 {
2596 sindex++;
2597 /* An IFS character that is not IFS white space, along with any adjacent
2598 IFS white space, shall delimit a field. */
2599 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2600 sindex++;
2601 }
2602
2603 /* Update STRING to point to the next field. */
2604 *stringp = s + sindex;
2605 return (current_word);
2606 }
2607
2608 /* Remove IFS white space at the end of STRING. Start at the end
2609 of the string and walk backwards until the beginning of the string
2610 or we find a character that's not IFS white space and not CTLESC.
2611 Only let CTLESC escape a white space character if SAW_ESCAPE is
2612 non-zero. */
2613 char *
2614 strip_trailing_ifs_whitespace (string, separators, saw_escape)
2615 char *string, *separators;
2616 int saw_escape;
2617 {
2618 char *s;
2619
2620 s = string + STRLEN (string) - 1;
2621 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
2622 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2623 s--;
2624 *++s = '\0';
2625 return string;
2626 }
2627
2628 #if 0
2629 /* UNUSED */
2630 /* Split STRING into words at whitespace. Obeys shell-style quoting with
2631 backslashes, single and double quotes. */
2632 WORD_LIST *
2633 list_string_with_quotes (string)
2634 char *string;
2635 {
2636 WORD_LIST *list;
2637 char *token, *s;
2638 size_t s_len;
2639 int c, i, tokstart, len;
2640
2641 for (s = string; s && *s && spctabnl (*s); s++)
2642 ;
2643 if (s == 0 || *s == 0)
2644 return ((WORD_LIST *)NULL);
2645
2646 s_len = strlen (s);
2647 tokstart = i = 0;
2648 list = (WORD_LIST *)NULL;
2649 while (1)
2650 {
2651 c = s[i];
2652 if (c == '\\')
2653 {
2654 i++;
2655 if (s[i])
2656 i++;
2657 }
2658 else if (c == '\'')
2659 i = skip_single_quoted (s, s_len, ++i);
2660 else if (c == '"')
2661 i = skip_double_quoted (s, s_len, ++i);
2662 else if (c == 0 || spctabnl (c))
2663 {
2664 /* We have found the end of a token. Make a word out of it and
2665 add it to the word list. */
2666 token = substring (s, tokstart, i);
2667 list = add_string_to_list (token, list);
2668 free (token);
2669 while (spctabnl (s[i]))
2670 i++;
2671 if (s[i])
2672 tokstart = i;
2673 else
2674 break;
2675 }
2676 else
2677 i++; /* normal character */
2678 }
2679 return (REVERSE_LIST (list, WORD_LIST *));
2680 }
2681 #endif
2682
2683 /********************************************************/
2684 /* */
2685 /* Functions to perform assignment statements */
2686 /* */
2687 /********************************************************/
2688
2689 #if defined (ARRAY_VARS)
2690 static SHELL_VAR *
2691 do_compound_assignment (name, value, flags)
2692 char *name, *value;
2693 int flags;
2694 {
2695 SHELL_VAR *v;
2696 int mklocal, mkassoc;
2697 WORD_LIST *list;
2698
2699 mklocal = flags & ASS_MKLOCAL;
2700 mkassoc = flags & ASS_MKASSOC;
2701
2702 if (mklocal && variable_context)
2703 {
2704 v = find_variable (name);
2705 list = expand_compound_array_assignment (v, value, flags);
2706 if (mkassoc)
2707 v = make_local_assoc_variable (name);
2708 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
2709 v = make_local_array_variable (name);
2710 assign_compound_array_list (v, list, flags);
2711 }
2712 else
2713 v = assign_array_from_string (name, value, flags);
2714
2715 return (v);
2716 }
2717 #endif
2718
2719 /* Given STRING, an assignment string, get the value of the right side
2720 of the `=', and bind it to the left side. If EXPAND is true, then
2721 perform parameter expansion, command substitution, and arithmetic
2722 expansion on the right-hand side. Perform tilde expansion in any
2723 case. Do not perform word splitting on the result of expansion. */
2724 static int
2725 do_assignment_internal (word, expand)
2726 const WORD_DESC *word;
2727 int expand;
2728 {
2729 int offset, appendop, assign_list, aflags, retval;
2730 char *name, *value, *temp;
2731 SHELL_VAR *entry;
2732 #if defined (ARRAY_VARS)
2733 char *t;
2734 int ni;
2735 #endif
2736 const char *string;
2737
2738 if (word == 0 || word->word == 0)
2739 return 0;
2740
2741 appendop = assign_list = aflags = 0;
2742 string = word->word;
2743 offset = assignment (string, 0);
2744 name = savestring (string);
2745 value = (char *)NULL;
2746
2747 if (name[offset] == '=')
2748 {
2749 if (name[offset - 1] == '+')
2750 {
2751 appendop = 1;
2752 name[offset - 1] = '\0';
2753 }
2754
2755 name[offset] = 0; /* might need this set later */
2756 temp = name + offset + 1;
2757
2758 #if defined (ARRAY_VARS)
2759 if (expand && (word->flags & W_COMPASSIGN))
2760 {
2761 assign_list = ni = 1;
2762 value = extract_array_assignment_list (temp, &ni);
2763 }
2764 else
2765 #endif
2766 if (expand && temp[0])
2767 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
2768 else
2769 value = savestring (temp);
2770 }
2771
2772 if (value == 0)
2773 {
2774 value = (char *)xmalloc (1);
2775 value[0] = '\0';
2776 }
2777
2778 if (echo_command_at_execute)
2779 {
2780 if (appendop)
2781 name[offset - 1] = '+';
2782 xtrace_print_assignment (name, value, assign_list, 1);
2783 if (appendop)
2784 name[offset - 1] = '\0';
2785 }
2786
2787 #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
2788
2789 if (appendop)
2790 aflags |= ASS_APPEND;
2791
2792 #if defined (ARRAY_VARS)
2793 if (t = mbschr (name, '[')) /*]*/
2794 {
2795 if (assign_list)
2796 {
2797 report_error (_("%s: cannot assign list to array member"), name);
2798 ASSIGN_RETURN (0);
2799 }
2800 entry = assign_array_element (name, value, aflags);
2801 if (entry == 0)
2802 ASSIGN_RETURN (0);
2803 }
2804 else if (assign_list)
2805 {
2806 if (word->flags & W_ASSIGNARG)
2807 aflags |= ASS_MKLOCAL;
2808 if (word->flags & W_ASSIGNASSOC)
2809 aflags |= ASS_MKASSOC;
2810 entry = do_compound_assignment (name, value, aflags);
2811 }
2812 else
2813 #endif /* ARRAY_VARS */
2814 entry = bind_variable (name, value, aflags);
2815
2816 stupidly_hack_special_variables (name);
2817
2818 #if 1
2819 /* Return 1 if the assignment seems to have been performed correctly. */
2820 if (entry == 0 || readonly_p (entry))
2821 retval = 0; /* assignment failure */
2822 else if (noassign_p (entry))
2823 {
2824 last_command_exit_value = EXECUTION_FAILURE;
2825 retval = 1; /* error status, but not assignment failure */
2826 }
2827 else
2828 retval = 1;
2829
2830 if (entry && retval != 0 && noassign_p (entry) == 0)
2831 VUNSETATTR (entry, att_invisible);
2832
2833 ASSIGN_RETURN (retval);
2834 #else
2835 if (entry)
2836 VUNSETATTR (entry, att_invisible);
2837
2838 ASSIGN_RETURN (entry ? ((readonly_p (entry) == 0) && noassign_p (entry) == 0) : 0);
2839 #endif
2840 }
2841
2842 /* Perform the assignment statement in STRING, and expand the
2843 right side by doing tilde, command and parameter expansion. */
2844 int
2845 do_assignment (string)
2846 char *string;
2847 {
2848 WORD_DESC td;
2849
2850 td.flags = W_ASSIGNMENT;
2851 td.word = string;
2852
2853 return do_assignment_internal (&td, 1);
2854 }
2855
2856 int
2857 do_word_assignment (word, flags)
2858 WORD_DESC *word;
2859 int flags;
2860 {
2861 return do_assignment_internal (word, 1);
2862 }
2863
2864 /* Given STRING, an assignment string, get the value of the right side
2865 of the `=', and bind it to the left side. Do not perform any word
2866 expansions on the right hand side. */
2867 int
2868 do_assignment_no_expand (string)
2869 char *string;
2870 {
2871 WORD_DESC td;
2872
2873 td.flags = W_ASSIGNMENT;
2874 td.word = string;
2875
2876 return (do_assignment_internal (&td, 0));
2877 }
2878
2879 /***************************************************
2880 * *
2881 * Functions to manage the positional parameters *
2882 * *
2883 ***************************************************/
2884
2885 /* Return the word list that corresponds to `$*'. */
2886 WORD_LIST *
2887 list_rest_of_args ()
2888 {
2889 register WORD_LIST *list, *args;
2890 int i;
2891
2892 /* Break out of the loop as soon as one of the dollar variables is null. */
2893 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
2894 list = make_word_list (make_bare_word (dollar_vars[i]), list);
2895
2896 for (args = rest_of_args; args; args = args->next)
2897 list = make_word_list (make_bare_word (args->word->word), list);
2898
2899 return (REVERSE_LIST (list, WORD_LIST *));
2900 }
2901
2902 int
2903 number_of_args ()
2904 {
2905 register WORD_LIST *list;
2906 int n;
2907
2908 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
2909 ;
2910 for (list = rest_of_args; list; list = list->next)
2911 n++;
2912 return n;
2913 }
2914
2915 /* Return the value of a positional parameter. This handles values > 10. */
2916 char *
2917 get_dollar_var_value (ind)
2918 intmax_t ind;
2919 {
2920 char *temp;
2921 WORD_LIST *p;
2922
2923 if (ind < 10)
2924 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
2925 else /* We want something like ${11} */
2926 {
2927 ind -= 10;
2928 for (p = rest_of_args; p && ind--; p = p->next)
2929 ;
2930 temp = p ? savestring (p->word->word) : (char *)NULL;
2931 }
2932 return (temp);
2933 }
2934
2935 /* Make a single large string out of the dollar digit variables,
2936 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
2937 case of "$*" with respect to IFS. */
2938 char *
2939 string_rest_of_args (dollar_star)
2940 int dollar_star;
2941 {
2942 register WORD_LIST *list;
2943 char *string;
2944
2945 list = list_rest_of_args ();
2946 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
2947 dispose_words (list);
2948 return (string);
2949 }
2950
2951 /* Return a string containing the positional parameters from START to
2952 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
2953 which only makes a difference if QUOTED is non-zero. If QUOTED includes
2954 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
2955 no quoting chars are added. */
2956 static char *
2957 pos_params (string, start, end, quoted)
2958 char *string;
2959 int start, end, quoted;
2960 {
2961 WORD_LIST *save, *params, *h, *t;
2962 char *ret;
2963 int i;
2964
2965 /* see if we can short-circuit. if start == end, we want 0 parameters. */
2966 if (start == end)
2967 return ((char *)NULL);
2968
2969 save = params = list_rest_of_args ();
2970 if (save == 0)
2971 return ((char *)NULL);
2972
2973 if (start == 0) /* handle ${@:0[:x]} specially */
2974 {
2975 t = make_word_list (make_word (dollar_vars[0]), params);
2976 save = params = t;
2977 }
2978
2979 for (i = start ? 1 : 0; params && i < start; i++)
2980 params = params->next;
2981 if (params == 0)
2982 return ((char *)NULL);
2983 for (h = t = params; params && i < end; i++)
2984 {
2985 t = params;
2986 params = params->next;
2987 }
2988
2989 t->next = (WORD_LIST *)NULL;
2990
2991 ret = string_list_pos_params (string[0], h, quoted);
2992
2993 if (t != params)
2994 t->next = params;
2995
2996 dispose_words (save);
2997 return (ret);
2998 }
2999
3000 /******************************************************************/
3001 /* */
3002 /* Functions to expand strings to strings or WORD_LISTs */
3003 /* */
3004 /******************************************************************/
3005
3006 #if defined (PROCESS_SUBSTITUTION)
3007 #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
3008 #else
3009 #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
3010 #endif
3011
3012 /* If there are any characters in STRING that require full expansion,
3013 then call FUNC to expand STRING; otherwise just perform quote
3014 removal if necessary. This returns a new string. */
3015 static char *
3016 expand_string_if_necessary (string, quoted, func)
3017 char *string;
3018 int quoted;
3019 EXPFUNC *func;
3020 {
3021 WORD_LIST *list;
3022 size_t slen;
3023 int i, saw_quote;
3024 char *ret;
3025 DECLARE_MBSTATE;
3026
3027 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3028 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3029 i = saw_quote = 0;
3030 while (string[i])
3031 {
3032 if (EXP_CHAR (string[i]))
3033 break;
3034 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3035 saw_quote = 1;
3036 ADVANCE_CHAR (string, slen, i);
3037 }
3038
3039 if (string[i])
3040 {
3041 list = (*func) (string, quoted);
3042 if (list)
3043 {
3044 ret = string_list (list);
3045 dispose_words (list);
3046 }
3047 else
3048 ret = (char *)NULL;
3049 }
3050 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3051 ret = string_quote_removal (string, quoted);
3052 else
3053 ret = savestring (string);
3054
3055 return ret;
3056 }
3057
3058 static inline char *
3059 expand_string_to_string_internal (string, quoted, func)
3060 char *string;
3061 int quoted;
3062 EXPFUNC *func;
3063 {
3064 WORD_LIST *list;
3065 char *ret;
3066
3067 if (string == 0 || *string == '\0')
3068 return ((char *)NULL);
3069
3070 list = (*func) (string, quoted);
3071 if (list)
3072 {
3073 ret = string_list (list);
3074 dispose_words (list);
3075 }
3076 else
3077 ret = (char *)NULL;
3078
3079 return (ret);
3080 }
3081
3082 char *
3083 expand_string_to_string (string, quoted)
3084 char *string;
3085 int quoted;
3086 {
3087 return (expand_string_to_string_internal (string, quoted, expand_string));
3088 }
3089
3090 char *
3091 expand_string_unsplit_to_string (string, quoted)
3092 char *string;
3093 int quoted;
3094 {
3095 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3096 }
3097
3098 char *
3099 expand_assignment_string_to_string (string, quoted)
3100 char *string;
3101 int quoted;
3102 {
3103 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3104 }
3105
3106 char *
3107 expand_arith_string (string, quoted)
3108 char *string;
3109 int quoted;
3110 {
3111 return (expand_string_if_necessary (string, quoted, expand_string));
3112 }
3113
3114 #if defined (COND_COMMAND)
3115 /* Just remove backslashes in STRING. Returns a new string. */
3116 char *
3117 remove_backslashes (string)
3118 char *string;
3119 {
3120 char *r, *ret, *s;
3121
3122 r = ret = (char *)xmalloc (strlen (string) + 1);
3123 for (s = string; s && *s; )
3124 {
3125 if (*s == '\\')
3126 s++;
3127 if (*s == 0)
3128 break;
3129 *r++ = *s++;
3130 }
3131 *r = '\0';
3132 return ret;
3133 }
3134
3135 /* This needs better error handling. */
3136 /* Expand W for use as an argument to a unary or binary operator in a
3137 [[...]] expression. If SPECIAL is 1, this is the rhs argument
3138 to the != or == operator, and should be treated as a pattern. In
3139 this case, we quote the string specially for the globbing code. If
3140 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3141 be quoted appropriately for regcomp/regexec. The caller is responsible
3142 for removing the backslashes if the unquoted word is needed later. */
3143 char *
3144 cond_expand_word (w, special)
3145 WORD_DESC *w;
3146 int special;
3147 {
3148 char *r, *p;
3149 WORD_LIST *l;
3150 int qflags;
3151
3152 if (w->word == 0 || w->word[0] == '\0')
3153 return ((char *)NULL);
3154
3155 w->flags |= W_NOSPLIT2;
3156 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3157 if (l)
3158 {
3159 if (special == 0)
3160 {
3161 dequote_list (l);
3162 r = string_list (l);
3163 }
3164 else
3165 {
3166 qflags = QGLOB_CVTNULL;
3167 if (special == 2)
3168 qflags |= QGLOB_REGEXP;
3169 p = string_list (l);
3170 r = quote_string_for_globbing (p, qflags);
3171 free (p);
3172 }
3173 dispose_words (l);
3174 }
3175 else
3176 r = (char *)NULL;
3177
3178 return r;
3179 }
3180 #endif
3181
3182 /* Call expand_word_internal to expand W and handle error returns.
3183 A convenience function for functions that don't want to handle
3184 any errors or free any memory before aborting. */
3185 static WORD_LIST *
3186 call_expand_word_internal (w, q, i, c, e)
3187 WORD_DESC *w;
3188 int q, i, *c, *e;
3189 {
3190 WORD_LIST *result;
3191
3192 result = expand_word_internal (w, q, i, c, e);
3193 if (result == &expand_word_error || result == &expand_word_fatal)
3194 {
3195 /* By convention, each time this error is returned, w->word has
3196 already been freed (it sometimes may not be in the fatal case,
3197 but that doesn't result in a memory leak because we're going
3198 to exit in most cases). */
3199 w->word = (char *)NULL;
3200 last_command_exit_value = EXECUTION_FAILURE;
3201 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
3202 /* NOTREACHED */
3203 }
3204 else
3205 return (result);
3206 }
3207
3208 /* Perform parameter expansion, command substitution, and arithmetic
3209 expansion on STRING, as if it were a word. Leave the result quoted. */
3210 static WORD_LIST *
3211 expand_string_internal (string, quoted)
3212 char *string;
3213 int quoted;
3214 {
3215 WORD_DESC td;
3216 WORD_LIST *tresult;
3217
3218 if (string == 0 || *string == 0)
3219 return ((WORD_LIST *)NULL);
3220
3221 td.flags = 0;
3222 td.word = savestring (string);
3223
3224 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3225
3226 FREE (td.word);
3227 return (tresult);
3228 }
3229
3230 /* Expand STRING by performing parameter expansion, command substitution,
3231 and arithmetic expansion. Dequote the resulting WORD_LIST before
3232 returning it, but do not perform word splitting. The call to
3233 remove_quoted_nulls () is in here because word splitting normally
3234 takes care of quote removal. */
3235 WORD_LIST *
3236 expand_string_unsplit (string, quoted)
3237 char *string;
3238 int quoted;
3239 {
3240 WORD_LIST *value;
3241
3242 if (string == 0 || *string == '\0')
3243 return ((WORD_LIST *)NULL);
3244
3245 expand_no_split_dollar_star = 1;
3246 value = expand_string_internal (string, quoted);
3247 expand_no_split_dollar_star = 0;
3248
3249 if (value)
3250 {
3251 if (value->word)
3252 {
3253 remove_quoted_nulls (value->word->word);
3254 value->word->flags &= ~W_HASQUOTEDNULL;
3255 }
3256 dequote_list (value);
3257 }
3258 return (value);
3259 }
3260
3261 /* Expand the rhs of an assignment statement */
3262 WORD_LIST *
3263 expand_string_assignment (string, quoted)
3264 char *string;
3265 int quoted;
3266 {
3267 WORD_DESC td;
3268 WORD_LIST *value;
3269
3270 if (string == 0 || *string == '\0')
3271 return ((WORD_LIST *)NULL);
3272
3273 expand_no_split_dollar_star = 1;
3274
3275 td.flags = W_ASSIGNRHS;
3276 td.word = savestring (string);
3277 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3278 FREE (td.word);
3279
3280 expand_no_split_dollar_star = 0;
3281
3282 if (value)
3283 {
3284 if (value->word)
3285 {
3286 remove_quoted_nulls (value->word->word);
3287 value->word->flags &= ~W_HASQUOTEDNULL;
3288 }
3289 dequote_list (value);
3290 }
3291 return (value);
3292 }
3293
3294
3295 /* Expand one of the PS? prompt strings. This is a sort of combination of
3296 expand_string_unsplit and expand_string_internal, but returns the
3297 passed string when an error occurs. Might want to trap other calls
3298 to jump_to_top_level here so we don't endlessly loop. */
3299 WORD_LIST *
3300 expand_prompt_string (string, quoted, wflags)
3301 char *string;
3302 int quoted;
3303 int wflags;
3304 {
3305 WORD_LIST *value;
3306 WORD_DESC td;
3307
3308 if (string == 0 || *string == 0)
3309 return ((WORD_LIST *)NULL);
3310
3311 td.flags = wflags;
3312 td.word = savestring (string);
3313
3314 no_longjmp_on_fatal_error = 1;
3315 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3316 no_longjmp_on_fatal_error = 0;
3317
3318 if (value == &expand_word_error || value == &expand_word_fatal)
3319 {
3320 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3321 return value;
3322 }
3323 FREE (td.word);
3324 if (value)
3325 {
3326 if (value->word)
3327 {
3328 remove_quoted_nulls (value->word->word);
3329 value->word->flags &= ~W_HASQUOTEDNULL;
3330 }
3331 dequote_list (value);
3332 }
3333 return (value);
3334 }
3335
3336 /* Expand STRING just as if you were expanding a word, but do not dequote
3337 the resultant WORD_LIST. This is called only from within this file,
3338 and is used to correctly preserve quoted characters when expanding
3339 things like ${1+"$@"}. This does parameter expansion, command
3340 substitution, arithmetic expansion, and word splitting. */
3341 static WORD_LIST *
3342 expand_string_leave_quoted (string, quoted)
3343 char *string;
3344 int quoted;
3345 {
3346 WORD_LIST *tlist;
3347 WORD_LIST *tresult;
3348
3349 if (string == 0 || *string == '\0')
3350 return ((WORD_LIST *)NULL);
3351
3352 tlist = expand_string_internal (string, quoted);
3353
3354 if (tlist)
3355 {
3356 tresult = word_list_split (tlist);
3357 dispose_words (tlist);
3358 return (tresult);
3359 }
3360 return ((WORD_LIST *)NULL);
3361 }
3362
3363 /* This does not perform word splitting or dequote the WORD_LIST
3364 it returns. */
3365 static WORD_LIST *
3366 expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at)
3367 char *string;
3368 int quoted, *dollar_at_p, *has_dollar_at;
3369 {
3370 WORD_DESC td;
3371 WORD_LIST *tresult;
3372
3373 if (string == 0 || *string == '\0')
3374 return (WORD_LIST *)NULL;
3375
3376 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
3377 td.word = string;
3378 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at);
3379 return (tresult);
3380 }
3381
3382 /* Expand STRING just as if you were expanding a word. This also returns
3383 a list of words. Note that filename globbing is *NOT* done for word
3384 or string expansion, just when the shell is expanding a command. This
3385 does parameter expansion, command substitution, arithmetic expansion,
3386 and word splitting. Dequote the resultant WORD_LIST before returning. */
3387 WORD_LIST *
3388 expand_string (string, quoted)
3389 char *string;
3390 int quoted;
3391 {
3392 WORD_LIST *result;
3393
3394 if (string == 0 || *string == '\0')
3395 return ((WORD_LIST *)NULL);
3396
3397 result = expand_string_leave_quoted (string, quoted);
3398 return (result ? dequote_list (result) : result);
3399 }
3400
3401 /***************************************************
3402 * *
3403 * Functions to handle quoting chars *
3404 * *
3405 ***************************************************/
3406
3407 /* Conventions:
3408
3409 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3410 The parser passes CTLNUL as CTLESC CTLNUL. */
3411
3412 /* Quote escape characters in string s, but no other characters. This is
3413 used to protect CTLESC and CTLNUL in variable values from the rest of
3414 the word expansion process after the variable is expanded (word splitting
3415 and filename generation). If IFS is null, we quote spaces as well, just
3416 in case we split on spaces later (in the case of unquoted $@, we will
3417 eventually attempt to split the entire word on spaces). Corresponding
3418 code exists in dequote_escapes. Even if we don't end up splitting on
3419 spaces, quoting spaces is not a problem. This should never be called on
3420 a string that is quoted with single or double quotes or part of a here
3421 document (effectively double-quoted). */
3422 char *
3423 quote_escapes (string)
3424 char *string;
3425 {
3426 register char *s, *t;
3427 size_t slen;
3428 char *result, *send;
3429 int quote_spaces, skip_ctlesc, skip_ctlnul;
3430 DECLARE_MBSTATE;
3431
3432 slen = strlen (string);
3433 send = string + slen;
3434
3435 quote_spaces = (ifs_value && *ifs_value == 0);
3436
3437 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3438 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3439
3440 t = result = (char *)xmalloc ((slen * 2) + 1);
3441 s = string;
3442
3443 while (*s)
3444 {
3445 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
3446 *t++ = CTLESC;
3447 COPY_CHAR_P (t, s, send);
3448 }
3449 *t = '\0';
3450 return (result);
3451 }
3452
3453 static WORD_LIST *
3454 list_quote_escapes (list)
3455 WORD_LIST *list;
3456 {
3457 register WORD_LIST *w;
3458 char *t;
3459
3460 for (w = list; w; w = w->next)
3461 {
3462 t = w->word->word;
3463 w->word->word = quote_escapes (t);
3464 free (t);
3465 }
3466 return list;
3467 }
3468
3469 /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3470
3471 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3472 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3473 data stream pass through properly.
3474
3475 We need to remove doubled CTLESC characters inside quoted strings before
3476 quoting the entire string, so we do not double the number of CTLESC
3477 characters.
3478
3479 Also used by parts of the pattern substitution code. */
3480 char *
3481 dequote_escapes (string)
3482 char *string;
3483 {
3484 register char *s, *t, *s1;
3485 size_t slen;
3486 char *result, *send;
3487 int quote_spaces;
3488 DECLARE_MBSTATE;
3489
3490 if (string == 0)
3491 return string;
3492
3493 slen = strlen (string);
3494 send = string + slen;
3495
3496 t = result = (char *)xmalloc (slen + 1);
3497
3498 if (strchr (string, CTLESC) == 0)
3499 return (strcpy (result, string));
3500
3501 quote_spaces = (ifs_value && *ifs_value == 0);
3502
3503 s = string;
3504 while (*s)
3505 {
3506 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
3507 {
3508 s++;
3509 if (*s == '\0')
3510 break;
3511 }
3512 COPY_CHAR_P (t, s, send);
3513 }
3514 *t = '\0';
3515 return result;
3516 }
3517
3518 /* Return a new string with the quoted representation of character C.
3519 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3520 set in any resultant WORD_DESC where this value is the word. */
3521 static char *
3522 make_quoted_char (c)
3523 int c;
3524 {
3525 char *temp;
3526
3527 temp = (char *)xmalloc (3);
3528 if (c == 0)
3529 {
3530 temp[0] = CTLNUL;
3531 temp[1] = '\0';
3532 }
3533 else
3534 {
3535 temp[0] = CTLESC;
3536 temp[1] = c;
3537 temp[2] = '\0';
3538 }
3539 return (temp);
3540 }
3541
3542 /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3543 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3544 this value is the word. */
3545 char *
3546 quote_string (string)
3547 char *string;
3548 {
3549 register char *t;
3550 size_t slen;
3551 char *result, *send;
3552
3553 if (*string == 0)
3554 {
3555 result = (char *)xmalloc (2);
3556 result[0] = CTLNUL;
3557 result[1] = '\0';
3558 }
3559 else
3560 {
3561 DECLARE_MBSTATE;
3562
3563 slen = strlen (string);
3564 send = string + slen;
3565
3566 result = (char *)xmalloc ((slen * 2) + 1);
3567
3568 for (t = result; string < send; )
3569 {
3570 *t++ = CTLESC;
3571 COPY_CHAR_P (t, string, send);
3572 }
3573 *t = '\0';
3574 }
3575 return (result);
3576 }
3577
3578 /* De-quote quoted characters in STRING. */
3579 char *
3580 dequote_string (string)
3581 char *string;
3582 {
3583 register char *s, *t;
3584 size_t slen;
3585 char *result, *send;
3586 DECLARE_MBSTATE;
3587
3588 slen = strlen (string);
3589
3590 t = result = (char *)xmalloc (slen + 1);
3591
3592 if (QUOTED_NULL (string))
3593 {
3594 result[0] = '\0';
3595 return (result);
3596 }
3597
3598 /* If no character in the string can be quoted, don't bother examining
3599 each character. Just return a copy of the string passed to us. */
3600 if (strchr (string, CTLESC) == NULL)
3601 return (strcpy (result, string));
3602
3603 send = string + slen;
3604 s = string;
3605 while (*s)
3606 {
3607 if (*s == CTLESC)
3608 {
3609 s++;
3610 if (*s == '\0')
3611 break;
3612 }
3613 COPY_CHAR_P (t, s, send);
3614 }
3615
3616 *t = '\0';
3617 return (result);
3618 }
3619
3620 /* Quote the entire WORD_LIST list. */
3621 static WORD_LIST *
3622 quote_list (list)
3623 WORD_LIST *list;
3624 {
3625 register WORD_LIST *w;
3626 char *t;
3627
3628 for (w = list; w; w = w->next)
3629 {
3630 t = w->word->word;
3631 w->word->word = quote_string (t);
3632 if (*t == 0)
3633 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
3634 w->word->flags |= W_QUOTED;
3635 free (t);
3636 }
3637 return list;
3638 }
3639
3640 /* De-quote quoted characters in each word in LIST. */
3641 WORD_LIST *
3642 dequote_list (list)
3643 WORD_LIST *list;
3644 {
3645 register char *s;
3646 register WORD_LIST *tlist;
3647
3648 for (tlist = list; tlist; tlist = tlist->next)
3649 {
3650 s = dequote_string (tlist->word->word);
3651 if (QUOTED_NULL (tlist->word->word))
3652 tlist->word->flags &= ~W_HASQUOTEDNULL;
3653 free (tlist->word->word);
3654 tlist->word->word = s;
3655 }
3656 return list;
3657 }
3658
3659 /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
3660 string. */
3661 char *
3662 remove_quoted_escapes (string)
3663 char *string;
3664 {
3665 char *t;
3666
3667 if (string)
3668 {
3669 t = dequote_escapes (string);
3670 strcpy (string, t);
3671 free (t);
3672 }
3673
3674 return (string);
3675 }
3676
3677 /* Perform quoted null character removal on STRING. We don't allow any
3678 quoted null characters in the middle or at the ends of strings because
3679 of how expand_word_internal works. remove_quoted_nulls () turns
3680 STRING into an empty string iff it only consists of a quoted null,
3681 and removes all unquoted CTLNUL characters. */
3682 char *
3683 remove_quoted_nulls (string)
3684 char *string;
3685 {
3686 register size_t slen;
3687 register int i, j, prev_i;
3688 DECLARE_MBSTATE;
3689
3690 if (strchr (string, CTLNUL) == 0) /* XXX */
3691 return string; /* XXX */
3692
3693 slen = strlen (string);
3694 i = j = 0;
3695
3696 while (i < slen)
3697 {
3698 if (string[i] == CTLESC)
3699 {
3700 /* Old code had j++, but we cannot assume that i == j at this
3701 point -- what if a CTLNUL has already been removed from the
3702 string? We don't want to drop the CTLESC or recopy characters
3703 that we've already copied down. */
3704 i++; string[j++] = CTLESC;
3705 if (i == slen)
3706 break;
3707 }
3708 else if (string[i] == CTLNUL)
3709 i++;
3710
3711 prev_i = i;
3712 ADVANCE_CHAR (string, slen, i);
3713 if (j < prev_i)
3714 {
3715 do string[j++] = string[prev_i++]; while (prev_i < i);
3716 }
3717 else
3718 j = i;
3719 }
3720 string[j] = '\0';
3721
3722 return (string);
3723 }
3724
3725 /* Perform quoted null character removal on each element of LIST.
3726 This modifies LIST. */
3727 void
3728 word_list_remove_quoted_nulls (list)
3729 WORD_LIST *list;
3730 {
3731 register WORD_LIST *t;
3732
3733 for (t = list; t; t = t->next)
3734 {
3735 remove_quoted_nulls (t->word->word);
3736 t->word->flags &= ~W_HASQUOTEDNULL;
3737 }
3738 }
3739
3740 /* **************************************************************** */
3741 /* */
3742 /* Functions for Matching and Removing Patterns */
3743 /* */
3744 /* **************************************************************** */
3745
3746 #if defined (HANDLE_MULTIBYTE)
3747 #if 0 /* Currently unused */
3748 static unsigned char *
3749 mb_getcharlens (string, len)
3750 char *string;
3751 int len;
3752 {
3753 int i, offset, last;
3754 unsigned char *ret;
3755 char *p;
3756 DECLARE_MBSTATE;
3757
3758 i = offset = 0;
3759 last = 0;
3760 ret = (unsigned char *)xmalloc (len);
3761 memset (ret, 0, len);
3762 while (string[last])
3763 {
3764 ADVANCE_CHAR (string, len, offset);
3765 ret[last] = offset - last;
3766 last = offset;
3767 }
3768 return ret;
3769 }
3770 #endif
3771 #endif
3772
3773 /* Remove the portion of PARAM matched by PATTERN according to OP, where OP
3774 can have one of 4 values:
3775 RP_LONG_LEFT remove longest matching portion at start of PARAM
3776 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
3777 RP_LONG_RIGHT remove longest matching portion at end of PARAM
3778 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
3779 */
3780
3781 #define RP_LONG_LEFT 1
3782 #define RP_SHORT_LEFT 2
3783 #define RP_LONG_RIGHT 3
3784 #define RP_SHORT_RIGHT 4
3785
3786 /* Returns its first argument if nothing matched; new memory otherwise */
3787 static char *
3788 remove_upattern (param, pattern, op)
3789 char *param, *pattern;
3790 int op;
3791 {
3792 register int len;
3793 register char *end;
3794 register char *p, *ret, c;
3795
3796 len = STRLEN (param);
3797 end = param + len;
3798
3799 switch (op)
3800 {
3801 case RP_LONG_LEFT: /* remove longest match at start */
3802 for (p = end; p >= param; p--)
3803 {
3804 c = *p; *p = '\0';
3805 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3806 {
3807 *p = c;
3808 return (savestring (p));
3809 }
3810 *p = c;
3811
3812 }
3813 break;
3814
3815 case RP_SHORT_LEFT: /* remove shortest match at start */
3816 for (p = param; p <= end; p++)
3817 {
3818 c = *p; *p = '\0';
3819 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3820 {
3821 *p = c;
3822 return (savestring (p));
3823 }
3824 *p = c;
3825 }
3826 break;
3827
3828 case RP_LONG_RIGHT: /* remove longest match at end */
3829 for (p = param; p <= end; p++)
3830 {
3831 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3832 {
3833 c = *p; *p = '\0';
3834 ret = savestring (param);
3835 *p = c;
3836 return (ret);
3837 }
3838 }
3839 break;
3840
3841 case RP_SHORT_RIGHT: /* remove shortest match at end */
3842 for (p = end; p >= param; p--)
3843 {
3844 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3845 {
3846 c = *p; *p = '\0';
3847 ret = savestring (param);
3848 *p = c;
3849 return (ret);
3850 }
3851 }
3852 break;
3853 }
3854
3855 return (param); /* no match, return original string */
3856 }
3857
3858 #if defined (HANDLE_MULTIBYTE)
3859 /* Returns its first argument if nothing matched; new memory otherwise */
3860 static wchar_t *
3861 remove_wpattern (wparam, wstrlen, wpattern, op)
3862 wchar_t *wparam;
3863 size_t wstrlen;
3864 wchar_t *wpattern;
3865 int op;
3866 {
3867 wchar_t wc, *ret;
3868 int n;
3869
3870 switch (op)
3871 {
3872 case RP_LONG_LEFT: /* remove longest match at start */
3873 for (n = wstrlen; n >= 0; n--)
3874 {
3875 wc = wparam[n]; wparam[n] = L'\0';
3876 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3877 {
3878 wparam[n] = wc;
3879 return (wcsdup (wparam + n));
3880 }
3881 wparam[n] = wc;
3882 }
3883 break;
3884
3885 case RP_SHORT_LEFT: /* remove shortest match at start */
3886 for (n = 0; n <= wstrlen; n++)
3887 {
3888 wc = wparam[n]; wparam[n] = L'\0';
3889 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3890 {
3891 wparam[n] = wc;
3892 return (wcsdup (wparam + n));
3893 }
3894 wparam[n] = wc;
3895 }
3896 break;
3897
3898 case RP_LONG_RIGHT: /* remove longest match at end */
3899 for (n = 0; n <= wstrlen; n++)
3900 {
3901 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3902 {
3903 wc = wparam[n]; wparam[n] = L'\0';
3904 ret = wcsdup (wparam);
3905 wparam[n] = wc;
3906 return (ret);
3907 }
3908 }
3909 break;
3910
3911 case RP_SHORT_RIGHT: /* remove shortest match at end */
3912 for (n = wstrlen; n >= 0; n--)
3913 {
3914 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3915 {
3916 wc = wparam[n]; wparam[n] = L'\0';
3917 ret = wcsdup (wparam);
3918 wparam[n] = wc;
3919 return (ret);
3920 }
3921 }
3922 break;
3923 }
3924
3925 return (wparam); /* no match, return original string */
3926 }
3927 #endif /* HANDLE_MULTIBYTE */
3928
3929 static char *
3930 remove_pattern (param, pattern, op)
3931 char *param, *pattern;
3932 int op;
3933 {
3934 char *xret;
3935
3936 if (param == NULL)
3937 return (param);
3938 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
3939 return (savestring (param));
3940
3941 #if defined (HANDLE_MULTIBYTE)
3942 if (MB_CUR_MAX > 1)
3943 {
3944 wchar_t *ret, *oret;
3945 size_t n;
3946 wchar_t *wparam, *wpattern;
3947 mbstate_t ps;
3948
3949 n = xdupmbstowcs (&wpattern, NULL, pattern);
3950 if (n == (size_t)-1)
3951 {
3952 xret = remove_upattern (param, pattern, op);
3953 return ((xret == param) ? savestring (param) : xret);
3954 }
3955 n = xdupmbstowcs (&wparam, NULL, param);
3956 if (n == (size_t)-1)
3957 {
3958 free (wpattern);
3959 xret = remove_upattern (param, pattern, op);
3960 return ((xret == param) ? savestring (param) : xret);
3961 }
3962 oret = ret = remove_wpattern (wparam, n, wpattern, op);
3963 /* Don't bother to convert wparam back to multibyte string if nothing
3964 matched; just return copy of original string */
3965 if (ret == wparam)
3966 {
3967 free (wparam);
3968 free (wpattern);
3969 return (savestring (param));
3970 }
3971
3972 free (wparam);
3973 free (wpattern);
3974
3975 n = strlen (param);
3976 xret = (char *)xmalloc (n + 1);
3977 memset (&ps, '\0', sizeof (mbstate_t));
3978 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
3979 xret[n] = '\0'; /* just to make sure */
3980 free (oret);
3981 return xret;
3982 }
3983 else
3984 #endif
3985 {
3986 xret = remove_upattern (param, pattern, op);
3987 return ((xret == param) ? savestring (param) : xret);
3988 }
3989 }
3990
3991 /* Match PAT anywhere in STRING and return the match boundaries.
3992 This returns 1 in case of a successful match, 0 otherwise. SP
3993 and EP are pointers into the string where the match begins and
3994 ends, respectively. MTYPE controls what kind of match is attempted.
3995 MATCH_BEG and MATCH_END anchor the match at the beginning and end
3996 of the string, respectively. The longest match is returned. */
3997 static int
3998 match_upattern (string, pat, mtype, sp, ep)
3999 char *string, *pat;
4000 int mtype;
4001 char **sp, **ep;
4002 {
4003 int c, len, mlen;
4004 register char *p, *p1, *npat;
4005 char *end;
4006 int n1;
4007
4008 /* If the pattern doesn't match anywhere in the string, go ahead and
4009 short-circuit right away. A minor optimization, saves a bunch of
4010 unnecessary calls to strmatch (up to N calls for a string of N
4011 characters) if the match is unsuccessful. To preserve the semantics
4012 of the substring matches below, we make sure that the pattern has
4013 `*' as first and last character, making a new pattern if necessary. */
4014 /* XXX - check this later if I ever implement `**' with special meaning,
4015 since this will potentially result in `**' at the beginning or end */
4016 len = STRLEN (pat);
4017 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
4018 {
4019 p = npat = (char *)xmalloc (len + 3);
4020 p1 = pat;
4021 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
4022 *p++ = '*';
4023 while (*p1)
4024 *p++ = *p1++;
4025 if (p1[-1] != '*' || p[-2] == '\\')
4026 *p++ = '*';
4027 *p = '\0';
4028 }
4029 else
4030 npat = pat;
4031 c = strmatch (npat, string, FNMATCH_EXTFLAG);
4032 if (npat != pat)
4033 free (npat);
4034 if (c == FNM_NOMATCH)
4035 return (0);
4036
4037 len = STRLEN (string);
4038 end = string + len;
4039
4040 mlen = umatchlen (pat, len);
4041
4042 switch (mtype)
4043 {
4044 case MATCH_ANY:
4045 for (p = string; p <= end; p++)
4046 {
4047 if (match_pattern_char (pat, p))
4048 {
4049 #if 0
4050 for (p1 = end; p1 >= p; p1--)
4051 #else
4052 p1 = (mlen == -1) ? end : p + mlen;
4053 /* p1 - p = length of portion of string to be considered
4054 p = current position in string
4055 mlen = number of characters consumed by match (-1 for entire string)
4056 end = end of string
4057 we want to break immediately if the potential match len
4058 is greater than the number of characters remaining in the
4059 string
4060 */
4061 if (p1 > end)
4062 break;
4063 for ( ; p1 >= p; p1--)
4064 #endif
4065 {
4066 c = *p1; *p1 = '\0';
4067 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4068 {
4069 *p1 = c;
4070 *sp = p;
4071 *ep = p1;
4072 return 1;
4073 }
4074 *p1 = c;
4075 #if 1
4076 /* If MLEN != -1, we have a fixed length pattern. */
4077 if (mlen != -1)
4078 break;
4079 #endif
4080 }
4081 }
4082 }
4083
4084 return (0);
4085
4086 case MATCH_BEG:
4087 if (match_pattern_char (pat, string) == 0)
4088 return (0);
4089
4090 #if 0
4091 for (p = end; p >= string; p--)
4092 #else
4093 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
4094 #endif
4095 {
4096 c = *p; *p = '\0';
4097 if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0)
4098 {
4099 *p = c;
4100 *sp = string;
4101 *ep = p;
4102 return 1;
4103 }
4104 *p = c;
4105 #if 1
4106 /* If MLEN != -1, we have a fixed length pattern. */
4107 if (mlen != -1)
4108 break;
4109 #endif
4110 }
4111
4112 return (0);
4113
4114 case MATCH_END:
4115 #if 0
4116 for (p = string; p <= end; p++)
4117 #else
4118 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
4119 #endif
4120 {
4121 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4122 {
4123 *sp = p;
4124 *ep = end;
4125 return 1;
4126 }
4127 #if 1
4128 /* If MLEN != -1, we have a fixed length pattern. */
4129 if (mlen != -1)
4130 break;
4131 #endif
4132 }
4133
4134 return (0);
4135 }
4136
4137 return (0);
4138 }
4139
4140 #if defined (HANDLE_MULTIBYTE)
4141 /* Match WPAT anywhere in WSTRING and return the match boundaries.
4142 This returns 1 in case of a successful match, 0 otherwise. Wide
4143 character version. */
4144 static int
4145 match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4146 wchar_t *wstring;
4147 char **indices;
4148 size_t wstrlen;
4149 wchar_t *wpat;
4150 int mtype;
4151 char **sp, **ep;
4152 {
4153 wchar_t wc, *wp, *nwpat, *wp1;
4154 size_t len;
4155 int mlen;
4156 int n, n1, n2, simple;
4157
4158 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4159 #if defined (EXTENDED_GLOB)
4160 if (extended_glob)
4161 simple |= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
4162 #endif
4163
4164 /* If the pattern doesn't match anywhere in the string, go ahead and
4165 short-circuit right away. A minor optimization, saves a bunch of
4166 unnecessary calls to strmatch (up to N calls for a string of N
4167 characters) if the match is unsuccessful. To preserve the semantics
4168 of the substring matches below, we make sure that the pattern has
4169 `*' as first and last character, making a new pattern if necessary. */
4170 len = wcslen (wpat);
4171 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
4172 {
4173 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
4174 wp1 = wpat;
4175 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
4176 *wp++ = L'*';
4177 while (*wp1 != L'\0')
4178 *wp++ = *wp1++;
4179 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4180 *wp++ = L'*';
4181 *wp = '\0';
4182 }
4183 else
4184 nwpat = wpat;
4185 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG);
4186 if (nwpat != wpat)
4187 free (nwpat);
4188 if (len == FNM_NOMATCH)
4189 return (0);
4190
4191 mlen = wmatchlen (wpat, wstrlen);
4192
4193 /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
4194 switch (mtype)
4195 {
4196 case MATCH_ANY:
4197 for (n = 0; n <= wstrlen; n++)
4198 {
4199 #if 1
4200 n2 = simple ? (*wpat == wstring[n]) : match_pattern_wchar (wpat, wstring + n);
4201 #else
4202 n2 = match_pattern_wchar (wpat, wstring + n);
4203 #endif
4204 if (n2)
4205 {
4206 #if 0
4207 for (n1 = wstrlen; n1 >= n; n1--)
4208 #else
4209 n1 = (mlen == -1) ? wstrlen : n + mlen;
4210 if (n1 > wstrlen)
4211 break;
4212
4213 for ( ; n1 >= n; n1--)
4214 #endif
4215 {
4216 wc = wstring[n1]; wstring[n1] = L'\0';
4217 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4218 {
4219 wstring[n1] = wc;
4220 *sp = indices[n];
4221 *ep = indices[n1];
4222 return 1;
4223 }
4224 wstring[n1] = wc;
4225 #if 1
4226 /* If MLEN != -1, we have a fixed length pattern. */
4227 if (mlen != -1)
4228 break;
4229 #endif
4230 }
4231 }
4232 }
4233
4234 return (0);
4235
4236 case MATCH_BEG:
4237 if (match_pattern_wchar (wpat, wstring) == 0)
4238 return (0);
4239
4240 #if 0
4241 for (n = wstrlen; n >= 0; n--)
4242 #else
4243 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
4244 #endif
4245 {
4246 wc = wstring[n]; wstring[n] = L'\0';
4247 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0)
4248 {
4249 wstring[n] = wc;
4250 *sp = indices[0];
4251 *ep = indices[n];
4252 return 1;
4253 }
4254 wstring[n] = wc;
4255 #if 1
4256 /* If MLEN != -1, we have a fixed length pattern. */
4257 if (mlen != -1)
4258 break;
4259 #endif
4260 }
4261
4262 return (0);
4263
4264 case MATCH_END:
4265 #if 0
4266 for (n = 0; n <= wstrlen; n++)
4267 #else
4268 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
4269 #endif
4270 {
4271 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4272 {
4273 *sp = indices[n];
4274 *ep = indices[wstrlen];
4275 return 1;
4276 }
4277 #if 1
4278 /* If MLEN != -1, we have a fixed length pattern. */
4279 if (mlen != -1)
4280 break;
4281 #endif
4282 }
4283
4284 return (0);
4285 }
4286
4287 return (0);
4288 }
4289 #endif /* HANDLE_MULTIBYTE */
4290
4291 static int
4292 match_pattern (string, pat, mtype, sp, ep)
4293 char *string, *pat;
4294 int mtype;
4295 char **sp, **ep;
4296 {
4297 #if defined (HANDLE_MULTIBYTE)
4298 int ret;
4299 size_t n;
4300 wchar_t *wstring, *wpat;
4301 char **indices;
4302 size_t slen, plen, mslen, mplen;
4303 #endif
4304
4305 if (string == 0 || *string == 0 || pat == 0 || *pat == 0)
4306 return (0);
4307
4308 #if defined (HANDLE_MULTIBYTE)
4309 if (MB_CUR_MAX > 1)
4310 {
4311 #if 0
4312 slen = STRLEN (string);
4313 mslen = MBSLEN (string);
4314 plen = STRLEN (pat);
4315 mplen = MBSLEN (pat);
4316 if (slen == mslen && plen == mplen)
4317 #else
4318 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
4319 #endif
4320 return (match_upattern (string, pat, mtype, sp, ep));
4321
4322 n = xdupmbstowcs (&wpat, NULL, pat);
4323 if (n == (size_t)-1)
4324 return (match_upattern (string, pat, mtype, sp, ep));
4325 n = xdupmbstowcs (&wstring, &indices, string);
4326 if (n == (size_t)-1)
4327 {
4328 free (wpat);
4329 return (match_upattern (string, pat, mtype, sp, ep));
4330 }
4331 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4332
4333 free (wpat);
4334 free (wstring);
4335 free (indices);
4336
4337 return (ret);
4338 }
4339 else
4340 #endif
4341 return (match_upattern (string, pat, mtype, sp, ep));
4342 }
4343
4344 static int
4345 getpatspec (c, value)
4346 int c;
4347 char *value;
4348 {
4349 if (c == '#')
4350 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4351 else /* c == '%' */
4352 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4353 }
4354
4355 /* Posix.2 says that the WORD should be run through tilde expansion,
4356 parameter expansion, command substitution and arithmetic expansion.
4357 This leaves the result quoted, so quote_string_for_globbing () has
4358 to be called to fix it up for strmatch (). If QUOTED is non-zero,
4359 it means that the entire expression was enclosed in double quotes.
4360 This means that quoting characters in the pattern do not make any
4361 special pattern characters quoted. For example, the `*' in the
4362 following retains its special meaning: "${foo#'*'}". */
4363 static char *
4364 getpattern (value, quoted, expandpat)
4365 char *value;
4366 int quoted, expandpat;
4367 {
4368 char *pat, *tword;
4369 WORD_LIST *l;
4370 #if 0
4371 int i;
4372 #endif
4373 /* There is a problem here: how to handle single or double quotes in the
4374 pattern string when the whole expression is between double quotes?
4375 POSIX.2 says that enclosing double quotes do not cause the pattern to
4376 be quoted, but does that leave us a problem with @ and array[@] and their
4377 expansions inside a pattern? */
4378 #if 0
4379 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4380 {
4381 i = 0;
4382 pat = string_extract_double_quoted (tword, &i, 1);
4383 free (tword);
4384 tword = pat;
4385 }
4386 #endif
4387
4388 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4389 word splitting. */
4390 l = *value ? expand_string_for_rhs (value,
4391 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
4392 (int *)NULL, (int *)NULL)
4393 : (WORD_LIST *)0;
4394 pat = string_list (l);
4395 dispose_words (l);
4396 if (pat)
4397 {
4398 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4399 free (pat);
4400 pat = tword;
4401 }
4402 return (pat);
4403 }
4404
4405 #if 0
4406 /* Handle removing a pattern from a string as a result of ${name%[%]value}
4407 or ${name#[#]value}. */
4408 static char *
4409 variable_remove_pattern (value, pattern, patspec, quoted)
4410 char *value, *pattern;
4411 int patspec, quoted;
4412 {
4413 char *tword;
4414
4415 tword = remove_pattern (value, pattern, patspec);
4416
4417 return (tword);
4418 }
4419 #endif
4420
4421 static char *
4422 list_remove_pattern (list, pattern, patspec, itype, quoted)
4423 WORD_LIST *list;
4424 char *pattern;
4425 int patspec, itype, quoted;
4426 {
4427 WORD_LIST *new, *l;
4428 WORD_DESC *w;
4429 char *tword;
4430
4431 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4432 {
4433 tword = remove_pattern (l->word->word, pattern, patspec);
4434 w = alloc_word_desc ();
4435 w->word = tword ? tword : savestring ("");
4436 new = make_word_list (w, new);
4437 }
4438
4439 l = REVERSE_LIST (new, WORD_LIST *);
4440 tword = string_list_pos_params (itype, l, quoted);
4441 dispose_words (l);
4442
4443 return (tword);
4444 }
4445
4446 static char *
4447 parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4448 int itype;
4449 char *pattern;
4450 int patspec, quoted;
4451 {
4452 char *ret;
4453 WORD_LIST *list;
4454
4455 list = list_rest_of_args ();
4456 if (list == 0)
4457 return ((char *)NULL);
4458 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4459 dispose_words (list);
4460 return (ret);
4461 }
4462
4463 #if defined (ARRAY_VARS)
4464 static char *
4465 array_remove_pattern (var, pattern, patspec, varname, quoted)
4466 SHELL_VAR *var;
4467 char *pattern;
4468 int patspec;
4469 char *varname; /* so we can figure out how it's indexed */
4470 int quoted;
4471 {
4472 ARRAY *a;
4473 HASH_TABLE *h;
4474 int itype;
4475 char *ret;
4476 WORD_LIST *list;
4477 SHELL_VAR *v;
4478
4479 /* compute itype from varname here */
4480 v = array_variable_part (varname, &ret, 0);
4481 itype = ret[0];
4482
4483 a = (v && array_p (v)) ? array_cell (v) : 0;
4484 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4485
4486 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4487 if (list == 0)
4488 return ((char *)NULL);
4489 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4490 dispose_words (list);
4491
4492 return ret;
4493 }
4494 #endif /* ARRAY_VARS */
4495
4496 static char *
4497 parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4498 char *varname, *value;
4499 int ind;
4500 char *patstr;
4501 int rtype, quoted, flags;
4502 {
4503 int vtype, patspec, starsub;
4504 char *temp1, *val, *pattern;
4505 SHELL_VAR *v;
4506
4507 if (value == 0)
4508 return ((char *)NULL);
4509
4510 this_command_name = varname;
4511
4512 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4513 if (vtype == -1)
4514 return ((char *)NULL);
4515
4516 starsub = vtype & VT_STARSUB;
4517 vtype &= ~VT_STARSUB;
4518
4519 patspec = getpatspec (rtype, patstr);
4520 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4521 patstr++;
4522
4523 /* Need to pass getpattern newly-allocated memory in case of expansion --
4524 the expansion code will free the passed string on an error. */
4525 temp1 = savestring (patstr);
4526 pattern = getpattern (temp1, quoted, 1);
4527 free (temp1);
4528
4529 temp1 = (char *)NULL; /* shut up gcc */
4530 switch (vtype)
4531 {
4532 case VT_VARIABLE:
4533 case VT_ARRAYMEMBER:
4534 temp1 = remove_pattern (val, pattern, patspec);
4535 if (vtype == VT_VARIABLE)
4536 FREE (val);
4537 if (temp1)
4538 {
4539 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
4540 ? quote_string (temp1)
4541 : quote_escapes (temp1);
4542 free (temp1);
4543 temp1 = val;
4544 }
4545 break;
4546 #if defined (ARRAY_VARS)
4547 case VT_ARRAYVAR:
4548 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
4549 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4550 {
4551 val = quote_escapes (temp1);
4552 free (temp1);
4553 temp1 = val;
4554 }
4555 break;
4556 #endif
4557 case VT_POSPARMS:
4558 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
4559 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4560 {
4561 val = quote_escapes (temp1);
4562 free (temp1);
4563 temp1 = val;
4564 }
4565 break;
4566 }
4567
4568 FREE (pattern);
4569 return temp1;
4570 }
4571
4572 /*******************************************
4573 * *
4574 * Functions to expand WORD_DESCs *
4575 * *
4576 *******************************************/
4577
4578 /* Expand WORD, performing word splitting on the result. This does
4579 parameter expansion, command substitution, arithmetic expansion,
4580 word splitting, and quote removal. */
4581
4582 WORD_LIST *
4583 expand_word (word, quoted)
4584 WORD_DESC *word;
4585 int quoted;
4586 {
4587 WORD_LIST *result, *tresult;
4588
4589 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4590 result = word_list_split (tresult);
4591 dispose_words (tresult);
4592 return (result ? dequote_list (result) : result);
4593 }
4594
4595 /* Expand WORD, but do not perform word splitting on the result. This
4596 does parameter expansion, command substitution, arithmetic expansion,
4597 and quote removal. */
4598 WORD_LIST *
4599 expand_word_unsplit (word, quoted)
4600 WORD_DESC *word;
4601 int quoted;
4602 {
4603 WORD_LIST *result;
4604
4605 expand_no_split_dollar_star = 1;
4606 #if defined (HANDLE_MULTIBYTE)
4607 if (ifs_firstc[0] == 0)
4608 #else
4609 if (ifs_firstc == 0)
4610 #endif
4611 word->flags |= W_NOSPLIT;
4612 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4613 expand_no_split_dollar_star = 0;
4614
4615 return (result ? dequote_list (result) : result);
4616 }
4617
4618 /* Perform shell expansions on WORD, but do not perform word splitting or
4619 quote removal on the result. Virtually identical to expand_word_unsplit;
4620 could be combined if implementations don't diverge. */
4621 WORD_LIST *
4622 expand_word_leave_quoted (word, quoted)
4623 WORD_DESC *word;
4624 int quoted;
4625 {
4626 WORD_LIST *result;
4627
4628 expand_no_split_dollar_star = 1;
4629 #if defined (HANDLE_MULTIBYTE)
4630 if (ifs_firstc[0] == 0)
4631 #else
4632 if (ifs_firstc == 0)
4633 #endif
4634 word->flags |= W_NOSPLIT;
4635 word->flags |= W_NOSPLIT2;
4636 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4637 expand_no_split_dollar_star = 0;
4638
4639 return result;
4640 }
4641
4642 #if defined (PROCESS_SUBSTITUTION)
4643
4644 /*****************************************************************/
4645 /* */
4646 /* Hacking Process Substitution */
4647 /* */
4648 /*****************************************************************/
4649
4650 #if !defined (HAVE_DEV_FD)
4651 /* Named pipes must be removed explicitly with `unlink'. This keeps a list
4652 of FIFOs the shell has open. unlink_fifo_list will walk the list and
4653 unlink all of them. add_fifo_list adds the name of an open FIFO to the
4654 list. NFIFO is a count of the number of FIFOs in the list. */
4655 #define FIFO_INCR 20
4656
4657 struct temp_fifo {
4658 char *file;
4659 pid_t proc;
4660 };
4661
4662 static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
4663 static int nfifo;
4664 static int fifo_list_size;
4665
4666 char *
4667 copy_fifo_list (sizep)
4668 int *sizep;
4669 {
4670 if (sizep)
4671 *sizep = 0;
4672 return (char *)NULL;
4673 }
4674
4675 static void
4676 add_fifo_list (pathname)
4677 char *pathname;
4678 {
4679 if (nfifo >= fifo_list_size - 1)
4680 {
4681 fifo_list_size += FIFO_INCR;
4682 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
4683 fifo_list_size * sizeof (struct temp_fifo));
4684 }
4685
4686 fifo_list[nfifo].file = savestring (pathname);
4687 nfifo++;
4688 }
4689
4690 void
4691 unlink_fifo (i)
4692 int i;
4693 {
4694 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4695 {
4696 unlink (fifo_list[i].file);
4697 free (fifo_list[i].file);
4698 fifo_list[i].file = (char *)NULL;
4699 fifo_list[i].proc = -1;
4700 }
4701 }
4702
4703 void
4704 unlink_fifo_list ()
4705 {
4706 int saved, i, j;
4707
4708 if (nfifo == 0)
4709 return;
4710
4711 for (i = saved = 0; i < nfifo; i++)
4712 {
4713 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4714 {
4715 unlink (fifo_list[i].file);
4716 free (fifo_list[i].file);
4717 fifo_list[i].file = (char *)NULL;
4718 fifo_list[i].proc = -1;
4719 }
4720 else
4721 saved++;
4722 }
4723
4724 /* If we didn't remove some of the FIFOs, compact the list. */
4725 if (saved)
4726 {
4727 for (i = j = 0; i < nfifo; i++)
4728 if (fifo_list[i].file)
4729 {
4730 fifo_list[j].file = fifo_list[i].file;
4731 fifo_list[j].proc = fifo_list[i].proc;
4732 j++;
4733 }
4734 nfifo = j;
4735 }
4736 else
4737 nfifo = 0;
4738 }
4739
4740 /* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
4741 from some point in the past, and close all open FIFOs in fifo_list
4742 that are not marked as active in LIST. If LIST is NULL, close
4743 everything in fifo_list. LSIZE is the number of elements in LIST, in
4744 case it's larger than fifo_list_size (size of fifo_list). */
4745 void
4746 close_new_fifos (list, lsize)
4747 char *list;
4748 int lsize;
4749 {
4750 int i;
4751
4752 if (list == 0)
4753 {
4754 unlink_fifo_list ();
4755 return;
4756 }
4757
4758 for (i = 0; i < lsize; i++)
4759 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
4760 unlink_fifo (i);
4761
4762 for (i = lsize; i < fifo_list_size; i++)
4763 unlink_fifo (i);
4764 }
4765
4766 int
4767 fifos_pending ()
4768 {
4769 return nfifo;
4770 }
4771
4772 int
4773 num_fifos ()
4774 {
4775 return nfifo;
4776 }
4777
4778 static char *
4779 make_named_pipe ()
4780 {
4781 char *tname;
4782
4783 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
4784 if (mkfifo (tname, 0600) < 0)
4785 {
4786 free (tname);
4787 return ((char *)NULL);
4788 }
4789
4790 add_fifo_list (tname);
4791 return (tname);
4792 }
4793
4794 #else /* HAVE_DEV_FD */
4795
4796 /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
4797 has open to children. NFDS is a count of the number of bits currently
4798 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
4799 of open files. */
4800 static char *dev_fd_list = (char *)NULL;
4801 static int nfds;
4802 static int totfds; /* The highest possible number of open files. */
4803
4804 char *
4805 copy_fifo_list (sizep)
4806 int *sizep;
4807 {
4808 char *ret;
4809
4810 if (nfds == 0 || totfds == 0)
4811 {
4812 if (sizep)
4813 *sizep = 0;
4814 return (char *)NULL;
4815 }
4816
4817 if (sizep)
4818 *sizep = totfds;
4819 ret = (char *)xmalloc (totfds);
4820 return (memcpy (ret, dev_fd_list, totfds));
4821 }
4822
4823 static void
4824 add_fifo_list (fd)
4825 int fd;
4826 {
4827 if (dev_fd_list == 0 || fd >= totfds)
4828 {
4829 int ofds;
4830
4831 ofds = totfds;
4832 totfds = getdtablesize ();
4833 if (totfds < 0 || totfds > 256)
4834 totfds = 256;
4835 if (fd >= totfds)
4836 totfds = fd + 2;
4837
4838 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
4839 memset (dev_fd_list + ofds, '\0', totfds - ofds);
4840 }
4841
4842 dev_fd_list[fd] = 1;
4843 nfds++;
4844 }
4845
4846 int
4847 fifos_pending ()
4848 {
4849 return 0; /* used for cleanup; not needed with /dev/fd */
4850 }
4851
4852 int
4853 num_fifos ()
4854 {
4855 return nfds;
4856 }
4857
4858 void
4859 unlink_fifo (fd)
4860 int fd;
4861 {
4862 if (dev_fd_list[fd])
4863 {
4864 close (fd);
4865 dev_fd_list[fd] = 0;
4866 nfds--;
4867 }
4868 }
4869
4870 void
4871 unlink_fifo_list ()
4872 {
4873 register int i;
4874
4875 if (nfds == 0)
4876 return;
4877
4878 for (i = 0; nfds && i < totfds; i++)
4879 unlink_fifo (i);
4880
4881 nfds = 0;
4882 }
4883
4884 /* Take LIST, which is a snapshot copy of dev_fd_list from some point in
4885 the past, and close all open fds in dev_fd_list that are not marked
4886 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
4887 LSIZE is the number of elements in LIST, in case it's larger than
4888 totfds (size of dev_fd_list). */
4889 void
4890 close_new_fifos (list, lsize)
4891 char *list;
4892 int lsize;
4893 {
4894 int i;
4895
4896 if (list == 0)
4897 {
4898 unlink_fifo_list ();
4899 return;
4900 }
4901
4902 for (i = 0; i < lsize; i++)
4903 if (list[i] == 0 && i < totfds && dev_fd_list[i])
4904 unlink_fifo (i);
4905
4906 for (i = lsize; i < totfds; i++)
4907 unlink_fifo (i);
4908 }
4909
4910 #if defined (NOTDEF)
4911 print_dev_fd_list ()
4912 {
4913 register int i;
4914
4915 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
4916 fflush (stderr);
4917
4918 for (i = 0; i < totfds; i++)
4919 {
4920 if (dev_fd_list[i])
4921 fprintf (stderr, " %d", i);
4922 }
4923 fprintf (stderr, "\n");
4924 }
4925 #endif /* NOTDEF */
4926
4927 static char *
4928 make_dev_fd_filename (fd)
4929 int fd;
4930 {
4931 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
4932
4933 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
4934
4935 strcpy (ret, DEV_FD_PREFIX);
4936 p = inttostr (fd, intbuf, sizeof (intbuf));
4937 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
4938
4939 add_fifo_list (fd);
4940 return (ret);
4941 }
4942
4943 #endif /* HAVE_DEV_FD */
4944
4945 /* Return a filename that will open a connection to the process defined by
4946 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
4947 a filename in /dev/fd corresponding to a descriptor that is one of the
4948 ends of the pipe. If not defined, we use named pipes on systems that have
4949 them. Systems without /dev/fd and named pipes are out of luck.
4950
4951 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
4952 use the read end of the pipe and dup that file descriptor to fd 0 in
4953 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
4954 writing or use the write end of the pipe in the child, and dup that
4955 file descriptor to fd 1 in the child. The parent does the opposite. */
4956
4957 static char *
4958 process_substitute (string, open_for_read_in_child)
4959 char *string;
4960 int open_for_read_in_child;
4961 {
4962 char *pathname;
4963 int fd, result;
4964 pid_t old_pid, pid;
4965 #if defined (HAVE_DEV_FD)
4966 int parent_pipe_fd, child_pipe_fd;
4967 int fildes[2];
4968 #endif /* HAVE_DEV_FD */
4969 #if defined (JOB_CONTROL)
4970 pid_t old_pipeline_pgrp;
4971 #endif
4972
4973 if (!string || !*string || wordexp_only)
4974 return ((char *)NULL);
4975
4976 #if !defined (HAVE_DEV_FD)
4977 pathname = make_named_pipe ();
4978 #else /* HAVE_DEV_FD */
4979 if (pipe (fildes) < 0)
4980 {
4981 sys_error (_("cannot make pipe for process substitution"));
4982 return ((char *)NULL);
4983 }
4984 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
4985 the pipe in the parent, otherwise the read end. */
4986 parent_pipe_fd = fildes[open_for_read_in_child];
4987 child_pipe_fd = fildes[1 - open_for_read_in_child];
4988 /* Move the parent end of the pipe to some high file descriptor, to
4989 avoid clashes with FDs used by the script. */
4990 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
4991
4992 pathname = make_dev_fd_filename (parent_pipe_fd);
4993 #endif /* HAVE_DEV_FD */
4994
4995 if (pathname == 0)
4996 {
4997 sys_error (_("cannot make pipe for process substitution"));
4998 return ((char *)NULL);
4999 }
5000
5001 old_pid = last_made_pid;
5002
5003 #if defined (JOB_CONTROL)
5004 old_pipeline_pgrp = pipeline_pgrp;
5005 pipeline_pgrp = shell_pgrp;
5006 save_pipeline (1);
5007 #endif /* JOB_CONTROL */
5008
5009 pid = make_child ((char *)NULL, 1);
5010 if (pid == 0)
5011 {
5012 reset_terminating_signals (); /* XXX */
5013 free_pushed_string_input ();
5014 /* Cancel traps, in trap.c. */
5015 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
5016 setup_async_signals ();
5017 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
5018 }
5019
5020 #if defined (JOB_CONTROL)
5021 set_sigchld_handler ();
5022 stop_making_children ();
5023 /* XXX - should we only do this in the parent? (as in command subst) */
5024 pipeline_pgrp = old_pipeline_pgrp;
5025 #endif /* JOB_CONTROL */
5026
5027 if (pid < 0)
5028 {
5029 sys_error (_("cannot make child for process substitution"));
5030 free (pathname);
5031 #if defined (HAVE_DEV_FD)
5032 close (parent_pipe_fd);
5033 close (child_pipe_fd);
5034 #endif /* HAVE_DEV_FD */
5035 return ((char *)NULL);
5036 }
5037
5038 if (pid > 0)
5039 {
5040 #if defined (JOB_CONTROL)
5041 restore_pipeline (1);
5042 #endif
5043
5044 #if !defined (HAVE_DEV_FD)
5045 fifo_list[nfifo-1].proc = pid;
5046 #endif
5047
5048 last_made_pid = old_pid;
5049
5050 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5051 close_pgrp_pipe ();
5052 #endif /* JOB_CONTROL && PGRP_PIPE */
5053
5054 #if defined (HAVE_DEV_FD)
5055 close (child_pipe_fd);
5056 #endif /* HAVE_DEV_FD */
5057
5058 return (pathname);
5059 }
5060
5061 set_sigint_handler ();
5062
5063 #if defined (JOB_CONTROL)
5064 set_job_control (0);
5065 #endif /* JOB_CONTROL */
5066
5067 #if !defined (HAVE_DEV_FD)
5068 /* Open the named pipe in the child. */
5069 fd = open (pathname, open_for_read_in_child ? O_RDONLY|O_NONBLOCK : O_WRONLY);
5070 if (fd < 0)
5071 {
5072 /* Two separate strings for ease of translation. */
5073 if (open_for_read_in_child)
5074 sys_error (_("cannot open named pipe %s for reading"), pathname);
5075 else
5076 sys_error (_("cannot open named pipe %s for writing"), pathname);
5077
5078 exit (127);
5079 }
5080 if (open_for_read_in_child)
5081 {
5082 if (sh_unset_nodelay_mode (fd) < 0)
5083 {
5084 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
5085 exit (127);
5086 }
5087 }
5088 #else /* HAVE_DEV_FD */
5089 fd = child_pipe_fd;
5090 #endif /* HAVE_DEV_FD */
5091
5092 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5093 {
5094 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
5095 open_for_read_in_child ? 0 : 1);
5096 exit (127);
5097 }
5098
5099 if (fd != (open_for_read_in_child ? 0 : 1))
5100 close (fd);
5101
5102 /* Need to close any files that this process has open to pipes inherited
5103 from its parent. */
5104 if (current_fds_to_close)
5105 {
5106 close_fd_bitmap (current_fds_to_close);
5107 current_fds_to_close = (struct fd_bitmap *)NULL;
5108 }
5109
5110 #if defined (HAVE_DEV_FD)
5111 /* Make sure we close the parent's end of the pipe and clear the slot
5112 in the fd list so it is not closed later, if reallocated by, for
5113 instance, pipe(2). */
5114 close (parent_pipe_fd);
5115 dev_fd_list[parent_pipe_fd] = 0;
5116 #endif /* HAVE_DEV_FD */
5117
5118 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
5119
5120 #if !defined (HAVE_DEV_FD)
5121 /* Make sure we close the named pipe in the child before we exit. */
5122 close (open_for_read_in_child ? 0 : 1);
5123 #endif /* !HAVE_DEV_FD */
5124
5125 exit (result);
5126 /*NOTREACHED*/
5127 }
5128 #endif /* PROCESS_SUBSTITUTION */
5129
5130 /***********************************/
5131 /* */
5132 /* Command Substitution */
5133 /* */
5134 /***********************************/
5135
5136 static char *
5137 read_comsub (fd, quoted, rflag)
5138 int fd, quoted;
5139 int *rflag;
5140 {
5141 char *istring, buf[128], *bufp, *s;
5142 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
5143 ssize_t bufn;
5144
5145 istring = (char *)NULL;
5146 istring_index = istring_size = bufn = tflag = 0;
5147
5148 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5149 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
5150
5151 /* Read the output of the command through the pipe. This may need to be
5152 changed to understand multibyte characters in the future. */
5153 while (1)
5154 {
5155 if (fd < 0)
5156 break;
5157 if (--bufn <= 0)
5158 {
5159 bufn = zread (fd, buf, sizeof (buf));
5160 if (bufn <= 0)
5161 break;
5162 bufp = buf;
5163 }
5164 c = *bufp++;
5165
5166 if (c == 0)
5167 {
5168 #if 0
5169 internal_warning ("read_comsub: ignored null byte in input");
5170 #endif
5171 continue;
5172 }
5173
5174 /* Add the character to ISTRING, possibly after resizing it. */
5175 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5176
5177 /* This is essentially quote_string inline */
5178 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5179 istring[istring_index++] = CTLESC;
5180 /* Escape CTLESC and CTLNUL in the output to protect those characters
5181 from the rest of the word expansions (word splitting and globbing.)
5182 This is essentially quote_escapes inline. */
5183 else if (skip_ctlesc == 0 && c == CTLESC)
5184 {
5185 tflag |= W_HASCTLESC;
5186 istring[istring_index++] = CTLESC;
5187 }
5188 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
5189 istring[istring_index++] = CTLESC;
5190
5191 istring[istring_index++] = c;
5192
5193 #if 0
5194 #if defined (__CYGWIN__)
5195 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5196 {
5197 istring_index--;
5198 istring[istring_index - 1] = '\n';
5199 }
5200 #endif
5201 #endif
5202 }
5203
5204 if (istring)
5205 istring[istring_index] = '\0';
5206
5207 /* If we read no output, just return now and save ourselves some
5208 trouble. */
5209 if (istring_index == 0)
5210 {
5211 FREE (istring);
5212 if (rflag)
5213 *rflag = tflag;
5214 return (char *)NULL;
5215 }
5216
5217 /* Strip trailing newlines from the output of the command. */
5218 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5219 {
5220 while (istring_index > 0)
5221 {
5222 if (istring[istring_index - 1] == '\n')
5223 {
5224 --istring_index;
5225
5226 /* If the newline was quoted, remove the quoting char. */
5227 if (istring[istring_index - 1] == CTLESC)
5228 --istring_index;
5229 }
5230 else
5231 break;
5232 }
5233 istring[istring_index] = '\0';
5234 }
5235 else
5236 strip_trailing (istring, istring_index - 1, 1);
5237
5238 if (rflag)
5239 *rflag = tflag;
5240 return istring;
5241 }
5242
5243 /* Perform command substitution on STRING. This returns a WORD_DESC * with the
5244 contained string possibly quoted. */
5245 WORD_DESC *
5246 command_substitute (string, quoted)
5247 char *string;
5248 int quoted;
5249 {
5250 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
5251 char *istring;
5252 int result, fildes[2], function_value, pflags, rc, tflag;
5253 WORD_DESC *ret;
5254
5255 istring = (char *)NULL;
5256
5257 /* Don't fork () if there is no need to. In the case of no command to
5258 run, just return NULL. */
5259 if (!string || !*string || (string[0] == '\n' && !string[1]))
5260 return ((WORD_DESC *)NULL);
5261
5262 if (wordexp_only && read_but_dont_execute)
5263 {
5264 last_command_exit_value = EX_WEXPCOMSUB;
5265 jump_to_top_level (EXITPROG);
5266 }
5267
5268 /* We're making the assumption here that the command substitution will
5269 eventually run a command from the file system. Since we'll run
5270 maybe_make_export_env in this subshell before executing that command,
5271 the parent shell and any other shells it starts will have to remake
5272 the environment. If we make it before we fork, other shells won't
5273 have to. Don't bother if we have any temporary variable assignments,
5274 though, because the export environment will be remade after this
5275 command completes anyway, but do it if all the words to be expanded
5276 are variable assignments. */
5277 if (subst_assign_varlist == 0 || garglist == 0)
5278 maybe_make_export_env (); /* XXX */
5279
5280 /* Flags to pass to parse_and_execute() */
5281 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
5282
5283 /* Pipe the output of executing STRING into the current shell. */
5284 if (pipe (fildes) < 0)
5285 {
5286 sys_error (_("cannot make pipe for command substitution"));
5287 goto error_exit;
5288 }
5289
5290 old_pid = last_made_pid;
5291 #if defined (JOB_CONTROL)
5292 old_pipeline_pgrp = pipeline_pgrp;
5293 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
5294 if ((subshell_environment & SUBSHELL_PIPE) == 0)
5295 pipeline_pgrp = shell_pgrp;
5296 cleanup_the_pipeline ();
5297 #endif /* JOB_CONTROL */
5298
5299 old_async_pid = last_asynchronous_pid;
5300 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
5301 last_asynchronous_pid = old_async_pid;
5302
5303 if (pid == 0)
5304 {
5305 /* Reset the signal handlers in the child, but don't free the
5306 trap strings. Set a flag noting that we have to free the
5307 trap strings if we run trap to change a signal disposition. */
5308 reset_signal_handlers ();
5309 subshell_environment |= SUBSHELL_RESETTRAP;
5310 }
5311
5312 #if defined (JOB_CONTROL)
5313 /* XXX DO THIS ONLY IN PARENT ? XXX */
5314 set_sigchld_handler ();
5315 stop_making_children ();
5316 if (pid != 0)
5317 pipeline_pgrp = old_pipeline_pgrp;
5318 #else
5319 stop_making_children ();
5320 #endif /* JOB_CONTROL */
5321
5322 if (pid < 0)
5323 {
5324 sys_error (_("cannot make child for command substitution"));
5325 error_exit:
5326
5327 FREE (istring);
5328 close (fildes[0]);
5329 close (fildes[1]);
5330 return ((WORD_DESC *)NULL);
5331 }
5332
5333 if (pid == 0)
5334 {
5335 set_sigint_handler (); /* XXX */
5336
5337 free_pushed_string_input ();
5338
5339 if (dup2 (fildes[1], 1) < 0)
5340 {
5341 sys_error (_("command_substitute: cannot duplicate pipe as fd 1"));
5342 exit (EXECUTION_FAILURE);
5343 }
5344
5345 /* If standard output is closed in the parent shell
5346 (such as after `exec >&-'), file descriptor 1 will be
5347 the lowest available file descriptor, and end up in
5348 fildes[0]. This can happen for stdin and stderr as well,
5349 but stdout is more important -- it will cause no output
5350 to be generated from this command. */
5351 if ((fildes[1] != fileno (stdin)) &&
5352 (fildes[1] != fileno (stdout)) &&
5353 (fildes[1] != fileno (stderr)))
5354 close (fildes[1]);
5355
5356 if ((fildes[0] != fileno (stdin)) &&
5357 (fildes[0] != fileno (stdout)) &&
5358 (fildes[0] != fileno (stderr)))
5359 close (fildes[0]);
5360
5361 #ifdef __CYGWIN__
5362 /* Let stdio know the fd may have changed from text to binary mode, and
5363 make sure to preserve stdout line buffering. */
5364 freopen (NULL, "w", stdout);
5365 sh_setlinebuf (stdout);
5366 #endif /* __CYGWIN__ */
5367
5368 /* The currently executing shell is not interactive. */
5369 interactive = 0;
5370
5371 /* This is a subshell environment. */
5372 subshell_environment |= SUBSHELL_COMSUB;
5373
5374 /* When not in POSIX mode, command substitution does not inherit
5375 the -e flag. */
5376 if (posixly_correct == 0)
5377 exit_immediately_on_error = 0;
5378
5379 remove_quoted_escapes (string);
5380
5381 startup_state = 2; /* see if we can avoid a fork */
5382 /* Give command substitution a place to jump back to on failure,
5383 so we don't go back up to main (). */
5384 result = setjmp (top_level);
5385
5386 /* If we're running a command substitution inside a shell function,
5387 trap `return' so we don't return from the function in the subshell
5388 and go off to never-never land. */
5389 if (result == 0 && return_catch_flag)
5390 function_value = setjmp (return_catch);
5391 else
5392 function_value = 0;
5393
5394 if (result == ERREXIT)
5395 rc = last_command_exit_value;
5396 else if (result == EXITPROG)
5397 rc = last_command_exit_value;
5398 else if (result)
5399 rc = EXECUTION_FAILURE;
5400 else if (function_value)
5401 rc = return_catch_value;
5402 else
5403 {
5404 subshell_level++;
5405 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
5406 subshell_level--;
5407 }
5408
5409 last_command_exit_value = rc;
5410 rc = run_exit_trap ();
5411 #if defined (PROCESS_SUBSTITUTION)
5412 unlink_fifo_list ();
5413 #endif
5414 exit (rc);
5415 }
5416 else
5417 {
5418 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5419 close_pgrp_pipe ();
5420 #endif /* JOB_CONTROL && PGRP_PIPE */
5421
5422 close (fildes[1]);
5423
5424 tflag = 0;
5425 istring = read_comsub (fildes[0], quoted, &tflag);
5426
5427 close (fildes[0]);
5428
5429 current_command_subst_pid = pid;
5430 last_command_exit_value = wait_for (pid);
5431 last_command_subst_pid = pid;
5432 last_made_pid = old_pid;
5433
5434 #if defined (JOB_CONTROL)
5435 /* If last_command_exit_value > 128, then the substituted command
5436 was terminated by a signal. If that signal was SIGINT, then send
5437 SIGINT to ourselves. This will break out of loops, for instance. */
5438 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
5439 kill (getpid (), SIGINT);
5440
5441 /* wait_for gives the terminal back to shell_pgrp. If some other
5442 process group should have it, give it away to that group here.
5443 pipeline_pgrp is non-zero only while we are constructing a
5444 pipline, so what we are concerned about is whether or not that
5445 pipeline was started in the background. A pipeline started in
5446 the background should never get the tty back here. */
5447 if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0)
5448 give_terminal_to (pipeline_pgrp, 0);
5449 #endif /* JOB_CONTROL */
5450
5451 ret = alloc_word_desc ();
5452 ret->word = istring;
5453 ret->flags = tflag;
5454
5455 return ret;
5456 }
5457 }
5458
5459 /********************************************************
5460 * *
5461 * Utility functions for parameter expansion *
5462 * *
5463 ********************************************************/
5464
5465 #if defined (ARRAY_VARS)
5466
5467 static arrayind_t
5468 array_length_reference (s)
5469 char *s;
5470 {
5471 int len;
5472 arrayind_t ind;
5473 char *akey;
5474 char *t, c;
5475 ARRAY *array;
5476 HASH_TABLE *h;
5477 SHELL_VAR *var;
5478
5479 var = array_variable_part (s, &t, &len);
5480
5481 /* If unbound variables should generate an error, report one and return
5482 failure. */
5483 if ((var == 0 || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
5484 {
5485 c = *--t;
5486 *t = '\0';
5487 last_command_exit_value = EXECUTION_FAILURE;
5488 err_unboundvar (s);
5489 *t = c;
5490 return (-1);
5491 }
5492 else if (var == 0)
5493 return 0;
5494
5495 /* We support a couple of expansions for variables that are not arrays.
5496 We'll return the length of the value for v[0], and 1 for v[@] or
5497 v[*]. Return 0 for everything else. */
5498
5499 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
5500 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
5501
5502 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
5503 {
5504 if (assoc_p (var))
5505 return (h ? assoc_num_elements (h) : 0);
5506 else if (array_p (var))
5507 return (array ? array_num_elements (array) : 0);
5508 else
5509 return (var_isset (var) ? 1 : 0);
5510 }
5511
5512 if (assoc_p (var))
5513 {
5514 t[len - 1] = '\0';
5515 akey = expand_assignment_string_to_string (t, 0); /* [ */
5516 t[len - 1] = ']';
5517 if (akey == 0 || *akey == 0)
5518 {
5519 err_badarraysub (t);
5520 return (-1);
5521 }
5522 t = assoc_reference (assoc_cell (var), akey);
5523 }
5524 else
5525 {
5526 ind = array_expand_index (t, len);
5527 if (ind < 0)
5528 {
5529 err_badarraysub (t);
5530 return (-1);
5531 }
5532 if (array_p (var))
5533 t = array_reference (array, ind);
5534 else
5535 t = (ind == 0) ? value_cell (var) : (char *)NULL;
5536 }
5537
5538 len = MB_STRLEN (t);
5539 return (len);
5540 }
5541 #endif /* ARRAY_VARS */
5542
5543 static int
5544 valid_brace_expansion_word (name, var_is_special)
5545 char *name;
5546 int var_is_special;
5547 {
5548 if (DIGIT (*name) && all_digits (name))
5549 return 1;
5550 else if (var_is_special)
5551 return 1;
5552 #if defined (ARRAY_VARS)
5553 else if (valid_array_reference (name))
5554 return 1;
5555 #endif /* ARRAY_VARS */
5556 else if (legal_identifier (name))
5557 return 1;
5558 else
5559 return 0;
5560 }
5561
5562 static int
5563 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
5564 char *name;
5565 int quoted;
5566 int *quoted_dollar_atp, *contains_dollar_at;
5567 {
5568 char *temp1;
5569
5570 if (name == 0)
5571 {
5572 if (quoted_dollar_atp)
5573 *quoted_dollar_atp = 0;
5574 if (contains_dollar_at)
5575 *contains_dollar_at = 0;
5576 return 0;
5577 }
5578
5579 /* check for $@ and $* */
5580 if (name[0] == '@' && name[1] == 0)
5581 {
5582 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5583 *quoted_dollar_atp = 1;
5584 if (contains_dollar_at)
5585 *contains_dollar_at = 1;
5586 return 1;
5587 }
5588 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
5589 {
5590 if (contains_dollar_at)
5591 *contains_dollar_at = 1;
5592 return 1;
5593 }
5594
5595 /* Now check for ${array[@]} and ${array[*]} */
5596 #if defined (ARRAY_VARS)
5597 else if (valid_array_reference (name))
5598 {
5599 temp1 = mbschr (name, '[');
5600 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
5601 {
5602 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5603 *quoted_dollar_atp = 1;
5604 if (contains_dollar_at)
5605 *contains_dollar_at = 1;
5606 return 1;
5607 } /* [ */
5608 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
5609 which should result in separate words even when IFS is unset. */
5610 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
5611 {
5612 if (contains_dollar_at)
5613 *contains_dollar_at = 1;
5614 return 1;
5615 }
5616 }
5617 #endif
5618 return 0;
5619 }
5620
5621 /* Parameter expand NAME, and return a new string which is the expansion,
5622 or NULL if there was no expansion.
5623 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
5624 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
5625 NAME was found inside of a double-quoted expression. */
5626 static WORD_DESC *
5627 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
5628 char *name;
5629 int var_is_special, quoted, pflags;
5630 arrayind_t *indp;
5631 {
5632 WORD_DESC *ret;
5633 char *temp, *tt;
5634 intmax_t arg_index;
5635 SHELL_VAR *var;
5636 int atype, rflags;
5637 arrayind_t ind;
5638
5639 ret = 0;
5640 temp = 0;
5641 rflags = 0;
5642
5643 if (indp)
5644 *indp = INTMAX_MIN;
5645
5646 /* Handle multiple digit arguments, as in ${11}. */
5647 if (legal_number (name, &arg_index))
5648 {
5649 tt = get_dollar_var_value (arg_index);
5650 if (tt)
5651 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5652 ? quote_string (tt)
5653 : quote_escapes (tt);
5654 else
5655 temp = (char *)NULL;
5656 FREE (tt);
5657 }
5658 else if (var_is_special) /* ${@} */
5659 {
5660 int sindex;
5661 tt = (char *)xmalloc (2 + strlen (name));
5662 tt[sindex = 0] = '$';
5663 strcpy (tt + 1, name);
5664
5665 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
5666 (int *)NULL, (int *)NULL, pflags);
5667 free (tt);
5668 }
5669 #if defined (ARRAY_VARS)
5670 else if (valid_array_reference (name))
5671 {
5672 temp = array_value (name, quoted, 0, &atype, &ind);
5673 if (atype == 0 && temp)
5674 {
5675 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5676 ? quote_string (temp)
5677 : quote_escapes (temp);
5678 rflags |= W_ARRAYIND;
5679 if (indp)
5680 *indp = ind;
5681 }
5682 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5683 rflags |= W_HASQUOTEDNULL;
5684 }
5685 #endif
5686 else if (var = find_variable (name))
5687 {
5688 if (var_isset (var) && invisible_p (var) == 0)
5689 {
5690 #if defined (ARRAY_VARS)
5691 if (assoc_p (var))
5692 temp = assoc_reference (assoc_cell (var), "0");
5693 else if (array_p (var))
5694 temp = array_reference (array_cell (var), 0);
5695 else
5696 temp = value_cell (var);
5697 #else
5698 temp = value_cell (var);
5699 #endif
5700
5701 if (temp)
5702 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5703 ? quote_string (temp)
5704 : quote_escapes (temp);
5705 }
5706 else
5707 temp = (char *)NULL;
5708 }
5709 else
5710 temp = (char *)NULL;
5711
5712 if (ret == 0)
5713 {
5714 ret = alloc_word_desc ();
5715 ret->word = temp;
5716 ret->flags |= rflags;
5717 }
5718 return ret;
5719 }
5720
5721 /* Expand an indirect reference to a variable: ${!NAME} expands to the
5722 value of the variable whose name is the value of NAME. */
5723 static WORD_DESC *
5724 parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
5725 char *name;
5726 int var_is_special, quoted;
5727 int *quoted_dollar_atp, *contains_dollar_at;
5728 {
5729 char *temp, *t;
5730 WORD_DESC *w;
5731
5732 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
5733 t = w->word;
5734 /* Have to dequote here if necessary */
5735 if (t)
5736 {
5737 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
5738 ? dequote_string (t)
5739 : dequote_escapes (t);
5740 free (t);
5741 t = temp;
5742 }
5743 dispose_word_desc (w);
5744
5745 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
5746 if (t == 0)
5747 return (WORD_DESC *)NULL;
5748
5749 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
5750 free (t);
5751
5752 return w;
5753 }
5754
5755 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
5756 depending on the value of C, the separating character. C can be one of
5757 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
5758 between double quotes. */
5759 static WORD_DESC *
5760 parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat)
5761 char *name, *value;
5762 int c, quoted, *qdollaratp, *hasdollarat;
5763 {
5764 WORD_DESC *w;
5765 WORD_LIST *l;
5766 char *t, *t1, *temp;
5767 int hasdol;
5768
5769 /* If the entire expression is between double quotes, we want to treat
5770 the value as a double-quoted string, with the exception that we strip
5771 embedded unescaped double quotes (for sh backwards compatibility). */
5772 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
5773 {
5774 hasdol = 0;
5775 temp = string_extract_double_quoted (value, &hasdol, 1);
5776 }
5777 else
5778 temp = value;
5779
5780 w = alloc_word_desc ();
5781 hasdol = 0;
5782 /* XXX was 0 not quoted */
5783 l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL)
5784 : (WORD_LIST *)0;
5785 if (hasdollarat)
5786 *hasdollarat = hasdol || (l && l->next);
5787 if (temp != value)
5788 free (temp);
5789 if (l)
5790 {
5791 /* The expansion of TEMP returned something. We need to treat things
5792 slightly differently if HASDOL is non-zero. If we have "$@", the
5793 individual words have already been quoted. We need to turn them
5794 into a string with the words separated by the first character of
5795 $IFS without any additional quoting, so string_list_dollar_at won't
5796 do the right thing. We use string_list_dollar_star instead. */
5797 temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l);
5798
5799 /* If l->next is not null, we know that TEMP contained "$@", since that
5800 is the only expansion that creates more than one word. */
5801 if (qdollaratp && ((hasdol && quoted) || l->next))
5802 *qdollaratp = 1;
5803 dispose_words (l);
5804 }
5805 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol)
5806 {
5807 /* The brace expansion occurred between double quotes and there was
5808 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
5809 it does not expand to anything. In this case, we want to return
5810 a quoted empty string. */
5811 temp = make_quoted_char ('\0');
5812 w->flags |= W_HASQUOTEDNULL;
5813 }
5814 else
5815 temp = (char *)NULL;
5816
5817 if (c == '-' || c == '+')
5818 {
5819 w->word = temp;
5820 return w;
5821 }
5822
5823 /* c == '=' */
5824 t = temp ? savestring (temp) : savestring ("");
5825 t1 = dequote_string (t);
5826 free (t);
5827 #if defined (ARRAY_VARS)
5828 if (valid_array_reference (name))
5829 assign_array_element (name, t1, 0);
5830 else
5831 #endif /* ARRAY_VARS */
5832 bind_variable (name, t1, 0);
5833
5834 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
5835 free (temp);
5836
5837 w->word = t1;
5838 return w;
5839 }
5840
5841 /* Deal with the right hand side of a ${name:?value} expansion in the case
5842 that NAME is null or not set. If VALUE is non-null it is expanded and
5843 used as the error message to print, otherwise a standard message is
5844 printed. */
5845 static void
5846 parameter_brace_expand_error (name, value)
5847 char *name, *value;
5848 {
5849 WORD_LIST *l;
5850 char *temp;
5851
5852 if (value && *value)
5853 {
5854 l = expand_string (value, 0);
5855 temp = string_list (l);
5856 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
5857 FREE (temp);
5858 dispose_words (l);
5859 }
5860 else
5861 report_error (_("%s: parameter null or not set"), name);
5862
5863 /* Free the data we have allocated during this expansion, since we
5864 are about to longjmp out. */
5865 free (name);
5866 FREE (value);
5867 }
5868
5869 /* Return 1 if NAME is something for which parameter_brace_expand_length is
5870 OK to do. */
5871 static int
5872 valid_length_expression (name)
5873 char *name;
5874 {
5875 return (name[1] == '\0' || /* ${#} */
5876 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
5877 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
5878 #if defined (ARRAY_VARS)
5879 valid_array_reference (name + 1) || /* ${#a[7]} */
5880 #endif
5881 legal_identifier (name + 1)); /* ${#PS1} */
5882 }
5883
5884 /* Handle the parameter brace expansion that requires us to return the
5885 length of a parameter. */
5886 static intmax_t
5887 parameter_brace_expand_length (name)
5888 char *name;
5889 {
5890 char *t, *newname;
5891 intmax_t number, arg_index;
5892 WORD_LIST *list;
5893 #if defined (ARRAY_VARS)
5894 SHELL_VAR *var;
5895 #endif
5896
5897 if (name[1] == '\0') /* ${#} */
5898 number = number_of_args ();
5899 else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */
5900 number = number_of_args ();
5901 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
5902 {
5903 /* Take the lengths of some of the shell's special parameters. */
5904 switch (name[1])
5905 {
5906 case '-':
5907 t = which_set_flags ();
5908 break;
5909 case '?':
5910 t = itos (last_command_exit_value);
5911 break;
5912 case '$':
5913 t = itos (dollar_dollar_pid);
5914 break;
5915 case '!':
5916 if (last_asynchronous_pid == NO_PID)
5917 t = (char *)NULL; /* XXX - error if set -u set? */
5918 else
5919 t = itos (last_asynchronous_pid);
5920 break;
5921 case '#':
5922 t = itos (number_of_args ());
5923 break;
5924 }
5925 number = STRLEN (t);
5926 FREE (t);
5927 }
5928 #if defined (ARRAY_VARS)
5929 else if (valid_array_reference (name + 1))
5930 number = array_length_reference (name + 1);
5931 #endif /* ARRAY_VARS */
5932 else
5933 {
5934 number = 0;
5935
5936 if (legal_number (name + 1, &arg_index)) /* ${#1} */
5937 {
5938 t = get_dollar_var_value (arg_index);
5939 if (t == 0 && unbound_vars_is_error)
5940 return INTMAX_MIN;
5941 number = MB_STRLEN (t);
5942 FREE (t);
5943 }
5944 #if defined (ARRAY_VARS)
5945 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
5946 {
5947 if (assoc_p (var))
5948 t = assoc_reference (assoc_cell (var), "0");
5949 else
5950 t = array_reference (array_cell (var), 0);
5951 if (t == 0 && unbound_vars_is_error)
5952 return INTMAX_MIN;
5953 number = MB_STRLEN (t);
5954 }
5955 #endif
5956 else /* ${#PS1} */
5957 {
5958 newname = savestring (name);
5959 newname[0] = '$';
5960 list = expand_string (newname, Q_DOUBLE_QUOTES);
5961 t = list ? string_list (list) : (char *)NULL;
5962 free (newname);
5963 if (list)
5964 dispose_words (list);
5965
5966 number = t ? MB_STRLEN (t) : 0;
5967 FREE (t);
5968 }
5969 }
5970
5971 return (number);
5972 }
5973
5974 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
5975 so we do some ad-hoc parsing of an arithmetic expression to find
5976 the first DELIM, instead of using strchr(3). Two rules:
5977 1. If the substring contains a `(', read until closing `)'.
5978 2. If the substring contains a `?', read past one `:' for each `?'.
5979 */
5980
5981 static char *
5982 skiparith (substr, delim)
5983 char *substr;
5984 int delim;
5985 {
5986 size_t sublen;
5987 int skipcol, pcount, i;
5988 DECLARE_MBSTATE;
5989
5990 sublen = strlen (substr);
5991 i = skipcol = pcount = 0;
5992 while (substr[i])
5993 {
5994 /* Balance parens */
5995 if (substr[i] == LPAREN)
5996 {
5997 pcount++;
5998 i++;
5999 continue;
6000 }
6001 if (substr[i] == RPAREN && pcount)
6002 {
6003 pcount--;
6004 i++;
6005 continue;
6006 }
6007 if (pcount)
6008 {
6009 ADVANCE_CHAR (substr, sublen, i);
6010 continue;
6011 }
6012
6013 /* Skip one `:' for each `?' */
6014 if (substr[i] == ':' && skipcol)
6015 {
6016 skipcol--;
6017 i++;
6018 continue;
6019 }
6020 if (substr[i] == delim)
6021 break;
6022 if (substr[i] == '?')
6023 {
6024 skipcol++;
6025 i++;
6026 continue;
6027 }
6028 ADVANCE_CHAR (substr, sublen, i);
6029 }
6030
6031 return (substr + i);
6032 }
6033
6034 /* Verify and limit the start and end of the desired substring. If
6035 VTYPE == 0, a regular shell variable is being used; if it is 1,
6036 then the positional parameters are being used; if it is 2, then
6037 VALUE is really a pointer to an array variable that should be used.
6038 Return value is 1 if both values were OK, 0 if there was a problem
6039 with an invalid expression, or -1 if the values were out of range. */
6040 static int
6041 verify_substring_values (v, value, substr, vtype, e1p, e2p)
6042 SHELL_VAR *v;
6043 char *value, *substr;
6044 int vtype;
6045 intmax_t *e1p, *e2p;
6046 {
6047 char *t, *temp1, *temp2;
6048 arrayind_t len;
6049 int expok;
6050 #if defined (ARRAY_VARS)
6051 ARRAY *a;
6052 HASH_TABLE *h;
6053 #endif
6054
6055 /* duplicate behavior of strchr(3) */
6056 t = skiparith (substr, ':');
6057 if (*t && *t == ':')
6058 *t = '\0';
6059 else
6060 t = (char *)0;
6061
6062 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
6063 *e1p = evalexp (temp1, &expok);
6064 free (temp1);
6065 if (expok == 0)
6066 return (0);
6067
6068 len = -1; /* paranoia */
6069 switch (vtype)
6070 {
6071 case VT_VARIABLE:
6072 case VT_ARRAYMEMBER:
6073 len = MB_STRLEN (value);
6074 break;
6075 case VT_POSPARMS:
6076 len = number_of_args () + 1;
6077 if (*e1p == 0)
6078 len++; /* add one arg if counting from $0 */
6079 break;
6080 #if defined (ARRAY_VARS)
6081 case VT_ARRAYVAR:
6082 /* For arrays, the first value deals with array indices. Negative
6083 offsets count from one past the array's maximum index. Associative
6084 arrays treat the number of elements as the maximum index. */
6085 if (assoc_p (v))
6086 {
6087 h = assoc_cell (v);
6088 len = assoc_num_elements (h) + (*e1p < 0);
6089 }
6090 else
6091 {
6092 a = (ARRAY *)value;
6093 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
6094 }
6095 break;
6096 #endif
6097 }
6098
6099 if (len == -1) /* paranoia */
6100 return -1;
6101
6102 if (*e1p < 0) /* negative offsets count from end */
6103 *e1p += len;
6104
6105 if (*e1p > len || *e1p < 0)
6106 return (-1);
6107
6108 #if defined (ARRAY_VARS)
6109 /* For arrays, the second offset deals with the number of elements. */
6110 if (vtype == VT_ARRAYVAR)
6111 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
6112 #endif
6113
6114 if (t)
6115 {
6116 t++;
6117 temp2 = savestring (t);
6118 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
6119 free (temp2);
6120 t[-1] = ':';
6121 *e2p = evalexp (temp1, &expok);
6122 free (temp1);
6123 if (expok == 0)
6124 return (0);
6125 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
6126 {
6127 internal_error (_("%s: substring expression < 0"), t);
6128 return (0);
6129 }
6130 #if defined (ARRAY_VARS)
6131 /* In order to deal with sparse arrays, push the intelligence about how
6132 to deal with the number of elements desired down to the array-
6133 specific functions. */
6134 if (vtype != VT_ARRAYVAR)
6135 #endif
6136 {
6137 if (*e2p < 0)
6138 {
6139 *e2p += len;
6140 if (*e2p < 0 || *e2p < *e1p)
6141 {
6142 internal_error (_("%s: substring expression < 0"), t);
6143 return (0);
6144 }
6145 }
6146 else
6147 *e2p += *e1p; /* want E2 chars starting at E1 */
6148 if (*e2p > len)
6149 *e2p = len;
6150 }
6151 }
6152 else
6153 *e2p = len;
6154
6155 return (1);
6156 }
6157
6158 /* Return the type of variable specified by VARNAME (simple variable,
6159 positional param, or array variable). Also return the value specified
6160 by VARNAME (value of a variable or a reference to an array element).
6161 QUOTED is the standard description of quoting state, using Q_* defines.
6162 FLAGS is currently a set of flags to pass to array_value. If IND is
6163 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
6164 passed to array_value so the array index is not computed again.
6165 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
6166 characters in the value are quoted with CTLESC and takes appropriate
6167 steps. For convenience, *VALP is set to the dequoted VALUE. */
6168 static int
6169 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
6170 char *varname, *value;
6171 arrayind_t ind;
6172 int quoted, flags;
6173 SHELL_VAR **varp;
6174 char **valp;
6175 {
6176 int vtype;
6177 char *temp;
6178 #if defined (ARRAY_VARS)
6179 SHELL_VAR *v;
6180 #endif
6181 arrayind_t lind;
6182
6183 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
6184 vtype = (varname[0] == '@' || varname[0] == '*') && varname[1] == '\0';
6185 if (vtype == VT_POSPARMS && varname[0] == '*')
6186 vtype |= VT_STARSUB;
6187 *varp = (SHELL_VAR *)NULL;
6188
6189 #if defined (ARRAY_VARS)
6190 if (valid_array_reference (varname))
6191 {
6192 v = array_variable_part (varname, &temp, (int *)0);
6193 /* If we want to signal array_value to use an already-computed index,
6194 set LIND to that index */
6195 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
6196 if (v && (array_p (v) || assoc_p (v)))
6197 { /* [ */
6198 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
6199 {
6200 /* Callers have to differentiate betwen indexed and associative */
6201 vtype = VT_ARRAYVAR;
6202 if (temp[0] == '*')
6203 vtype |= VT_STARSUB;
6204 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
6205 }
6206 else
6207 {
6208 vtype = VT_ARRAYMEMBER;
6209 *valp = array_value (varname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6210 }
6211 *varp = v;
6212 }
6213 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
6214 {
6215 vtype = VT_VARIABLE;
6216 *varp = v;
6217 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6218 *valp = dequote_string (value);
6219 else
6220 *valp = dequote_escapes (value);
6221 }
6222 else
6223 {
6224 vtype = VT_ARRAYMEMBER;
6225 *varp = v;
6226 *valp = array_value (varname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6227 }
6228 }
6229 else if ((v = find_variable (varname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
6230 {
6231 vtype = VT_ARRAYMEMBER;
6232 *varp = v;
6233 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
6234 }
6235 else
6236 #endif
6237 {
6238 if (value && vtype == VT_VARIABLE)
6239 {
6240 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6241 *valp = dequote_string (value);
6242 else
6243 *valp = dequote_escapes (value);
6244 }
6245 else
6246 *valp = value;
6247 }
6248
6249 return vtype;
6250 }
6251
6252 /******************************************************/
6253 /* */
6254 /* Functions to extract substrings of variable values */
6255 /* */
6256 /******************************************************/
6257
6258 #if defined (HANDLE_MULTIBYTE)
6259 /* Character-oriented rather than strictly byte-oriented substrings. S and
6260 E, rather being strict indices into STRING, indicate character (possibly
6261 multibyte character) positions that require calculation.
6262 Used by the ${param:offset[:length]} expansion. */
6263 static char *
6264 mb_substring (string, s, e)
6265 char *string;
6266 int s, e;
6267 {
6268 char *tt;
6269 int start, stop, i, slen;
6270 DECLARE_MBSTATE;
6271
6272 start = 0;
6273 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
6274 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
6275
6276 i = s;
6277 while (string[start] && i--)
6278 ADVANCE_CHAR (string, slen, start);
6279 stop = start;
6280 i = e - s;
6281 while (string[stop] && i--)
6282 ADVANCE_CHAR (string, slen, stop);
6283 tt = substring (string, start, stop);
6284 return tt;
6285 }
6286 #endif
6287
6288 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
6289 is `@', use the positional parameters; otherwise, use the value of
6290 VARNAME. If VARNAME is an array variable, use the array elements. */
6291
6292 static char *
6293 parameter_brace_substring (varname, value, ind, substr, quoted, flags)
6294 char *varname, *value;
6295 int ind;
6296 char *substr;
6297 int quoted, flags;
6298 {
6299 intmax_t e1, e2;
6300 int vtype, r, starsub;
6301 char *temp, *val, *tt, *oname;
6302 SHELL_VAR *v;
6303
6304 if (value == 0)
6305 return ((char *)NULL);
6306
6307 oname = this_command_name;
6308 this_command_name = varname;
6309
6310 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6311 if (vtype == -1)
6312 {
6313 this_command_name = oname;
6314 return ((char *)NULL);
6315 }
6316
6317 starsub = vtype & VT_STARSUB;
6318 vtype &= ~VT_STARSUB;
6319
6320 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
6321 this_command_name = oname;
6322 if (r <= 0)
6323 return ((r == 0) ? &expand_param_error : (char *)NULL);
6324
6325 switch (vtype)
6326 {
6327 case VT_VARIABLE:
6328 case VT_ARRAYMEMBER:
6329 #if defined (HANDLE_MULTIBYTE)
6330 if (MB_CUR_MAX > 1)
6331 tt = mb_substring (val, e1, e2);
6332 else
6333 #endif
6334 tt = substring (val, e1, e2);
6335
6336 if (vtype == VT_VARIABLE)
6337 FREE (val);
6338 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6339 temp = quote_string (tt);
6340 else
6341 temp = tt ? quote_escapes (tt) : (char *)NULL;
6342 FREE (tt);
6343 break;
6344 case VT_POSPARMS:
6345 tt = pos_params (varname, e1, e2, quoted);
6346 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
6347 {
6348 temp = tt ? quote_escapes (tt) : (char *)NULL;
6349 FREE (tt);
6350 }
6351 else
6352 temp = tt;
6353 break;
6354 #if defined (ARRAY_VARS)
6355 case VT_ARRAYVAR:
6356 if (assoc_p (v))
6357 /* we convert to list and take first e2 elements starting at e1th
6358 element -- officially undefined for now */
6359 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
6360 else
6361 /* We want E2 to be the number of elements desired (arrays can be sparse,
6362 so verify_substring_values just returns the numbers specified and we
6363 rely on array_subrange to understand how to deal with them). */
6364 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
6365 /* array_subrange now calls array_quote_escapes as appropriate, so the
6366 caller no longer needs to. */
6367 break;
6368 #endif
6369 default:
6370 temp = (char *)NULL;
6371 }
6372
6373 return temp;
6374 }
6375
6376 /****************************************************************/
6377 /* */
6378 /* Functions to perform pattern substitution on variable values */
6379 /* */
6380 /****************************************************************/
6381
6382 static int
6383 shouldexp_replacement (s)
6384 char *s;
6385 {
6386 register char *p;
6387
6388 for (p = s; p && *p; p++)
6389 {
6390 if (*p == '\\')
6391 p++;
6392 else if (*p == '&')
6393 return 1;
6394 }
6395 return 0;
6396 }
6397
6398 char *
6399 pat_subst (string, pat, rep, mflags)
6400 char *string, *pat, *rep;
6401 int mflags;
6402 {
6403 char *ret, *s, *e, *str, *rstr, *mstr;
6404 int rsize, rptr, l, replen, mtype, rxpand, rslen, mlen;
6405
6406 if (string == 0)
6407 return (savestring (""));
6408
6409 mtype = mflags & MATCH_TYPEMASK;
6410
6411 #if 0 /* bash-4.2 ? */
6412 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
6413 #else
6414 rxpand = 0;
6415 #endif
6416
6417 /* Special cases:
6418 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
6419 * with REP and return the result.
6420 * 2. A null pattern with mtype == MATCH_END means to append REP to
6421 * STRING and return the result.
6422 * These don't understand or process `&' in the replacement string.
6423 */
6424 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
6425 {
6426 replen = STRLEN (rep);
6427 l = STRLEN (string);
6428 ret = (char *)xmalloc (replen + l + 2);
6429 if (replen == 0)
6430 strcpy (ret, string);
6431 else if (mtype == MATCH_BEG)
6432 {
6433 strcpy (ret, rep);
6434 strcpy (ret + replen, string);
6435 }
6436 else
6437 {
6438 strcpy (ret, string);
6439 strcpy (ret + l, rep);
6440 }
6441 return (ret);
6442 }
6443
6444 ret = (char *)xmalloc (rsize = 64);
6445 ret[0] = '\0';
6446
6447 for (replen = STRLEN (rep), rptr = 0, str = string;;)
6448 {
6449 if (match_pattern (str, pat, mtype, &s, &e) == 0)
6450 break;
6451 l = s - str;
6452
6453 if (rxpand)
6454 {
6455 int x;
6456 mlen = e - s;
6457 mstr = xmalloc (mlen + 1);
6458 for (x = 0; x < mlen; x++)
6459 mstr[x] = s[x];
6460 mstr[mlen] = '\0';
6461 rstr = strcreplace (rep, '&', mstr, 0);
6462 rslen = strlen (rstr);
6463 }
6464 else
6465 {
6466 rstr = rep;
6467 rslen = replen;
6468 }
6469
6470 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
6471
6472 /* OK, now copy the leading unmatched portion of the string (from
6473 str to s) to ret starting at rptr (the current offset). Then copy
6474 the replacement string at ret + rptr + (s - str). Increment
6475 rptr (if necessary) and str and go on. */
6476 if (l)
6477 {
6478 strncpy (ret + rptr, str, l);
6479 rptr += l;
6480 }
6481 if (replen)
6482 {
6483 strncpy (ret + rptr, rstr, rslen);
6484 rptr += rslen;
6485 }
6486 str = e; /* e == end of match */
6487
6488 if (rstr != rep)
6489 free (rstr);
6490
6491 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
6492 break;
6493
6494 if (s == e)
6495 {
6496 /* On a zero-length match, make sure we copy one character, since
6497 we increment one character to avoid infinite recursion. */
6498 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
6499 ret[rptr++] = *str++;
6500 e++; /* avoid infinite recursion on zero-length match */
6501 }
6502 }
6503
6504 /* Now copy the unmatched portion of the input string */
6505 if (str && *str)
6506 {
6507 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
6508 strcpy (ret + rptr, str);
6509 }
6510 else
6511 ret[rptr] = '\0';
6512
6513 return ret;
6514 }
6515
6516 /* Do pattern match and replacement on the positional parameters. */
6517 static char *
6518 pos_params_pat_subst (string, pat, rep, mflags)
6519 char *string, *pat, *rep;
6520 int mflags;
6521 {
6522 WORD_LIST *save, *params;
6523 WORD_DESC *w;
6524 char *ret;
6525 int pchar, qflags;
6526
6527 save = params = list_rest_of_args ();
6528 if (save == 0)
6529 return ((char *)NULL);
6530
6531 for ( ; params; params = params->next)
6532 {
6533 ret = pat_subst (params->word->word, pat, rep, mflags);
6534 w = alloc_word_desc ();
6535 w->word = ret ? ret : savestring ("");
6536 dispose_word (params->word);
6537 params->word = w;
6538 }
6539
6540 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6541 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6542
6543 #if 0
6544 if ((mflags & (MATCH_QUOTED|MATCH_STARSUB)) == (MATCH_QUOTED|MATCH_STARSUB))
6545 ret = string_list_dollar_star (quote_list (save));
6546 else if ((mflags & MATCH_STARSUB) == MATCH_STARSUB)
6547 ret = string_list_dollar_star (save);
6548 else if ((mflags & MATCH_QUOTED) == MATCH_QUOTED)
6549 ret = string_list_dollar_at (save, qflags);
6550 else
6551 ret = string_list_dollar_star (save);
6552 #else
6553 ret = string_list_pos_params (pchar, save, qflags);
6554 #endif
6555
6556 dispose_words (save);
6557
6558 return (ret);
6559 }
6560
6561 /* Perform pattern substitution on VALUE, which is the expansion of
6562 VARNAME. PATSUB is an expression supplying the pattern to match
6563 and the string to substitute. QUOTED is a flags word containing
6564 the type of quoting currently in effect. */
6565 static char *
6566 parameter_brace_patsub (varname, value, ind, patsub, quoted, flags)
6567 char *varname, *value;
6568 int ind;
6569 char *patsub;
6570 int quoted, flags;
6571 {
6572 int vtype, mflags, starsub, delim;
6573 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
6574 SHELL_VAR *v;
6575
6576 if (value == 0)
6577 return ((char *)NULL);
6578
6579 this_command_name = varname;
6580
6581 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6582 if (vtype == -1)
6583 return ((char *)NULL);
6584
6585 starsub = vtype & VT_STARSUB;
6586 vtype &= ~VT_STARSUB;
6587
6588 mflags = 0;
6589 if (patsub && *patsub == '/')
6590 {
6591 mflags |= MATCH_GLOBREP;
6592 patsub++;
6593 }
6594
6595 /* Malloc this because expand_string_if_necessary or one of the expansion
6596 functions in its call chain may free it on a substitution error. */
6597 lpatsub = savestring (patsub);
6598
6599 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6600 mflags |= MATCH_QUOTED;
6601
6602 if (starsub)
6603 mflags |= MATCH_STARSUB;
6604
6605 /* If the pattern starts with a `/', make sure we skip over it when looking
6606 for the replacement delimiter. */
6607 #if 0
6608 if (rep = quoted_strchr ((*patsub == '/') ? lpatsub+1 : lpatsub, '/', ST_BACKSL))
6609 *rep++ = '\0';
6610 else
6611 rep = (char *)NULL;
6612 #else
6613 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
6614 if (lpatsub[delim] == '/')
6615 {
6616 lpatsub[delim] = 0;
6617 rep = lpatsub + delim + 1;
6618 }
6619 else
6620 rep = (char *)NULL;
6621 #endif
6622
6623 if (rep && *rep == '\0')
6624 rep = (char *)NULL;
6625
6626 /* Perform the same expansions on the pattern as performed by the
6627 pattern removal expansions. */
6628 pat = getpattern (lpatsub, quoted, 1);
6629
6630 if (rep)
6631 {
6632 if ((mflags & MATCH_QUOTED) == 0)
6633 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
6634 else
6635 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
6636 }
6637
6638 /* ksh93 doesn't allow the match specifier to be a part of the expanded
6639 pattern. This is an extension. Make sure we don't anchor the pattern
6640 at the beginning or end of the string if we're doing global replacement,
6641 though. */
6642 p = pat;
6643 if (mflags & MATCH_GLOBREP)
6644 mflags |= MATCH_ANY;
6645 else if (pat && pat[0] == '#')
6646 {
6647 mflags |= MATCH_BEG;
6648 p++;
6649 }
6650 else if (pat && pat[0] == '%')
6651 {
6652 mflags |= MATCH_END;
6653 p++;
6654 }
6655 else
6656 mflags |= MATCH_ANY;
6657
6658 /* OK, we now want to substitute REP for PAT in VAL. If
6659 flags & MATCH_GLOBREP is non-zero, the substitution is done
6660 everywhere, otherwise only the first occurrence of PAT is
6661 replaced. The pattern matching code doesn't understand
6662 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
6663 values passed in (VT_VARIABLE) so the pattern substitution
6664 code works right. We need to requote special chars after
6665 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
6666 other cases if QUOTED == 0, since the posparams and arrays
6667 indexed by * or @ do special things when QUOTED != 0. */
6668
6669 switch (vtype)
6670 {
6671 case VT_VARIABLE:
6672 case VT_ARRAYMEMBER:
6673 temp = pat_subst (val, p, rep, mflags);
6674 if (vtype == VT_VARIABLE)
6675 FREE (val);
6676 if (temp)
6677 {
6678 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
6679 free (temp);
6680 temp = tt;
6681 }
6682 break;
6683 case VT_POSPARMS:
6684 temp = pos_params_pat_subst (val, p, rep, mflags);
6685 if (temp && (mflags & MATCH_QUOTED) == 0)
6686 {
6687 tt = quote_escapes (temp);
6688 free (temp);
6689 temp = tt;
6690 }
6691 break;
6692 #if defined (ARRAY_VARS)
6693 case VT_ARRAYVAR:
6694 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
6695 : array_patsub (array_cell (v), p, rep, mflags);
6696 /* Don't call quote_escapes anymore; array_patsub calls
6697 array_quote_escapes as appropriate before adding the
6698 space separators; ditto for assoc_patsub. */
6699 break;
6700 #endif
6701 }
6702
6703 FREE (pat);
6704 FREE (rep);
6705 free (lpatsub);
6706
6707 return temp;
6708 }
6709
6710 /****************************************************************/
6711 /* */
6712 /* Functions to perform case modification on variable values */
6713 /* */
6714 /****************************************************************/
6715
6716 /* Do case modification on the positional parameters. */
6717
6718 static char *
6719 pos_params_modcase (string, pat, modop, mflags)
6720 char *string, *pat;
6721 int modop;
6722 int mflags;
6723 {
6724 WORD_LIST *save, *params;
6725 WORD_DESC *w;
6726 char *ret;
6727 int pchar, qflags;
6728
6729 save = params = list_rest_of_args ();
6730 if (save == 0)
6731 return ((char *)NULL);
6732
6733 for ( ; params; params = params->next)
6734 {
6735 ret = sh_modcase (params->word->word, pat, modop);
6736 w = alloc_word_desc ();
6737 w->word = ret ? ret : savestring ("");
6738 dispose_word (params->word);
6739 params->word = w;
6740 }
6741
6742 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6743 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6744
6745 ret = string_list_pos_params (pchar, save, qflags);
6746 dispose_words (save);
6747
6748 return (ret);
6749 }
6750
6751 /* Perform case modification on VALUE, which is the expansion of
6752 VARNAME. MODSPEC is an expression supplying the type of modification
6753 to perform. QUOTED is a flags word containing the type of quoting
6754 currently in effect. */
6755 static char *
6756 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
6757 char *varname, *value;
6758 int ind, modspec;
6759 char *patspec;
6760 int quoted, flags;
6761 {
6762 int vtype, starsub, modop, mflags, x;
6763 char *val, *temp, *pat, *p, *lpat, *tt;
6764 SHELL_VAR *v;
6765
6766 if (value == 0)
6767 return ((char *)NULL);
6768
6769 this_command_name = varname;
6770
6771 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6772 if (vtype == -1)
6773 return ((char *)NULL);
6774
6775 starsub = vtype & VT_STARSUB;
6776 vtype &= ~VT_STARSUB;
6777
6778 modop = 0;
6779 mflags = 0;
6780 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6781 mflags |= MATCH_QUOTED;
6782 if (starsub)
6783 mflags |= MATCH_STARSUB;
6784
6785 p = patspec;
6786 if (modspec == '^')
6787 {
6788 x = p && p[0] == modspec;
6789 modop = x ? CASE_UPPER : CASE_UPFIRST;
6790 p += x;
6791 }
6792 else if (modspec == ',')
6793 {
6794 x = p && p[0] == modspec;
6795 modop = x ? CASE_LOWER : CASE_LOWFIRST;
6796 p += x;
6797 }
6798 else if (modspec == '~')
6799 {
6800 x = p && p[0] == modspec;
6801 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
6802 p += x;
6803 }
6804
6805 lpat = p ? savestring (p) : 0;
6806 /* Perform the same expansions on the pattern as performed by the
6807 pattern removal expansions. FOR LATER */
6808 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
6809
6810 /* OK, now we do the case modification. */
6811 switch (vtype)
6812 {
6813 case VT_VARIABLE:
6814 case VT_ARRAYMEMBER:
6815 temp = sh_modcase (val, pat, modop);
6816 if (vtype == VT_VARIABLE)
6817 FREE (val);
6818 if (temp)
6819 {
6820 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
6821 free (temp);
6822 temp = tt;
6823 }
6824 break;
6825
6826 case VT_POSPARMS:
6827 temp = pos_params_modcase (val, pat, modop, mflags);
6828 if (temp && (mflags & MATCH_QUOTED) == 0)
6829 {
6830 tt = quote_escapes (temp);
6831 free (temp);
6832 temp = tt;
6833 }
6834 break;
6835
6836 #if defined (ARRAY_VARS)
6837 case VT_ARRAYVAR:
6838 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
6839 : array_modcase (array_cell (v), pat, modop, mflags);
6840 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
6841 as appropriate before adding the space separators; ditto for
6842 assoc_modcase. */
6843 break;
6844 #endif
6845 }
6846
6847 FREE (pat);
6848 free (lpat);
6849
6850 return temp;
6851 }
6852
6853 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
6854 any occur, this must be a nested command substitution, so return 0.
6855 Otherwise, return 1. A valid arithmetic expression must always have a
6856 ( before a matching ), so any cases where there are more right parens
6857 means that this must not be an arithmetic expression, though the parser
6858 will not accept it without a balanced total number of parens. */
6859 static int
6860 chk_arithsub (s, len)
6861 const char *s;
6862 int len;
6863 {
6864 int i, count;
6865 DECLARE_MBSTATE;
6866
6867 i = count = 0;
6868 while (i < len)
6869 {
6870 if (s[i] == LPAREN)
6871 count++;
6872 else if (s[i] == RPAREN)
6873 {
6874 count--;
6875 if (count < 0)
6876 return 0;
6877 }
6878
6879 switch (s[i])
6880 {
6881 default:
6882 ADVANCE_CHAR (s, len, i);
6883 break;
6884
6885 case '\\':
6886 i++;
6887 if (s[i])
6888 ADVANCE_CHAR (s, len, i);
6889 break;
6890
6891 case '\'':
6892 i = skip_single_quoted (s, len, ++i);
6893 break;
6894
6895 case '"':
6896 i = skip_double_quoted ((char *)s, len, ++i);
6897 break;
6898 }
6899 }
6900
6901 return (count == 0);
6902 }
6903
6904 /****************************************************************/
6905 /* */
6906 /* Functions to perform parameter expansion on a string */
6907 /* */
6908 /****************************************************************/
6909
6910 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
6911 static WORD_DESC *
6912 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
6913 char *string;
6914 int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at, pflags;
6915 {
6916 int check_nullness, var_is_set, var_is_null, var_is_special;
6917 int want_substring, want_indir, want_patsub, want_casemod;
6918 char *name, *value, *temp, *temp1;
6919 WORD_DESC *tdesc, *ret;
6920 int t_index, sindex, c, tflag, modspec;
6921 intmax_t number;
6922 arrayind_t ind;
6923
6924 temp = temp1 = value = (char *)NULL;
6925 var_is_set = var_is_null = var_is_special = check_nullness = 0;
6926 want_substring = want_indir = want_patsub = want_casemod = 0;
6927
6928 sindex = *indexp;
6929 t_index = ++sindex;
6930 /* ${#var} doesn't have any of the other parameter expansions on it. */
6931 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
6932 name = string_extract (string, &t_index, "}", SX_VARNAME);
6933 else
6934 #if defined (CASEMOD_EXPANSIONS)
6935 /* To enable case-toggling expansions using the `~' operator character
6936 change the 1 to 0. */
6937 # if defined (CASEMOD_CAPCASE)
6938 name = string_extract (string, &t_index, "#%^,~:-=?+/}", SX_VARNAME);
6939 # else
6940 name = string_extract (string, &t_index, "#%^,:-=?+/}", SX_VARNAME);
6941 # endif /* CASEMOD_CAPCASE */
6942 #else
6943 name = string_extract (string, &t_index, "#%:-=?+/}", SX_VARNAME);
6944 #endif /* CASEMOD_EXPANSIONS */
6945
6946 ret = 0;
6947 tflag = 0;
6948
6949 ind = INTMAX_MIN;
6950
6951 /* If the name really consists of a special variable, then make sure
6952 that we have the entire name. We don't allow indirect references
6953 to special variables except `#', `?', `@' and `*'. */
6954 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
6955 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
6956 {
6957 t_index++;
6958 free (name);
6959 temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0);
6960 name = (char *)xmalloc (3 + (strlen (temp1)));
6961 *name = string[sindex];
6962 if (string[sindex] == '!')
6963 {
6964 /* indirect reference of $#, $?, $@, or $* */
6965 name[1] = string[sindex + 1];
6966 strcpy (name + 2, temp1);
6967 }
6968 else
6969 strcpy (name + 1, temp1);
6970 free (temp1);
6971 }
6972 sindex = t_index;
6973
6974 /* Find out what character ended the variable name. Then
6975 do the appropriate thing. */
6976 if (c = string[sindex])
6977 sindex++;
6978
6979 /* If c is followed by one of the valid parameter expansion
6980 characters, move past it as normal. If not, assume that
6981 a substring specification is being given, and do not move
6982 past it. */
6983 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
6984 {
6985 check_nullness++;
6986 if (c = string[sindex])
6987 sindex++;
6988 }
6989 else if (c == ':' && string[sindex] != RBRACE)
6990 want_substring = 1;
6991 else if (c == '/' && string[sindex] != RBRACE)
6992 want_patsub = 1;
6993 #if defined (CASEMOD_EXPANSIONS)
6994 else if (c == '^' || c == ',' || c == '~')
6995 {
6996 modspec = c;
6997 want_casemod = 1;
6998 }
6999 #endif
7000
7001 /* Catch the valid and invalid brace expressions that made it through the
7002 tests above. */
7003 /* ${#-} is a valid expansion and means to take the length of $-.
7004 Similarly for ${#?} and ${##}... */
7005 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7006 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
7007 {
7008 name = (char *)xrealloc (name, 3);
7009 name[1] = c;
7010 name[2] = '\0';
7011 c = string[sindex++];
7012 }
7013
7014 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
7015 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7016 member (c, "%:=+/") && string[sindex] == RBRACE)
7017 {
7018 temp = (char *)NULL;
7019 goto bad_substitution;
7020 }
7021
7022 /* Indirect expansion begins with a `!'. A valid indirect expansion is
7023 either a variable name, one of the positional parameters or a special
7024 variable that expands to one of the positional parameters. */
7025 want_indir = *name == '!' &&
7026 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7027 || VALID_INDIR_PARAM (name[1]));
7028
7029 /* Determine the value of this variable. */
7030
7031 /* Check for special variables, directly referenced. */
7032 if (SPECIAL_VAR (name, want_indir))
7033 var_is_special++;
7034
7035 /* Check for special expansion things, like the length of a parameter */
7036 if (*name == '#' && name[1])
7037 {
7038 /* If we are not pointing at the character just after the
7039 closing brace, then we haven't gotten all of the name.
7040 Since it begins with a special character, this is a bad
7041 substitution. Also check NAME for validity before trying
7042 to go on. */
7043 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
7044 {
7045 temp = (char *)NULL;
7046 goto bad_substitution;
7047 }
7048
7049 number = parameter_brace_expand_length (name);
7050 if (number == INTMAX_MIN && unbound_vars_is_error)
7051 {
7052 last_command_exit_value = EXECUTION_FAILURE;
7053 err_unboundvar (name+1);
7054 free (name);
7055 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7056 }
7057 free (name);
7058
7059 *indexp = sindex;
7060 if (number < 0)
7061 return (&expand_wdesc_error);
7062 else
7063 {
7064 ret = alloc_word_desc ();
7065 ret->word = itos (number);
7066 return ret;
7067 }
7068 }
7069
7070 /* ${@} is identical to $@. */
7071 if (name[0] == '@' && name[1] == '\0')
7072 {
7073 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7074 *quoted_dollar_atp = 1;
7075
7076 if (contains_dollar_at)
7077 *contains_dollar_at = 1;
7078 }
7079
7080 /* Process ${!PREFIX*} expansion. */
7081 if (want_indir && string[sindex - 1] == RBRACE &&
7082 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
7083 legal_variable_starter ((unsigned char) name[1]))
7084 {
7085 char **x;
7086 WORD_LIST *xlist;
7087
7088 temp1 = savestring (name + 1);
7089 number = strlen (temp1);
7090 temp1[number - 1] = '\0';
7091 x = all_variables_matching_prefix (temp1);
7092 xlist = strvec_to_word_list (x, 0, 0);
7093 if (string[sindex - 2] == '*')
7094 temp = string_list_dollar_star (xlist);
7095 else
7096 {
7097 temp = string_list_dollar_at (xlist, quoted);
7098 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7099 *quoted_dollar_atp = 1;
7100 if (contains_dollar_at)
7101 *contains_dollar_at = 1;
7102 }
7103 free (x);
7104 dispose_words (xlist);
7105 free (temp1);
7106 *indexp = sindex;
7107
7108 ret = alloc_word_desc ();
7109 ret->word = temp;
7110 return ret;
7111 }
7112
7113 #if defined (ARRAY_VARS)
7114 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
7115 if (want_indir && string[sindex - 1] == RBRACE &&
7116 string[sindex - 2] == ']' && valid_array_reference (name+1))
7117 {
7118 char *x, *x1;
7119
7120 temp1 = savestring (name + 1);
7121 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
7122 FREE (x);
7123 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
7124 {
7125 temp = array_keys (temp1, quoted); /* handles assoc vars too */
7126 if (x1[0] == '@')
7127 {
7128 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7129 *quoted_dollar_atp = 1;
7130 if (contains_dollar_at)
7131 *contains_dollar_at = 1;
7132 }
7133
7134 free (temp1);
7135 *indexp = sindex;
7136
7137 ret = alloc_word_desc ();
7138 ret->word = temp;
7139 return ret;
7140 }
7141
7142 free (temp1);
7143 }
7144 #endif /* ARRAY_VARS */
7145
7146 /* Make sure that NAME is valid before trying to go on. */
7147 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
7148 var_is_special) == 0)
7149 {
7150 temp = (char *)NULL;
7151 goto bad_substitution;
7152 }
7153
7154 if (want_indir)
7155 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
7156 else
7157 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&PF_NOSPLIT2), &ind);
7158
7159 if (tdesc)
7160 {
7161 temp = tdesc->word;
7162 tflag = tdesc->flags;
7163 dispose_word_desc (tdesc);
7164 }
7165 else
7166 temp = (char *)0;
7167
7168 #if defined (ARRAY_VARS)
7169 if (valid_array_reference (name))
7170 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at);
7171 #endif
7172
7173 var_is_set = temp != (char *)0;
7174 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
7175
7176 /* Get the rest of the stuff inside the braces. */
7177 if (c && c != RBRACE)
7178 {
7179 /* Extract the contents of the ${ ... } expansion
7180 according to the Posix.2 rules. */
7181 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
7182 if (string[sindex] == RBRACE)
7183 sindex++;
7184 else
7185 goto bad_substitution;
7186 }
7187 else
7188 value = (char *)NULL;
7189
7190 *indexp = sindex;
7191
7192 /* All the cases where an expansion can possibly generate an unbound
7193 variable error. */
7194 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
7195 {
7196 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]))
7197 {
7198 last_command_exit_value = EXECUTION_FAILURE;
7199 err_unboundvar (name);
7200 FREE (value);
7201 FREE (temp);
7202 free (name);
7203 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7204 }
7205 }
7206
7207 /* If this is a substring spec, process it and add the result. */
7208 if (want_substring)
7209 {
7210 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7211 FREE (name);
7212 FREE (value);
7213 FREE (temp);
7214
7215 if (temp1 == &expand_param_error)
7216 return (&expand_wdesc_error);
7217 else if (temp1 == &expand_param_fatal)
7218 return (&expand_wdesc_fatal);
7219
7220 ret = alloc_word_desc ();
7221 ret->word = temp1;
7222 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7223 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7224 return ret;
7225 }
7226 else if (want_patsub)
7227 {
7228 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7229 FREE (name);
7230 FREE (value);
7231 FREE (temp);
7232
7233 if (temp1 == &expand_param_error)
7234 return (&expand_wdesc_error);
7235 else if (temp1 == &expand_param_fatal)
7236 return (&expand_wdesc_fatal);
7237
7238 ret = alloc_word_desc ();
7239 ret->word = temp1;
7240 ret = alloc_word_desc ();
7241 ret->word = temp1;
7242 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7243 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7244 return ret;
7245 }
7246 #if defined (CASEMOD_EXPANSIONS)
7247 else if (want_casemod)
7248 {
7249 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7250 FREE (name);
7251 FREE (value);
7252 FREE (temp);
7253
7254 if (temp1 == &expand_param_error)
7255 return (&expand_wdesc_error);
7256 else if (temp1 == &expand_param_fatal)
7257 return (&expand_wdesc_fatal);
7258
7259 ret = alloc_word_desc ();
7260 ret->word = temp1;
7261 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7262 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7263 return ret;
7264 }
7265 #endif
7266
7267 /* Do the right thing based on which character ended the variable name. */
7268 switch (c)
7269 {
7270 default:
7271 case '\0':
7272 bad_substitution:
7273 report_error (_("%s: bad substitution"), string ? string : "??");
7274 FREE (value);
7275 FREE (temp);
7276 free (name);
7277 return &expand_wdesc_error;
7278
7279 case RBRACE:
7280 break;
7281
7282 case '#': /* ${param#[#]pattern} */
7283 case '%': /* ${param%[%]pattern} */
7284 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
7285 {
7286 FREE (value);
7287 break;
7288 }
7289 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7290 free (temp);
7291 free (value);
7292 free (name);
7293
7294 ret = alloc_word_desc ();
7295 ret->word = temp1;
7296 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7297 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7298 return ret;
7299
7300 case '-':
7301 case '=':
7302 case '?':
7303 case '+':
7304 if (var_is_set && var_is_null == 0)
7305 {
7306 /* If the operator is `+', we don't want the value of the named
7307 variable for anything, just the value of the right hand side. */
7308 if (c == '+')
7309 {
7310 /* XXX -- if we're double-quoted and the named variable is "$@",
7311 we want to turn off any special handling of "$@" --
7312 we're not using it, so whatever is on the rhs applies. */
7313 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7314 *quoted_dollar_atp = 0;
7315 if (contains_dollar_at)
7316 *contains_dollar_at = 0;
7317
7318 FREE (temp);
7319 if (value)
7320 {
7321 /* From Posix discussion on austin-group list. Issue 221
7322 requires that backslashes escaping `}' inside
7323 double-quoted ${...} be removed. */
7324 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7325 quoted |= Q_DOLBRACE;
7326 ret = parameter_brace_expand_rhs (name, value, c,
7327 quoted,
7328 quoted_dollar_atp,
7329 contains_dollar_at);
7330 /* XXX - fix up later, esp. noting presence of
7331 W_HASQUOTEDNULL in ret->flags */
7332 free (value);
7333 }
7334 else
7335 temp = (char *)NULL;
7336 }
7337 else
7338 {
7339 FREE (value);
7340 }
7341 /* Otherwise do nothing; just use the value in TEMP. */
7342 }
7343 else /* VAR not set or VAR is NULL. */
7344 {
7345 FREE (temp);
7346 temp = (char *)NULL;
7347 if (c == '=' && var_is_special)
7348 {
7349 report_error (_("$%s: cannot assign in this way"), name);
7350 free (name);
7351 free (value);
7352 return &expand_wdesc_error;
7353 }
7354 else if (c == '?')
7355 {
7356 parameter_brace_expand_error (name, value);
7357 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7358 }
7359 else if (c != '+')
7360 {
7361 /* XXX -- if we're double-quoted and the named variable is "$@",
7362 we want to turn off any special handling of "$@" --
7363 we're not using it, so whatever is on the rhs applies. */
7364 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7365 *quoted_dollar_atp = 0;
7366 if (contains_dollar_at)
7367 *contains_dollar_at = 0;
7368
7369 /* From Posix discussion on austin-group list. Issue 221 requires
7370 that backslashes escaping `}' inside double-quoted ${...} be
7371 removed. */
7372 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7373 quoted |= Q_DOLBRACE;
7374 ret = parameter_brace_expand_rhs (name, value, c, quoted,
7375 quoted_dollar_atp,
7376 contains_dollar_at);
7377 /* XXX - fix up later, esp. noting presence of
7378 W_HASQUOTEDNULL in tdesc->flags */
7379 }
7380 free (value);
7381 }
7382
7383 break;
7384 }
7385 free (name);
7386
7387 if (ret == 0)
7388 {
7389 ret = alloc_word_desc ();
7390 ret->flags = tflag;
7391 ret->word = temp;
7392 }
7393 return (ret);
7394 }
7395
7396 /* Expand a single ${xxx} expansion. The braces are optional. When
7397 the braces are used, parameter_brace_expand() does the work,
7398 possibly calling param_expand recursively. */
7399 static WORD_DESC *
7400 param_expand (string, sindex, quoted, expanded_something,
7401 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
7402 pflags)
7403 char *string;
7404 int *sindex, quoted, *expanded_something, *contains_dollar_at;
7405 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
7406 {
7407 char *temp, *temp1, uerror[3];
7408 int zindex, t_index, expok;
7409 unsigned char c;
7410 intmax_t number;
7411 SHELL_VAR *var;
7412 WORD_LIST *list;
7413 WORD_DESC *tdesc, *ret;
7414 int tflag;
7415
7416 zindex = *sindex;
7417 c = string[++zindex];
7418
7419 temp = (char *)NULL;
7420 ret = tdesc = (WORD_DESC *)NULL;
7421 tflag = 0;
7422
7423 /* Do simple cases first. Switch on what follows '$'. */
7424 switch (c)
7425 {
7426 /* $0 .. $9? */
7427 case '0':
7428 case '1':
7429 case '2':
7430 case '3':
7431 case '4':
7432 case '5':
7433 case '6':
7434 case '7':
7435 case '8':
7436 case '9':
7437 temp1 = dollar_vars[TODIGIT (c)];
7438 if (unbound_vars_is_error && temp1 == (char *)NULL)
7439 {
7440 uerror[0] = '$';
7441 uerror[1] = c;
7442 uerror[2] = '\0';
7443 last_command_exit_value = EXECUTION_FAILURE;
7444 err_unboundvar (uerror);
7445 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7446 }
7447 if (temp1)
7448 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7449 ? quote_string (temp1)
7450 : quote_escapes (temp1);
7451 else
7452 temp = (char *)NULL;
7453
7454 break;
7455
7456 /* $$ -- pid of the invoking shell. */
7457 case '$':
7458 temp = itos (dollar_dollar_pid);
7459 break;
7460
7461 /* $# -- number of positional parameters. */
7462 case '#':
7463 temp = itos (number_of_args ());
7464 break;
7465
7466 /* $? -- return value of the last synchronous command. */
7467 case '?':
7468 temp = itos (last_command_exit_value);
7469 break;
7470
7471 /* $- -- flags supplied to the shell on invocation or by `set'. */
7472 case '-':
7473 temp = which_set_flags ();
7474 break;
7475
7476 /* $! -- Pid of the last asynchronous command. */
7477 case '!':
7478 /* If no asynchronous pids have been created, expand to nothing.
7479 If `set -u' has been executed, and no async processes have
7480 been created, this is an expansion error. */
7481 if (last_asynchronous_pid == NO_PID)
7482 {
7483 if (expanded_something)
7484 *expanded_something = 0;
7485 temp = (char *)NULL;
7486 if (unbound_vars_is_error)
7487 {
7488 uerror[0] = '$';
7489 uerror[1] = c;
7490 uerror[2] = '\0';
7491 last_command_exit_value = EXECUTION_FAILURE;
7492 err_unboundvar (uerror);
7493 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7494 }
7495 }
7496 else
7497 temp = itos (last_asynchronous_pid);
7498 break;
7499
7500 /* The only difference between this and $@ is when the arg is quoted. */
7501 case '*': /* `$*' */
7502 list = list_rest_of_args ();
7503
7504 #if 0
7505 /* According to austin-group posix proposal by Geoff Clare in
7506 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7507
7508 "The shell shall write a message to standard error and
7509 immediately exit when it tries to expand an unset parameter
7510 other than the '@' and '*' special parameters."
7511 */
7512
7513 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7514 {
7515 uerror[0] = '$';
7516 uerror[1] = '*';
7517 uerror[2] = '\0';
7518 last_command_exit_value = EXECUTION_FAILURE;
7519 err_unboundvar (uerror);
7520 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7521 }
7522 #endif
7523
7524 /* If there are no command-line arguments, this should just
7525 disappear if there are other characters in the expansion,
7526 even if it's quoted. */
7527 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
7528 temp = (char *)NULL;
7529 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
7530 {
7531 /* If we have "$*" we want to make a string of the positional
7532 parameters, separated by the first character of $IFS, and
7533 quote the whole string, including the separators. If IFS
7534 is unset, the parameters are separated by ' '; if $IFS is
7535 null, the parameters are concatenated. */
7536 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
7537 if (temp)
7538 {
7539 temp1 = quote_string (temp);
7540 if (*temp == 0)
7541 tflag |= W_HASQUOTEDNULL;
7542 free (temp);
7543 temp = temp1;
7544 }
7545 }
7546 else
7547 {
7548 /* We check whether or not we're eventually going to split $* here,
7549 for example when IFS is empty and we are processing the rhs of
7550 an assignment statement. In that case, we don't separate the
7551 arguments at all. Otherwise, if the $* is not quoted it is
7552 identical to $@ */
7553 #if 1
7554 # if defined (HANDLE_MULTIBYTE)
7555 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
7556 # else
7557 if (expand_no_split_dollar_star && ifs_firstc == 0)
7558 # endif
7559 temp = string_list_dollar_star (list);
7560 else
7561 temp = string_list_dollar_at (list, quoted);
7562 #else
7563 temp = string_list_dollar_at (list, quoted);
7564 #endif
7565 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
7566 *contains_dollar_at = 1;
7567 }
7568
7569 dispose_words (list);
7570 break;
7571
7572 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
7573 means that we have to turn quoting off after we split into
7574 the individually quoted arguments so that the final split
7575 on the first character of $IFS is still done. */
7576 case '@': /* `$@' */
7577 list = list_rest_of_args ();
7578
7579 #if 0
7580 /* According to austin-group posix proposal by Geoff Clare in
7581 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7582
7583 "The shell shall write a message to standard error and
7584 immediately exit when it tries to expand an unset parameter
7585 other than the '@' and '*' special parameters."
7586 */
7587
7588 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7589 {
7590 uerror[0] = '$';
7591 uerror[1] = '@';
7592 uerror[2] = '\0';
7593 last_command_exit_value = EXECUTION_FAILURE;
7594 err_unboundvar (uerror);
7595 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7596 }
7597 #endif
7598
7599 /* We want to flag the fact that we saw this. We can't turn
7600 off quoting entirely, because other characters in the
7601 string might need it (consider "\"$@\""), but we need some
7602 way to signal that the final split on the first character
7603 of $IFS should be done, even though QUOTED is 1. */
7604 /* XXX - should this test include Q_PATQUOTE? */
7605 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7606 *quoted_dollar_at_p = 1;
7607 if (contains_dollar_at)
7608 *contains_dollar_at = 1;
7609
7610 #if 0
7611 if (pflags & PF_NOSPLIT2)
7612 temp = string_list_internal (quoted ? quote_list (list) : list, " ");
7613 else
7614 #endif
7615 /* We want to separate the positional parameters with the first
7616 character of $IFS in case $IFS is something other than a space.
7617 We also want to make sure that splitting is done no matter what --
7618 according to POSIX.2, this expands to a list of the positional
7619 parameters no matter what IFS is set to. */
7620 temp = string_list_dollar_at (list, quoted);
7621
7622 dispose_words (list);
7623 break;
7624
7625 case LBRACE:
7626 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
7627 quoted_dollar_at_p,
7628 contains_dollar_at);
7629
7630 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
7631 return (tdesc);
7632 temp = tdesc ? tdesc->word : (char *)0;
7633
7634 /* XXX */
7635 /* Quoted nulls should be removed if there is anything else
7636 in the string. */
7637 /* Note that we saw the quoted null so we can add one back at
7638 the end of this function if there are no other characters
7639 in the string, discard TEMP, and go on. The exception to
7640 this is when we have "${@}" and $1 is '', since $@ needs
7641 special handling. */
7642 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
7643 {
7644 if (had_quoted_null_p)
7645 *had_quoted_null_p = 1;
7646 if (*quoted_dollar_at_p == 0)
7647 {
7648 free (temp);
7649 tdesc->word = temp = (char *)NULL;
7650 }
7651
7652 }
7653
7654 ret = tdesc;
7655 goto return0;
7656
7657 /* Do command or arithmetic substitution. */
7658 case LPAREN:
7659 /* We have to extract the contents of this paren substitution. */
7660 t_index = zindex + 1;
7661 temp = extract_command_subst (string, &t_index, 0);
7662 zindex = t_index;
7663
7664 /* For Posix.2-style `$(( ))' arithmetic substitution,
7665 extract the expression and pass it to the evaluator. */
7666 if (temp && *temp == LPAREN)
7667 {
7668 char *temp2;
7669 temp1 = temp + 1;
7670 temp2 = savestring (temp1);
7671 t_index = strlen (temp2) - 1;
7672
7673 if (temp2[t_index] != RPAREN)
7674 {
7675 free (temp2);
7676 goto comsub;
7677 }
7678
7679 /* Cut off ending `)' */
7680 temp2[t_index] = '\0';
7681
7682 if (chk_arithsub (temp2, t_index) == 0)
7683 {
7684 free (temp2);
7685 #if 0
7686 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
7687 #endif
7688 goto comsub;
7689 }
7690
7691 /* Expand variables found inside the expression. */
7692 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7693 free (temp2);
7694
7695 arithsub:
7696 /* No error messages. */
7697 this_command_name = (char *)NULL;
7698 number = evalexp (temp1, &expok);
7699 free (temp);
7700 free (temp1);
7701 if (expok == 0)
7702 {
7703 if (interactive_shell == 0 && posixly_correct)
7704 {
7705 last_command_exit_value = EXECUTION_FAILURE;
7706 return (&expand_wdesc_fatal);
7707 }
7708 else
7709 return (&expand_wdesc_error);
7710 }
7711 temp = itos (number);
7712 break;
7713 }
7714
7715 comsub:
7716 if (pflags & PF_NOCOMSUB)
7717 /* we need zindex+1 because string[zindex] == RPAREN */
7718 temp1 = substring (string, *sindex, zindex+1);
7719 else
7720 {
7721 tdesc = command_substitute (temp, quoted);
7722 temp1 = tdesc ? tdesc->word : (char *)NULL;
7723 if (tdesc)
7724 dispose_word_desc (tdesc);
7725 }
7726 FREE (temp);
7727 temp = temp1;
7728 break;
7729
7730 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
7731 away in a future bash release. */
7732 case '[':
7733 /* Extract the contents of this arithmetic substitution. */
7734 t_index = zindex + 1;
7735 temp = extract_arithmetic_subst (string, &t_index);
7736 zindex = t_index;
7737 if (temp == 0)
7738 {
7739 temp = savestring (string);
7740 if (expanded_something)
7741 *expanded_something = 0;
7742 goto return0;
7743 }
7744
7745 /* Do initial variable expansion. */
7746 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES);
7747
7748 goto arithsub;
7749
7750 default:
7751 /* Find the variable in VARIABLE_LIST. */
7752 temp = (char *)NULL;
7753
7754 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
7755 ;
7756 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
7757
7758 /* If this isn't a variable name, then just output the `$'. */
7759 if (temp1 == 0 || *temp1 == '\0')
7760 {
7761 FREE (temp1);
7762 temp = (char *)xmalloc (2);
7763 temp[0] = '$';
7764 temp[1] = '\0';
7765 if (expanded_something)
7766 *expanded_something = 0;
7767 goto return0;
7768 }
7769
7770 /* If the variable exists, return its value cell. */
7771 var = find_variable (temp1);
7772
7773 if (var && invisible_p (var) == 0 && var_isset (var))
7774 {
7775 #if defined (ARRAY_VARS)
7776 if (assoc_p (var) || array_p (var))
7777 {
7778 temp = array_p (var) ? array_reference (array_cell (var), 0)
7779 : assoc_reference (assoc_cell (var), "0");
7780 if (temp)
7781 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7782 ? quote_string (temp)
7783 : quote_escapes (temp);
7784 else if (unbound_vars_is_error)
7785 goto unbound_variable;
7786 }
7787 else
7788 #endif
7789 {
7790 temp = value_cell (var);
7791
7792 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7793 ? quote_string (temp)
7794 : quote_escapes (temp);
7795 }
7796
7797 free (temp1);
7798
7799 goto return0;
7800 }
7801
7802 temp = (char *)NULL;
7803
7804 unbound_variable:
7805 if (unbound_vars_is_error)
7806 {
7807 last_command_exit_value = EXECUTION_FAILURE;
7808 err_unboundvar (temp1);
7809 }
7810 else
7811 {
7812 free (temp1);
7813 goto return0;
7814 }
7815
7816 free (temp1);
7817 last_command_exit_value = EXECUTION_FAILURE;
7818 return ((unbound_vars_is_error && interactive_shell == 0)
7819 ? &expand_wdesc_fatal
7820 : &expand_wdesc_error);
7821 }
7822
7823 if (string[zindex])
7824 zindex++;
7825
7826 return0:
7827 *sindex = zindex;
7828
7829 if (ret == 0)
7830 {
7831 ret = alloc_word_desc ();
7832 ret->flags = tflag; /* XXX */
7833 ret->word = temp;
7834 }
7835 return ret;
7836 }
7837
7838 /* Make a word list which is the result of parameter and variable
7839 expansion, command substitution, arithmetic substitution, and
7840 quote removal of WORD. Return a pointer to a WORD_LIST which is
7841 the result of the expansion. If WORD contains a null word, the
7842 word list returned is also null.
7843
7844 QUOTED contains flag values defined in shell.h.
7845
7846 ISEXP is used to tell expand_word_internal that the word should be
7847 treated as the result of an expansion. This has implications for
7848 how IFS characters in the word are treated.
7849
7850 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
7851 they point to an integer value which receives information about expansion.
7852 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
7853 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
7854 else zero.
7855
7856 This only does word splitting in the case of $@ expansion. In that
7857 case, we split on ' '. */
7858
7859 /* Values for the local variable quoted_state. */
7860 #define UNQUOTED 0
7861 #define PARTIALLY_QUOTED 1
7862 #define WHOLLY_QUOTED 2
7863
7864 static WORD_LIST *
7865 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
7866 WORD_DESC *word;
7867 int quoted, isexp;
7868 int *contains_dollar_at;
7869 int *expanded_something;
7870 {
7871 WORD_LIST *list;
7872 WORD_DESC *tword;
7873
7874 /* The intermediate string that we build while expanding. */
7875 char *istring;
7876
7877 /* The current size of the above object. */
7878 int istring_size;
7879
7880 /* Index into ISTRING. */
7881 int istring_index;
7882
7883 /* Temporary string storage. */
7884 char *temp, *temp1;
7885
7886 /* The text of WORD. */
7887 register char *string;
7888
7889 /* The size of STRING. */
7890 size_t string_size;
7891
7892 /* The index into STRING. */
7893 int sindex;
7894
7895 /* This gets 1 if we see a $@ while quoted. */
7896 int quoted_dollar_at;
7897
7898 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
7899 whether WORD contains no quoting characters, a partially quoted
7900 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
7901 int quoted_state;
7902
7903 /* State flags */
7904 int had_quoted_null;
7905 int has_dollar_at;
7906 int tflag;
7907 int pflags; /* flags passed to param_expand */
7908
7909 int assignoff; /* If assignment, offset of `=' */
7910
7911 register unsigned char c; /* Current character. */
7912 int t_index; /* For calls to string_extract_xxx. */
7913
7914 char twochars[2];
7915
7916 DECLARE_MBSTATE;
7917
7918 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
7919 istring[istring_index = 0] = '\0';
7920 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
7921 quoted_state = UNQUOTED;
7922
7923 string = word->word;
7924 if (string == 0)
7925 goto finished_with_string;
7926 /* Don't need the string length for the SADD... and COPY_ macros unless
7927 multibyte characters are possible. */
7928 string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1;
7929
7930 if (contains_dollar_at)
7931 *contains_dollar_at = 0;
7932
7933 assignoff = -1;
7934
7935 /* Begin the expansion. */
7936
7937 for (sindex = 0; ;)
7938 {
7939 c = string[sindex];
7940
7941 /* Case on toplevel character. */
7942 switch (c)
7943 {
7944 case '\0':
7945 goto finished_with_string;
7946
7947 case CTLESC:
7948 sindex++;
7949 #if HANDLE_MULTIBYTE
7950 if (MB_CUR_MAX > 1 && string[sindex])
7951 {
7952 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
7953 }
7954 else
7955 #endif
7956 {
7957 temp = (char *)xmalloc (3);
7958 temp[0] = CTLESC;
7959 temp[1] = c = string[sindex];
7960 temp[2] = '\0';
7961 }
7962
7963 dollar_add_string:
7964 if (string[sindex])
7965 sindex++;
7966
7967 add_string:
7968 if (temp)
7969 {
7970 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
7971 temp = (char *)0;
7972 }
7973
7974 break;
7975
7976 #if defined (PROCESS_SUBSTITUTION)
7977 /* Process substitution. */
7978 case '<':
7979 case '>':
7980 {
7981 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
7982 {
7983 sindex--; /* add_character: label increments sindex */
7984 goto add_character;
7985 }
7986 else
7987 t_index = sindex + 1; /* skip past both '<' and LPAREN */
7988
7989 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index); /*))*/
7990 sindex = t_index;
7991
7992 /* If the process substitution specification is `<()', we want to
7993 open the pipe for writing in the child and produce output; if
7994 it is `>()', we want to open the pipe for reading in the child
7995 and consume input. */
7996 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
7997
7998 FREE (temp1);
7999
8000 goto dollar_add_string;
8001 }
8002 #endif /* PROCESS_SUBSTITUTION */
8003
8004 case '=':
8005 /* Posix.2 section 3.6.1 says that tildes following `=' in words
8006 which are not assignment statements are not expanded. If the
8007 shell isn't in posix mode, though, we perform tilde expansion
8008 on `likely candidate' unquoted assignment statements (flags
8009 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
8010 contains an unquoted :~ or =~. Something to think about: we
8011 now have a flag that says to perform tilde expansion on arguments
8012 to `assignment builtins' like declare and export that look like
8013 assignment statements. We now do tilde expansion on such words
8014 even in POSIX mode. */
8015 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
8016 {
8017 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8018 goto add_ifs_character;
8019 else
8020 goto add_character;
8021 }
8022 /* If we're not in posix mode or forcing assignment-statement tilde
8023 expansion, note where the `=' appears in the word and prepare to
8024 do tilde expansion following the first `='. */
8025 if ((word->flags & W_ASSIGNMENT) &&
8026 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8027 assignoff == -1 && sindex > 0)
8028 assignoff = sindex;
8029 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
8030 word->flags |= W_ITILDE;
8031 #if 0
8032 else if ((word->flags & W_ASSIGNMENT) &&
8033 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8034 string[sindex+1] == '~')
8035 word->flags |= W_ITILDE;
8036 #endif
8037 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8038 goto add_ifs_character;
8039 else
8040 goto add_character;
8041
8042 case ':':
8043 if (word->flags & W_NOTILDE)
8044 {
8045 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8046 goto add_ifs_character;
8047 else
8048 goto add_character;
8049 }
8050
8051 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
8052 string[sindex+1] == '~')
8053 word->flags |= W_ITILDE;
8054
8055 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8056 goto add_ifs_character;
8057 else
8058 goto add_character;
8059
8060 case '~':
8061 /* If the word isn't supposed to be tilde expanded, or we're not
8062 at the start of a word or after an unquoted : or = in an
8063 assignment statement, we don't do tilde expansion. */
8064 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
8065 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
8066 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8067 {
8068 word->flags &= ~W_ITILDE;
8069 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
8070 goto add_ifs_character;
8071 else
8072 goto add_character;
8073 }
8074
8075 if (word->flags & W_ASSIGNRHS)
8076 tflag = 2;
8077 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
8078 tflag = 1;
8079 else
8080 tflag = 0;
8081
8082 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
8083
8084 word->flags &= ~W_ITILDE;
8085
8086 if (temp && *temp && t_index > 0)
8087 {
8088 temp1 = bash_tilde_expand (temp, tflag);
8089 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
8090 {
8091 FREE (temp);
8092 FREE (temp1);
8093 goto add_character; /* tilde expansion failed */
8094 }
8095 free (temp);
8096 temp = temp1;
8097 sindex += t_index;
8098 goto add_quoted_string; /* XXX was add_string */
8099 }
8100 else
8101 {
8102 FREE (temp);
8103 goto add_character;
8104 }
8105
8106 case '$':
8107 if (expanded_something)
8108 *expanded_something = 1;
8109
8110 has_dollar_at = 0;
8111 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
8112 if (word->flags & W_NOSPLIT2)
8113 pflags |= PF_NOSPLIT2;
8114 tword = param_expand (string, &sindex, quoted, expanded_something,
8115 &has_dollar_at, &quoted_dollar_at,
8116 &had_quoted_null, pflags);
8117
8118 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
8119 {
8120 free (string);
8121 free (istring);
8122 return ((tword == &expand_wdesc_error) ? &expand_word_error
8123 : &expand_word_fatal);
8124 }
8125 if (contains_dollar_at && has_dollar_at)
8126 *contains_dollar_at = 1;
8127
8128 if (tword && (tword->flags & W_HASQUOTEDNULL))
8129 had_quoted_null = 1;
8130
8131 temp = tword->word;
8132 dispose_word_desc (tword);
8133
8134 goto add_string;
8135 break;
8136
8137 case '`': /* Backquoted command substitution. */
8138 {
8139 t_index = sindex++;
8140
8141 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
8142 /* The test of sindex against t_index is to allow bare instances of
8143 ` to pass through, for backwards compatibility. */
8144 if (temp == &extract_string_error || temp == &extract_string_fatal)
8145 {
8146 if (sindex - 1 == t_index)
8147 {
8148 sindex = t_index;
8149 goto add_character;
8150 }
8151 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
8152 free (string);
8153 free (istring);
8154 return ((temp == &extract_string_error) ? &expand_word_error
8155 : &expand_word_fatal);
8156 }
8157
8158 if (expanded_something)
8159 *expanded_something = 1;
8160
8161 if (word->flags & W_NOCOMSUB)
8162 /* sindex + 1 because string[sindex] == '`' */
8163 temp1 = substring (string, t_index, sindex + 1);
8164 else
8165 {
8166 de_backslash (temp);
8167 tword = command_substitute (temp, quoted);
8168 temp1 = tword ? tword->word : (char *)NULL;
8169 if (tword)
8170 dispose_word_desc (tword);
8171 }
8172 FREE (temp);
8173 temp = temp1;
8174 goto dollar_add_string;
8175 }
8176
8177 case '\\':
8178 if (string[sindex + 1] == '\n')
8179 {
8180 sindex += 2;
8181 continue;
8182 }
8183
8184 c = string[++sindex];
8185
8186 if (quoted & Q_HERE_DOCUMENT)
8187 tflag = CBSHDOC;
8188 else if (quoted & Q_DOUBLE_QUOTES)
8189 tflag = CBSDQUOTE;
8190 else
8191 tflag = 0;
8192
8193 /* From Posix discussion on austin-group list: Backslash escaping
8194 a } in ${...} is removed. Issue 0000221 */
8195 if ((quoted & Q_DOLBRACE) && c == RBRACE)
8196 {
8197 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8198 }
8199 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
8200 {
8201 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
8202 }
8203 else if (c == 0)
8204 {
8205 c = CTLNUL;
8206 sindex--; /* add_character: label increments sindex */
8207 goto add_character;
8208 }
8209 else
8210 {
8211 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8212 }
8213
8214 sindex++;
8215 add_twochars:
8216 /* BEFORE jumping here, we need to increment sindex if appropriate */
8217 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8218 DEFAULT_ARRAY_SIZE);
8219 istring[istring_index++] = twochars[0];
8220 istring[istring_index++] = twochars[1];
8221 istring[istring_index] = '\0';
8222
8223 break;
8224
8225 case '"':
8226 #if 0
8227 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE))
8228 #else
8229 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8230 #endif
8231 goto add_character;
8232
8233 t_index = ++sindex;
8234 temp = string_extract_double_quoted (string, &sindex, 0);
8235
8236 /* If the quotes surrounded the entire string, then the
8237 whole word was quoted. */
8238 quoted_state = (t_index == 1 && string[sindex] == '\0')
8239 ? WHOLLY_QUOTED
8240 : PARTIALLY_QUOTED;
8241
8242 if (temp && *temp)
8243 {
8244 tword = alloc_word_desc ();
8245 tword->word = temp;
8246
8247 temp = (char *)NULL;
8248
8249 has_dollar_at = 0;
8250 /* Need to get W_HASQUOTEDNULL flag through this function. */
8251 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &has_dollar_at, (int *)NULL);
8252
8253 if (list == &expand_word_error || list == &expand_word_fatal)
8254 {
8255 free (istring);
8256 free (string);
8257 /* expand_word_internal has already freed temp_word->word
8258 for us because of the way it prints error messages. */
8259 tword->word = (char *)NULL;
8260 dispose_word (tword);
8261 return list;
8262 }
8263
8264 dispose_word (tword);
8265
8266 /* "$@" (a double-quoted dollar-at) expands into nothing,
8267 not even a NULL word, when there are no positional
8268 parameters. */
8269 if (list == 0 && has_dollar_at)
8270 {
8271 quoted_dollar_at++;
8272 break;
8273 }
8274
8275 /* If we get "$@", we know we have expanded something, so we
8276 need to remember it for the final split on $IFS. This is
8277 a special case; it's the only case where a quoted string
8278 can expand into more than one word. It's going to come back
8279 from the above call to expand_word_internal as a list with
8280 a single word, in which all characters are quoted and
8281 separated by blanks. What we want to do is to turn it back
8282 into a list for the next piece of code. */
8283 if (list)
8284 dequote_list (list);
8285
8286 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
8287 had_quoted_null = 1;
8288
8289 if (has_dollar_at)
8290 {
8291 quoted_dollar_at++;
8292 if (contains_dollar_at)
8293 *contains_dollar_at = 1;
8294 if (expanded_something)
8295 *expanded_something = 1;
8296 }
8297 }
8298 else
8299 {
8300 /* What we have is "". This is a minor optimization. */
8301 FREE (temp);
8302 list = (WORD_LIST *)NULL;
8303 }
8304
8305 /* The code above *might* return a list (consider the case of "$@",
8306 where it returns "$1", "$2", etc.). We can't throw away the
8307 rest of the list, and we have to make sure each word gets added
8308 as quoted. We test on tresult->next: if it is non-NULL, we
8309 quote the whole list, save it to a string with string_list, and
8310 add that string. We don't need to quote the results of this
8311 (and it would be wrong, since that would quote the separators
8312 as well), so we go directly to add_string. */
8313 if (list)
8314 {
8315 if (list->next)
8316 {
8317 #if 0
8318 if (quoted_dollar_at && (word->flags & W_NOSPLIT2))
8319 temp = string_list_internal (quote_list (list), " ");
8320 else
8321 #endif
8322 /* Testing quoted_dollar_at makes sure that "$@" is
8323 split correctly when $IFS does not contain a space. */
8324 temp = quoted_dollar_at
8325 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES)
8326 : string_list (quote_list (list));
8327 dispose_words (list);
8328 goto add_string;
8329 }
8330 else
8331 {
8332 temp = savestring (list->word->word);
8333 tflag = list->word->flags;
8334 dispose_words (list);
8335
8336 /* If the string is not a quoted null string, we want
8337 to remove any embedded unquoted CTLNUL characters.
8338 We do not want to turn quoted null strings back into
8339 the empty string, though. We do this because we
8340 want to remove any quoted nulls from expansions that
8341 contain other characters. For example, if we have
8342 x"$*"y or "x$*y" and there are no positional parameters,
8343 the $* should expand into nothing. */
8344 /* We use the W_HASQUOTEDNULL flag to differentiate the
8345 cases: a quoted null character as above and when
8346 CTLNUL is contained in the (non-null) expansion
8347 of some variable. We use the had_quoted_null flag to
8348 pass the value through this function to its caller. */
8349 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
8350 remove_quoted_nulls (temp); /* XXX */
8351 }
8352 }
8353 else
8354 temp = (char *)NULL;
8355
8356 /* We do not want to add quoted nulls to strings that are only
8357 partially quoted; we can throw them away. */
8358 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
8359 continue;
8360
8361 add_quoted_string:
8362
8363 if (temp)
8364 {
8365 temp1 = temp;
8366 temp = quote_string (temp);
8367 free (temp1);
8368 goto add_string;
8369 }
8370 else
8371 {
8372 /* Add NULL arg. */
8373 c = CTLNUL;
8374 sindex--; /* add_character: label increments sindex */
8375 goto add_character;
8376 }
8377
8378 /* break; */
8379
8380 case '\'':
8381 #if 0
8382 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE))
8383 #else
8384 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8385 #endif
8386 goto add_character;
8387
8388 t_index = ++sindex;
8389 temp = string_extract_single_quoted (string, &sindex);
8390
8391 /* If the entire STRING was surrounded by single quotes,
8392 then the string is wholly quoted. */
8393 quoted_state = (t_index == 1 && string[sindex] == '\0')
8394 ? WHOLLY_QUOTED
8395 : PARTIALLY_QUOTED;
8396
8397 /* If all we had was '', it is a null expansion. */
8398 if (*temp == '\0')
8399 {
8400 free (temp);
8401 temp = (char *)NULL;
8402 }
8403 else
8404 remove_quoted_escapes (temp); /* ??? */
8405
8406 /* We do not want to add quoted nulls to strings that are only
8407 partially quoted; such nulls are discarded. */
8408 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED))
8409 continue;
8410
8411 /* If we have a quoted null expansion, add a quoted NULL to istring. */
8412 if (temp == 0)
8413 {
8414 c = CTLNUL;
8415 sindex--; /* add_character: label increments sindex */
8416 goto add_character;
8417 }
8418 else
8419 goto add_quoted_string;
8420
8421 /* break; */
8422
8423 default:
8424 /* This is the fix for " $@ " */
8425 add_ifs_character:
8426 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c)))
8427 {
8428 if (string[sindex]) /* from old goto dollar_add_string */
8429 sindex++;
8430 if (c == 0)
8431 {
8432 c = CTLNUL;
8433 goto add_character;
8434 }
8435 else
8436 {
8437 #if HANDLE_MULTIBYTE
8438 if (MB_CUR_MAX > 1)
8439 sindex--;
8440
8441 if (MB_CUR_MAX > 1)
8442 {
8443 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8444 }
8445 else
8446 #endif
8447 {
8448 twochars[0] = CTLESC;
8449 twochars[1] = c;
8450 goto add_twochars;
8451 }
8452 }
8453 }
8454
8455 SADD_MBCHAR (temp, string, sindex, string_size);
8456
8457 add_character:
8458 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
8459 DEFAULT_ARRAY_SIZE);
8460 istring[istring_index++] = c;
8461 istring[istring_index] = '\0';
8462
8463 /* Next character. */
8464 sindex++;
8465 }
8466 }
8467
8468 finished_with_string:
8469 /* OK, we're ready to return. If we have a quoted string, and
8470 quoted_dollar_at is not set, we do no splitting at all; otherwise
8471 we split on ' '. The routines that call this will handle what to
8472 do if nothing has been expanded. */
8473
8474 /* Partially and wholly quoted strings which expand to the empty
8475 string are retained as an empty arguments. Unquoted strings
8476 which expand to the empty string are discarded. The single
8477 exception is the case of expanding "$@" when there are no
8478 positional parameters. In that case, we discard the expansion. */
8479
8480 /* Because of how the code that handles "" and '' in partially
8481 quoted strings works, we need to make ISTRING into a QUOTED_NULL
8482 if we saw quoting characters, but the expansion was empty.
8483 "" and '' are tossed away before we get to this point when
8484 processing partially quoted strings. This makes "" and $xxx""
8485 equivalent when xxx is unset. We also look to see whether we
8486 saw a quoted null from a ${} expansion and add one back if we
8487 need to. */
8488
8489 /* If we expand to nothing and there were no single or double quotes
8490 in the word, we throw it away. Otherwise, we return a NULL word.
8491 The single exception is for $@ surrounded by double quotes when
8492 there are no positional parameters. In that case, we also throw
8493 the word away. */
8494
8495 if (*istring == '\0')
8496 {
8497 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
8498 {
8499 istring[0] = CTLNUL;
8500 istring[1] = '\0';
8501 tword = make_bare_word (istring);
8502 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8503 list = make_word_list (tword, (WORD_LIST *)NULL);
8504 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8505 tword->flags |= W_QUOTED;
8506 }
8507 /* According to sh, ksh, and Posix.2, if a word expands into nothing
8508 and a double-quoted "$@" appears anywhere in it, then the entire
8509 word is removed. */
8510 else if (quoted_state == UNQUOTED || quoted_dollar_at)
8511 list = (WORD_LIST *)NULL;
8512 #if 0
8513 else
8514 {
8515 tword = make_bare_word (istring);
8516 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8517 tword->flags |= W_QUOTED;
8518 list = make_word_list (tword, (WORD_LIST *)NULL);
8519 }
8520 #else
8521 else
8522 list = (WORD_LIST *)NULL;
8523 #endif
8524 }
8525 else if (word->flags & W_NOSPLIT)
8526 {
8527 tword = make_bare_word (istring);
8528 if (word->flags & W_ASSIGNMENT)
8529 tword->flags |= W_ASSIGNMENT; /* XXX */
8530 if (word->flags & W_COMPASSIGN)
8531 tword->flags |= W_COMPASSIGN; /* XXX */
8532 if (word->flags & W_NOGLOB)
8533 tword->flags |= W_NOGLOB; /* XXX */
8534 if (word->flags & W_NOEXPAND)
8535 tword->flags |= W_NOEXPAND; /* XXX */
8536 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8537 tword->flags |= W_QUOTED;
8538 if (had_quoted_null)
8539 tword->flags |= W_HASQUOTEDNULL;
8540 list = make_word_list (tword, (WORD_LIST *)NULL);
8541 }
8542 else
8543 {
8544 char *ifs_chars;
8545
8546 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
8547
8548 /* If we have $@, we need to split the results no matter what. If
8549 IFS is unset or NULL, string_list_dollar_at has separated the
8550 positional parameters with a space, so we split on space (we have
8551 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
8552 string_list_dollar_at has separated the positional parameters
8553 with the first character of $IFS, so we split on $IFS. */
8554 if (has_dollar_at && ifs_chars)
8555 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
8556 else
8557 {
8558 tword = make_bare_word (istring);
8559 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
8560 tword->flags |= W_QUOTED;
8561 if (word->flags & W_ASSIGNMENT)
8562 tword->flags |= W_ASSIGNMENT;
8563 if (word->flags & W_COMPASSIGN)
8564 tword->flags |= W_COMPASSIGN;
8565 if (word->flags & W_NOGLOB)
8566 tword->flags |= W_NOGLOB;
8567 if (word->flags & W_NOEXPAND)
8568 tword->flags |= W_NOEXPAND;
8569 if (had_quoted_null)
8570 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8571 list = make_word_list (tword, (WORD_LIST *)NULL);
8572 }
8573 }
8574
8575 free (istring);
8576 return (list);
8577 }
8578
8579 /* **************************************************************** */
8580 /* */
8581 /* Functions for Quote Removal */
8582 /* */
8583 /* **************************************************************** */
8584
8585 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
8586 backslash quoting rules for within double quotes or a here document. */
8587 char *
8588 string_quote_removal (string, quoted)
8589 char *string;
8590 int quoted;
8591 {
8592 size_t slen;
8593 char *r, *result_string, *temp, *send;
8594 int sindex, tindex, dquote;
8595 unsigned char c;
8596 DECLARE_MBSTATE;
8597
8598 /* The result can be no longer than the original string. */
8599 slen = strlen (string);
8600 send = string + slen;
8601
8602 r = result_string = (char *)xmalloc (slen + 1);
8603
8604 for (dquote = sindex = 0; c = string[sindex];)
8605 {
8606 switch (c)
8607 {
8608 case '\\':
8609 c = string[++sindex];
8610 if (c == 0)
8611 {
8612 *r++ = '\\';
8613 break;
8614 }
8615 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
8616 *r++ = '\\';
8617 /* FALLTHROUGH */
8618
8619 default:
8620 SCOPY_CHAR_M (r, string, send, sindex);
8621 break;
8622
8623 case '\'':
8624 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
8625 {
8626 *r++ = c;
8627 sindex++;
8628 break;
8629 }
8630 tindex = sindex + 1;
8631 temp = string_extract_single_quoted (string, &tindex);
8632 if (temp)
8633 {
8634 strcpy (r, temp);
8635 r += strlen (r);
8636 free (temp);
8637 }
8638 sindex = tindex;
8639 break;
8640
8641 case '"':
8642 dquote = 1 - dquote;
8643 sindex++;
8644 break;
8645 }
8646 }
8647 *r = '\0';
8648 return (result_string);
8649 }
8650
8651 #if 0
8652 /* UNUSED */
8653 /* Perform quote removal on word WORD. This allocates and returns a new
8654 WORD_DESC *. */
8655 WORD_DESC *
8656 word_quote_removal (word, quoted)
8657 WORD_DESC *word;
8658 int quoted;
8659 {
8660 WORD_DESC *w;
8661 char *t;
8662
8663 t = string_quote_removal (word->word, quoted);
8664 w = alloc_word_desc ();
8665 w->word = t ? t : savestring ("");
8666 return (w);
8667 }
8668
8669 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
8670 the members of the list are treated as if they are surrounded by
8671 double quotes. Return a new list, or NULL if LIST is NULL. */
8672 WORD_LIST *
8673 word_list_quote_removal (list, quoted)
8674 WORD_LIST *list;
8675 int quoted;
8676 {
8677 WORD_LIST *result, *t, *tresult, *e;
8678
8679 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
8680 {
8681 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
8682 #if 0
8683 result = (WORD_LIST *) list_append (result, tresult);
8684 #else
8685 if (result == 0)
8686 result = e = tresult;
8687 else
8688 {
8689 e->next = tresult;
8690 while (e->next)
8691 e = e->next;
8692 }
8693 #endif
8694 }
8695 return (result);
8696 }
8697 #endif
8698
8699 /*******************************************
8700 * *
8701 * Functions to perform word splitting *
8702 * *
8703 *******************************************/
8704
8705 void
8706 setifs (v)
8707 SHELL_VAR *v;
8708 {
8709 char *t;
8710 unsigned char uc;
8711
8712 ifs_var = v;
8713 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
8714
8715 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
8716 handle multibyte chars in IFS */
8717 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
8718 for (t = ifs_value ; t && *t; t++)
8719 {
8720 uc = *t;
8721 ifs_cmap[uc] = 1;
8722 }
8723
8724 #if defined (HANDLE_MULTIBYTE)
8725 if (ifs_value == 0)
8726 {
8727 ifs_firstc[0] = '\0';
8728 ifs_firstc_len = 1;
8729 }
8730 else
8731 {
8732 size_t ifs_len;
8733 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
8734 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
8735 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
8736 {
8737 ifs_firstc[0] = ifs_value[0];
8738 ifs_firstc[1] = '\0';
8739 ifs_firstc_len = 1;
8740 }
8741 else
8742 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
8743 }
8744 #else
8745 ifs_firstc = ifs_value ? *ifs_value : 0;
8746 #endif
8747 }
8748
8749 char *
8750 getifs ()
8751 {
8752 return ifs_value;
8753 }
8754
8755 /* This splits a single word into a WORD LIST on $IFS, but only if the word
8756 is not quoted. list_string () performs quote removal for us, even if we
8757 don't do any splitting. */
8758 WORD_LIST *
8759 word_split (w, ifs_chars)
8760 WORD_DESC *w;
8761 char *ifs_chars;
8762 {
8763 WORD_LIST *result;
8764
8765 if (w)
8766 {
8767 char *xifs;
8768
8769 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
8770 result = list_string (w->word, xifs, w->flags & W_QUOTED);
8771 }
8772 else
8773 result = (WORD_LIST *)NULL;
8774
8775 return (result);
8776 }
8777
8778 /* Perform word splitting on LIST and return the RESULT. It is possible
8779 to return (WORD_LIST *)NULL. */
8780 static WORD_LIST *
8781 word_list_split (list)
8782 WORD_LIST *list;
8783 {
8784 WORD_LIST *result, *t, *tresult, *e;
8785
8786 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
8787 {
8788 tresult = word_split (t->word, ifs_value);
8789 if (result == 0)
8790 result = e = tresult;
8791 else
8792 {
8793 e->next = tresult;
8794 while (e->next)
8795 e = e->next;
8796 }
8797 }
8798 return (result);
8799 }
8800
8801 /**************************************************
8802 * *
8803 * Functions to expand an entire WORD_LIST *
8804 * *
8805 **************************************************/
8806
8807 /* Do any word-expansion-specific cleanup and jump to top_level */
8808 static void
8809 exp_jump_to_top_level (v)
8810 int v;
8811 {
8812 set_pipestatus_from_exit (last_command_exit_value);
8813
8814 /* Cleanup code goes here. */
8815 expand_no_split_dollar_star = 0; /* XXX */
8816 expanding_redir = 0;
8817 assigning_in_environment = 0;
8818
8819 if (parse_and_execute_level == 0)
8820 top_level_cleanup (); /* from sig.c */
8821
8822 jump_to_top_level (v);
8823 }
8824
8825 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
8826 ELIST, and set ELIST to the new list. */
8827 #define PREPEND_LIST(nlist, elist) \
8828 do { nlist->next = elist; elist = nlist; } while (0)
8829
8830 /* Separate out any initial variable assignments from TLIST. If set -k has
8831 been executed, remove all assignment statements from TLIST. Initial
8832 variable assignments and other environment assignments are placed
8833 on SUBST_ASSIGN_VARLIST. */
8834 static WORD_LIST *
8835 separate_out_assignments (tlist)
8836 WORD_LIST *tlist;
8837 {
8838 register WORD_LIST *vp, *lp;
8839
8840 if (tlist == 0)
8841 return ((WORD_LIST *)NULL);
8842
8843 if (subst_assign_varlist)
8844 dispose_words (subst_assign_varlist); /* Clean up after previous error */
8845
8846 subst_assign_varlist = (WORD_LIST *)NULL;
8847 vp = lp = tlist;
8848
8849 /* Separate out variable assignments at the start of the command.
8850 Loop invariant: vp->next == lp
8851 Loop postcondition:
8852 lp = list of words left after assignment statements skipped
8853 tlist = original list of words
8854 */
8855 while (lp && (lp->word->flags & W_ASSIGNMENT))
8856 {
8857 vp = lp;
8858 lp = lp->next;
8859 }
8860
8861 /* If lp != tlist, we have some initial assignment statements.
8862 We make SUBST_ASSIGN_VARLIST point to the list of assignment
8863 words and TLIST point to the remaining words. */
8864 if (lp != tlist)
8865 {
8866 subst_assign_varlist = tlist;
8867 /* ASSERT(vp->next == lp); */
8868 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
8869 tlist = lp; /* remainder of word list */
8870 }
8871
8872 /* vp == end of variable list */
8873 /* tlist == remainder of original word list without variable assignments */
8874 if (!tlist)
8875 /* All the words in tlist were assignment statements */
8876 return ((WORD_LIST *)NULL);
8877
8878 /* ASSERT(tlist != NULL); */
8879 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
8880
8881 /* If the -k option is in effect, we need to go through the remaining
8882 words, separate out the assignment words, and place them on
8883 SUBST_ASSIGN_VARLIST. */
8884 if (place_keywords_in_env)
8885 {
8886 WORD_LIST *tp; /* tp == running pointer into tlist */
8887
8888 tp = tlist;
8889 lp = tlist->next;
8890
8891 /* Loop Invariant: tp->next == lp */
8892 /* Loop postcondition: tlist == word list without assignment statements */
8893 while (lp)
8894 {
8895 if (lp->word->flags & W_ASSIGNMENT)
8896 {
8897 /* Found an assignment statement, add this word to end of
8898 subst_assign_varlist (vp). */
8899 if (!subst_assign_varlist)
8900 subst_assign_varlist = vp = lp;
8901 else
8902 {
8903 vp->next = lp;
8904 vp = lp;
8905 }
8906
8907 /* Remove the word pointed to by LP from TLIST. */
8908 tp->next = lp->next;
8909 /* ASSERT(vp == lp); */
8910 lp->next = (WORD_LIST *)NULL;
8911 lp = tp->next;
8912 }
8913 else
8914 {
8915 tp = lp;
8916 lp = lp->next;
8917 }
8918 }
8919 }
8920 return (tlist);
8921 }
8922
8923 #define WEXP_VARASSIGN 0x001
8924 #define WEXP_BRACEEXP 0x002
8925 #define WEXP_TILDEEXP 0x004
8926 #define WEXP_PARAMEXP 0x008
8927 #define WEXP_PATHEXP 0x010
8928
8929 /* All of the expansions, including variable assignments at the start of
8930 the list. */
8931 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
8932
8933 /* All of the expansions except variable assignments at the start of
8934 the list. */
8935 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
8936
8937 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
8938 expansion, command substitution, arithmetic expansion, word splitting, and
8939 quote removal. */
8940 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
8941
8942 /* Take the list of words in LIST and do the various substitutions. Return
8943 a new list of words which is the expanded list, and without things like
8944 variable assignments. */
8945
8946 WORD_LIST *
8947 expand_words (list)
8948 WORD_LIST *list;
8949 {
8950 return (expand_word_list_internal (list, WEXP_ALL));
8951 }
8952
8953 /* Same as expand_words (), but doesn't hack variable or environment
8954 variables. */
8955 WORD_LIST *
8956 expand_words_no_vars (list)
8957 WORD_LIST *list;
8958 {
8959 return (expand_word_list_internal (list, WEXP_NOVARS));
8960 }
8961
8962 WORD_LIST *
8963 expand_words_shellexp (list)
8964 WORD_LIST *list;
8965 {
8966 return (expand_word_list_internal (list, WEXP_SHELLEXP));
8967 }
8968
8969 static WORD_LIST *
8970 glob_expand_word_list (tlist, eflags)
8971 WORD_LIST *tlist;
8972 int eflags;
8973 {
8974 char **glob_array, *temp_string;
8975 register int glob_index;
8976 WORD_LIST *glob_list, *output_list, *disposables, *next;
8977 WORD_DESC *tword;
8978
8979 output_list = disposables = (WORD_LIST *)NULL;
8980 glob_array = (char **)NULL;
8981 while (tlist)
8982 {
8983 /* For each word, either globbing is attempted or the word is
8984 added to orig_list. If globbing succeeds, the results are
8985 added to orig_list and the word (tlist) is added to the list
8986 of disposable words. If globbing fails and failed glob
8987 expansions are left unchanged (the shell default), the
8988 original word is added to orig_list. If globbing fails and
8989 failed glob expansions are removed, the original word is
8990 added to the list of disposable words. orig_list ends up
8991 in reverse order and requires a call to REVERSE_LIST to
8992 be set right. After all words are examined, the disposable
8993 words are freed. */
8994 next = tlist->next;
8995
8996 /* If the word isn't an assignment and contains an unquoted
8997 pattern matching character, then glob it. */
8998 if ((tlist->word->flags & W_NOGLOB) == 0 &&
8999 unquoted_glob_pattern_p (tlist->word->word))
9000 {
9001 glob_array = shell_glob_filename (tlist->word->word);
9002
9003 /* Handle error cases.
9004 I don't think we should report errors like "No such file
9005 or directory". However, I would like to report errors
9006 like "Read failed". */
9007
9008 if (glob_array == 0 || GLOB_FAILED (glob_array))
9009 {
9010 glob_array = (char **)xmalloc (sizeof (char *));
9011 glob_array[0] = (char *)NULL;
9012 }
9013
9014 /* Dequote the current word in case we have to use it. */
9015 if (glob_array[0] == NULL)
9016 {
9017 temp_string = dequote_string (tlist->word->word);
9018 free (tlist->word->word);
9019 tlist->word->word = temp_string;
9020 }
9021
9022 /* Make the array into a word list. */
9023 glob_list = (WORD_LIST *)NULL;
9024 for (glob_index = 0; glob_array[glob_index]; glob_index++)
9025 {
9026 tword = make_bare_word (glob_array[glob_index]);
9027 tword->flags |= W_GLOBEXP; /* XXX */
9028 glob_list = make_word_list (tword, glob_list);
9029 }
9030
9031 if (glob_list)
9032 {
9033 output_list = (WORD_LIST *)list_append (glob_list, output_list);
9034 PREPEND_LIST (tlist, disposables);
9035 }
9036 else if (fail_glob_expansion != 0)
9037 {
9038 report_error (_("no match: %s"), tlist->word->word);
9039 exp_jump_to_top_level (DISCARD);
9040 }
9041 else if (allow_null_glob_expansion == 0)
9042 {
9043 /* Failed glob expressions are left unchanged. */
9044 PREPEND_LIST (tlist, output_list);
9045 }
9046 else
9047 {
9048 /* Failed glob expressions are removed. */
9049 PREPEND_LIST (tlist, disposables);
9050 }
9051 }
9052 else
9053 {
9054 /* Dequote the string. */
9055 temp_string = dequote_string (tlist->word->word);
9056 free (tlist->word->word);
9057 tlist->word->word = temp_string;
9058 PREPEND_LIST (tlist, output_list);
9059 }
9060
9061 strvec_dispose (glob_array);
9062 glob_array = (char **)NULL;
9063
9064 tlist = next;
9065 }
9066
9067 if (disposables)
9068 dispose_words (disposables);
9069
9070 if (output_list)
9071 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9072
9073 return (output_list);
9074 }
9075
9076 #if defined (BRACE_EXPANSION)
9077 static WORD_LIST *
9078 brace_expand_word_list (tlist, eflags)
9079 WORD_LIST *tlist;
9080 int eflags;
9081 {
9082 register char **expansions;
9083 char *temp_string;
9084 WORD_LIST *disposables, *output_list, *next;
9085 WORD_DESC *w;
9086 int eindex;
9087
9088 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
9089 {
9090 next = tlist->next;
9091
9092 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9093 {
9094 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
9095 PREPEND_LIST (tlist, output_list);
9096 continue;
9097 }
9098
9099 /* Only do brace expansion if the word has a brace character. If
9100 not, just add the word list element to BRACES and continue. In
9101 the common case, at least when running shell scripts, this will
9102 degenerate to a bunch of calls to `mbschr', and then what is
9103 basically a reversal of TLIST into BRACES, which is corrected
9104 by a call to REVERSE_LIST () on BRACES when the end of TLIST
9105 is reached. */
9106 if (mbschr (tlist->word->word, LBRACE))
9107 {
9108 expansions = brace_expand (tlist->word->word);
9109
9110 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
9111 {
9112 w = make_word (temp_string);
9113 /* If brace expansion didn't change the word, preserve
9114 the flags. We may want to preserve the flags
9115 unconditionally someday -- XXX */
9116 if (STREQ (temp_string, tlist->word->word))
9117 w->flags = tlist->word->flags;
9118 output_list = make_word_list (w, output_list);
9119 free (expansions[eindex]);
9120 }
9121 free (expansions);
9122
9123 /* Add TLIST to the list of words to be freed after brace
9124 expansion has been performed. */
9125 PREPEND_LIST (tlist, disposables);
9126 }
9127 else
9128 PREPEND_LIST (tlist, output_list);
9129 }
9130
9131 if (disposables)
9132 dispose_words (disposables);
9133
9134 if (output_list)
9135 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9136
9137 return (output_list);
9138 }
9139 #endif
9140
9141 #if defined (ARRAY_VARS)
9142 /* Take WORD, a compound associative array assignment, and internally run
9143 'declare -A w', where W is the variable name portion of WORD. */
9144 static int
9145 make_internal_declare (word, option)
9146 char *word;
9147 char *option;
9148 {
9149 int t;
9150 WORD_LIST *wl;
9151 WORD_DESC *w;
9152
9153 w = make_word (word);
9154
9155 t = assignment (w->word, 0);
9156 w->word[t] = '\0';
9157
9158 wl = make_word_list (w, (WORD_LIST *)NULL);
9159 wl = make_word_list (make_word (option), wl);
9160
9161 return (declare_builtin (wl));
9162 }
9163 #endif
9164
9165 static WORD_LIST *
9166 shell_expand_word_list (tlist, eflags)
9167 WORD_LIST *tlist;
9168 int eflags;
9169 {
9170 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list;
9171 int expanded_something, has_dollar_at;
9172 char *temp_string;
9173
9174 /* We do tilde expansion all the time. This is what 1003.2 says. */
9175 new_list = (WORD_LIST *)NULL;
9176 for (orig_list = tlist; tlist; tlist = next)
9177 {
9178 temp_string = tlist->word->word;
9179
9180 next = tlist->next;
9181
9182 #if defined (ARRAY_VARS)
9183 /* If this is a compound array assignment to a builtin that accepts
9184 such assignments (e.g., `declare'), take the assignment and perform
9185 it separately, handling the semantics of declarations inside shell
9186 functions. This avoids the double-evaluation of such arguments,
9187 because `declare' does some evaluation of compound assignments on
9188 its own. */
9189 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9190 {
9191 int t;
9192
9193 if (tlist->word->flags & W_ASSIGNASSOC)
9194 make_internal_declare (tlist->word->word, "-A");
9195
9196 t = do_word_assignment (tlist->word, 0);
9197 if (t == 0)
9198 {
9199 last_command_exit_value = EXECUTION_FAILURE;
9200 exp_jump_to_top_level (DISCARD);
9201 }
9202
9203 /* Now transform the word as ksh93 appears to do and go on */
9204 t = assignment (tlist->word->word, 0);
9205 tlist->word->word[t] = '\0';
9206 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC);
9207 }
9208 #endif
9209
9210 expanded_something = 0;
9211 expanded = expand_word_internal
9212 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
9213
9214 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
9215 {
9216 /* By convention, each time this error is returned,
9217 tlist->word->word has already been freed. */
9218 tlist->word->word = (char *)NULL;
9219
9220 /* Dispose our copy of the original list. */
9221 dispose_words (orig_list);
9222 /* Dispose the new list we're building. */
9223 dispose_words (new_list);
9224
9225 last_command_exit_value = EXECUTION_FAILURE;
9226 if (expanded == &expand_word_error)
9227 exp_jump_to_top_level (DISCARD);
9228 else
9229 exp_jump_to_top_level (FORCE_EOF);
9230 }
9231
9232 /* Don't split words marked W_NOSPLIT. */
9233 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
9234 {
9235 temp_list = word_list_split (expanded);
9236 dispose_words (expanded);
9237 }
9238 else
9239 {
9240 /* If no parameter expansion, command substitution, process
9241 substitution, or arithmetic substitution took place, then
9242 do not do word splitting. We still have to remove quoted
9243 null characters from the result. */
9244 word_list_remove_quoted_nulls (expanded);
9245 temp_list = expanded;
9246 }
9247
9248 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
9249 new_list = (WORD_LIST *)list_append (expanded, new_list);
9250 }
9251
9252 if (orig_list)
9253 dispose_words (orig_list);
9254
9255 if (new_list)
9256 new_list = REVERSE_LIST (new_list, WORD_LIST *);
9257
9258 return (new_list);
9259 }
9260
9261 /* The workhorse for expand_words () and expand_words_no_vars ().
9262 First arg is LIST, a WORD_LIST of words.
9263 Second arg EFLAGS is a flags word controlling which expansions are
9264 performed.
9265
9266 This does all of the substitutions: brace expansion, tilde expansion,
9267 parameter expansion, command substitution, arithmetic expansion,
9268 process substitution, word splitting, and pathname expansion, according
9269 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
9270 set, or for which no expansion is done, do not undergo word splitting.
9271 Words with the W_NOGLOB bit set do not undergo pathname expansion. */
9272 static WORD_LIST *
9273 expand_word_list_internal (list, eflags)
9274 WORD_LIST *list;
9275 int eflags;
9276 {
9277 WORD_LIST *new_list, *temp_list;
9278 int tint;
9279
9280 if (list == 0)
9281 return ((WORD_LIST *)NULL);
9282
9283 garglist = new_list = copy_word_list (list);
9284 if (eflags & WEXP_VARASSIGN)
9285 {
9286 garglist = new_list = separate_out_assignments (new_list);
9287 if (new_list == 0)
9288 {
9289 if (subst_assign_varlist)
9290 {
9291 /* All the words were variable assignments, so they are placed
9292 into the shell's environment. */
9293 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9294 {
9295 this_command_name = (char *)NULL; /* no arithmetic errors */
9296 tint = do_word_assignment (temp_list->word, 0);
9297 /* Variable assignment errors in non-interactive shells
9298 running in Posix.2 mode cause the shell to exit. */
9299 if (tint == 0)
9300 {
9301 last_command_exit_value = EXECUTION_FAILURE;
9302 if (interactive_shell == 0 && posixly_correct)
9303 exp_jump_to_top_level (FORCE_EOF);
9304 else
9305 exp_jump_to_top_level (DISCARD);
9306 }
9307 }
9308 dispose_words (subst_assign_varlist);
9309 subst_assign_varlist = (WORD_LIST *)NULL;
9310 }
9311 return ((WORD_LIST *)NULL);
9312 }
9313 }
9314
9315 /* Begin expanding the words that remain. The expansions take place on
9316 things that aren't really variable assignments. */
9317
9318 #if defined (BRACE_EXPANSION)
9319 /* Do brace expansion on this word if there are any brace characters
9320 in the string. */
9321 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
9322 new_list = brace_expand_word_list (new_list, eflags);
9323 #endif /* BRACE_EXPANSION */
9324
9325 /* Perform the `normal' shell expansions: tilde expansion, parameter and
9326 variable substitution, command substitution, arithmetic expansion,
9327 and word splitting. */
9328 new_list = shell_expand_word_list (new_list, eflags);
9329
9330 /* Okay, we're almost done. Now let's just do some filename
9331 globbing. */
9332 if (new_list)
9333 {
9334 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
9335 /* Glob expand the word list unless globbing has been disabled. */
9336 new_list = glob_expand_word_list (new_list, eflags);
9337 else
9338 /* Dequote the words, because we're not performing globbing. */
9339 new_list = dequote_list (new_list);
9340 }
9341
9342 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
9343 {
9344 sh_wassign_func_t *assign_func;
9345 int is_special_builtin, is_builtin_or_func;
9346
9347 /* If the remainder of the words expand to nothing, Posix.2 requires
9348 that the variable and environment assignments affect the shell's
9349 environment. */
9350 assign_func = new_list ? assign_in_env : do_word_assignment;
9351 tempenv_assign_error = 0;
9352
9353 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
9354 /* Posix says that special builtins exit if a variable assignment error
9355 occurs in an assignment preceding it. */
9356 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
9357
9358 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9359 {
9360 this_command_name = (char *)NULL;
9361 assigning_in_environment = (assign_func == assign_in_env);
9362 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
9363 assigning_in_environment = 0;
9364 /* Variable assignment errors in non-interactive shells running
9365 in Posix.2 mode cause the shell to exit. */
9366 if (tint == 0)
9367 {
9368 if (assign_func == do_word_assignment)
9369 {
9370 last_command_exit_value = EXECUTION_FAILURE;
9371 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
9372 exp_jump_to_top_level (FORCE_EOF);
9373 else
9374 exp_jump_to_top_level (DISCARD);
9375 }
9376 else
9377 tempenv_assign_error++;
9378 }
9379 }
9380
9381 dispose_words (subst_assign_varlist);
9382 subst_assign_varlist = (WORD_LIST *)NULL;
9383 }
9384
9385 #if 0
9386 tint = list_length (new_list) + 1;
9387 RESIZE_MALLOCED_BUFFER (glob_argv_flags, 0, tint, glob_argv_flags_size, 16);
9388 for (tint = 0, temp_list = new_list; temp_list; temp_list = temp_list->next)
9389 glob_argv_flags[tint++] = (temp_list->word->flags & W_GLOBEXP) ? '1' : '0';
9390 glob_argv_flags[tint] = '\0';
9391 #endif
9392
9393 return (new_list);
9394 }