]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
Bash-4.3 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) 1987-2013 Free Software Foundation, Inc.
8
9 This file is part of GNU Bash, the Bourne Again SHell.
10
11 Bash is free software: you can redistribute it and/or modify
12 it under the terms of the GNU General Public License as published by
13 the Free Software Foundation, either version 3 of the License, or
14 (at your option) any later version.
15
16 Bash is distributed in the hope that it will be useful,
17 but WITHOUT ANY WARRANTY; without even the implied warranty of
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 GNU General Public License for more details.
20
21 You should have received a copy of the GNU General Public License
22 along with Bash. If not, see <http://www.gnu.org/licenses/>.
23 */
24
25 #include "config.h"
26
27 #include "bashtypes.h"
28 #include <stdio.h>
29 #include "chartypes.h"
30 #if defined (HAVE_PWD_H)
31 # include <pwd.h>
32 #endif
33 #include <signal.h>
34 #include <errno.h>
35
36 #if defined (HAVE_UNISTD_H)
37 # include <unistd.h>
38 #endif
39
40 #include "bashansi.h"
41 #include "posixstat.h"
42 #include "bashintl.h"
43
44 #include "shell.h"
45 #include "parser.h"
46 #include "flags.h"
47 #include "jobs.h"
48 #include "execute_cmd.h"
49 #include "filecntl.h"
50 #include "trap.h"
51 #include "pathexp.h"
52 #include "mailcheck.h"
53
54 #include "shmbutil.h"
55 #include "typemax.h"
56
57 #include "builtins/getopt.h"
58 #include "builtins/common.h"
59
60 #include "builtins/builtext.h"
61
62 #include <tilde/tilde.h>
63 #include <glob/strmatch.h>
64
65 #if !defined (errno)
66 extern int errno;
67 #endif /* !errno */
68
69 /* The size that strings change by. */
70 #define DEFAULT_INITIAL_ARRAY_SIZE 112
71 #define DEFAULT_ARRAY_SIZE 128
72
73 /* Variable types. */
74 #define VT_VARIABLE 0
75 #define VT_POSPARMS 1
76 #define VT_ARRAYVAR 2
77 #define VT_ARRAYMEMBER 3
78 #define VT_ASSOCVAR 4
79
80 #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */
81
82 /* Flags for quoted_strchr */
83 #define ST_BACKSL 0x01
84 #define ST_CTLESC 0x02
85 #define ST_SQUOTE 0x04 /* unused yet */
86 #define ST_DQUOTE 0x08 /* unused yet */
87
88 /* Flags for the `pflags' argument to param_expand() */
89 #define PF_NOCOMSUB 0x01 /* Do not perform command substitution */
90 #define PF_IGNUNBOUND 0x02 /* ignore unbound vars even if -u set */
91 #define PF_NOSPLIT2 0x04 /* same as W_NOSPLIT2 */
92 #define PF_ASSIGNRHS 0x08 /* same as W_ASSIGNRHS */
93
94 /* These defs make it easier to use the editor. */
95 #define LBRACE '{'
96 #define RBRACE '}'
97 #define LPAREN '('
98 #define RPAREN ')'
99 #define LBRACK '['
100 #define RBRACK ']'
101
102 #if defined (HANDLE_MULTIBYTE)
103 #define WLPAREN L'('
104 #define WRPAREN L')'
105 #endif
106
107 /* Evaluates to 1 if C is one of the shell's special parameters whose length
108 can be taken, but is also one of the special expansion characters. */
109 #define VALID_SPECIAL_LENGTH_PARAM(c) \
110 ((c) == '-' || (c) == '?' || (c) == '#')
111
112 /* Evaluates to 1 if C is one of the shell's special parameters for which an
113 indirect variable reference may be made. */
114 #define VALID_INDIR_PARAM(c) \
115 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
116
117 /* Evaluates to 1 if C is one of the OP characters that follows the parameter
118 in ${parameter[:]OPword}. */
119 #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
120
121 /* Evaluates to 1 if this is one of the shell's special variables. */
122 #define SPECIAL_VAR(name, wi) \
123 ((DIGIT (*name) && all_digits (name)) || \
124 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
125 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1])))
126
127 /* An expansion function that takes a string and a quoted flag and returns
128 a WORD_LIST *. Used as the type of the third argument to
129 expand_string_if_necessary(). */
130 typedef WORD_LIST *EXPFUNC __P((char *, int));
131
132 /* Process ID of the last command executed within command substitution. */
133 pid_t last_command_subst_pid = NO_PID;
134 pid_t current_command_subst_pid = NO_PID;
135
136 /* Variables used to keep track of the characters in IFS. */
137 SHELL_VAR *ifs_var;
138 char *ifs_value;
139 unsigned char ifs_cmap[UCHAR_MAX + 1];
140 int ifs_is_set, ifs_is_null;
141
142 #if defined (HANDLE_MULTIBYTE)
143 unsigned char ifs_firstc[MB_LEN_MAX];
144 size_t ifs_firstc_len;
145 #else
146 unsigned char ifs_firstc;
147 #endif
148
149 /* Sentinel to tell when we are performing variable assignments preceding a
150 command name and putting them into the environment. Used to make sure
151 we use the temporary environment when looking up variable values. */
152 int assigning_in_environment;
153
154 /* Used to hold a list of variable assignments preceding a command. Global
155 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
156 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
157 WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
158
159 /* Extern functions and variables from different files. */
160 extern int last_command_exit_value, last_command_exit_signal;
161 extern int subshell_environment, line_number;
162 extern int subshell_level, parse_and_execute_level, sourcelevel;
163 extern int eof_encountered;
164 extern int return_catch_flag, return_catch_value;
165 extern pid_t dollar_dollar_pid;
166 extern int posixly_correct;
167 extern char *this_command_name;
168 extern struct fd_bitmap *current_fds_to_close;
169 extern int wordexp_only;
170 extern int expanding_redir;
171 extern int tempenv_assign_error;
172 extern int builtin_ignoring_errexit;
173
174 #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
175 extern wchar_t *wcsdup __P((const wchar_t *));
176 #endif
177
178 /* Non-zero means to allow unmatched globbed filenames to expand to
179 a null file. */
180 int allow_null_glob_expansion;
181
182 /* Non-zero means to throw an error when globbing fails to match anything. */
183 int fail_glob_expansion;
184
185 #if 0
186 /* Variables to keep track of which words in an expanded word list (the
187 output of expand_word_list_internal) are the result of globbing
188 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
189 (CURRENTLY UNUSED). */
190 char *glob_argv_flags;
191 static int glob_argv_flags_size;
192 #endif
193
194 static WORD_LIST expand_word_error, expand_word_fatal;
195 static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
196 static char expand_param_error, expand_param_fatal;
197 static char extract_string_error, extract_string_fatal;
198
199 /* Tell the expansion functions to not longjmp back to top_level on fatal
200 errors. Enabled when doing completion and prompt string expansion. */
201 static int no_longjmp_on_fatal_error = 0;
202
203 /* Set by expand_word_unsplit; used to inhibit splitting and re-joining
204 $* on $IFS, primarily when doing assignment statements. */
205 static int expand_no_split_dollar_star = 0;
206
207 /* A WORD_LIST of words to be expanded by expand_word_list_internal,
208 without any leading variable assignments. */
209 static WORD_LIST *garglist = (WORD_LIST *)NULL;
210
211 static char *quoted_substring __P((char *, int, int));
212 static int quoted_strlen __P((char *));
213 static char *quoted_strchr __P((char *, int, int));
214
215 static char *expand_string_if_necessary __P((char *, int, EXPFUNC *));
216 static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *));
217 static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
218 static WORD_LIST *expand_string_internal __P((char *, int));
219 static WORD_LIST *expand_string_leave_quoted __P((char *, int));
220 static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *));
221
222 static WORD_LIST *list_quote_escapes __P((WORD_LIST *));
223 static WORD_LIST *list_dequote_escapes __P((WORD_LIST *));
224
225 static char *make_quoted_char __P((int));
226 static WORD_LIST *quote_list __P((WORD_LIST *));
227
228 static int unquoted_substring __P((char *, char *));
229 static int unquoted_member __P((int, char *));
230
231 #if defined (ARRAY_VARS)
232 static SHELL_VAR *do_compound_assignment __P((char *, char *, int));
233 #endif
234 static int do_assignment_internal __P((const WORD_DESC *, int));
235
236 static char *string_extract_verbatim __P((char *, size_t, int *, char *, int));
237 static char *string_extract __P((char *, int *, char *, int));
238 static char *string_extract_double_quoted __P((char *, int *, int));
239 static inline char *string_extract_single_quoted __P((char *, int *));
240 static inline int skip_single_quoted __P((const char *, size_t, int));
241 static int skip_double_quoted __P((char *, size_t, int));
242 static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int));
243 static char *extract_dollar_brace_string __P((char *, int *, int, int));
244 static int skip_matched_pair __P((const char *, int, int, int, int));
245
246 static char *pos_params __P((char *, int, int, int));
247
248 static unsigned char *mb_getcharlens __P((char *, int));
249
250 static char *remove_upattern __P((char *, char *, int));
251 #if defined (HANDLE_MULTIBYTE)
252 static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int));
253 #endif
254 static char *remove_pattern __P((char *, char *, int));
255
256 static int match_upattern __P((char *, char *, int, char **, char **));
257 #if defined (HANDLE_MULTIBYTE)
258 static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
259 #endif
260 static int match_pattern __P((char *, char *, int, char **, char **));
261 static int getpatspec __P((int, char *));
262 static char *getpattern __P((char *, int, int));
263 static char *variable_remove_pattern __P((char *, char *, int, int));
264 static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int));
265 static char *parameter_list_remove_pattern __P((int, char *, int, int));
266 #ifdef ARRAY_VARS
267 static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int));
268 #endif
269 static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int));
270
271 static char *process_substitute __P((char *, int));
272
273 static char *read_comsub __P((int, int, int *));
274
275 #ifdef ARRAY_VARS
276 static arrayind_t array_length_reference __P((char *));
277 #endif
278
279 static int valid_brace_expansion_word __P((char *, int));
280 static int chk_atstar __P((char *, int, int *, int *));
281 static int chk_arithsub __P((const char *, int));
282
283 static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *));
284 static char *parameter_brace_find_indir __P((char *, int, int, int));
285 static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *));
286 static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *));
287 static void parameter_brace_expand_error __P((char *, char *));
288
289 static int valid_length_expression __P((char *));
290 static intmax_t parameter_brace_expand_length __P((char *));
291
292 static char *skiparith __P((char *, int));
293 static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
294 static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
295 static char *mb_substring __P((char *, int, int));
296 static char *parameter_brace_substring __P((char *, char *, int, char *, int, int));
297
298 static int shouldexp_replacement __P((char *));
299
300 static char *pos_params_pat_subst __P((char *, char *, char *, int));
301
302 static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int));
303
304 static char *pos_params_casemod __P((char *, char *, int, int));
305 static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int));
306
307 static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *));
308 static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int));
309
310 static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
311
312 static WORD_LIST *word_list_split __P((WORD_LIST *));
313
314 static void exp_jump_to_top_level __P((int));
315
316 static WORD_LIST *separate_out_assignments __P((WORD_LIST *));
317 static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int));
318 #ifdef BRACE_EXPANSION
319 static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int));
320 #endif
321 #if defined (ARRAY_VARS)
322 static int make_internal_declare __P((char *, char *));
323 #endif
324 static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int));
325 static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int));
326
327 /* **************************************************************** */
328 /* */
329 /* Utility Functions */
330 /* */
331 /* **************************************************************** */
332
333 #if defined (DEBUG)
334 void
335 dump_word_flags (flags)
336 int flags;
337 {
338 int f;
339
340 f = flags;
341 fprintf (stderr, "%d -> ", f);
342 if (f & W_ASSIGNASSOC)
343 {
344 f &= ~W_ASSIGNASSOC;
345 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
346 }
347 if (f & W_ASSIGNARRAY)
348 {
349 f &= ~W_ASSIGNARRAY;
350 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
351 }
352 if (f & W_HASCTLESC)
353 {
354 f &= ~W_HASCTLESC;
355 fprintf (stderr, "W_HASCTLESC%s", f ? "|" : "");
356 }
357 if (f & W_NOPROCSUB)
358 {
359 f &= ~W_NOPROCSUB;
360 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
361 }
362 if (f & W_DQUOTE)
363 {
364 f &= ~W_DQUOTE;
365 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
366 }
367 if (f & W_HASQUOTEDNULL)
368 {
369 f &= ~W_HASQUOTEDNULL;
370 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
371 }
372 if (f & W_ASSIGNARG)
373 {
374 f &= ~W_ASSIGNARG;
375 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
376 }
377 if (f & W_ASSNBLTIN)
378 {
379 f &= ~W_ASSNBLTIN;
380 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
381 }
382 if (f & W_ASSNGLOBAL)
383 {
384 f &= ~W_ASSNGLOBAL;
385 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
386 }
387 if (f & W_ASSIGNINT)
388 {
389 f &= ~W_ASSIGNINT;
390 fprintf (stderr, "W_ASSIGNINT%s", f ? "|" : "");
391 }
392 if (f & W_COMPASSIGN)
393 {
394 f &= ~W_COMPASSIGN;
395 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
396 }
397 if (f & W_NOEXPAND)
398 {
399 f &= ~W_NOEXPAND;
400 fprintf (stderr, "W_NOEXPAND%s", f ? "|" : "");
401 }
402 if (f & W_ITILDE)
403 {
404 f &= ~W_ITILDE;
405 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
406 }
407 if (f & W_NOTILDE)
408 {
409 f &= ~W_NOTILDE;
410 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
411 }
412 if (f & W_ASSIGNRHS)
413 {
414 f &= ~W_ASSIGNRHS;
415 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
416 }
417 if (f & W_NOCOMSUB)
418 {
419 f &= ~W_NOCOMSUB;
420 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
421 }
422 if (f & W_DOLLARSTAR)
423 {
424 f &= ~W_DOLLARSTAR;
425 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
426 }
427 if (f & W_DOLLARAT)
428 {
429 f &= ~W_DOLLARAT;
430 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
431 }
432 if (f & W_TILDEEXP)
433 {
434 f &= ~W_TILDEEXP;
435 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
436 }
437 if (f & W_NOSPLIT2)
438 {
439 f &= ~W_NOSPLIT2;
440 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
441 }
442 if (f & W_NOSPLIT)
443 {
444 f &= ~W_NOSPLIT;
445 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
446 }
447 if (f & W_NOBRACE)
448 {
449 f &= ~W_NOBRACE;
450 fprintf (stderr, "W_NOBRACE%s", f ? "|" : "");
451 }
452 if (f & W_NOGLOB)
453 {
454 f &= ~W_NOGLOB;
455 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
456 }
457 if (f & W_SPLITSPACE)
458 {
459 f &= ~W_SPLITSPACE;
460 fprintf (stderr, "W_SPLITSPACE%s", f ? "|" : "");
461 }
462 if (f & W_ASSIGNMENT)
463 {
464 f &= ~W_ASSIGNMENT;
465 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
466 }
467 if (f & W_QUOTED)
468 {
469 f &= ~W_QUOTED;
470 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
471 }
472 if (f & W_HASDOLLAR)
473 {
474 f &= ~W_HASDOLLAR;
475 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
476 }
477 fprintf (stderr, "\n");
478 fflush (stderr);
479 }
480 #endif
481
482 #ifdef INCLUDE_UNUSED
483 static char *
484 quoted_substring (string, start, end)
485 char *string;
486 int start, end;
487 {
488 register int len, l;
489 register char *result, *s, *r;
490
491 len = end - start;
492
493 /* Move to string[start], skipping quoted characters. */
494 for (s = string, l = 0; *s && l < start; )
495 {
496 if (*s == CTLESC)
497 {
498 s++;
499 continue;
500 }
501 l++;
502 if (*s == 0)
503 break;
504 }
505
506 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
507
508 /* Copy LEN characters, including quote characters. */
509 s = string + l;
510 for (l = 0; l < len; s++)
511 {
512 if (*s == CTLESC)
513 *r++ = *s++;
514 *r++ = *s;
515 l++;
516 if (*s == 0)
517 break;
518 }
519 *r = '\0';
520 return result;
521 }
522 #endif
523
524 #ifdef INCLUDE_UNUSED
525 /* Return the length of S, skipping over quoted characters */
526 static int
527 quoted_strlen (s)
528 char *s;
529 {
530 register char *p;
531 int i;
532
533 i = 0;
534 for (p = s; *p; p++)
535 {
536 if (*p == CTLESC)
537 {
538 p++;
539 if (*p == 0)
540 return (i + 1);
541 }
542 i++;
543 }
544
545 return i;
546 }
547 #endif
548
549 /* Find the first occurrence of character C in string S, obeying shell
550 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
551 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
552 escaped with CTLESC are skipped. */
553 static char *
554 quoted_strchr (s, c, flags)
555 char *s;
556 int c, flags;
557 {
558 register char *p;
559
560 for (p = s; *p; p++)
561 {
562 if (((flags & ST_BACKSL) && *p == '\\')
563 || ((flags & ST_CTLESC) && *p == CTLESC))
564 {
565 p++;
566 if (*p == '\0')
567 return ((char *)NULL);
568 continue;
569 }
570 else if (*p == c)
571 return p;
572 }
573 return ((char *)NULL);
574 }
575
576 /* Return 1 if CHARACTER appears in an unquoted portion of
577 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
578 static int
579 unquoted_member (character, string)
580 int character;
581 char *string;
582 {
583 size_t slen;
584 int sindex, c;
585 DECLARE_MBSTATE;
586
587 slen = strlen (string);
588 sindex = 0;
589 while (c = string[sindex])
590 {
591 if (c == character)
592 return (1);
593
594 switch (c)
595 {
596 default:
597 ADVANCE_CHAR (string, slen, sindex);
598 break;
599
600 case '\\':
601 sindex++;
602 if (string[sindex])
603 ADVANCE_CHAR (string, slen, sindex);
604 break;
605
606 case '\'':
607 sindex = skip_single_quoted (string, slen, ++sindex);
608 break;
609
610 case '"':
611 sindex = skip_double_quoted (string, slen, ++sindex);
612 break;
613 }
614 }
615 return (0);
616 }
617
618 /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
619 static int
620 unquoted_substring (substr, string)
621 char *substr, *string;
622 {
623 size_t slen;
624 int sindex, c, sublen;
625 DECLARE_MBSTATE;
626
627 if (substr == 0 || *substr == '\0')
628 return (0);
629
630 slen = strlen (string);
631 sublen = strlen (substr);
632 for (sindex = 0; c = string[sindex]; )
633 {
634 if (STREQN (string + sindex, substr, sublen))
635 return (1);
636
637 switch (c)
638 {
639 case '\\':
640 sindex++;
641 if (string[sindex])
642 ADVANCE_CHAR (string, slen, sindex);
643 break;
644
645 case '\'':
646 sindex = skip_single_quoted (string, slen, ++sindex);
647 break;
648
649 case '"':
650 sindex = skip_double_quoted (string, slen, ++sindex);
651 break;
652
653 default:
654 ADVANCE_CHAR (string, slen, sindex);
655 break;
656 }
657 }
658 return (0);
659 }
660
661 /* Most of the substitutions must be done in parallel. In order
662 to avoid using tons of unclear goto's, I have some functions
663 for manipulating malloc'ed strings. They all take INDX, a
664 pointer to an integer which is the offset into the string
665 where manipulation is taking place. They also take SIZE, a
666 pointer to an integer which is the current length of the
667 character array for this string. */
668
669 /* Append SOURCE to TARGET at INDEX. SIZE is the current amount
670 of space allocated to TARGET. SOURCE can be NULL, in which
671 case nothing happens. Gets rid of SOURCE by freeing it.
672 Returns TARGET in case the location has changed. */
673 INLINE char *
674 sub_append_string (source, target, indx, size)
675 char *source, *target;
676 int *indx, *size;
677 {
678 if (source)
679 {
680 int srclen, n;
681
682 srclen = STRLEN (source);
683 if (srclen >= (int)(*size - *indx))
684 {
685 n = srclen + *indx;
686 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
687 target = (char *)xrealloc (target, (*size = n));
688 }
689
690 FASTCOPY (source, target + *indx, srclen);
691 *indx += srclen;
692 target[*indx] = '\0';
693
694 free (source);
695 }
696 return (target);
697 }
698
699 #if 0
700 /* UNUSED */
701 /* Append the textual representation of NUMBER to TARGET.
702 INDX and SIZE are as in SUB_APPEND_STRING. */
703 char *
704 sub_append_number (number, target, indx, size)
705 intmax_t number;
706 int *indx, *size;
707 char *target;
708 {
709 char *temp;
710
711 temp = itos (number);
712 return (sub_append_string (temp, target, indx, size));
713 }
714 #endif
715
716 /* Extract a substring from STRING, starting at SINDEX and ending with
717 one of the characters in CHARLIST. Don't make the ending character
718 part of the string. Leave SINDEX pointing at the ending character.
719 Understand about backslashes in the string. If (flags & SX_VARNAME)
720 is non-zero, and array variables have been compiled into the shell,
721 everything between a `[' and a corresponding `]' is skipped over.
722 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
723 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
724 contain a closing character from CHARLIST. */
725 static char *
726 string_extract (string, sindex, charlist, flags)
727 char *string;
728 int *sindex;
729 char *charlist;
730 int flags;
731 {
732 register int c, i;
733 int found;
734 size_t slen;
735 char *temp;
736 DECLARE_MBSTATE;
737
738 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
739 i = *sindex;
740 found = 0;
741 while (c = string[i])
742 {
743 if (c == '\\')
744 {
745 if (string[i + 1])
746 i++;
747 else
748 break;
749 }
750 #if defined (ARRAY_VARS)
751 else if ((flags & SX_VARNAME) && c == '[')
752 {
753 int ni;
754 /* If this is an array subscript, skip over it and continue. */
755 ni = skipsubscript (string, i, 0);
756 if (string[ni] == ']')
757 i = ni;
758 }
759 #endif
760 else if (MEMBER (c, charlist))
761 {
762 found = 1;
763 break;
764 }
765
766 ADVANCE_CHAR (string, slen, i);
767 }
768
769 /* If we had to have a matching delimiter and didn't find one, return an
770 error and let the caller deal with it. */
771 if ((flags & SX_REQMATCH) && found == 0)
772 {
773 *sindex = i;
774 return (&extract_string_error);
775 }
776
777 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
778 *sindex = i;
779
780 return (temp);
781 }
782
783 /* Extract the contents of STRING as if it is enclosed in double quotes.
784 SINDEX, when passed in, is the offset of the character immediately
785 following the opening double quote; on exit, SINDEX is left pointing after
786 the closing double quote. If STRIPDQ is non-zero, unquoted double
787 quotes are stripped and the string is terminated by a null byte.
788 Backslashes between the embedded double quotes are processed. If STRIPDQ
789 is zero, an unquoted `"' terminates the string. */
790 static char *
791 string_extract_double_quoted (string, sindex, stripdq)
792 char *string;
793 int *sindex, stripdq;
794 {
795 size_t slen;
796 char *send;
797 int j, i, t;
798 unsigned char c;
799 char *temp, *ret; /* The new string we return. */
800 int pass_next, backquote, si; /* State variables for the machine. */
801 int dquote;
802 DECLARE_MBSTATE;
803
804 slen = strlen (string + *sindex) + *sindex;
805 send = string + slen;
806
807 pass_next = backquote = dquote = 0;
808 temp = (char *)xmalloc (1 + slen - *sindex);
809
810 j = 0;
811 i = *sindex;
812 while (c = string[i])
813 {
814 /* Process a character that was quoted by a backslash. */
815 if (pass_next)
816 {
817 /* XXX - take another look at this in light of Interp 221 */
818 /* Posix.2 sez:
819
820 ``The backslash shall retain its special meaning as an escape
821 character only when followed by one of the characters:
822 $ ` " \ <newline>''.
823
824 If STRIPDQ is zero, we handle the double quotes here and let
825 expand_word_internal handle the rest. If STRIPDQ is non-zero,
826 we have already been through one round of backslash stripping,
827 and want to strip these backslashes only if DQUOTE is non-zero,
828 indicating that we are inside an embedded double-quoted string. */
829
830 /* If we are in an embedded quoted string, then don't strip
831 backslashes before characters for which the backslash
832 retains its special meaning, but remove backslashes in
833 front of other characters. If we are not in an
834 embedded quoted string, don't strip backslashes at all.
835 This mess is necessary because the string was already
836 surrounded by double quotes (and sh has some really weird
837 quoting rules).
838 The returned string will be run through expansion as if
839 it were double-quoted. */
840 if ((stripdq == 0 && c != '"') ||
841 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
842 temp[j++] = '\\';
843 pass_next = 0;
844
845 add_one_character:
846 COPY_CHAR_I (temp, j, string, send, i);
847 continue;
848 }
849
850 /* A backslash protects the next character. The code just above
851 handles preserving the backslash in front of any character but
852 a double quote. */
853 if (c == '\\')
854 {
855 pass_next++;
856 i++;
857 continue;
858 }
859
860 /* Inside backquotes, ``the portion of the quoted string from the
861 initial backquote and the characters up to the next backquote
862 that is not preceded by a backslash, having escape characters
863 removed, defines that command''. */
864 if (backquote)
865 {
866 if (c == '`')
867 backquote = 0;
868 temp[j++] = c;
869 i++;
870 continue;
871 }
872
873 if (c == '`')
874 {
875 temp[j++] = c;
876 backquote++;
877 i++;
878 continue;
879 }
880
881 /* Pass everything between `$(' and the matching `)' or a quoted
882 ${ ... } pair through according to the Posix.2 specification. */
883 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
884 {
885 int free_ret = 1;
886
887 si = i + 2;
888 if (string[i + 1] == LPAREN)
889 ret = extract_command_subst (string, &si, 0);
890 else
891 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
892
893 temp[j++] = '$';
894 temp[j++] = string[i + 1];
895
896 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
897 is set. */
898 if (ret == 0 && no_longjmp_on_fatal_error)
899 {
900 free_ret = 0;
901 ret = string + i + 2;
902 }
903
904 for (t = 0; ret[t]; t++, j++)
905 temp[j] = ret[t];
906 temp[j] = string[si];
907
908 if (string[si])
909 {
910 j++;
911 i = si + 1;
912 }
913 else
914 i = si;
915
916 if (free_ret)
917 free (ret);
918 continue;
919 }
920
921 /* Add any character but a double quote to the quoted string we're
922 accumulating. */
923 if (c != '"')
924 goto add_one_character;
925
926 /* c == '"' */
927 if (stripdq)
928 {
929 dquote ^= 1;
930 i++;
931 continue;
932 }
933
934 break;
935 }
936 temp[j] = '\0';
937
938 /* Point to after the closing quote. */
939 if (c)
940 i++;
941 *sindex = i;
942
943 return (temp);
944 }
945
946 /* This should really be another option to string_extract_double_quoted. */
947 static int
948 skip_double_quoted (string, slen, sind)
949 char *string;
950 size_t slen;
951 int sind;
952 {
953 int c, i;
954 char *ret;
955 int pass_next, backquote, si;
956 DECLARE_MBSTATE;
957
958 pass_next = backquote = 0;
959 i = sind;
960 while (c = string[i])
961 {
962 if (pass_next)
963 {
964 pass_next = 0;
965 ADVANCE_CHAR (string, slen, i);
966 continue;
967 }
968 else if (c == '\\')
969 {
970 pass_next++;
971 i++;
972 continue;
973 }
974 else if (backquote)
975 {
976 if (c == '`')
977 backquote = 0;
978 ADVANCE_CHAR (string, slen, i);
979 continue;
980 }
981 else if (c == '`')
982 {
983 backquote++;
984 i++;
985 continue;
986 }
987 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
988 {
989 si = i + 2;
990 if (string[i + 1] == LPAREN)
991 ret = extract_command_subst (string, &si, SX_NOALLOC);
992 else
993 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
994
995 i = si + 1;
996 continue;
997 }
998 else if (c != '"')
999 {
1000 ADVANCE_CHAR (string, slen, i);
1001 continue;
1002 }
1003 else
1004 break;
1005 }
1006
1007 if (c)
1008 i++;
1009
1010 return (i);
1011 }
1012
1013 /* Extract the contents of STRING as if it is enclosed in single quotes.
1014 SINDEX, when passed in, is the offset of the character immediately
1015 following the opening single quote; on exit, SINDEX is left pointing after
1016 the closing single quote. */
1017 static inline char *
1018 string_extract_single_quoted (string, sindex)
1019 char *string;
1020 int *sindex;
1021 {
1022 register int i;
1023 size_t slen;
1024 char *t;
1025 DECLARE_MBSTATE;
1026
1027 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1028 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
1029 i = *sindex;
1030 while (string[i] && string[i] != '\'')
1031 ADVANCE_CHAR (string, slen, i);
1032
1033 t = substring (string, *sindex, i);
1034
1035 if (string[i])
1036 i++;
1037 *sindex = i;
1038
1039 return (t);
1040 }
1041
1042 static inline int
1043 skip_single_quoted (string, slen, sind)
1044 const char *string;
1045 size_t slen;
1046 int sind;
1047 {
1048 register int c;
1049 DECLARE_MBSTATE;
1050
1051 c = sind;
1052 while (string[c] && string[c] != '\'')
1053 ADVANCE_CHAR (string, slen, c);
1054
1055 if (string[c])
1056 c++;
1057 return c;
1058 }
1059
1060 /* Just like string_extract, but doesn't hack backslashes or any of
1061 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
1062 static char *
1063 string_extract_verbatim (string, slen, sindex, charlist, flags)
1064 char *string;
1065 size_t slen;
1066 int *sindex;
1067 char *charlist;
1068 int flags;
1069 {
1070 register int i;
1071 #if defined (HANDLE_MULTIBYTE)
1072 size_t clen;
1073 wchar_t *wcharlist;
1074 #endif
1075 int c;
1076 char *temp;
1077 DECLARE_MBSTATE;
1078
1079 if (charlist[0] == '\'' && charlist[1] == '\0')
1080 {
1081 temp = string_extract_single_quoted (string, sindex);
1082 --*sindex; /* leave *sindex at separator character */
1083 return temp;
1084 }
1085
1086 i = *sindex;
1087 #if 0
1088 /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need
1089 this only if MB_CUR_MAX > 1. */
1090 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1;
1091 #endif
1092 #if defined (HANDLE_MULTIBYTE)
1093 clen = strlen (charlist);
1094 wcharlist = 0;
1095 #endif
1096 while (c = string[i])
1097 {
1098 #if defined (HANDLE_MULTIBYTE)
1099 size_t mblength;
1100 #endif
1101 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1102 {
1103 i += 2;
1104 continue;
1105 }
1106 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1107 through, to protect the CTLNULs from later calls to
1108 remove_quoted_nulls. */
1109 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
1110 {
1111 i += 2;
1112 continue;
1113 }
1114
1115 #if defined (HANDLE_MULTIBYTE)
1116 mblength = MBLEN (string + i, slen - i);
1117 if (mblength > 1)
1118 {
1119 wchar_t wc;
1120 mblength = mbtowc (&wc, string + i, slen - i);
1121 if (MB_INVALIDCH (mblength))
1122 {
1123 if (MEMBER (c, charlist))
1124 break;
1125 }
1126 else
1127 {
1128 if (wcharlist == 0)
1129 {
1130 size_t len;
1131 len = mbstowcs (wcharlist, charlist, 0);
1132 if (len == -1)
1133 len = 0;
1134 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1135 mbstowcs (wcharlist, charlist, len + 1);
1136 }
1137
1138 if (wcschr (wcharlist, wc))
1139 break;
1140 }
1141 }
1142 else
1143 #endif
1144 if (MEMBER (c, charlist))
1145 break;
1146
1147 ADVANCE_CHAR (string, slen, i);
1148 }
1149
1150 #if defined (HANDLE_MULTIBYTE)
1151 FREE (wcharlist);
1152 #endif
1153
1154 temp = substring (string, *sindex, i);
1155 *sindex = i;
1156
1157 return (temp);
1158 }
1159
1160 /* Extract the $( construct in STRING, and return a new string.
1161 Start extracting at (SINDEX) as if we had just seen "$(".
1162 Make (SINDEX) get the position of the matching ")". )
1163 XFLAGS is additional flags to pass to other extraction functions. */
1164 char *
1165 extract_command_subst (string, sindex, xflags)
1166 char *string;
1167 int *sindex;
1168 int xflags;
1169 {
1170 if (string[*sindex] == LPAREN)
1171 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1172 else
1173 {
1174 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1175 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1176 }
1177 }
1178
1179 /* Extract the $[ construct in STRING, and return a new string. (])
1180 Start extracting at (SINDEX) as if we had just seen "$[".
1181 Make (SINDEX) get the position of the matching "]". */
1182 char *
1183 extract_arithmetic_subst (string, sindex)
1184 char *string;
1185 int *sindex;
1186 {
1187 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
1188 }
1189
1190 #if defined (PROCESS_SUBSTITUTION)
1191 /* Extract the <( or >( construct in STRING, and return a new string.
1192 Start extracting at (SINDEX) as if we had just seen "<(".
1193 Make (SINDEX) get the position of the matching ")". */ /*))*/
1194 char *
1195 extract_process_subst (string, starter, sindex)
1196 char *string;
1197 char *starter;
1198 int *sindex;
1199 {
1200 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
1201 }
1202 #endif /* PROCESS_SUBSTITUTION */
1203
1204 #if defined (ARRAY_VARS)
1205 /* This can be fooled by unquoted right parens in the passed string. If
1206 each caller verifies that the last character in STRING is a right paren,
1207 we don't even need to call extract_delimited_string. */
1208 char *
1209 extract_array_assignment_list (string, sindex)
1210 char *string;
1211 int *sindex;
1212 {
1213 int slen;
1214 char *ret;
1215
1216 slen = strlen (string); /* ( */
1217 if (string[slen - 1] == ')')
1218 {
1219 ret = substring (string, *sindex, slen - 1);
1220 *sindex = slen - 1;
1221 return ret;
1222 }
1223 return 0;
1224 }
1225 #endif
1226
1227 /* Extract and create a new string from the contents of STRING, a
1228 character string delimited with OPENER and CLOSER. SINDEX is
1229 the address of an int describing the current offset in STRING;
1230 it should point to just after the first OPENER found. On exit,
1231 SINDEX gets the position of the last character of the matching CLOSER.
1232 If OPENER is more than a single character, ALT_OPENER, if non-null,
1233 contains a character string that can also match CLOSER and thus
1234 needs to be skipped. */
1235 static char *
1236 extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
1237 char *string;
1238 int *sindex;
1239 char *opener, *alt_opener, *closer;
1240 int flags;
1241 {
1242 int i, c, si;
1243 size_t slen;
1244 char *t, *result;
1245 int pass_character, nesting_level, in_comment;
1246 int len_closer, len_opener, len_alt_opener;
1247 DECLARE_MBSTATE;
1248
1249 slen = strlen (string + *sindex) + *sindex;
1250 len_opener = STRLEN (opener);
1251 len_alt_opener = STRLEN (alt_opener);
1252 len_closer = STRLEN (closer);
1253
1254 pass_character = in_comment = 0;
1255
1256 nesting_level = 1;
1257 i = *sindex;
1258
1259 while (nesting_level)
1260 {
1261 c = string[i];
1262
1263 if (c == 0)
1264 break;
1265
1266 if (in_comment)
1267 {
1268 if (c == '\n')
1269 in_comment = 0;
1270 ADVANCE_CHAR (string, slen, i);
1271 continue;
1272 }
1273
1274 if (pass_character) /* previous char was backslash */
1275 {
1276 pass_character = 0;
1277 ADVANCE_CHAR (string, slen, i);
1278 continue;
1279 }
1280
1281 /* Not exactly right yet; should handle shell metacharacters and
1282 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
1283 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
1284 {
1285 in_comment = 1;
1286 ADVANCE_CHAR (string, slen, i);
1287 continue;
1288 }
1289
1290 if (c == CTLESC || c == '\\')
1291 {
1292 pass_character++;
1293 i++;
1294 continue;
1295 }
1296
1297 /* Process a nested command substitution, but only if we're parsing an
1298 arithmetic substitution. */
1299 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1300 {
1301 si = i + 2;
1302 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1303 i = si + 1;
1304 continue;
1305 }
1306
1307 /* Process a nested OPENER. */
1308 if (STREQN (string + i, opener, len_opener))
1309 {
1310 si = i + len_opener;
1311 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
1312 i = si + 1;
1313 continue;
1314 }
1315
1316 /* Process a nested ALT_OPENER */
1317 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
1318 {
1319 si = i + len_alt_opener;
1320 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
1321 i = si + 1;
1322 continue;
1323 }
1324
1325 /* If the current substring terminates the delimited string, decrement
1326 the nesting level. */
1327 if (STREQN (string + i, closer, len_closer))
1328 {
1329 i += len_closer - 1; /* move to last byte of the closer */
1330 nesting_level--;
1331 if (nesting_level == 0)
1332 break;
1333 }
1334
1335 /* Pass old-style command substitution through verbatim. */
1336 if (c == '`')
1337 {
1338 si = i + 1;
1339 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1340 i = si + 1;
1341 continue;
1342 }
1343
1344 /* Pass single-quoted and double-quoted strings through verbatim. */
1345 if (c == '\'' || c == '"')
1346 {
1347 si = i + 1;
1348 i = (c == '\'') ? skip_single_quoted (string, slen, si)
1349 : skip_double_quoted (string, slen, si);
1350 continue;
1351 }
1352
1353 /* move past this character, which was not special. */
1354 ADVANCE_CHAR (string, slen, i);
1355 }
1356
1357 if (c == 0 && nesting_level)
1358 {
1359 if (no_longjmp_on_fatal_error == 0)
1360 {
1361 last_command_exit_value = EXECUTION_FAILURE;
1362 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
1363 exp_jump_to_top_level (DISCARD);
1364 }
1365 else
1366 {
1367 *sindex = i;
1368 return (char *)NULL;
1369 }
1370 }
1371
1372 si = i - *sindex - len_closer + 1;
1373 if (flags & SX_NOALLOC)
1374 result = (char *)NULL;
1375 else
1376 {
1377 result = (char *)xmalloc (1 + si);
1378 strncpy (result, string + *sindex, si);
1379 result[si] = '\0';
1380 }
1381 *sindex = i;
1382
1383 return (result);
1384 }
1385
1386 /* Extract a parameter expansion expression within ${ and } from STRING.
1387 Obey the Posix.2 rules for finding the ending `}': count braces while
1388 skipping over enclosed quoted strings and command substitutions.
1389 SINDEX is the address of an int describing the current offset in STRING;
1390 it should point to just after the first `{' found. On exit, SINDEX
1391 gets the position of the matching `}'. QUOTED is non-zero if this
1392 occurs inside double quotes. */
1393 /* XXX -- this is very similar to extract_delimited_string -- XXX */
1394 static char *
1395 extract_dollar_brace_string (string, sindex, quoted, flags)
1396 char *string;
1397 int *sindex, quoted, flags;
1398 {
1399 register int i, c;
1400 size_t slen;
1401 int pass_character, nesting_level, si, dolbrace_state;
1402 char *result, *t;
1403 DECLARE_MBSTATE;
1404
1405 pass_character = 0;
1406 nesting_level = 1;
1407 slen = strlen (string + *sindex) + *sindex;
1408
1409 /* The handling of dolbrace_state needs to agree with the code in parse.y:
1410 parse_matched_pair(). The different initial value is to handle the
1411 case where this function is called to parse the word in
1412 ${param op word} (SX_WORD). */
1413 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1414 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1415 dolbrace_state = DOLBRACE_QUOTE;
1416
1417 i = *sindex;
1418 while (c = string[i])
1419 {
1420 if (pass_character)
1421 {
1422 pass_character = 0;
1423 ADVANCE_CHAR (string, slen, i);
1424 continue;
1425 }
1426
1427 /* CTLESCs and backslashes quote the next character. */
1428 if (c == CTLESC || c == '\\')
1429 {
1430 pass_character++;
1431 i++;
1432 continue;
1433 }
1434
1435 if (string[i] == '$' && string[i+1] == LBRACE)
1436 {
1437 nesting_level++;
1438 i += 2;
1439 continue;
1440 }
1441
1442 if (c == RBRACE)
1443 {
1444 nesting_level--;
1445 if (nesting_level == 0)
1446 break;
1447 i++;
1448 continue;
1449 }
1450
1451 /* Pass the contents of old-style command substitutions through
1452 verbatim. */
1453 if (c == '`')
1454 {
1455 si = i + 1;
1456 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1457 i = si + 1;
1458 continue;
1459 }
1460
1461 /* Pass the contents of new-style command substitutions and
1462 arithmetic substitutions through verbatim. */
1463 if (string[i] == '$' && string[i+1] == LPAREN)
1464 {
1465 si = i + 2;
1466 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1467 i = si + 1;
1468 continue;
1469 }
1470
1471 /* Pass the contents of double-quoted strings through verbatim. */
1472 if (c == '"')
1473 {
1474 si = i + 1;
1475 i = skip_double_quoted (string, slen, si);
1476 /* skip_XXX_quoted leaves index one past close quote */
1477 continue;
1478 }
1479
1480 if (c == '\'')
1481 {
1482 /*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
1483 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
1484 ADVANCE_CHAR (string, slen, i);
1485 else
1486 {
1487 si = i + 1;
1488 i = skip_single_quoted (string, slen, si);
1489 }
1490
1491 continue;
1492 }
1493
1494 /* move past this character, which was not special. */
1495 ADVANCE_CHAR (string, slen, i);
1496
1497 /* This logic must agree with parse.y:parse_matched_pair, since they
1498 share the same defines. */
1499 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1500 dolbrace_state = DOLBRACE_QUOTE;
1501 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1502 dolbrace_state = DOLBRACE_QUOTE;
1503 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
1504 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
1505 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1506 dolbrace_state = DOLBRACE_QUOTE;
1507 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1508 dolbrace_state = DOLBRACE_QUOTE;
1509 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1510 dolbrace_state = DOLBRACE_OP;
1511 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1512 dolbrace_state = DOLBRACE_WORD;
1513 }
1514
1515 if (c == 0 && nesting_level)
1516 {
1517 if (no_longjmp_on_fatal_error == 0)
1518 { /* { */
1519 last_command_exit_value = EXECUTION_FAILURE;
1520 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
1521 exp_jump_to_top_level (DISCARD);
1522 }
1523 else
1524 {
1525 *sindex = i;
1526 return ((char *)NULL);
1527 }
1528 }
1529
1530 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
1531 *sindex = i;
1532
1533 return (result);
1534 }
1535
1536 /* Remove backslashes which are quoting backquotes from STRING. Modifies
1537 STRING, and returns a pointer to it. */
1538 char *
1539 de_backslash (string)
1540 char *string;
1541 {
1542 register size_t slen;
1543 register int i, j, prev_i;
1544 DECLARE_MBSTATE;
1545
1546 slen = strlen (string);
1547 i = j = 0;
1548
1549 /* Loop copying string[i] to string[j], i >= j. */
1550 while (i < slen)
1551 {
1552 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
1553 string[i + 1] == '$'))
1554 i++;
1555 prev_i = i;
1556 ADVANCE_CHAR (string, slen, i);
1557 if (j < prev_i)
1558 do string[j++] = string[prev_i++]; while (prev_i < i);
1559 else
1560 j = i;
1561 }
1562 string[j] = '\0';
1563
1564 return (string);
1565 }
1566
1567 #if 0
1568 /*UNUSED*/
1569 /* Replace instances of \! in a string with !. */
1570 void
1571 unquote_bang (string)
1572 char *string;
1573 {
1574 register int i, j;
1575 register char *temp;
1576
1577 temp = (char *)xmalloc (1 + strlen (string));
1578
1579 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1580 {
1581 if (string[i] == '\\' && string[i + 1] == '!')
1582 {
1583 temp[j] = '!';
1584 i++;
1585 }
1586 }
1587 strcpy (string, temp);
1588 free (temp);
1589 }
1590 #endif
1591
1592 #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0)
1593
1594 /* This function assumes s[i] == open; returns with s[ret] == close; used to
1595 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1596 matched pairs of quotes or backquotes, or skip word expansions; it is
1597 intended to be used after expansion has been performed and during final
1598 assignment parsing (see arrayfunc.c:assign_compound_array_list()). */
1599 static int
1600 skip_matched_pair (string, start, open, close, flags)
1601 const char *string;
1602 int start, open, close, flags;
1603 {
1604 int i, pass_next, backq, si, c, count;
1605 size_t slen;
1606 char *temp, *ss;
1607 DECLARE_MBSTATE;
1608
1609 slen = strlen (string + start) + start;
1610 no_longjmp_on_fatal_error = 1;
1611
1612 i = start + 1; /* skip over leading bracket */
1613 count = 1;
1614 pass_next = backq = 0;
1615 ss = (char *)string;
1616 while (c = string[i])
1617 {
1618 if (pass_next)
1619 {
1620 pass_next = 0;
1621 if (c == 0)
1622 CQ_RETURN(i);
1623 ADVANCE_CHAR (string, slen, i);
1624 continue;
1625 }
1626 else if (c == '\\')
1627 {
1628 pass_next = 1;
1629 i++;
1630 continue;
1631 }
1632 else if (backq)
1633 {
1634 if (c == '`')
1635 backq = 0;
1636 ADVANCE_CHAR (string, slen, i);
1637 continue;
1638 }
1639 else if ((flags & 1) == 0 && c == '`')
1640 {
1641 backq = 1;
1642 i++;
1643 continue;
1644 }
1645 else if ((flags & 1) == 0 && c == open)
1646 {
1647 count++;
1648 i++;
1649 continue;
1650 }
1651 else if (c == close)
1652 {
1653 count--;
1654 if (count == 0)
1655 break;
1656 i++;
1657 continue;
1658 }
1659 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
1660 {
1661 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i)
1662 : skip_double_quoted (ss, slen, ++i);
1663 /* no increment, the skip functions increment past the closing quote. */
1664 }
1665 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
1666 {
1667 si = i + 2;
1668 if (string[si] == '\0')
1669 CQ_RETURN(si);
1670
1671 if (string[i+1] == LPAREN)
1672 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1673 else
1674 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
1675 i = si;
1676 if (string[i] == '\0') /* don't increment i past EOS in loop */
1677 break;
1678 i++;
1679 continue;
1680 }
1681 else
1682 ADVANCE_CHAR (string, slen, i);
1683 }
1684
1685 CQ_RETURN(i);
1686 }
1687
1688 #if defined (ARRAY_VARS)
1689 int
1690 skipsubscript (string, start, flags)
1691 const char *string;
1692 int start, flags;
1693 {
1694 return (skip_matched_pair (string, start, '[', ']', flags));
1695 }
1696 #endif
1697
1698 /* Skip characters in STRING until we find a character in DELIMS, and return
1699 the index of that character. START is the index into string at which we
1700 begin. This is similar in spirit to strpbrk, but it returns an index into
1701 STRING and takes a starting index. This little piece of code knows quite
1702 a lot of shell syntax. It's very similar to skip_double_quoted and other
1703 functions of that ilk. */
1704 int
1705 skip_to_delim (string, start, delims, flags)
1706 char *string;
1707 int start;
1708 char *delims;
1709 int flags;
1710 {
1711 int i, pass_next, backq, si, c, invert, skipquote, skipcmd;
1712 size_t slen;
1713 char *temp, open[3];
1714 DECLARE_MBSTATE;
1715
1716 slen = strlen (string + start) + start;
1717 if (flags & SD_NOJMP)
1718 no_longjmp_on_fatal_error = 1;
1719 invert = (flags & SD_INVERT);
1720 skipcmd = (flags & SD_NOSKIPCMD) == 0;
1721
1722 i = start;
1723 pass_next = backq = 0;
1724 while (c = string[i])
1725 {
1726 /* If this is non-zero, we should not let quote characters be delimiters
1727 and the current character is a single or double quote. We should not
1728 test whether or not it's a delimiter until after we skip single- or
1729 double-quoted strings. */
1730 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
1731 if (pass_next)
1732 {
1733 pass_next = 0;
1734 if (c == 0)
1735 CQ_RETURN(i);
1736 ADVANCE_CHAR (string, slen, i);
1737 continue;
1738 }
1739 else if (c == '\\')
1740 {
1741 pass_next = 1;
1742 i++;
1743 continue;
1744 }
1745 else if (backq)
1746 {
1747 if (c == '`')
1748 backq = 0;
1749 ADVANCE_CHAR (string, slen, i);
1750 continue;
1751 }
1752 else if (c == '`')
1753 {
1754 backq = 1;
1755 i++;
1756 continue;
1757 }
1758 else if (skipquote == 0 && invert == 0 && member (c, delims))
1759 break;
1760 else if (c == '\'' || c == '"')
1761 {
1762 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1763 : skip_double_quoted (string, slen, ++i);
1764 /* no increment, the skip functions increment past the closing quote. */
1765 }
1766 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
1767 {
1768 si = i + 2;
1769 if (string[si] == '\0')
1770 CQ_RETURN(si);
1771
1772 if (string[i+1] == LPAREN)
1773 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1774 else
1775 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1776 i = si;
1777 if (string[i] == '\0') /* don't increment i past EOS in loop */
1778 break;
1779 i++;
1780 continue;
1781 }
1782 #if defined (PROCESS_SUBSTITUTION)
1783 else if (skipcmd && (c == '<' || c == '>') && string[i+1] == LPAREN)
1784 {
1785 si = i + 2;
1786 if (string[si] == '\0')
1787 CQ_RETURN(si);
1788 temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si);
1789 free (temp); /* no SX_ALLOC here */
1790 i = si;
1791 if (string[i] == '\0')
1792 break;
1793 i++;
1794 continue;
1795 }
1796 #endif /* PROCESS_SUBSTITUTION */
1797 #if defined (EXTENDED_GLOB)
1798 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1799 {
1800 si = i + 2;
1801 if (string[si] == '\0')
1802 CQ_RETURN(si);
1803
1804 open[0] = c;
1805 open[1] = LPAREN;
1806 open[2] = '\0';
1807 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
1808
1809 i = si;
1810 if (string[i] == '\0') /* don't increment i past EOS in loop */
1811 break;
1812 i++;
1813 continue;
1814 }
1815 #endif
1816 else if ((flags & SD_GLOB) && c == LBRACK)
1817 {
1818 si = i + 1;
1819 if (string[si] == '\0')
1820 CQ_RETURN(si);
1821
1822 temp = extract_delimited_string (string, &si, "[", "[", "]", SX_NOALLOC); /* ] */
1823
1824 i = si;
1825 if (string[i] == '\0') /* don't increment i past EOS in loop */
1826 break;
1827 i++;
1828 continue;
1829 }
1830 else if ((skipquote || invert) && (member (c, delims) == 0))
1831 break;
1832 else
1833 ADVANCE_CHAR (string, slen, i);
1834 }
1835
1836 CQ_RETURN(i);
1837 }
1838
1839 #if defined (READLINE)
1840 /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
1841 an unclosed quoted string), or if the character at EINDEX is quoted
1842 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
1843 single and double-quoted string parsing functions should not return an
1844 error if there are unclosed quotes or braces. The characters that this
1845 recognizes need to be the same as the contents of
1846 rl_completer_quote_characters. */
1847
1848 int
1849 char_is_quoted (string, eindex)
1850 char *string;
1851 int eindex;
1852 {
1853 int i, pass_next, c;
1854 size_t slen;
1855 DECLARE_MBSTATE;
1856
1857 slen = strlen (string);
1858 no_longjmp_on_fatal_error = 1;
1859 i = pass_next = 0;
1860 while (i <= eindex)
1861 {
1862 c = string[i];
1863
1864 if (pass_next)
1865 {
1866 pass_next = 0;
1867 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1868 CQ_RETURN(1);
1869 ADVANCE_CHAR (string, slen, i);
1870 continue;
1871 }
1872 else if (c == '\\')
1873 {
1874 pass_next = 1;
1875 i++;
1876 continue;
1877 }
1878 else if (c == '\'' || c == '"')
1879 {
1880 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1881 : skip_double_quoted (string, slen, ++i);
1882 if (i > eindex)
1883 CQ_RETURN(1);
1884 /* no increment, the skip_xxx functions go one past end */
1885 }
1886 else
1887 ADVANCE_CHAR (string, slen, i);
1888 }
1889
1890 CQ_RETURN(0);
1891 }
1892
1893 int
1894 unclosed_pair (string, eindex, openstr)
1895 char *string;
1896 int eindex;
1897 char *openstr;
1898 {
1899 int i, pass_next, openc, olen;
1900 size_t slen;
1901 DECLARE_MBSTATE;
1902
1903 slen = strlen (string);
1904 olen = strlen (openstr);
1905 i = pass_next = openc = 0;
1906 while (i <= eindex)
1907 {
1908 if (pass_next)
1909 {
1910 pass_next = 0;
1911 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1912 return 0;
1913 ADVANCE_CHAR (string, slen, i);
1914 continue;
1915 }
1916 else if (string[i] == '\\')
1917 {
1918 pass_next = 1;
1919 i++;
1920 continue;
1921 }
1922 else if (STREQN (string + i, openstr, olen))
1923 {
1924 openc = 1 - openc;
1925 i += olen;
1926 }
1927 else if (string[i] == '\'' || string[i] == '"')
1928 {
1929 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i)
1930 : skip_double_quoted (string, slen, i);
1931 if (i > eindex)
1932 return 0;
1933 }
1934 else
1935 ADVANCE_CHAR (string, slen, i);
1936 }
1937 return (openc);
1938 }
1939
1940 /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
1941 individual words. If DELIMS is NULL, the current value of $IFS is used
1942 to split the string, and the function follows the shell field splitting
1943 rules. SENTINEL is an index to look for. NWP, if non-NULL,
1944 gets the number of words in the returned list. CWP, if non-NULL, gets
1945 the index of the word containing SENTINEL. Non-whitespace chars in
1946 DELIMS delimit separate fields. */
1947 WORD_LIST *
1948 split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
1949 char *string;
1950 int slen;
1951 char *delims;
1952 int sentinel, flags;
1953 int *nwp, *cwp;
1954 {
1955 int ts, te, i, nw, cw, ifs_split, dflags;
1956 char *token, *d, *d2;
1957 WORD_LIST *ret, *tl;
1958
1959 if (string == 0 || *string == '\0')
1960 {
1961 if (nwp)
1962 *nwp = 0;
1963 if (cwp)
1964 *cwp = 0;
1965 return ((WORD_LIST *)NULL);
1966 }
1967
1968 d = (delims == 0) ? ifs_value : delims;
1969 ifs_split = delims == 0;
1970
1971 /* Make d2 the non-whitespace characters in delims */
1972 d2 = 0;
1973 if (delims)
1974 {
1975 size_t slength;
1976 #if defined (HANDLE_MULTIBYTE)
1977 size_t mblength = 1;
1978 #endif
1979 DECLARE_MBSTATE;
1980
1981 slength = strlen (delims);
1982 d2 = (char *)xmalloc (slength + 1);
1983 i = ts = 0;
1984 while (delims[i])
1985 {
1986 #if defined (HANDLE_MULTIBYTE)
1987 mbstate_t state_bak;
1988 state_bak = state;
1989 mblength = MBRLEN (delims + i, slength, &state);
1990 if (MB_INVALIDCH (mblength))
1991 state = state_bak;
1992 else if (mblength > 1)
1993 {
1994 memcpy (d2 + ts, delims + i, mblength);
1995 ts += mblength;
1996 i += mblength;
1997 slength -= mblength;
1998 continue;
1999 }
2000 #endif
2001 if (whitespace (delims[i]) == 0)
2002 d2[ts++] = delims[i];
2003
2004 i++;
2005 slength--;
2006 }
2007 d2[ts] = '\0';
2008 }
2009
2010 ret = (WORD_LIST *)NULL;
2011
2012 /* Remove sequences of whitespace characters at the start of the string, as
2013 long as those characters are delimiters. */
2014 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
2015 ;
2016 if (string[i] == '\0')
2017 return (ret);
2018
2019 ts = i;
2020 nw = 0;
2021 cw = -1;
2022 dflags = flags|SD_NOJMP;
2023 while (1)
2024 {
2025 te = skip_to_delim (string, ts, d, dflags);
2026
2027 /* If we have a non-whitespace delimiter character, use it to make a
2028 separate field. This is just about what $IFS splitting does and
2029 is closer to the behavior of the shell parser. */
2030 if (ts == te && d2 && member (string[ts], d2))
2031 {
2032 te = ts + 1;
2033 /* If we're using IFS splitting, the non-whitespace delimiter char
2034 and any additional IFS whitespace delimits a field. */
2035 if (ifs_split)
2036 while (member (string[te], d) && spctabnl (string[te]))
2037 te++;
2038 else
2039 while (member (string[te], d2))
2040 te++;
2041 }
2042
2043 token = substring (string, ts, te);
2044
2045 ret = add_string_to_list (token, ret);
2046 free (token);
2047 nw++;
2048
2049 if (sentinel >= ts && sentinel <= te)
2050 cw = nw;
2051
2052 /* If the cursor is at whitespace just before word start, set the
2053 sentinel word to the current word. */
2054 if (cwp && cw == -1 && sentinel == ts-1)
2055 cw = nw;
2056
2057 /* If the cursor is at whitespace between two words, make a new, empty
2058 word, add it before (well, after, since the list is in reverse order)
2059 the word we just added, and set the current word to that one. */
2060 if (cwp && cw == -1 && sentinel < ts)
2061 {
2062 tl = make_word_list (make_word (""), ret->next);
2063 ret->next = tl;
2064 cw = nw;
2065 nw++;
2066 }
2067
2068 if (string[te] == 0)
2069 break;
2070
2071 i = te;
2072 while (member (string[i], d) && (ifs_split || spctabnl(string[i])))
2073 i++;
2074
2075 if (string[i])
2076 ts = i;
2077 else
2078 break;
2079 }
2080
2081 /* Special case for SENTINEL at the end of STRING. If we haven't found
2082 the word containing SENTINEL yet, and the index we're looking for is at
2083 the end of STRING (or past the end of the previously-found token,
2084 possible if the end of the line is composed solely of IFS whitespace)
2085 add an additional null argument and set the current word pointer to that. */
2086 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
2087 {
2088 if (whitespace (string[sentinel - 1]))
2089 {
2090 token = "";
2091 ret = add_string_to_list (token, ret);
2092 nw++;
2093 }
2094 cw = nw;
2095 }
2096
2097 if (nwp)
2098 *nwp = nw;
2099 if (cwp)
2100 *cwp = cw;
2101
2102 FREE (d2);
2103
2104 return (REVERSE_LIST (ret, WORD_LIST *));
2105 }
2106 #endif /* READLINE */
2107
2108 #if 0
2109 /* UNUSED */
2110 /* Extract the name of the variable to bind to from the assignment string. */
2111 char *
2112 assignment_name (string)
2113 char *string;
2114 {
2115 int offset;
2116 char *temp;
2117
2118 offset = assignment (string, 0);
2119 if (offset == 0)
2120 return (char *)NULL;
2121 temp = substring (string, 0, offset);
2122 return (temp);
2123 }
2124 #endif
2125
2126 /* **************************************************************** */
2127 /* */
2128 /* Functions to convert strings to WORD_LISTs and vice versa */
2129 /* */
2130 /* **************************************************************** */
2131
2132 /* Return a single string of all the words in LIST. SEP is the separator
2133 to put between individual elements of LIST in the output string. */
2134 char *
2135 string_list_internal (list, sep)
2136 WORD_LIST *list;
2137 char *sep;
2138 {
2139 register WORD_LIST *t;
2140 char *result, *r;
2141 int word_len, sep_len, result_size;
2142
2143 if (list == 0)
2144 return ((char *)NULL);
2145
2146 /* Short-circuit quickly if we don't need to separate anything. */
2147 if (list->next == 0)
2148 return (savestring (list->word->word));
2149
2150 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2151 sep_len = STRLEN (sep);
2152 result_size = 0;
2153
2154 for (t = list; t; t = t->next)
2155 {
2156 if (t != list)
2157 result_size += sep_len;
2158 result_size += strlen (t->word->word);
2159 }
2160
2161 r = result = (char *)xmalloc (result_size + 1);
2162
2163 for (t = list; t; t = t->next)
2164 {
2165 if (t != list && sep_len)
2166 {
2167 if (sep_len > 1)
2168 {
2169 FASTCOPY (sep, r, sep_len);
2170 r += sep_len;
2171 }
2172 else
2173 *r++ = sep[0];
2174 }
2175
2176 word_len = strlen (t->word->word);
2177 FASTCOPY (t->word->word, r, word_len);
2178 r += word_len;
2179 }
2180
2181 *r = '\0';
2182 return (result);
2183 }
2184
2185 /* Return a single string of all the words present in LIST, separating
2186 each word with a space. */
2187 char *
2188 string_list (list)
2189 WORD_LIST *list;
2190 {
2191 return (string_list_internal (list, " "));
2192 }
2193
2194 /* An external interface that can be used by the rest of the shell to
2195 obtain a string containing the first character in $IFS. Handles all
2196 the multibyte complications. If LENP is non-null, it is set to the
2197 length of the returned string. */
2198 char *
2199 ifs_firstchar (lenp)
2200 int *lenp;
2201 {
2202 char *ret;
2203 int len;
2204
2205 ret = xmalloc (MB_LEN_MAX + 1);
2206 #if defined (HANDLE_MULTIBYTE)
2207 if (ifs_firstc_len == 1)
2208 {
2209 ret[0] = ifs_firstc[0];
2210 ret[1] = '\0';
2211 len = ret[0] ? 1 : 0;
2212 }
2213 else
2214 {
2215 memcpy (ret, ifs_firstc, ifs_firstc_len);
2216 ret[len = ifs_firstc_len] = '\0';
2217 }
2218 #else
2219 ret[0] = ifs_firstc;
2220 ret[1] = '\0';
2221 len = ret[0] ? 0 : 1;
2222 #endif
2223
2224 if (lenp)
2225 *lenp = len;
2226
2227 return ret;
2228 }
2229
2230 /* Return a single string of all the words present in LIST, obeying the
2231 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2232 expansion [of $*] appears within a double quoted string, it expands
2233 to a single field with the value of each parameter separated by the
2234 first character of the IFS variable, or by a <space> if IFS is unset." */
2235 char *
2236 string_list_dollar_star (list)
2237 WORD_LIST *list;
2238 {
2239 char *ret;
2240 #if defined (HANDLE_MULTIBYTE)
2241 # if defined (__GNUC__)
2242 char sep[MB_CUR_MAX + 1];
2243 # else
2244 char *sep = 0;
2245 # endif
2246 #else
2247 char sep[2];
2248 #endif
2249
2250 #if defined (HANDLE_MULTIBYTE)
2251 # if !defined (__GNUC__)
2252 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2253 # endif /* !__GNUC__ */
2254 if (ifs_firstc_len == 1)
2255 {
2256 sep[0] = ifs_firstc[0];
2257 sep[1] = '\0';
2258 }
2259 else
2260 {
2261 memcpy (sep, ifs_firstc, ifs_firstc_len);
2262 sep[ifs_firstc_len] = '\0';
2263 }
2264 #else
2265 sep[0] = ifs_firstc;
2266 sep[1] = '\0';
2267 #endif
2268
2269 ret = string_list_internal (list, sep);
2270 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2271 free (sep);
2272 #endif
2273 return ret;
2274 }
2275
2276 /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2277 is non-zero, the $@ appears within double quotes, and we should quote
2278 the list before converting it into a string. If IFS is unset, and the
2279 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2280 in the words in the list, because the default value of $IFS is
2281 <space><tab><newline>, IFS characters in the words in the list should
2282 also be split. If IFS is null, and the word is not quoted, we need
2283 to quote the words in the list to preserve the positional parameters
2284 exactly. */
2285 char *
2286 string_list_dollar_at (list, quoted)
2287 WORD_LIST *list;
2288 int quoted;
2289 {
2290 char *ifs, *ret;
2291 #if defined (HANDLE_MULTIBYTE)
2292 # if defined (__GNUC__)
2293 char sep[MB_CUR_MAX + 1];
2294 # else
2295 char *sep = 0;
2296 # endif /* !__GNUC__ */
2297 #else
2298 char sep[2];
2299 #endif
2300 WORD_LIST *tlist;
2301
2302 /* XXX this could just be ifs = ifs_value; */
2303 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
2304
2305 #if defined (HANDLE_MULTIBYTE)
2306 # if !defined (__GNUC__)
2307 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2308 # endif /* !__GNUC__ */
2309 if (ifs && *ifs)
2310 {
2311 if (ifs_firstc_len == 1)
2312 {
2313 sep[0] = ifs_firstc[0];
2314 sep[1] = '\0';
2315 }
2316 else
2317 {
2318 memcpy (sep, ifs_firstc, ifs_firstc_len);
2319 sep[ifs_firstc_len] = '\0';
2320 }
2321 }
2322 else
2323 {
2324 sep[0] = ' ';
2325 sep[1] = '\0';
2326 }
2327 #else
2328 sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs;
2329 sep[1] = '\0';
2330 #endif
2331
2332 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2333 it now that quote_escapes quotes spaces */
2334 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
2335 ? quote_list (list)
2336 : list_quote_escapes (list);
2337
2338 ret = string_list_internal (tlist, sep);
2339 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2340 free (sep);
2341 #endif
2342 return ret;
2343 }
2344
2345 /* Turn the positional parameters into a string, understanding quoting and
2346 the various subtleties of using the first character of $IFS as the
2347 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2348 string_list as appropriate. */
2349 char *
2350 string_list_pos_params (pchar, list, quoted)
2351 int pchar;
2352 WORD_LIST *list;
2353 int quoted;
2354 {
2355 char *ret;
2356 WORD_LIST *tlist;
2357
2358 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2359 {
2360 tlist = quote_list (list);
2361 word_list_remove_quoted_nulls (tlist);
2362 ret = string_list_dollar_star (tlist);
2363 }
2364 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2365 {
2366 tlist = quote_list (list);
2367 word_list_remove_quoted_nulls (tlist);
2368 ret = string_list (tlist);
2369 }
2370 else if (pchar == '*')
2371 {
2372 /* Even when unquoted, string_list_dollar_star does the right thing
2373 making sure that the first character of $IFS is used as the
2374 separator. */
2375 ret = string_list_dollar_star (list);
2376 }
2377 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2378 /* We use string_list_dollar_at, but only if the string is quoted, since
2379 that quotes the escapes if it's not, which we don't want. We could
2380 use string_list (the old code did), but that doesn't do the right
2381 thing if the first character of $IFS is not a space. We use
2382 string_list_dollar_star if the string is unquoted so we make sure that
2383 the elements of $@ are separated by the first character of $IFS for
2384 later splitting. */
2385 ret = string_list_dollar_at (list, quoted);
2386 else if (pchar == '@')
2387 ret = string_list_dollar_star (list);
2388 else
2389 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2390
2391 return ret;
2392 }
2393
2394 /* Return the list of words present in STRING. Separate the string into
2395 words at any of the characters found in SEPARATORS. If QUOTED is
2396 non-zero then word in the list will have its quoted flag set, otherwise
2397 the quoted flag is left as make_word () deemed fit.
2398
2399 This obeys the P1003.2 word splitting semantics. If `separators' is
2400 exactly <space><tab><newline>, then the splitting algorithm is that of
2401 the Bourne shell, which treats any sequence of characters from `separators'
2402 as a delimiter. If IFS is unset, which results in `separators' being set
2403 to "", no splitting occurs. If separators has some other value, the
2404 following rules are applied (`IFS white space' means zero or more
2405 occurrences of <space>, <tab>, or <newline>, as long as those characters
2406 are in `separators'):
2407
2408 1) IFS white space is ignored at the start and the end of the
2409 string.
2410 2) Each occurrence of a character in `separators' that is not
2411 IFS white space, along with any adjacent occurrences of
2412 IFS white space delimits a field.
2413 3) Any nonzero-length sequence of IFS white space delimits a field.
2414 */
2415
2416 /* BEWARE! list_string strips null arguments. Don't call it twice and
2417 expect to have "" preserved! */
2418
2419 /* This performs word splitting and quoted null character removal on
2420 STRING. */
2421 #define issep(c) \
2422 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2423 : (c) == (separators)[0]) \
2424 : 0)
2425
2426 WORD_LIST *
2427 list_string (string, separators, quoted)
2428 register char *string, *separators;
2429 int quoted;
2430 {
2431 WORD_LIST *result;
2432 WORD_DESC *t;
2433 char *current_word, *s;
2434 int sindex, sh_style_split, whitesep, xflags;
2435 size_t slen;
2436
2437 if (!string || !*string)
2438 return ((WORD_LIST *)NULL);
2439
2440 sh_style_split = separators && separators[0] == ' ' &&
2441 separators[1] == '\t' &&
2442 separators[2] == '\n' &&
2443 separators[3] == '\0';
2444 for (xflags = 0, s = ifs_value; s && *s; s++)
2445 {
2446 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2447 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2448 }
2449
2450 slen = 0;
2451 /* Remove sequences of whitespace at the beginning of STRING, as
2452 long as those characters appear in IFS. Do not do this if
2453 STRING is quoted or if there are no separator characters. */
2454 if (!quoted || !separators || !*separators)
2455 {
2456 for (s = string; *s && spctabnl (*s) && issep (*s); s++);
2457
2458 if (!*s)
2459 return ((WORD_LIST *)NULL);
2460
2461 string = s;
2462 }
2463
2464 /* OK, now STRING points to a word that does not begin with white space.
2465 The splitting algorithm is:
2466 extract a word, stopping at a separator
2467 skip sequences of spc, tab, or nl as long as they are separators
2468 This obeys the field splitting rules in Posix.2. */
2469 slen = (MB_CUR_MAX > 1) ? strlen (string) : 1;
2470 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
2471 {
2472 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2473 unless multibyte chars are possible. */
2474 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
2475 if (current_word == 0)
2476 break;
2477
2478 /* If we have a quoted empty string, add a quoted null argument. We
2479 want to preserve the quoted null character iff this is a quoted
2480 empty string; otherwise the quoted null characters are removed
2481 below. */
2482 if (QUOTED_NULL (current_word))
2483 {
2484 t = alloc_word_desc ();
2485 t->word = make_quoted_char ('\0');
2486 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2487 result = make_word_list (t, result);
2488 }
2489 else if (current_word[0] != '\0')
2490 {
2491 /* If we have something, then add it regardless. However,
2492 perform quoted null character removal on the current word. */
2493 remove_quoted_nulls (current_word);
2494 result = add_string_to_list (current_word, result);
2495 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
2496 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2497 result->word->flags |= W_QUOTED;
2498 }
2499
2500 /* If we're not doing sequences of separators in the traditional
2501 Bourne shell style, then add a quoted null argument. */
2502 else if (!sh_style_split && !spctabnl (string[sindex]))
2503 {
2504 t = alloc_word_desc ();
2505 t->word = make_quoted_char ('\0');
2506 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2507 result = make_word_list (t, result);
2508 }
2509
2510 free (current_word);
2511
2512 /* Note whether or not the separator is IFS whitespace, used later. */
2513 whitesep = string[sindex] && spctabnl (string[sindex]);
2514
2515 /* Move past the current separator character. */
2516 if (string[sindex])
2517 {
2518 DECLARE_MBSTATE;
2519 ADVANCE_CHAR (string, slen, sindex);
2520 }
2521
2522 /* Now skip sequences of space, tab, or newline characters if they are
2523 in the list of separators. */
2524 while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex]))
2525 sindex++;
2526
2527 /* If the first separator was IFS whitespace and the current character
2528 is a non-whitespace IFS character, it should be part of the current
2529 field delimiter, not a separate delimiter that would result in an
2530 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
2531 if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex]))
2532 {
2533 sindex++;
2534 /* An IFS character that is not IFS white space, along with any
2535 adjacent IFS white space, shall delimit a field. (SUSv3) */
2536 while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex]))
2537 sindex++;
2538 }
2539 }
2540 return (REVERSE_LIST (result, WORD_LIST *));
2541 }
2542
2543 /* Parse a single word from STRING, using SEPARATORS to separate fields.
2544 ENDPTR is set to the first character after the word. This is used by
2545 the `read' builtin. This is never called with SEPARATORS != $IFS;
2546 it should be simplified.
2547
2548 XXX - this function is very similar to list_string; they should be
2549 combined - XXX */
2550 char *
2551 get_word_from_string (stringp, separators, endptr)
2552 char **stringp, *separators, **endptr;
2553 {
2554 register char *s;
2555 char *current_word;
2556 int sindex, sh_style_split, whitesep, xflags;
2557 size_t slen;
2558
2559 if (!stringp || !*stringp || !**stringp)
2560 return ((char *)NULL);
2561
2562 sh_style_split = separators && separators[0] == ' ' &&
2563 separators[1] == '\t' &&
2564 separators[2] == '\n' &&
2565 separators[3] == '\0';
2566 for (xflags = 0, s = ifs_value; s && *s; s++)
2567 {
2568 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2569 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2570 }
2571
2572 s = *stringp;
2573 slen = 0;
2574
2575 /* Remove sequences of whitespace at the beginning of STRING, as
2576 long as those characters appear in IFS. */
2577 if (sh_style_split || !separators || !*separators)
2578 {
2579 for (; *s && spctabnl (*s) && isifs (*s); s++);
2580
2581 /* If the string is nothing but whitespace, update it and return. */
2582 if (!*s)
2583 {
2584 *stringp = s;
2585 if (endptr)
2586 *endptr = s;
2587 return ((char *)NULL);
2588 }
2589 }
2590
2591 /* OK, S points to a word that does not begin with white space.
2592 Now extract a word, stopping at a separator, save a pointer to
2593 the first character after the word, then skip sequences of spc,
2594 tab, or nl as long as they are separators.
2595
2596 This obeys the field splitting rules in Posix.2. */
2597 sindex = 0;
2598 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2599 unless multibyte chars are possible. */
2600 slen = (MB_CUR_MAX > 1) ? strlen (s) : 1;
2601 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
2602
2603 /* Set ENDPTR to the first character after the end of the word. */
2604 if (endptr)
2605 *endptr = s + sindex;
2606
2607 /* Note whether or not the separator is IFS whitespace, used later. */
2608 whitesep = s[sindex] && spctabnl (s[sindex]);
2609
2610 /* Move past the current separator character. */
2611 if (s[sindex])
2612 {
2613 DECLARE_MBSTATE;
2614 ADVANCE_CHAR (s, slen, sindex);
2615 }
2616
2617 /* Now skip sequences of space, tab, or newline characters if they are
2618 in the list of separators. */
2619 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2620 sindex++;
2621
2622 /* If the first separator was IFS whitespace and the current character is
2623 a non-whitespace IFS character, it should be part of the current field
2624 delimiter, not a separate delimiter that would result in an empty field.
2625 Look at POSIX.2, 3.6.5, (3)(b). */
2626 if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex]))
2627 {
2628 sindex++;
2629 /* An IFS character that is not IFS white space, along with any adjacent
2630 IFS white space, shall delimit a field. */
2631 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2632 sindex++;
2633 }
2634
2635 /* Update STRING to point to the next field. */
2636 *stringp = s + sindex;
2637 return (current_word);
2638 }
2639
2640 /* Remove IFS white space at the end of STRING. Start at the end
2641 of the string and walk backwards until the beginning of the string
2642 or we find a character that's not IFS white space and not CTLESC.
2643 Only let CTLESC escape a white space character if SAW_ESCAPE is
2644 non-zero. */
2645 char *
2646 strip_trailing_ifs_whitespace (string, separators, saw_escape)
2647 char *string, *separators;
2648 int saw_escape;
2649 {
2650 char *s;
2651
2652 s = string + STRLEN (string) - 1;
2653 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
2654 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2655 s--;
2656 *++s = '\0';
2657 return string;
2658 }
2659
2660 #if 0
2661 /* UNUSED */
2662 /* Split STRING into words at whitespace. Obeys shell-style quoting with
2663 backslashes, single and double quotes. */
2664 WORD_LIST *
2665 list_string_with_quotes (string)
2666 char *string;
2667 {
2668 WORD_LIST *list;
2669 char *token, *s;
2670 size_t s_len;
2671 int c, i, tokstart, len;
2672
2673 for (s = string; s && *s && spctabnl (*s); s++)
2674 ;
2675 if (s == 0 || *s == 0)
2676 return ((WORD_LIST *)NULL);
2677
2678 s_len = strlen (s);
2679 tokstart = i = 0;
2680 list = (WORD_LIST *)NULL;
2681 while (1)
2682 {
2683 c = s[i];
2684 if (c == '\\')
2685 {
2686 i++;
2687 if (s[i])
2688 i++;
2689 }
2690 else if (c == '\'')
2691 i = skip_single_quoted (s, s_len, ++i);
2692 else if (c == '"')
2693 i = skip_double_quoted (s, s_len, ++i);
2694 else if (c == 0 || spctabnl (c))
2695 {
2696 /* We have found the end of a token. Make a word out of it and
2697 add it to the word list. */
2698 token = substring (s, tokstart, i);
2699 list = add_string_to_list (token, list);
2700 free (token);
2701 while (spctabnl (s[i]))
2702 i++;
2703 if (s[i])
2704 tokstart = i;
2705 else
2706 break;
2707 }
2708 else
2709 i++; /* normal character */
2710 }
2711 return (REVERSE_LIST (list, WORD_LIST *));
2712 }
2713 #endif
2714
2715 /********************************************************/
2716 /* */
2717 /* Functions to perform assignment statements */
2718 /* */
2719 /********************************************************/
2720
2721 #if defined (ARRAY_VARS)
2722 static SHELL_VAR *
2723 do_compound_assignment (name, value, flags)
2724 char *name, *value;
2725 int flags;
2726 {
2727 SHELL_VAR *v;
2728 int mklocal, mkassoc, mkglobal;
2729 WORD_LIST *list;
2730
2731 mklocal = flags & ASS_MKLOCAL;
2732 mkassoc = flags & ASS_MKASSOC;
2733 mkglobal = flags & ASS_MKGLOBAL;
2734
2735 if (mklocal && variable_context)
2736 {
2737 v = find_variable (name);
2738 list = expand_compound_array_assignment (v, value, flags);
2739 if (mkassoc)
2740 v = make_local_assoc_variable (name);
2741 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
2742 v = make_local_array_variable (name, 0);
2743 if (v)
2744 assign_compound_array_list (v, list, flags);
2745 }
2746 /* In a function but forcing assignment in global context */
2747 else if (mkglobal && variable_context)
2748 {
2749 v = find_global_variable (name);
2750 list = expand_compound_array_assignment (v, value, flags);
2751 if (v == 0 && mkassoc)
2752 v = make_new_assoc_variable (name);
2753 else if (v && mkassoc && assoc_p (v) == 0)
2754 v = convert_var_to_assoc (v);
2755 else if (v == 0)
2756 v = make_new_array_variable (name);
2757 else if (v && mkassoc == 0 && array_p (v) == 0)
2758 v = convert_var_to_array (v);
2759 if (v)
2760 assign_compound_array_list (v, list, flags);
2761 }
2762 else
2763 v = assign_array_from_string (name, value, flags);
2764
2765 return (v);
2766 }
2767 #endif
2768
2769 /* Given STRING, an assignment string, get the value of the right side
2770 of the `=', and bind it to the left side. If EXPAND is true, then
2771 perform parameter expansion, command substitution, and arithmetic
2772 expansion on the right-hand side. Perform tilde expansion in any
2773 case. Do not perform word splitting on the result of expansion. */
2774 static int
2775 do_assignment_internal (word, expand)
2776 const WORD_DESC *word;
2777 int expand;
2778 {
2779 int offset, appendop, assign_list, aflags, retval;
2780 char *name, *value, *temp;
2781 SHELL_VAR *entry;
2782 #if defined (ARRAY_VARS)
2783 char *t;
2784 int ni;
2785 #endif
2786 const char *string;
2787
2788 if (word == 0 || word->word == 0)
2789 return 0;
2790
2791 appendop = assign_list = aflags = 0;
2792 string = word->word;
2793 offset = assignment (string, 0);
2794 name = savestring (string);
2795 value = (char *)NULL;
2796
2797 if (name[offset] == '=')
2798 {
2799 if (name[offset - 1] == '+')
2800 {
2801 appendop = 1;
2802 name[offset - 1] = '\0';
2803 }
2804
2805 name[offset] = 0; /* might need this set later */
2806 temp = name + offset + 1;
2807
2808 #if defined (ARRAY_VARS)
2809 if (expand && (word->flags & W_COMPASSIGN))
2810 {
2811 assign_list = ni = 1;
2812 value = extract_array_assignment_list (temp, &ni);
2813 }
2814 else
2815 #endif
2816 if (expand && temp[0])
2817 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
2818 else
2819 value = savestring (temp);
2820 }
2821
2822 if (value == 0)
2823 {
2824 value = (char *)xmalloc (1);
2825 value[0] = '\0';
2826 }
2827
2828 if (echo_command_at_execute)
2829 {
2830 if (appendop)
2831 name[offset - 1] = '+';
2832 xtrace_print_assignment (name, value, assign_list, 1);
2833 if (appendop)
2834 name[offset - 1] = '\0';
2835 }
2836
2837 #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
2838
2839 if (appendop)
2840 aflags |= ASS_APPEND;
2841
2842 #if defined (ARRAY_VARS)
2843 if (t = mbschr (name, '[')) /*]*/
2844 {
2845 if (assign_list)
2846 {
2847 report_error (_("%s: cannot assign list to array member"), name);
2848 ASSIGN_RETURN (0);
2849 }
2850 entry = assign_array_element (name, value, aflags);
2851 if (entry == 0)
2852 ASSIGN_RETURN (0);
2853 }
2854 else if (assign_list)
2855 {
2856 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
2857 aflags |= ASS_MKLOCAL;
2858 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
2859 aflags |= ASS_MKGLOBAL;
2860 if (word->flags & W_ASSIGNASSOC)
2861 aflags |= ASS_MKASSOC;
2862 entry = do_compound_assignment (name, value, aflags);
2863 }
2864 else
2865 #endif /* ARRAY_VARS */
2866 entry = bind_variable (name, value, aflags);
2867
2868 stupidly_hack_special_variables (name);
2869
2870 /* Return 1 if the assignment seems to have been performed correctly. */
2871 if (entry == 0 || readonly_p (entry))
2872 retval = 0; /* assignment failure */
2873 else if (noassign_p (entry))
2874 {
2875 last_command_exit_value = EXECUTION_FAILURE;
2876 retval = 1; /* error status, but not assignment failure */
2877 }
2878 else
2879 retval = 1;
2880
2881 if (entry && retval != 0 && noassign_p (entry) == 0)
2882 VUNSETATTR (entry, att_invisible);
2883
2884 ASSIGN_RETURN (retval);
2885 }
2886
2887 /* Perform the assignment statement in STRING, and expand the
2888 right side by doing tilde, command and parameter expansion. */
2889 int
2890 do_assignment (string)
2891 char *string;
2892 {
2893 WORD_DESC td;
2894
2895 td.flags = W_ASSIGNMENT;
2896 td.word = string;
2897
2898 return do_assignment_internal (&td, 1);
2899 }
2900
2901 int
2902 do_word_assignment (word, flags)
2903 WORD_DESC *word;
2904 int flags;
2905 {
2906 return do_assignment_internal (word, 1);
2907 }
2908
2909 /* Given STRING, an assignment string, get the value of the right side
2910 of the `=', and bind it to the left side. Do not perform any word
2911 expansions on the right hand side. */
2912 int
2913 do_assignment_no_expand (string)
2914 char *string;
2915 {
2916 WORD_DESC td;
2917
2918 td.flags = W_ASSIGNMENT;
2919 td.word = string;
2920
2921 return (do_assignment_internal (&td, 0));
2922 }
2923
2924 /***************************************************
2925 * *
2926 * Functions to manage the positional parameters *
2927 * *
2928 ***************************************************/
2929
2930 /* Return the word list that corresponds to `$*'. */
2931 WORD_LIST *
2932 list_rest_of_args ()
2933 {
2934 register WORD_LIST *list, *args;
2935 int i;
2936
2937 /* Break out of the loop as soon as one of the dollar variables is null. */
2938 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
2939 list = make_word_list (make_bare_word (dollar_vars[i]), list);
2940
2941 for (args = rest_of_args; args; args = args->next)
2942 list = make_word_list (make_bare_word (args->word->word), list);
2943
2944 return (REVERSE_LIST (list, WORD_LIST *));
2945 }
2946
2947 int
2948 number_of_args ()
2949 {
2950 register WORD_LIST *list;
2951 int n;
2952
2953 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
2954 ;
2955 for (list = rest_of_args; list; list = list->next)
2956 n++;
2957 return n;
2958 }
2959
2960 /* Return the value of a positional parameter. This handles values > 10. */
2961 char *
2962 get_dollar_var_value (ind)
2963 intmax_t ind;
2964 {
2965 char *temp;
2966 WORD_LIST *p;
2967
2968 if (ind < 10)
2969 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
2970 else /* We want something like ${11} */
2971 {
2972 ind -= 10;
2973 for (p = rest_of_args; p && ind--; p = p->next)
2974 ;
2975 temp = p ? savestring (p->word->word) : (char *)NULL;
2976 }
2977 return (temp);
2978 }
2979
2980 /* Make a single large string out of the dollar digit variables,
2981 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
2982 case of "$*" with respect to IFS. */
2983 char *
2984 string_rest_of_args (dollar_star)
2985 int dollar_star;
2986 {
2987 register WORD_LIST *list;
2988 char *string;
2989
2990 list = list_rest_of_args ();
2991 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
2992 dispose_words (list);
2993 return (string);
2994 }
2995
2996 /* Return a string containing the positional parameters from START to
2997 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
2998 which only makes a difference if QUOTED is non-zero. If QUOTED includes
2999 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3000 no quoting chars are added. */
3001 static char *
3002 pos_params (string, start, end, quoted)
3003 char *string;
3004 int start, end, quoted;
3005 {
3006 WORD_LIST *save, *params, *h, *t;
3007 char *ret;
3008 int i;
3009
3010 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3011 if (start == end)
3012 return ((char *)NULL);
3013
3014 save = params = list_rest_of_args ();
3015 if (save == 0)
3016 return ((char *)NULL);
3017
3018 if (start == 0) /* handle ${@:0[:x]} specially */
3019 {
3020 t = make_word_list (make_word (dollar_vars[0]), params);
3021 save = params = t;
3022 }
3023
3024 for (i = start ? 1 : 0; params && i < start; i++)
3025 params = params->next;
3026 if (params == 0)
3027 return ((char *)NULL);
3028 for (h = t = params; params && i < end; i++)
3029 {
3030 t = params;
3031 params = params->next;
3032 }
3033
3034 t->next = (WORD_LIST *)NULL;
3035
3036 ret = string_list_pos_params (string[0], h, quoted);
3037
3038 if (t != params)
3039 t->next = params;
3040
3041 dispose_words (save);
3042 return (ret);
3043 }
3044
3045 /******************************************************************/
3046 /* */
3047 /* Functions to expand strings to strings or WORD_LISTs */
3048 /* */
3049 /******************************************************************/
3050
3051 #if defined (PROCESS_SUBSTITUTION)
3052 #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
3053 #else
3054 #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
3055 #endif
3056
3057 /* If there are any characters in STRING that require full expansion,
3058 then call FUNC to expand STRING; otherwise just perform quote
3059 removal if necessary. This returns a new string. */
3060 static char *
3061 expand_string_if_necessary (string, quoted, func)
3062 char *string;
3063 int quoted;
3064 EXPFUNC *func;
3065 {
3066 WORD_LIST *list;
3067 size_t slen;
3068 int i, saw_quote;
3069 char *ret;
3070 DECLARE_MBSTATE;
3071
3072 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3073 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3074 i = saw_quote = 0;
3075 while (string[i])
3076 {
3077 if (EXP_CHAR (string[i]))
3078 break;
3079 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3080 saw_quote = 1;
3081 ADVANCE_CHAR (string, slen, i);
3082 }
3083
3084 if (string[i])
3085 {
3086 list = (*func) (string, quoted);
3087 if (list)
3088 {
3089 ret = string_list (list);
3090 dispose_words (list);
3091 }
3092 else
3093 ret = (char *)NULL;
3094 }
3095 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3096 ret = string_quote_removal (string, quoted);
3097 else
3098 ret = savestring (string);
3099
3100 return ret;
3101 }
3102
3103 static inline char *
3104 expand_string_to_string_internal (string, quoted, func)
3105 char *string;
3106 int quoted;
3107 EXPFUNC *func;
3108 {
3109 WORD_LIST *list;
3110 char *ret;
3111
3112 if (string == 0 || *string == '\0')
3113 return ((char *)NULL);
3114
3115 list = (*func) (string, quoted);
3116 if (list)
3117 {
3118 ret = string_list (list);
3119 dispose_words (list);
3120 }
3121 else
3122 ret = (char *)NULL;
3123
3124 return (ret);
3125 }
3126
3127 char *
3128 expand_string_to_string (string, quoted)
3129 char *string;
3130 int quoted;
3131 {
3132 return (expand_string_to_string_internal (string, quoted, expand_string));
3133 }
3134
3135 char *
3136 expand_string_unsplit_to_string (string, quoted)
3137 char *string;
3138 int quoted;
3139 {
3140 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3141 }
3142
3143 char *
3144 expand_assignment_string_to_string (string, quoted)
3145 char *string;
3146 int quoted;
3147 {
3148 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3149 }
3150
3151 char *
3152 expand_arith_string (string, quoted)
3153 char *string;
3154 int quoted;
3155 {
3156 WORD_DESC td;
3157 WORD_LIST *list, *tlist;
3158 size_t slen;
3159 int i, saw_quote;
3160 char *ret;
3161 DECLARE_MBSTATE;
3162
3163 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3164 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3165 i = saw_quote = 0;
3166 while (string[i])
3167 {
3168 if (EXP_CHAR (string[i]))
3169 break;
3170 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3171 saw_quote = 1;
3172 ADVANCE_CHAR (string, slen, i);
3173 }
3174
3175 if (string[i])
3176 {
3177 /* This is expanded version of expand_string_internal as it's called by
3178 expand_string_leave_quoted */
3179 td.flags = W_NOPROCSUB; /* don't want process substitution */
3180 td.word = savestring (string);
3181 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3182 /* This takes care of the calls from expand_string_leave_quoted and
3183 expand_string */
3184 if (list)
3185 {
3186 tlist = word_list_split (list);
3187 dispose_words (list);
3188 list = tlist;
3189 if (list)
3190 dequote_list (list);
3191 }
3192 /* This comes from expand_string_if_necessary */
3193 if (list)
3194 {
3195 ret = string_list (list);
3196 dispose_words (list);
3197 }
3198 else
3199 ret = (char *)NULL;
3200 FREE (td.word);
3201 }
3202 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3203 ret = string_quote_removal (string, quoted);
3204 else
3205 ret = savestring (string);
3206
3207 return ret;
3208 }
3209
3210 #if defined (COND_COMMAND)
3211 /* Just remove backslashes in STRING. Returns a new string. */
3212 char *
3213 remove_backslashes (string)
3214 char *string;
3215 {
3216 char *r, *ret, *s;
3217
3218 r = ret = (char *)xmalloc (strlen (string) + 1);
3219 for (s = string; s && *s; )
3220 {
3221 if (*s == '\\')
3222 s++;
3223 if (*s == 0)
3224 break;
3225 *r++ = *s++;
3226 }
3227 *r = '\0';
3228 return ret;
3229 }
3230
3231 /* This needs better error handling. */
3232 /* Expand W for use as an argument to a unary or binary operator in a
3233 [[...]] expression. If SPECIAL is 1, this is the rhs argument
3234 to the != or == operator, and should be treated as a pattern. In
3235 this case, we quote the string specially for the globbing code. If
3236 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3237 be quoted appropriately for regcomp/regexec. The caller is responsible
3238 for removing the backslashes if the unquoted word is needed later. */
3239 char *
3240 cond_expand_word (w, special)
3241 WORD_DESC *w;
3242 int special;
3243 {
3244 char *r, *p;
3245 WORD_LIST *l;
3246 int qflags;
3247
3248 if (w->word == 0 || w->word[0] == '\0')
3249 return ((char *)NULL);
3250
3251 w->flags |= W_NOSPLIT2;
3252 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3253 if (l)
3254 {
3255 if (special == 0) /* LHS */
3256 {
3257 dequote_list (l);
3258 r = string_list (l);
3259 }
3260 else
3261 {
3262 /* Need to figure out whether or not we should call dequote_escapes
3263 or a new dequote_ctlnul function here, and under what
3264 circumstances. */
3265 qflags = QGLOB_CVTNULL;
3266 if (special == 2)
3267 qflags |= QGLOB_REGEXP;
3268 p = string_list (l);
3269 r = quote_string_for_globbing (p, qflags);
3270 free (p);
3271 }
3272 dispose_words (l);
3273 }
3274 else
3275 r = (char *)NULL;
3276
3277 return r;
3278 }
3279 #endif
3280
3281 /* Call expand_word_internal to expand W and handle error returns.
3282 A convenience function for functions that don't want to handle
3283 any errors or free any memory before aborting. */
3284 static WORD_LIST *
3285 call_expand_word_internal (w, q, i, c, e)
3286 WORD_DESC *w;
3287 int q, i, *c, *e;
3288 {
3289 WORD_LIST *result;
3290
3291 result = expand_word_internal (w, q, i, c, e);
3292 if (result == &expand_word_error || result == &expand_word_fatal)
3293 {
3294 /* By convention, each time this error is returned, w->word has
3295 already been freed (it sometimes may not be in the fatal case,
3296 but that doesn't result in a memory leak because we're going
3297 to exit in most cases). */
3298 w->word = (char *)NULL;
3299 last_command_exit_value = EXECUTION_FAILURE;
3300 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
3301 /* NOTREACHED */
3302 return (NULL);
3303 }
3304 else
3305 return (result);
3306 }
3307
3308 /* Perform parameter expansion, command substitution, and arithmetic
3309 expansion on STRING, as if it were a word. Leave the result quoted.
3310 Since this does not perform word splitting, it leaves quoted nulls
3311 in the result. */
3312 static WORD_LIST *
3313 expand_string_internal (string, quoted)
3314 char *string;
3315 int quoted;
3316 {
3317 WORD_DESC td;
3318 WORD_LIST *tresult;
3319
3320 if (string == 0 || *string == 0)
3321 return ((WORD_LIST *)NULL);
3322
3323 td.flags = 0;
3324 td.word = savestring (string);
3325
3326 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3327
3328 FREE (td.word);
3329 return (tresult);
3330 }
3331
3332 /* Expand STRING by performing parameter expansion, command substitution,
3333 and arithmetic expansion. Dequote the resulting WORD_LIST before
3334 returning it, but do not perform word splitting. The call to
3335 remove_quoted_nulls () is in here because word splitting normally
3336 takes care of quote removal. */
3337 WORD_LIST *
3338 expand_string_unsplit (string, quoted)
3339 char *string;
3340 int quoted;
3341 {
3342 WORD_LIST *value;
3343
3344 if (string == 0 || *string == '\0')
3345 return ((WORD_LIST *)NULL);
3346
3347 expand_no_split_dollar_star = 1;
3348 value = expand_string_internal (string, quoted);
3349 expand_no_split_dollar_star = 0;
3350
3351 if (value)
3352 {
3353 if (value->word)
3354 {
3355 remove_quoted_nulls (value->word->word);
3356 value->word->flags &= ~W_HASQUOTEDNULL;
3357 }
3358 dequote_list (value);
3359 }
3360 return (value);
3361 }
3362
3363 /* Expand the rhs of an assignment statement */
3364 WORD_LIST *
3365 expand_string_assignment (string, quoted)
3366 char *string;
3367 int quoted;
3368 {
3369 WORD_DESC td;
3370 WORD_LIST *value;
3371
3372 if (string == 0 || *string == '\0')
3373 return ((WORD_LIST *)NULL);
3374
3375 expand_no_split_dollar_star = 1;
3376
3377 td.flags = W_ASSIGNRHS;
3378 td.word = savestring (string);
3379 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3380 FREE (td.word);
3381
3382 expand_no_split_dollar_star = 0;
3383
3384 if (value)
3385 {
3386 if (value->word)
3387 {
3388 remove_quoted_nulls (value->word->word);
3389 value->word->flags &= ~W_HASQUOTEDNULL;
3390 }
3391 dequote_list (value);
3392 }
3393 return (value);
3394 }
3395
3396
3397 /* Expand one of the PS? prompt strings. This is a sort of combination of
3398 expand_string_unsplit and expand_string_internal, but returns the
3399 passed string when an error occurs. Might want to trap other calls
3400 to jump_to_top_level here so we don't endlessly loop. */
3401 WORD_LIST *
3402 expand_prompt_string (string, quoted, wflags)
3403 char *string;
3404 int quoted;
3405 int wflags;
3406 {
3407 WORD_LIST *value;
3408 WORD_DESC td;
3409
3410 if (string == 0 || *string == 0)
3411 return ((WORD_LIST *)NULL);
3412
3413 td.flags = wflags;
3414 td.word = savestring (string);
3415
3416 no_longjmp_on_fatal_error = 1;
3417 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3418 no_longjmp_on_fatal_error = 0;
3419
3420 if (value == &expand_word_error || value == &expand_word_fatal)
3421 {
3422 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3423 return value;
3424 }
3425 FREE (td.word);
3426 if (value)
3427 {
3428 if (value->word)
3429 {
3430 remove_quoted_nulls (value->word->word);
3431 value->word->flags &= ~W_HASQUOTEDNULL;
3432 }
3433 dequote_list (value);
3434 }
3435 return (value);
3436 }
3437
3438 /* Expand STRING just as if you were expanding a word, but do not dequote
3439 the resultant WORD_LIST. This is called only from within this file,
3440 and is used to correctly preserve quoted characters when expanding
3441 things like ${1+"$@"}. This does parameter expansion, command
3442 substitution, arithmetic expansion, and word splitting. */
3443 static WORD_LIST *
3444 expand_string_leave_quoted (string, quoted)
3445 char *string;
3446 int quoted;
3447 {
3448 WORD_LIST *tlist;
3449 WORD_LIST *tresult;
3450
3451 if (string == 0 || *string == '\0')
3452 return ((WORD_LIST *)NULL);
3453
3454 tlist = expand_string_internal (string, quoted);
3455
3456 if (tlist)
3457 {
3458 tresult = word_list_split (tlist);
3459 dispose_words (tlist);
3460 return (tresult);
3461 }
3462 return ((WORD_LIST *)NULL);
3463 }
3464
3465 /* This does not perform word splitting or dequote the WORD_LIST
3466 it returns. */
3467 static WORD_LIST *
3468 expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at)
3469 char *string;
3470 int quoted, *dollar_at_p, *has_dollar_at;
3471 {
3472 WORD_DESC td;
3473 WORD_LIST *tresult;
3474
3475 if (string == 0 || *string == '\0')
3476 return (WORD_LIST *)NULL;
3477
3478 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
3479 td.word = string;
3480 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at);
3481 return (tresult);
3482 }
3483
3484 /* Expand STRING just as if you were expanding a word. This also returns
3485 a list of words. Note that filename globbing is *NOT* done for word
3486 or string expansion, just when the shell is expanding a command. This
3487 does parameter expansion, command substitution, arithmetic expansion,
3488 and word splitting. Dequote the resultant WORD_LIST before returning. */
3489 WORD_LIST *
3490 expand_string (string, quoted)
3491 char *string;
3492 int quoted;
3493 {
3494 WORD_LIST *result;
3495
3496 if (string == 0 || *string == '\0')
3497 return ((WORD_LIST *)NULL);
3498
3499 result = expand_string_leave_quoted (string, quoted);
3500 return (result ? dequote_list (result) : result);
3501 }
3502
3503 /***************************************************
3504 * *
3505 * Functions to handle quoting chars *
3506 * *
3507 ***************************************************/
3508
3509 /* Conventions:
3510
3511 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3512 The parser passes CTLNUL as CTLESC CTLNUL. */
3513
3514 /* Quote escape characters in string s, but no other characters. This is
3515 used to protect CTLESC and CTLNUL in variable values from the rest of
3516 the word expansion process after the variable is expanded (word splitting
3517 and filename generation). If IFS is null, we quote spaces as well, just
3518 in case we split on spaces later (in the case of unquoted $@, we will
3519 eventually attempt to split the entire word on spaces). Corresponding
3520 code exists in dequote_escapes. Even if we don't end up splitting on
3521 spaces, quoting spaces is not a problem. This should never be called on
3522 a string that is quoted with single or double quotes or part of a here
3523 document (effectively double-quoted). */
3524 char *
3525 quote_escapes (string)
3526 char *string;
3527 {
3528 register char *s, *t;
3529 size_t slen;
3530 char *result, *send;
3531 int quote_spaces, skip_ctlesc, skip_ctlnul;
3532 DECLARE_MBSTATE;
3533
3534 slen = strlen (string);
3535 send = string + slen;
3536
3537 quote_spaces = (ifs_value && *ifs_value == 0);
3538
3539 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3540 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3541
3542 t = result = (char *)xmalloc ((slen * 2) + 1);
3543 s = string;
3544
3545 while (*s)
3546 {
3547 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
3548 *t++ = CTLESC;
3549 COPY_CHAR_P (t, s, send);
3550 }
3551 *t = '\0';
3552
3553 return (result);
3554 }
3555
3556 static WORD_LIST *
3557 list_quote_escapes (list)
3558 WORD_LIST *list;
3559 {
3560 register WORD_LIST *w;
3561 char *t;
3562
3563 for (w = list; w; w = w->next)
3564 {
3565 t = w->word->word;
3566 w->word->word = quote_escapes (t);
3567 free (t);
3568 }
3569 return list;
3570 }
3571
3572 /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3573
3574 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3575 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3576 data stream pass through properly.
3577
3578 We need to remove doubled CTLESC characters inside quoted strings before
3579 quoting the entire string, so we do not double the number of CTLESC
3580 characters.
3581
3582 Also used by parts of the pattern substitution code. */
3583 char *
3584 dequote_escapes (string)
3585 char *string;
3586 {
3587 register char *s, *t, *s1;
3588 size_t slen;
3589 char *result, *send;
3590 int quote_spaces;
3591 DECLARE_MBSTATE;
3592
3593 if (string == 0)
3594 return string;
3595
3596 slen = strlen (string);
3597 send = string + slen;
3598
3599 t = result = (char *)xmalloc (slen + 1);
3600
3601 if (strchr (string, CTLESC) == 0)
3602 return (strcpy (result, string));
3603
3604 quote_spaces = (ifs_value && *ifs_value == 0);
3605
3606 s = string;
3607 while (*s)
3608 {
3609 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
3610 {
3611 s++;
3612 if (*s == '\0')
3613 break;
3614 }
3615 COPY_CHAR_P (t, s, send);
3616 }
3617 *t = '\0';
3618
3619 return result;
3620 }
3621
3622 static WORD_LIST *
3623 list_dequote_escapes (list)
3624 WORD_LIST *list;
3625 {
3626 register WORD_LIST *w;
3627 char *t;
3628
3629 for (w = list; w; w = w->next)
3630 {
3631 t = w->word->word;
3632 w->word->word = dequote_escapes (t);
3633 free (t);
3634 }
3635 return list;
3636 }
3637
3638 /* Return a new string with the quoted representation of character C.
3639 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3640 set in any resultant WORD_DESC where this value is the word. */
3641 static char *
3642 make_quoted_char (c)
3643 int c;
3644 {
3645 char *temp;
3646
3647 temp = (char *)xmalloc (3);
3648 if (c == 0)
3649 {
3650 temp[0] = CTLNUL;
3651 temp[1] = '\0';
3652 }
3653 else
3654 {
3655 temp[0] = CTLESC;
3656 temp[1] = c;
3657 temp[2] = '\0';
3658 }
3659 return (temp);
3660 }
3661
3662 /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3663 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3664 this value is the word. */
3665 char *
3666 quote_string (string)
3667 char *string;
3668 {
3669 register char *t;
3670 size_t slen;
3671 char *result, *send;
3672
3673 if (*string == 0)
3674 {
3675 result = (char *)xmalloc (2);
3676 result[0] = CTLNUL;
3677 result[1] = '\0';
3678 }
3679 else
3680 {
3681 DECLARE_MBSTATE;
3682
3683 slen = strlen (string);
3684 send = string + slen;
3685
3686 result = (char *)xmalloc ((slen * 2) + 1);
3687
3688 for (t = result; string < send; )
3689 {
3690 *t++ = CTLESC;
3691 COPY_CHAR_P (t, string, send);
3692 }
3693 *t = '\0';
3694 }
3695 return (result);
3696 }
3697
3698 /* De-quote quoted characters in STRING. */
3699 char *
3700 dequote_string (string)
3701 char *string;
3702 {
3703 register char *s, *t;
3704 size_t slen;
3705 char *result, *send;
3706 DECLARE_MBSTATE;
3707
3708 slen = strlen (string);
3709
3710 t = result = (char *)xmalloc (slen + 1);
3711
3712 if (QUOTED_NULL (string))
3713 {
3714 result[0] = '\0';
3715 return (result);
3716 }
3717
3718 /* If no character in the string can be quoted, don't bother examining
3719 each character. Just return a copy of the string passed to us. */
3720 if (strchr (string, CTLESC) == NULL)
3721 return (strcpy (result, string));
3722
3723 send = string + slen;
3724 s = string;
3725 while (*s)
3726 {
3727 if (*s == CTLESC)
3728 {
3729 s++;
3730 if (*s == '\0')
3731 break;
3732 }
3733 COPY_CHAR_P (t, s, send);
3734 }
3735
3736 *t = '\0';
3737 return (result);
3738 }
3739
3740 /* Quote the entire WORD_LIST list. */
3741 static WORD_LIST *
3742 quote_list (list)
3743 WORD_LIST *list;
3744 {
3745 register WORD_LIST *w;
3746 char *t;
3747
3748 for (w = list; w; w = w->next)
3749 {
3750 t = w->word->word;
3751 w->word->word = quote_string (t);
3752 if (*t == 0)
3753 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
3754 w->word->flags |= W_QUOTED;
3755 free (t);
3756 }
3757 return list;
3758 }
3759
3760 /* De-quote quoted characters in each word in LIST. */
3761 WORD_LIST *
3762 dequote_list (list)
3763 WORD_LIST *list;
3764 {
3765 register char *s;
3766 register WORD_LIST *tlist;
3767
3768 for (tlist = list; tlist; tlist = tlist->next)
3769 {
3770 s = dequote_string (tlist->word->word);
3771 if (QUOTED_NULL (tlist->word->word))
3772 tlist->word->flags &= ~W_HASQUOTEDNULL;
3773 free (tlist->word->word);
3774 tlist->word->word = s;
3775 }
3776 return list;
3777 }
3778
3779 /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
3780 string. */
3781 char *
3782 remove_quoted_escapes (string)
3783 char *string;
3784 {
3785 char *t;
3786
3787 if (string)
3788 {
3789 t = dequote_escapes (string);
3790 strcpy (string, t);
3791 free (t);
3792 }
3793
3794 return (string);
3795 }
3796
3797 /* Perform quoted null character removal on STRING. We don't allow any
3798 quoted null characters in the middle or at the ends of strings because
3799 of how expand_word_internal works. remove_quoted_nulls () turns
3800 STRING into an empty string iff it only consists of a quoted null,
3801 and removes all unquoted CTLNUL characters. */
3802 char *
3803 remove_quoted_nulls (string)
3804 char *string;
3805 {
3806 register size_t slen;
3807 register int i, j, prev_i;
3808 DECLARE_MBSTATE;
3809
3810 if (strchr (string, CTLNUL) == 0) /* XXX */
3811 return string; /* XXX */
3812
3813 slen = strlen (string);
3814 i = j = 0;
3815
3816 while (i < slen)
3817 {
3818 if (string[i] == CTLESC)
3819 {
3820 /* Old code had j++, but we cannot assume that i == j at this
3821 point -- what if a CTLNUL has already been removed from the
3822 string? We don't want to drop the CTLESC or recopy characters
3823 that we've already copied down. */
3824 i++; string[j++] = CTLESC;
3825 if (i == slen)
3826 break;
3827 }
3828 else if (string[i] == CTLNUL)
3829 {
3830 i++;
3831 continue;
3832 }
3833
3834 prev_i = i;
3835 ADVANCE_CHAR (string, slen, i);
3836 if (j < prev_i)
3837 {
3838 do string[j++] = string[prev_i++]; while (prev_i < i);
3839 }
3840 else
3841 j = i;
3842 }
3843 string[j] = '\0';
3844
3845 return (string);
3846 }
3847
3848 /* Perform quoted null character removal on each element of LIST.
3849 This modifies LIST. */
3850 void
3851 word_list_remove_quoted_nulls (list)
3852 WORD_LIST *list;
3853 {
3854 register WORD_LIST *t;
3855
3856 for (t = list; t; t = t->next)
3857 {
3858 remove_quoted_nulls (t->word->word);
3859 t->word->flags &= ~W_HASQUOTEDNULL;
3860 }
3861 }
3862
3863 /* **************************************************************** */
3864 /* */
3865 /* Functions for Matching and Removing Patterns */
3866 /* */
3867 /* **************************************************************** */
3868
3869 #if defined (HANDLE_MULTIBYTE)
3870 #if 0 /* Currently unused */
3871 static unsigned char *
3872 mb_getcharlens (string, len)
3873 char *string;
3874 int len;
3875 {
3876 int i, offset, last;
3877 unsigned char *ret;
3878 char *p;
3879 DECLARE_MBSTATE;
3880
3881 i = offset = 0;
3882 last = 0;
3883 ret = (unsigned char *)xmalloc (len);
3884 memset (ret, 0, len);
3885 while (string[last])
3886 {
3887 ADVANCE_CHAR (string, len, offset);
3888 ret[last] = offset - last;
3889 last = offset;
3890 }
3891 return ret;
3892 }
3893 #endif
3894 #endif
3895
3896 /* Remove the portion of PARAM matched by PATTERN according to OP, where OP
3897 can have one of 4 values:
3898 RP_LONG_LEFT remove longest matching portion at start of PARAM
3899 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
3900 RP_LONG_RIGHT remove longest matching portion at end of PARAM
3901 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
3902 */
3903
3904 #define RP_LONG_LEFT 1
3905 #define RP_SHORT_LEFT 2
3906 #define RP_LONG_RIGHT 3
3907 #define RP_SHORT_RIGHT 4
3908
3909 /* Returns its first argument if nothing matched; new memory otherwise */
3910 static char *
3911 remove_upattern (param, pattern, op)
3912 char *param, *pattern;
3913 int op;
3914 {
3915 register int len;
3916 register char *end;
3917 register char *p, *ret, c;
3918
3919 len = STRLEN (param);
3920 end = param + len;
3921
3922 switch (op)
3923 {
3924 case RP_LONG_LEFT: /* remove longest match at start */
3925 for (p = end; p >= param; p--)
3926 {
3927 c = *p; *p = '\0';
3928 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3929 {
3930 *p = c;
3931 return (savestring (p));
3932 }
3933 *p = c;
3934
3935 }
3936 break;
3937
3938 case RP_SHORT_LEFT: /* remove shortest match at start */
3939 for (p = param; p <= end; p++)
3940 {
3941 c = *p; *p = '\0';
3942 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3943 {
3944 *p = c;
3945 return (savestring (p));
3946 }
3947 *p = c;
3948 }
3949 break;
3950
3951 case RP_LONG_RIGHT: /* remove longest match at end */
3952 for (p = param; p <= end; p++)
3953 {
3954 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3955 {
3956 c = *p; *p = '\0';
3957 ret = savestring (param);
3958 *p = c;
3959 return (ret);
3960 }
3961 }
3962 break;
3963
3964 case RP_SHORT_RIGHT: /* remove shortest match at end */
3965 for (p = end; p >= param; p--)
3966 {
3967 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3968 {
3969 c = *p; *p = '\0';
3970 ret = savestring (param);
3971 *p = c;
3972 return (ret);
3973 }
3974 }
3975 break;
3976 }
3977
3978 return (param); /* no match, return original string */
3979 }
3980
3981 #if defined (HANDLE_MULTIBYTE)
3982 /* Returns its first argument if nothing matched; new memory otherwise */
3983 static wchar_t *
3984 remove_wpattern (wparam, wstrlen, wpattern, op)
3985 wchar_t *wparam;
3986 size_t wstrlen;
3987 wchar_t *wpattern;
3988 int op;
3989 {
3990 wchar_t wc, *ret;
3991 int n;
3992
3993 switch (op)
3994 {
3995 case RP_LONG_LEFT: /* remove longest match at start */
3996 for (n = wstrlen; n >= 0; n--)
3997 {
3998 wc = wparam[n]; wparam[n] = L'\0';
3999 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4000 {
4001 wparam[n] = wc;
4002 return (wcsdup (wparam + n));
4003 }
4004 wparam[n] = wc;
4005 }
4006 break;
4007
4008 case RP_SHORT_LEFT: /* remove shortest match at start */
4009 for (n = 0; n <= wstrlen; n++)
4010 {
4011 wc = wparam[n]; wparam[n] = L'\0';
4012 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4013 {
4014 wparam[n] = wc;
4015 return (wcsdup (wparam + n));
4016 }
4017 wparam[n] = wc;
4018 }
4019 break;
4020
4021 case RP_LONG_RIGHT: /* remove longest match at end */
4022 for (n = 0; n <= wstrlen; n++)
4023 {
4024 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4025 {
4026 wc = wparam[n]; wparam[n] = L'\0';
4027 ret = wcsdup (wparam);
4028 wparam[n] = wc;
4029 return (ret);
4030 }
4031 }
4032 break;
4033
4034 case RP_SHORT_RIGHT: /* remove shortest match at end */
4035 for (n = wstrlen; n >= 0; n--)
4036 {
4037 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4038 {
4039 wc = wparam[n]; wparam[n] = L'\0';
4040 ret = wcsdup (wparam);
4041 wparam[n] = wc;
4042 return (ret);
4043 }
4044 }
4045 break;
4046 }
4047
4048 return (wparam); /* no match, return original string */
4049 }
4050 #endif /* HANDLE_MULTIBYTE */
4051
4052 static char *
4053 remove_pattern (param, pattern, op)
4054 char *param, *pattern;
4055 int op;
4056 {
4057 char *xret;
4058
4059 if (param == NULL)
4060 return (param);
4061 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4062 return (savestring (param));
4063
4064 #if defined (HANDLE_MULTIBYTE)
4065 if (MB_CUR_MAX > 1)
4066 {
4067 wchar_t *ret, *oret;
4068 size_t n;
4069 wchar_t *wparam, *wpattern;
4070 mbstate_t ps;
4071
4072 n = xdupmbstowcs (&wpattern, NULL, pattern);
4073 if (n == (size_t)-1)
4074 {
4075 xret = remove_upattern (param, pattern, op);
4076 return ((xret == param) ? savestring (param) : xret);
4077 }
4078 n = xdupmbstowcs (&wparam, NULL, param);
4079
4080 if (n == (size_t)-1)
4081 {
4082 free (wpattern);
4083 xret = remove_upattern (param, pattern, op);
4084 return ((xret == param) ? savestring (param) : xret);
4085 }
4086 oret = ret = remove_wpattern (wparam, n, wpattern, op);
4087 /* Don't bother to convert wparam back to multibyte string if nothing
4088 matched; just return copy of original string */
4089 if (ret == wparam)
4090 {
4091 free (wparam);
4092 free (wpattern);
4093 return (savestring (param));
4094 }
4095
4096 free (wparam);
4097 free (wpattern);
4098
4099 n = strlen (param);
4100 xret = (char *)xmalloc (n + 1);
4101 memset (&ps, '\0', sizeof (mbstate_t));
4102 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4103 xret[n] = '\0'; /* just to make sure */
4104 free (oret);
4105 return xret;
4106 }
4107 else
4108 #endif
4109 {
4110 xret = remove_upattern (param, pattern, op);
4111 return ((xret == param) ? savestring (param) : xret);
4112 }
4113 }
4114
4115 /* Match PAT anywhere in STRING and return the match boundaries.
4116 This returns 1 in case of a successful match, 0 otherwise. SP
4117 and EP are pointers into the string where the match begins and
4118 ends, respectively. MTYPE controls what kind of match is attempted.
4119 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4120 of the string, respectively. The longest match is returned. */
4121 static int
4122 match_upattern (string, pat, mtype, sp, ep)
4123 char *string, *pat;
4124 int mtype;
4125 char **sp, **ep;
4126 {
4127 int c, len, mlen;
4128 register char *p, *p1, *npat;
4129 char *end;
4130 int n1;
4131
4132 /* If the pattern doesn't match anywhere in the string, go ahead and
4133 short-circuit right away. A minor optimization, saves a bunch of
4134 unnecessary calls to strmatch (up to N calls for a string of N
4135 characters) if the match is unsuccessful. To preserve the semantics
4136 of the substring matches below, we make sure that the pattern has
4137 `*' as first and last character, making a new pattern if necessary. */
4138 /* XXX - check this later if I ever implement `**' with special meaning,
4139 since this will potentially result in `**' at the beginning or end */
4140 len = STRLEN (pat);
4141 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
4142 {
4143 p = npat = (char *)xmalloc (len + 3);
4144 p1 = pat;
4145 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
4146 *p++ = '*';
4147 while (*p1)
4148 *p++ = *p1++;
4149 if (p1[-1] != '*' || p[-2] == '\\')
4150 *p++ = '*';
4151 *p = '\0';
4152 }
4153 else
4154 npat = pat;
4155 c = strmatch (npat, string, FNMATCH_EXTFLAG);
4156 if (npat != pat)
4157 free (npat);
4158 if (c == FNM_NOMATCH)
4159 return (0);
4160
4161 len = STRLEN (string);
4162 end = string + len;
4163
4164 mlen = umatchlen (pat, len);
4165
4166 switch (mtype)
4167 {
4168 case MATCH_ANY:
4169 for (p = string; p <= end; p++)
4170 {
4171 if (match_pattern_char (pat, p))
4172 {
4173 p1 = (mlen == -1) ? end : p + mlen;
4174 /* p1 - p = length of portion of string to be considered
4175 p = current position in string
4176 mlen = number of characters consumed by match (-1 for entire string)
4177 end = end of string
4178 we want to break immediately if the potential match len
4179 is greater than the number of characters remaining in the
4180 string
4181 */
4182 if (p1 > end)
4183 break;
4184 for ( ; p1 >= p; p1--)
4185 {
4186 c = *p1; *p1 = '\0';
4187 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4188 {
4189 *p1 = c;
4190 *sp = p;
4191 *ep = p1;
4192 return 1;
4193 }
4194 *p1 = c;
4195 #if 1
4196 /* If MLEN != -1, we have a fixed length pattern. */
4197 if (mlen != -1)
4198 break;
4199 #endif
4200 }
4201 }
4202 }
4203
4204 return (0);
4205
4206 case MATCH_BEG:
4207 if (match_pattern_char (pat, string) == 0)
4208 return (0);
4209
4210 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
4211 {
4212 c = *p; *p = '\0';
4213 if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0)
4214 {
4215 *p = c;
4216 *sp = string;
4217 *ep = p;
4218 return 1;
4219 }
4220 *p = c;
4221 /* If MLEN != -1, we have a fixed length pattern. */
4222 if (mlen != -1)
4223 break;
4224 }
4225
4226 return (0);
4227
4228 case MATCH_END:
4229 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
4230 {
4231 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4232 {
4233 *sp = p;
4234 *ep = end;
4235 return 1;
4236 }
4237 /* If MLEN != -1, we have a fixed length pattern. */
4238 if (mlen != -1)
4239 break;
4240 }
4241
4242 return (0);
4243 }
4244
4245 return (0);
4246 }
4247
4248 #if defined (HANDLE_MULTIBYTE)
4249 /* Match WPAT anywhere in WSTRING and return the match boundaries.
4250 This returns 1 in case of a successful match, 0 otherwise. Wide
4251 character version. */
4252 static int
4253 match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4254 wchar_t *wstring;
4255 char **indices;
4256 size_t wstrlen;
4257 wchar_t *wpat;
4258 int mtype;
4259 char **sp, **ep;
4260 {
4261 wchar_t wc, *wp, *nwpat, *wp1;
4262 size_t len;
4263 int mlen;
4264 int n, n1, n2, simple;
4265
4266 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4267 #if defined (EXTENDED_GLOB)
4268 if (extended_glob)
4269 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
4270 #endif
4271
4272 /* If the pattern doesn't match anywhere in the string, go ahead and
4273 short-circuit right away. A minor optimization, saves a bunch of
4274 unnecessary calls to strmatch (up to N calls for a string of N
4275 characters) if the match is unsuccessful. To preserve the semantics
4276 of the substring matches below, we make sure that the pattern has
4277 `*' as first and last character, making a new pattern if necessary. */
4278 len = wcslen (wpat);
4279 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
4280 {
4281 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
4282 wp1 = wpat;
4283 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
4284 *wp++ = L'*';
4285 while (*wp1 != L'\0')
4286 *wp++ = *wp1++;
4287 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4288 *wp++ = L'*';
4289 *wp = '\0';
4290 }
4291 else
4292 nwpat = wpat;
4293 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG);
4294 if (nwpat != wpat)
4295 free (nwpat);
4296 if (len == FNM_NOMATCH)
4297 return (0);
4298
4299 mlen = wmatchlen (wpat, wstrlen);
4300
4301 /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
4302 switch (mtype)
4303 {
4304 case MATCH_ANY:
4305 for (n = 0; n <= wstrlen; n++)
4306 {
4307 n2 = simple ? (*wpat == wstring[n]) : match_pattern_wchar (wpat, wstring + n);
4308 if (n2)
4309 {
4310 n1 = (mlen == -1) ? wstrlen : n + mlen;
4311 if (n1 > wstrlen)
4312 break;
4313
4314 for ( ; n1 >= n; n1--)
4315 {
4316 wc = wstring[n1]; wstring[n1] = L'\0';
4317 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4318 {
4319 wstring[n1] = wc;
4320 *sp = indices[n];
4321 *ep = indices[n1];
4322 return 1;
4323 }
4324 wstring[n1] = wc;
4325 /* If MLEN != -1, we have a fixed length pattern. */
4326 if (mlen != -1)
4327 break;
4328 }
4329 }
4330 }
4331
4332 return (0);
4333
4334 case MATCH_BEG:
4335 if (match_pattern_wchar (wpat, wstring) == 0)
4336 return (0);
4337
4338 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
4339 {
4340 wc = wstring[n]; wstring[n] = L'\0';
4341 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0)
4342 {
4343 wstring[n] = wc;
4344 *sp = indices[0];
4345 *ep = indices[n];
4346 return 1;
4347 }
4348 wstring[n] = wc;
4349 /* If MLEN != -1, we have a fixed length pattern. */
4350 if (mlen != -1)
4351 break;
4352 }
4353
4354 return (0);
4355
4356 case MATCH_END:
4357 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
4358 {
4359 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4360 {
4361 *sp = indices[n];
4362 *ep = indices[wstrlen];
4363 return 1;
4364 }
4365 /* If MLEN != -1, we have a fixed length pattern. */
4366 if (mlen != -1)
4367 break;
4368 }
4369
4370 return (0);
4371 }
4372
4373 return (0);
4374 }
4375 #endif /* HANDLE_MULTIBYTE */
4376
4377 static int
4378 match_pattern (string, pat, mtype, sp, ep)
4379 char *string, *pat;
4380 int mtype;
4381 char **sp, **ep;
4382 {
4383 #if defined (HANDLE_MULTIBYTE)
4384 int ret;
4385 size_t n;
4386 wchar_t *wstring, *wpat;
4387 char **indices;
4388 size_t slen, plen, mslen, mplen;
4389 #endif
4390
4391 if (string == 0 || *string == 0 || pat == 0 || *pat == 0)
4392 return (0);
4393
4394 #if defined (HANDLE_MULTIBYTE)
4395 if (MB_CUR_MAX > 1)
4396 {
4397 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
4398 return (match_upattern (string, pat, mtype, sp, ep));
4399
4400 n = xdupmbstowcs (&wpat, NULL, pat);
4401 if (n == (size_t)-1)
4402 return (match_upattern (string, pat, mtype, sp, ep));
4403 n = xdupmbstowcs (&wstring, &indices, string);
4404 if (n == (size_t)-1)
4405 {
4406 free (wpat);
4407 return (match_upattern (string, pat, mtype, sp, ep));
4408 }
4409 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4410
4411 free (wpat);
4412 free (wstring);
4413 free (indices);
4414
4415 return (ret);
4416 }
4417 else
4418 #endif
4419 return (match_upattern (string, pat, mtype, sp, ep));
4420 }
4421
4422 static int
4423 getpatspec (c, value)
4424 int c;
4425 char *value;
4426 {
4427 if (c == '#')
4428 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4429 else /* c == '%' */
4430 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4431 }
4432
4433 /* Posix.2 says that the WORD should be run through tilde expansion,
4434 parameter expansion, command substitution and arithmetic expansion.
4435 This leaves the result quoted, so quote_string_for_globbing () has
4436 to be called to fix it up for strmatch (). If QUOTED is non-zero,
4437 it means that the entire expression was enclosed in double quotes.
4438 This means that quoting characters in the pattern do not make any
4439 special pattern characters quoted. For example, the `*' in the
4440 following retains its special meaning: "${foo#'*'}". */
4441 static char *
4442 getpattern (value, quoted, expandpat)
4443 char *value;
4444 int quoted, expandpat;
4445 {
4446 char *pat, *tword;
4447 WORD_LIST *l;
4448 #if 0
4449 int i;
4450 #endif
4451 /* There is a problem here: how to handle single or double quotes in the
4452 pattern string when the whole expression is between double quotes?
4453 POSIX.2 says that enclosing double quotes do not cause the pattern to
4454 be quoted, but does that leave us a problem with @ and array[@] and their
4455 expansions inside a pattern? */
4456 #if 0
4457 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4458 {
4459 i = 0;
4460 pat = string_extract_double_quoted (tword, &i, 1);
4461 free (tword);
4462 tword = pat;
4463 }
4464 #endif
4465
4466 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4467 word splitting. */
4468 l = *value ? expand_string_for_rhs (value,
4469 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
4470 (int *)NULL, (int *)NULL)
4471 : (WORD_LIST *)0;
4472 pat = string_list (l);
4473 dispose_words (l);
4474 if (pat)
4475 {
4476 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4477 free (pat);
4478 pat = tword;
4479 }
4480 return (pat);
4481 }
4482
4483 #if 0
4484 /* Handle removing a pattern from a string as a result of ${name%[%]value}
4485 or ${name#[#]value}. */
4486 static char *
4487 variable_remove_pattern (value, pattern, patspec, quoted)
4488 char *value, *pattern;
4489 int patspec, quoted;
4490 {
4491 char *tword;
4492
4493 tword = remove_pattern (value, pattern, patspec);
4494
4495 return (tword);
4496 }
4497 #endif
4498
4499 static char *
4500 list_remove_pattern (list, pattern, patspec, itype, quoted)
4501 WORD_LIST *list;
4502 char *pattern;
4503 int patspec, itype, quoted;
4504 {
4505 WORD_LIST *new, *l;
4506 WORD_DESC *w;
4507 char *tword;
4508
4509 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4510 {
4511 tword = remove_pattern (l->word->word, pattern, patspec);
4512 w = alloc_word_desc ();
4513 w->word = tword ? tword : savestring ("");
4514 new = make_word_list (w, new);
4515 }
4516
4517 l = REVERSE_LIST (new, WORD_LIST *);
4518 tword = string_list_pos_params (itype, l, quoted);
4519 dispose_words (l);
4520
4521 return (tword);
4522 }
4523
4524 static char *
4525 parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4526 int itype;
4527 char *pattern;
4528 int patspec, quoted;
4529 {
4530 char *ret;
4531 WORD_LIST *list;
4532
4533 list = list_rest_of_args ();
4534 if (list == 0)
4535 return ((char *)NULL);
4536 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4537 dispose_words (list);
4538 return (ret);
4539 }
4540
4541 #if defined (ARRAY_VARS)
4542 static char *
4543 array_remove_pattern (var, pattern, patspec, varname, quoted)
4544 SHELL_VAR *var;
4545 char *pattern;
4546 int patspec;
4547 char *varname; /* so we can figure out how it's indexed */
4548 int quoted;
4549 {
4550 ARRAY *a;
4551 HASH_TABLE *h;
4552 int itype;
4553 char *ret;
4554 WORD_LIST *list;
4555 SHELL_VAR *v;
4556
4557 /* compute itype from varname here */
4558 v = array_variable_part (varname, &ret, 0);
4559
4560 /* XXX */
4561 if (v && invisible_p (var))
4562 return ((char *)NULL);
4563
4564 itype = ret[0];
4565
4566 a = (v && array_p (v)) ? array_cell (v) : 0;
4567 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4568
4569 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4570 if (list == 0)
4571 return ((char *)NULL);
4572 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4573 dispose_words (list);
4574
4575 return ret;
4576 }
4577 #endif /* ARRAY_VARS */
4578
4579 static char *
4580 parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4581 char *varname, *value;
4582 int ind;
4583 char *patstr;
4584 int rtype, quoted, flags;
4585 {
4586 int vtype, patspec, starsub;
4587 char *temp1, *val, *pattern;
4588 SHELL_VAR *v;
4589
4590 if (value == 0)
4591 return ((char *)NULL);
4592
4593 this_command_name = varname;
4594
4595 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4596 if (vtype == -1)
4597 return ((char *)NULL);
4598
4599 starsub = vtype & VT_STARSUB;
4600 vtype &= ~VT_STARSUB;
4601
4602 patspec = getpatspec (rtype, patstr);
4603 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4604 patstr++;
4605
4606 /* Need to pass getpattern newly-allocated memory in case of expansion --
4607 the expansion code will free the passed string on an error. */
4608 temp1 = savestring (patstr);
4609 pattern = getpattern (temp1, quoted, 1);
4610 free (temp1);
4611
4612 temp1 = (char *)NULL; /* shut up gcc */
4613 switch (vtype)
4614 {
4615 case VT_VARIABLE:
4616 case VT_ARRAYMEMBER:
4617 temp1 = remove_pattern (val, pattern, patspec);
4618 if (vtype == VT_VARIABLE)
4619 FREE (val);
4620 if (temp1)
4621 {
4622 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
4623 ? quote_string (temp1)
4624 : quote_escapes (temp1);
4625 free (temp1);
4626 temp1 = val;
4627 }
4628 break;
4629 #if defined (ARRAY_VARS)
4630 case VT_ARRAYVAR:
4631 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
4632 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4633 {
4634 val = quote_escapes (temp1);
4635 free (temp1);
4636 temp1 = val;
4637 }
4638 break;
4639 #endif
4640 case VT_POSPARMS:
4641 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
4642 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4643 {
4644 val = quote_escapes (temp1);
4645 free (temp1);
4646 temp1 = val;
4647 }
4648 break;
4649 }
4650
4651 FREE (pattern);
4652 return temp1;
4653 }
4654
4655 /*******************************************
4656 * *
4657 * Functions to expand WORD_DESCs *
4658 * *
4659 *******************************************/
4660
4661 /* Expand WORD, performing word splitting on the result. This does
4662 parameter expansion, command substitution, arithmetic expansion,
4663 word splitting, and quote removal. */
4664
4665 WORD_LIST *
4666 expand_word (word, quoted)
4667 WORD_DESC *word;
4668 int quoted;
4669 {
4670 WORD_LIST *result, *tresult;
4671
4672 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4673 result = word_list_split (tresult);
4674 dispose_words (tresult);
4675 return (result ? dequote_list (result) : result);
4676 }
4677
4678 /* Expand WORD, but do not perform word splitting on the result. This
4679 does parameter expansion, command substitution, arithmetic expansion,
4680 and quote removal. */
4681 WORD_LIST *
4682 expand_word_unsplit (word, quoted)
4683 WORD_DESC *word;
4684 int quoted;
4685 {
4686 WORD_LIST *result;
4687
4688 expand_no_split_dollar_star = 1;
4689 #if defined (HANDLE_MULTIBYTE)
4690 if (ifs_firstc[0] == 0)
4691 #else
4692 if (ifs_firstc == 0)
4693 #endif
4694 word->flags |= W_NOSPLIT;
4695 word->flags |= W_NOSPLIT2;
4696 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4697 expand_no_split_dollar_star = 0;
4698
4699 return (result ? dequote_list (result) : result);
4700 }
4701
4702 /* Perform shell expansions on WORD, but do not perform word splitting or
4703 quote removal on the result. Virtually identical to expand_word_unsplit;
4704 could be combined if implementations don't diverge. */
4705 WORD_LIST *
4706 expand_word_leave_quoted (word, quoted)
4707 WORD_DESC *word;
4708 int quoted;
4709 {
4710 WORD_LIST *result;
4711
4712 expand_no_split_dollar_star = 1;
4713 #if defined (HANDLE_MULTIBYTE)
4714 if (ifs_firstc[0] == 0)
4715 #else
4716 if (ifs_firstc == 0)
4717 #endif
4718 word->flags |= W_NOSPLIT;
4719 word->flags |= W_NOSPLIT2;
4720 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4721 expand_no_split_dollar_star = 0;
4722
4723 return result;
4724 }
4725
4726 #if defined (PROCESS_SUBSTITUTION)
4727
4728 /*****************************************************************/
4729 /* */
4730 /* Hacking Process Substitution */
4731 /* */
4732 /*****************************************************************/
4733
4734 #if !defined (HAVE_DEV_FD)
4735 /* Named pipes must be removed explicitly with `unlink'. This keeps a list
4736 of FIFOs the shell has open. unlink_fifo_list will walk the list and
4737 unlink all of them. add_fifo_list adds the name of an open FIFO to the
4738 list. NFIFO is a count of the number of FIFOs in the list. */
4739 #define FIFO_INCR 20
4740
4741 struct temp_fifo {
4742 char *file;
4743 pid_t proc;
4744 };
4745
4746 static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
4747 static int nfifo;
4748 static int fifo_list_size;
4749
4750 char *
4751 copy_fifo_list (sizep)
4752 int *sizep;
4753 {
4754 if (sizep)
4755 *sizep = 0;
4756 return (char *)NULL;
4757 }
4758
4759 static void
4760 add_fifo_list (pathname)
4761 char *pathname;
4762 {
4763 if (nfifo >= fifo_list_size - 1)
4764 {
4765 fifo_list_size += FIFO_INCR;
4766 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
4767 fifo_list_size * sizeof (struct temp_fifo));
4768 }
4769
4770 fifo_list[nfifo].file = savestring (pathname);
4771 nfifo++;
4772 }
4773
4774 void
4775 unlink_fifo (i)
4776 int i;
4777 {
4778 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4779 {
4780 unlink (fifo_list[i].file);
4781 free (fifo_list[i].file);
4782 fifo_list[i].file = (char *)NULL;
4783 fifo_list[i].proc = -1;
4784 }
4785 }
4786
4787 void
4788 unlink_fifo_list ()
4789 {
4790 int saved, i, j;
4791
4792 if (nfifo == 0)
4793 return;
4794
4795 for (i = saved = 0; i < nfifo; i++)
4796 {
4797 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4798 {
4799 unlink (fifo_list[i].file);
4800 free (fifo_list[i].file);
4801 fifo_list[i].file = (char *)NULL;
4802 fifo_list[i].proc = -1;
4803 }
4804 else
4805 saved++;
4806 }
4807
4808 /* If we didn't remove some of the FIFOs, compact the list. */
4809 if (saved)
4810 {
4811 for (i = j = 0; i < nfifo; i++)
4812 if (fifo_list[i].file)
4813 {
4814 fifo_list[j].file = fifo_list[i].file;
4815 fifo_list[j].proc = fifo_list[i].proc;
4816 j++;
4817 }
4818 nfifo = j;
4819 }
4820 else
4821 nfifo = 0;
4822 }
4823
4824 /* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
4825 from some point in the past, and close all open FIFOs in fifo_list
4826 that are not marked as active in LIST. If LIST is NULL, close
4827 everything in fifo_list. LSIZE is the number of elements in LIST, in
4828 case it's larger than fifo_list_size (size of fifo_list). */
4829 void
4830 close_new_fifos (list, lsize)
4831 char *list;
4832 int lsize;
4833 {
4834 int i;
4835
4836 if (list == 0)
4837 {
4838 unlink_fifo_list ();
4839 return;
4840 }
4841
4842 for (i = 0; i < lsize; i++)
4843 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
4844 unlink_fifo (i);
4845
4846 for (i = lsize; i < fifo_list_size; i++)
4847 unlink_fifo (i);
4848 }
4849
4850 int
4851 fifos_pending ()
4852 {
4853 return nfifo;
4854 }
4855
4856 int
4857 num_fifos ()
4858 {
4859 return nfifo;
4860 }
4861
4862 static char *
4863 make_named_pipe ()
4864 {
4865 char *tname;
4866
4867 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
4868 if (mkfifo (tname, 0600) < 0)
4869 {
4870 free (tname);
4871 return ((char *)NULL);
4872 }
4873
4874 add_fifo_list (tname);
4875 return (tname);
4876 }
4877
4878 #else /* HAVE_DEV_FD */
4879
4880 /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
4881 has open to children. NFDS is a count of the number of bits currently
4882 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
4883 of open files. */
4884 static char *dev_fd_list = (char *)NULL;
4885 static int nfds;
4886 static int totfds; /* The highest possible number of open files. */
4887
4888 char *
4889 copy_fifo_list (sizep)
4890 int *sizep;
4891 {
4892 char *ret;
4893
4894 if (nfds == 0 || totfds == 0)
4895 {
4896 if (sizep)
4897 *sizep = 0;
4898 return (char *)NULL;
4899 }
4900
4901 if (sizep)
4902 *sizep = totfds;
4903 ret = (char *)xmalloc (totfds);
4904 return (memcpy (ret, dev_fd_list, totfds));
4905 }
4906
4907 static void
4908 add_fifo_list (fd)
4909 int fd;
4910 {
4911 if (dev_fd_list == 0 || fd >= totfds)
4912 {
4913 int ofds;
4914
4915 ofds = totfds;
4916 totfds = getdtablesize ();
4917 if (totfds < 0 || totfds > 256)
4918 totfds = 256;
4919 if (fd >= totfds)
4920 totfds = fd + 2;
4921
4922 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
4923 memset (dev_fd_list + ofds, '\0', totfds - ofds);
4924 }
4925
4926 dev_fd_list[fd] = 1;
4927 nfds++;
4928 }
4929
4930 int
4931 fifos_pending ()
4932 {
4933 return 0; /* used for cleanup; not needed with /dev/fd */
4934 }
4935
4936 int
4937 num_fifos ()
4938 {
4939 return nfds;
4940 }
4941
4942 void
4943 unlink_fifo (fd)
4944 int fd;
4945 {
4946 if (dev_fd_list[fd])
4947 {
4948 close (fd);
4949 dev_fd_list[fd] = 0;
4950 nfds--;
4951 }
4952 }
4953
4954 void
4955 unlink_fifo_list ()
4956 {
4957 register int i;
4958
4959 if (nfds == 0)
4960 return;
4961
4962 for (i = 0; nfds && i < totfds; i++)
4963 unlink_fifo (i);
4964
4965 nfds = 0;
4966 }
4967
4968 /* Take LIST, which is a snapshot copy of dev_fd_list from some point in
4969 the past, and close all open fds in dev_fd_list that are not marked
4970 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
4971 LSIZE is the number of elements in LIST, in case it's larger than
4972 totfds (size of dev_fd_list). */
4973 void
4974 close_new_fifos (list, lsize)
4975 char *list;
4976 int lsize;
4977 {
4978 int i;
4979
4980 if (list == 0)
4981 {
4982 unlink_fifo_list ();
4983 return;
4984 }
4985
4986 for (i = 0; i < lsize; i++)
4987 if (list[i] == 0 && i < totfds && dev_fd_list[i])
4988 unlink_fifo (i);
4989
4990 for (i = lsize; i < totfds; i++)
4991 unlink_fifo (i);
4992 }
4993
4994 #if defined (NOTDEF)
4995 print_dev_fd_list ()
4996 {
4997 register int i;
4998
4999 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
5000 fflush (stderr);
5001
5002 for (i = 0; i < totfds; i++)
5003 {
5004 if (dev_fd_list[i])
5005 fprintf (stderr, " %d", i);
5006 }
5007 fprintf (stderr, "\n");
5008 }
5009 #endif /* NOTDEF */
5010
5011 static char *
5012 make_dev_fd_filename (fd)
5013 int fd;
5014 {
5015 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
5016
5017 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
5018
5019 strcpy (ret, DEV_FD_PREFIX);
5020 p = inttostr (fd, intbuf, sizeof (intbuf));
5021 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5022
5023 add_fifo_list (fd);
5024 return (ret);
5025 }
5026
5027 #endif /* HAVE_DEV_FD */
5028
5029 /* Return a filename that will open a connection to the process defined by
5030 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5031 a filename in /dev/fd corresponding to a descriptor that is one of the
5032 ends of the pipe. If not defined, we use named pipes on systems that have
5033 them. Systems without /dev/fd and named pipes are out of luck.
5034
5035 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5036 use the read end of the pipe and dup that file descriptor to fd 0 in
5037 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5038 writing or use the write end of the pipe in the child, and dup that
5039 file descriptor to fd 1 in the child. The parent does the opposite. */
5040
5041 static char *
5042 process_substitute (string, open_for_read_in_child)
5043 char *string;
5044 int open_for_read_in_child;
5045 {
5046 char *pathname;
5047 int fd, result;
5048 pid_t old_pid, pid;
5049 #if defined (HAVE_DEV_FD)
5050 int parent_pipe_fd, child_pipe_fd;
5051 int fildes[2];
5052 #endif /* HAVE_DEV_FD */
5053 #if defined (JOB_CONTROL)
5054 pid_t old_pipeline_pgrp;
5055 #endif
5056
5057 if (!string || !*string || wordexp_only)
5058 return ((char *)NULL);
5059
5060 #if !defined (HAVE_DEV_FD)
5061 pathname = make_named_pipe ();
5062 #else /* HAVE_DEV_FD */
5063 if (pipe (fildes) < 0)
5064 {
5065 sys_error (_("cannot make pipe for process substitution"));
5066 return ((char *)NULL);
5067 }
5068 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5069 the pipe in the parent, otherwise the read end. */
5070 parent_pipe_fd = fildes[open_for_read_in_child];
5071 child_pipe_fd = fildes[1 - open_for_read_in_child];
5072 /* Move the parent end of the pipe to some high file descriptor, to
5073 avoid clashes with FDs used by the script. */
5074 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5075
5076 pathname = make_dev_fd_filename (parent_pipe_fd);
5077 #endif /* HAVE_DEV_FD */
5078
5079 if (pathname == 0)
5080 {
5081 sys_error (_("cannot make pipe for process substitution"));
5082 return ((char *)NULL);
5083 }
5084
5085 old_pid = last_made_pid;
5086
5087 #if defined (JOB_CONTROL)
5088 old_pipeline_pgrp = pipeline_pgrp;
5089 pipeline_pgrp = shell_pgrp;
5090 save_pipeline (1);
5091 #endif /* JOB_CONTROL */
5092
5093 pid = make_child ((char *)NULL, 1);
5094 if (pid == 0)
5095 {
5096 reset_terminating_signals (); /* XXX */
5097 free_pushed_string_input ();
5098 /* Cancel traps, in trap.c. */
5099 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
5100 setup_async_signals ();
5101 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
5102 }
5103
5104 #if defined (JOB_CONTROL)
5105 set_sigchld_handler ();
5106 stop_making_children ();
5107 /* XXX - should we only do this in the parent? (as in command subst) */
5108 pipeline_pgrp = old_pipeline_pgrp;
5109 #endif /* JOB_CONTROL */
5110
5111 if (pid < 0)
5112 {
5113 sys_error (_("cannot make child for process substitution"));
5114 free (pathname);
5115 #if defined (HAVE_DEV_FD)
5116 close (parent_pipe_fd);
5117 close (child_pipe_fd);
5118 #endif /* HAVE_DEV_FD */
5119 return ((char *)NULL);
5120 }
5121
5122 if (pid > 0)
5123 {
5124 #if defined (JOB_CONTROL)
5125 restore_pipeline (1);
5126 #endif
5127
5128 #if !defined (HAVE_DEV_FD)
5129 fifo_list[nfifo-1].proc = pid;
5130 #endif
5131
5132 last_made_pid = old_pid;
5133
5134 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5135 close_pgrp_pipe ();
5136 #endif /* JOB_CONTROL && PGRP_PIPE */
5137
5138 #if defined (HAVE_DEV_FD)
5139 close (child_pipe_fd);
5140 #endif /* HAVE_DEV_FD */
5141
5142 return (pathname);
5143 }
5144
5145 set_sigint_handler ();
5146
5147 #if defined (JOB_CONTROL)
5148 set_job_control (0);
5149 #endif /* JOB_CONTROL */
5150
5151 #if !defined (HAVE_DEV_FD)
5152 /* Open the named pipe in the child. */
5153 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
5154 if (fd < 0)
5155 {
5156 /* Two separate strings for ease of translation. */
5157 if (open_for_read_in_child)
5158 sys_error (_("cannot open named pipe %s for reading"), pathname);
5159 else
5160 sys_error (_("cannot open named pipe %s for writing"), pathname);
5161
5162 exit (127);
5163 }
5164 if (open_for_read_in_child)
5165 {
5166 if (sh_unset_nodelay_mode (fd) < 0)
5167 {
5168 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
5169 exit (127);
5170 }
5171 }
5172 #else /* HAVE_DEV_FD */
5173 fd = child_pipe_fd;
5174 #endif /* HAVE_DEV_FD */
5175
5176 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5177 {
5178 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
5179 open_for_read_in_child ? 0 : 1);
5180 exit (127);
5181 }
5182
5183 if (fd != (open_for_read_in_child ? 0 : 1))
5184 close (fd);
5185
5186 /* Need to close any files that this process has open to pipes inherited
5187 from its parent. */
5188 if (current_fds_to_close)
5189 {
5190 close_fd_bitmap (current_fds_to_close);
5191 current_fds_to_close = (struct fd_bitmap *)NULL;
5192 }
5193
5194 #if defined (HAVE_DEV_FD)
5195 /* Make sure we close the parent's end of the pipe and clear the slot
5196 in the fd list so it is not closed later, if reallocated by, for
5197 instance, pipe(2). */
5198 close (parent_pipe_fd);
5199 dev_fd_list[parent_pipe_fd] = 0;
5200 #endif /* HAVE_DEV_FD */
5201
5202 /* subshells shouldn't have this flag, which controls using the temporary
5203 environment for variable lookups. */
5204 expanding_redir = 0;
5205
5206 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
5207
5208 #if !defined (HAVE_DEV_FD)
5209 /* Make sure we close the named pipe in the child before we exit. */
5210 close (open_for_read_in_child ? 0 : 1);
5211 #endif /* !HAVE_DEV_FD */
5212
5213 last_command_exit_value = result;
5214 result = run_exit_trap ();
5215 exit (result);
5216 /*NOTREACHED*/
5217 }
5218 #endif /* PROCESS_SUBSTITUTION */
5219
5220 /***********************************/
5221 /* */
5222 /* Command Substitution */
5223 /* */
5224 /***********************************/
5225
5226 static char *
5227 read_comsub (fd, quoted, rflag)
5228 int fd, quoted;
5229 int *rflag;
5230 {
5231 char *istring, buf[128], *bufp, *s;
5232 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
5233 ssize_t bufn;
5234
5235 istring = (char *)NULL;
5236 istring_index = istring_size = bufn = tflag = 0;
5237
5238 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5239 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
5240
5241 /* Read the output of the command through the pipe. This may need to be
5242 changed to understand multibyte characters in the future. */
5243 while (1)
5244 {
5245 if (fd < 0)
5246 break;
5247 if (--bufn <= 0)
5248 {
5249 bufn = zread (fd, buf, sizeof (buf));
5250 if (bufn <= 0)
5251 break;
5252 bufp = buf;
5253 }
5254 c = *bufp++;
5255
5256 if (c == 0)
5257 {
5258 #if 0
5259 internal_warning ("read_comsub: ignored null byte in input");
5260 #endif
5261 continue;
5262 }
5263
5264 /* Add the character to ISTRING, possibly after resizing it. */
5265 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5266
5267 /* This is essentially quote_string inline */
5268 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5269 istring[istring_index++] = CTLESC;
5270 /* Escape CTLESC and CTLNUL in the output to protect those characters
5271 from the rest of the word expansions (word splitting and globbing.)
5272 This is essentially quote_escapes inline. */
5273 else if (skip_ctlesc == 0 && c == CTLESC)
5274 {
5275 tflag |= W_HASCTLESC;
5276 istring[istring_index++] = CTLESC;
5277 }
5278 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
5279 istring[istring_index++] = CTLESC;
5280
5281 istring[istring_index++] = c;
5282
5283 #if 0
5284 #if defined (__CYGWIN__)
5285 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5286 {
5287 istring_index--;
5288 istring[istring_index - 1] = '\n';
5289 }
5290 #endif
5291 #endif
5292 }
5293
5294 if (istring)
5295 istring[istring_index] = '\0';
5296
5297 /* If we read no output, just return now and save ourselves some
5298 trouble. */
5299 if (istring_index == 0)
5300 {
5301 FREE (istring);
5302 if (rflag)
5303 *rflag = tflag;
5304 return (char *)NULL;
5305 }
5306
5307 /* Strip trailing newlines from the output of the command. */
5308 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5309 {
5310 while (istring_index > 0)
5311 {
5312 if (istring[istring_index - 1] == '\n')
5313 {
5314 --istring_index;
5315
5316 /* If the newline was quoted, remove the quoting char. */
5317 if (istring[istring_index - 1] == CTLESC)
5318 --istring_index;
5319 }
5320 else
5321 break;
5322 }
5323 istring[istring_index] = '\0';
5324 }
5325 else
5326 strip_trailing (istring, istring_index - 1, 1);
5327
5328 if (rflag)
5329 *rflag = tflag;
5330 return istring;
5331 }
5332
5333 /* Perform command substitution on STRING. This returns a WORD_DESC * with the
5334 contained string possibly quoted. */
5335 WORD_DESC *
5336 command_substitute (string, quoted)
5337 char *string;
5338 int quoted;
5339 {
5340 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
5341 char *istring;
5342 int result, fildes[2], function_value, pflags, rc, tflag;
5343 WORD_DESC *ret;
5344
5345 istring = (char *)NULL;
5346
5347 /* Don't fork () if there is no need to. In the case of no command to
5348 run, just return NULL. */
5349 if (!string || !*string || (string[0] == '\n' && !string[1]))
5350 return ((WORD_DESC *)NULL);
5351
5352 if (wordexp_only && read_but_dont_execute)
5353 {
5354 last_command_exit_value = EX_WEXPCOMSUB;
5355 jump_to_top_level (EXITPROG);
5356 }
5357
5358 /* We're making the assumption here that the command substitution will
5359 eventually run a command from the file system. Since we'll run
5360 maybe_make_export_env in this subshell before executing that command,
5361 the parent shell and any other shells it starts will have to remake
5362 the environment. If we make it before we fork, other shells won't
5363 have to. Don't bother if we have any temporary variable assignments,
5364 though, because the export environment will be remade after this
5365 command completes anyway, but do it if all the words to be expanded
5366 are variable assignments. */
5367 if (subst_assign_varlist == 0 || garglist == 0)
5368 maybe_make_export_env (); /* XXX */
5369
5370 /* Flags to pass to parse_and_execute() */
5371 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
5372
5373 /* Pipe the output of executing STRING into the current shell. */
5374 if (pipe (fildes) < 0)
5375 {
5376 sys_error (_("cannot make pipe for command substitution"));
5377 goto error_exit;
5378 }
5379
5380 old_pid = last_made_pid;
5381 #if defined (JOB_CONTROL)
5382 old_pipeline_pgrp = pipeline_pgrp;
5383 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
5384 if ((subshell_environment & SUBSHELL_PIPE) == 0)
5385 pipeline_pgrp = shell_pgrp;
5386 cleanup_the_pipeline ();
5387 #endif /* JOB_CONTROL */
5388
5389 old_async_pid = last_asynchronous_pid;
5390 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
5391 last_asynchronous_pid = old_async_pid;
5392
5393 if (pid == 0)
5394 {
5395 /* Reset the signal handlers in the child, but don't free the
5396 trap strings. Set a flag noting that we have to free the
5397 trap strings if we run trap to change a signal disposition. */
5398 reset_signal_handlers ();
5399 subshell_environment |= SUBSHELL_RESETTRAP;
5400 }
5401
5402 #if defined (JOB_CONTROL)
5403 /* XXX DO THIS ONLY IN PARENT ? XXX */
5404 set_sigchld_handler ();
5405 stop_making_children ();
5406 if (pid != 0)
5407 pipeline_pgrp = old_pipeline_pgrp;
5408 #else
5409 stop_making_children ();
5410 #endif /* JOB_CONTROL */
5411
5412 if (pid < 0)
5413 {
5414 sys_error (_("cannot make child for command substitution"));
5415 error_exit:
5416
5417 last_made_pid = old_pid;
5418
5419 FREE (istring);
5420 close (fildes[0]);
5421 close (fildes[1]);
5422 return ((WORD_DESC *)NULL);
5423 }
5424
5425 if (pid == 0)
5426 {
5427 set_sigint_handler (); /* XXX */
5428
5429 free_pushed_string_input ();
5430
5431 if (dup2 (fildes[1], 1) < 0)
5432 {
5433 sys_error (_("command_substitute: cannot duplicate pipe as fd 1"));
5434 exit (EXECUTION_FAILURE);
5435 }
5436
5437 /* If standard output is closed in the parent shell
5438 (such as after `exec >&-'), file descriptor 1 will be
5439 the lowest available file descriptor, and end up in
5440 fildes[0]. This can happen for stdin and stderr as well,
5441 but stdout is more important -- it will cause no output
5442 to be generated from this command. */
5443 if ((fildes[1] != fileno (stdin)) &&
5444 (fildes[1] != fileno (stdout)) &&
5445 (fildes[1] != fileno (stderr)))
5446 close (fildes[1]);
5447
5448 if ((fildes[0] != fileno (stdin)) &&
5449 (fildes[0] != fileno (stdout)) &&
5450 (fildes[0] != fileno (stderr)))
5451 close (fildes[0]);
5452
5453 #ifdef __CYGWIN__
5454 /* Let stdio know the fd may have changed from text to binary mode, and
5455 make sure to preserve stdout line buffering. */
5456 freopen (NULL, "w", stdout);
5457 sh_setlinebuf (stdout);
5458 #endif /* __CYGWIN__ */
5459
5460 /* The currently executing shell is not interactive. */
5461 interactive = 0;
5462
5463 /* This is a subshell environment. */
5464 subshell_environment |= SUBSHELL_COMSUB;
5465
5466 /* When not in POSIX mode, command substitution does not inherit
5467 the -e flag. */
5468 if (posixly_correct == 0)
5469 {
5470 builtin_ignoring_errexit = 0;
5471 change_flag ('e', FLAG_OFF);
5472 set_shellopts ();
5473 }
5474
5475 remove_quoted_escapes (string);
5476
5477 startup_state = 2; /* see if we can avoid a fork */
5478 /* Give command substitution a place to jump back to on failure,
5479 so we don't go back up to main (). */
5480 result = setjmp_nosigs (top_level);
5481
5482 /* If we're running a command substitution inside a shell function,
5483 trap `return' so we don't return from the function in the subshell
5484 and go off to never-never land. */
5485 if (result == 0 && return_catch_flag)
5486 function_value = setjmp_nosigs (return_catch);
5487 else
5488 function_value = 0;
5489
5490 if (result == ERREXIT)
5491 rc = last_command_exit_value;
5492 else if (result == EXITPROG)
5493 rc = last_command_exit_value;
5494 else if (result)
5495 rc = EXECUTION_FAILURE;
5496 else if (function_value)
5497 rc = return_catch_value;
5498 else
5499 {
5500 subshell_level++;
5501 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
5502 subshell_level--;
5503 }
5504
5505 last_command_exit_value = rc;
5506 rc = run_exit_trap ();
5507 #if defined (PROCESS_SUBSTITUTION)
5508 unlink_fifo_list ();
5509 #endif
5510 exit (rc);
5511 }
5512 else
5513 {
5514 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5515 close_pgrp_pipe ();
5516 #endif /* JOB_CONTROL && PGRP_PIPE */
5517
5518 close (fildes[1]);
5519
5520 tflag = 0;
5521 istring = read_comsub (fildes[0], quoted, &tflag);
5522
5523 close (fildes[0]);
5524
5525 current_command_subst_pid = pid;
5526 last_command_exit_value = wait_for (pid);
5527 last_command_subst_pid = pid;
5528 last_made_pid = old_pid;
5529
5530 #if defined (JOB_CONTROL)
5531 /* If last_command_exit_value > 128, then the substituted command
5532 was terminated by a signal. If that signal was SIGINT, then send
5533 SIGINT to ourselves. This will break out of loops, for instance. */
5534 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
5535 kill (getpid (), SIGINT);
5536
5537 /* wait_for gives the terminal back to shell_pgrp. If some other
5538 process group should have it, give it away to that group here.
5539 pipeline_pgrp is non-zero only while we are constructing a
5540 pipeline, so what we are concerned about is whether or not that
5541 pipeline was started in the background. A pipeline started in
5542 the background should never get the tty back here. */
5543 if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0)
5544 give_terminal_to (pipeline_pgrp, 0);
5545 #endif /* JOB_CONTROL */
5546
5547 ret = alloc_word_desc ();
5548 ret->word = istring;
5549 ret->flags = tflag;
5550
5551 return ret;
5552 }
5553 }
5554
5555 /********************************************************
5556 * *
5557 * Utility functions for parameter expansion *
5558 * *
5559 ********************************************************/
5560
5561 #if defined (ARRAY_VARS)
5562
5563 static arrayind_t
5564 array_length_reference (s)
5565 char *s;
5566 {
5567 int len;
5568 arrayind_t ind;
5569 char *akey;
5570 char *t, c;
5571 ARRAY *array;
5572 HASH_TABLE *h;
5573 SHELL_VAR *var;
5574
5575 var = array_variable_part (s, &t, &len);
5576
5577 /* If unbound variables should generate an error, report one and return
5578 failure. */
5579 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
5580 {
5581 c = *--t;
5582 *t = '\0';
5583 last_command_exit_value = EXECUTION_FAILURE;
5584 err_unboundvar (s);
5585 *t = c;
5586 return (-1);
5587 }
5588 else if (var == 0 || invisible_p (var))
5589 return 0;
5590
5591 /* We support a couple of expansions for variables that are not arrays.
5592 We'll return the length of the value for v[0], and 1 for v[@] or
5593 v[*]. Return 0 for everything else. */
5594
5595 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
5596 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
5597
5598 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
5599 {
5600 if (assoc_p (var))
5601 return (h ? assoc_num_elements (h) : 0);
5602 else if (array_p (var))
5603 return (array ? array_num_elements (array) : 0);
5604 else
5605 return (var_isset (var) ? 1 : 0);
5606 }
5607
5608 if (assoc_p (var))
5609 {
5610 t[len - 1] = '\0';
5611 akey = expand_assignment_string_to_string (t, 0); /* [ */
5612 t[len - 1] = ']';
5613 if (akey == 0 || *akey == 0)
5614 {
5615 err_badarraysub (t);
5616 FREE (akey);
5617 return (-1);
5618 }
5619 t = assoc_reference (assoc_cell (var), akey);
5620 free (akey);
5621 }
5622 else
5623 {
5624 ind = array_expand_index (var, t, len);
5625 /* negative subscripts to indexed arrays count back from end */
5626 if (var && array_p (var) && ind < 0)
5627 ind = array_max_index (array_cell (var)) + 1 + ind;
5628 if (ind < 0)
5629 {
5630 err_badarraysub (t);
5631 return (-1);
5632 }
5633 if (array_p (var))
5634 t = array_reference (array, ind);
5635 else
5636 t = (ind == 0) ? value_cell (var) : (char *)NULL;
5637 }
5638
5639 len = MB_STRLEN (t);
5640 return (len);
5641 }
5642 #endif /* ARRAY_VARS */
5643
5644 static int
5645 valid_brace_expansion_word (name, var_is_special)
5646 char *name;
5647 int var_is_special;
5648 {
5649 if (DIGIT (*name) && all_digits (name))
5650 return 1;
5651 else if (var_is_special)
5652 return 1;
5653 #if defined (ARRAY_VARS)
5654 else if (valid_array_reference (name))
5655 return 1;
5656 #endif /* ARRAY_VARS */
5657 else if (legal_identifier (name))
5658 return 1;
5659 else
5660 return 0;
5661 }
5662
5663 static int
5664 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
5665 char *name;
5666 int quoted;
5667 int *quoted_dollar_atp, *contains_dollar_at;
5668 {
5669 char *temp1;
5670
5671 if (name == 0)
5672 {
5673 if (quoted_dollar_atp)
5674 *quoted_dollar_atp = 0;
5675 if (contains_dollar_at)
5676 *contains_dollar_at = 0;
5677 return 0;
5678 }
5679
5680 /* check for $@ and $* */
5681 if (name[0] == '@' && name[1] == 0)
5682 {
5683 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5684 *quoted_dollar_atp = 1;
5685 if (contains_dollar_at)
5686 *contains_dollar_at = 1;
5687 return 1;
5688 }
5689 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
5690 {
5691 if (contains_dollar_at)
5692 *contains_dollar_at = 1;
5693 return 1;
5694 }
5695
5696 /* Now check for ${array[@]} and ${array[*]} */
5697 #if defined (ARRAY_VARS)
5698 else if (valid_array_reference (name))
5699 {
5700 temp1 = mbschr (name, '[');
5701 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
5702 {
5703 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5704 *quoted_dollar_atp = 1;
5705 if (contains_dollar_at)
5706 *contains_dollar_at = 1;
5707 return 1;
5708 } /* [ */
5709 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
5710 which should result in separate words even when IFS is unset. */
5711 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
5712 {
5713 if (contains_dollar_at)
5714 *contains_dollar_at = 1;
5715 return 1;
5716 }
5717 }
5718 #endif
5719 return 0;
5720 }
5721
5722 /* Parameter expand NAME, and return a new string which is the expansion,
5723 or NULL if there was no expansion.
5724 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
5725 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
5726 NAME was found inside of a double-quoted expression. */
5727 static WORD_DESC *
5728 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
5729 char *name;
5730 int var_is_special, quoted, pflags;
5731 arrayind_t *indp;
5732 {
5733 WORD_DESC *ret;
5734 char *temp, *tt;
5735 intmax_t arg_index;
5736 SHELL_VAR *var;
5737 int atype, rflags;
5738 arrayind_t ind;
5739
5740 ret = 0;
5741 temp = 0;
5742 rflags = 0;
5743
5744 if (indp)
5745 *indp = INTMAX_MIN;
5746
5747 /* Handle multiple digit arguments, as in ${11}. */
5748 if (legal_number (name, &arg_index))
5749 {
5750 tt = get_dollar_var_value (arg_index);
5751 if (tt)
5752 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5753 ? quote_string (tt)
5754 : quote_escapes (tt);
5755 else
5756 temp = (char *)NULL;
5757 FREE (tt);
5758 }
5759 else if (var_is_special) /* ${@} */
5760 {
5761 int sindex;
5762 tt = (char *)xmalloc (2 + strlen (name));
5763 tt[sindex = 0] = '$';
5764 strcpy (tt + 1, name);
5765
5766 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
5767 (int *)NULL, (int *)NULL, pflags);
5768 free (tt);
5769 }
5770 #if defined (ARRAY_VARS)
5771 else if (valid_array_reference (name))
5772 {
5773 expand_arrayref:
5774 /* XXX - does this leak if name[@] or name[*]? */
5775 if (pflags & PF_ASSIGNRHS)
5776 {
5777 temp = array_variable_name (name, &tt, (int *)0);
5778 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
5779 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, 0, &atype, &ind);
5780 else
5781 temp = array_value (name, quoted, 0, &atype, &ind);
5782 }
5783 else
5784 temp = array_value (name, quoted, 0, &atype, &ind);
5785 if (atype == 0 && temp)
5786 {
5787 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5788 ? quote_string (temp)
5789 : quote_escapes (temp);
5790 rflags |= W_ARRAYIND;
5791 if (indp)
5792 *indp = ind;
5793 }
5794 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5795 rflags |= W_HASQUOTEDNULL;
5796 }
5797 #endif
5798 else if (var = find_variable (name))
5799 {
5800 if (var_isset (var) && invisible_p (var) == 0)
5801 {
5802 #if defined (ARRAY_VARS)
5803 if (assoc_p (var))
5804 temp = assoc_reference (assoc_cell (var), "0");
5805 else if (array_p (var))
5806 temp = array_reference (array_cell (var), 0);
5807 else
5808 temp = value_cell (var);
5809 #else
5810 temp = value_cell (var);
5811 #endif
5812
5813 if (temp)
5814 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5815 ? quote_string (temp)
5816 : quote_escapes (temp);
5817 }
5818 else
5819 temp = (char *)NULL;
5820 }
5821 else if (var = find_variable_last_nameref (name))
5822 {
5823 temp = nameref_cell (var);
5824 #if defined (ARRAY_VARS)
5825 /* Handle expanding nameref whose value is x[n] */
5826 if (temp && *temp && valid_array_reference (temp))
5827 {
5828 name = temp;
5829 goto expand_arrayref;
5830 }
5831 else
5832 #endif
5833 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
5834 if (temp && *temp && legal_identifier (temp) == 0)
5835 {
5836 last_command_exit_value = EXECUTION_FAILURE;
5837 report_error (_("%s: invalid variable name for name reference"), temp);
5838 temp = &expand_param_error;
5839 }
5840 else
5841 temp = (char *)NULL;
5842 }
5843 else
5844 temp = (char *)NULL;
5845
5846 if (ret == 0)
5847 {
5848 ret = alloc_word_desc ();
5849 ret->word = temp;
5850 ret->flags |= rflags;
5851 }
5852 return ret;
5853 }
5854
5855 static char *
5856 parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
5857 char *name;
5858 int var_is_special, quoted, find_nameref;
5859 {
5860 char *temp, *t;
5861 WORD_DESC *w;
5862 SHELL_VAR *v;
5863
5864 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name)) &&
5865 nameref_p (v) && (t = nameref_cell (v)) && *t)
5866 return (savestring (t));
5867
5868 /* If var_is_special == 0, and name is not an array reference, this does
5869 more expansion than necessary. It should really look up the variable's
5870 value and not try to expand it. */
5871 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
5872 t = w->word;
5873 /* Have to dequote here if necessary */
5874 if (t)
5875 {
5876 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
5877 ? dequote_string (t)
5878 : dequote_escapes (t);
5879 free (t);
5880 t = temp;
5881 }
5882 dispose_word_desc (w);
5883
5884 return t;
5885 }
5886
5887 /* Expand an indirect reference to a variable: ${!NAME} expands to the
5888 value of the variable whose name is the value of NAME. */
5889 static WORD_DESC *
5890 parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
5891 char *name;
5892 int var_is_special, quoted;
5893 int *quoted_dollar_atp, *contains_dollar_at;
5894 {
5895 char *temp, *t;
5896 WORD_DESC *w;
5897 SHELL_VAR *v;
5898
5899 /* See if it's a nameref first, behave in ksh93-compatible fashion.
5900 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
5901 bash performs an indirect lookup on foo[0] and expands the result;
5902 ksh93 expands bar[0]. We could do that here -- there are enough usable
5903 primitives to do that -- but do not at this point. */
5904 if (var_is_special == 0 && (v = find_variable_last_nameref (name)))
5905 {
5906 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
5907 {
5908 w = alloc_word_desc ();
5909 w->word = savestring (t);
5910 w->flags = 0;
5911 return w;
5912 }
5913 }
5914
5915 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
5916
5917 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
5918 if (t == 0)
5919 return (WORD_DESC *)NULL;
5920
5921 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
5922 free (t);
5923
5924 return w;
5925 }
5926
5927 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
5928 depending on the value of C, the separating character. C can be one of
5929 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
5930 between double quotes. */
5931 static WORD_DESC *
5932 parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat)
5933 char *name, *value;
5934 int c, quoted, *qdollaratp, *hasdollarat;
5935 {
5936 WORD_DESC *w;
5937 WORD_LIST *l;
5938 char *t, *t1, *temp;
5939 int hasdol;
5940
5941 /* If the entire expression is between double quotes, we want to treat
5942 the value as a double-quoted string, with the exception that we strip
5943 embedded unescaped double quotes (for sh backwards compatibility). */
5944 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
5945 {
5946 hasdol = 0;
5947 temp = string_extract_double_quoted (value, &hasdol, 1);
5948 }
5949 else
5950 temp = value;
5951
5952 w = alloc_word_desc ();
5953 hasdol = 0;
5954 /* XXX was 0 not quoted */
5955 l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL)
5956 : (WORD_LIST *)0;
5957 if (hasdollarat)
5958 *hasdollarat = hasdol || (l && l->next);
5959 if (temp != value)
5960 free (temp);
5961 if (l)
5962 {
5963 /* The expansion of TEMP returned something. We need to treat things
5964 slightly differently if HASDOL is non-zero. If we have "$@", the
5965 individual words have already been quoted. We need to turn them
5966 into a string with the words separated by the first character of
5967 $IFS without any additional quoting, so string_list_dollar_at won't
5968 do the right thing. We use string_list_dollar_star instead. */
5969 temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l);
5970
5971 /* If l->next is not null, we know that TEMP contained "$@", since that
5972 is the only expansion that creates more than one word. */
5973 if (qdollaratp && ((hasdol && quoted) || l->next))
5974 *qdollaratp = 1;
5975 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
5976 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
5977 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
5978 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5979 (which is more paranoia than anything else), we need to return the
5980 quoted null string and set the flags to indicate it. */
5981 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
5982 {
5983 w->flags |= W_HASQUOTEDNULL;
5984 }
5985 dispose_words (l);
5986 }
5987 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol)
5988 {
5989 /* The brace expansion occurred between double quotes and there was
5990 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
5991 it does not expand to anything. In this case, we want to return
5992 a quoted empty string. */
5993 temp = make_quoted_char ('\0');
5994 w->flags |= W_HASQUOTEDNULL;
5995 }
5996 else
5997 temp = (char *)NULL;
5998
5999 if (c == '-' || c == '+')
6000 {
6001 w->word = temp;
6002 return w;
6003 }
6004
6005 /* c == '=' */
6006 t = temp ? savestring (temp) : savestring ("");
6007 t1 = dequote_string (t);
6008 free (t);
6009 #if defined (ARRAY_VARS)
6010 if (valid_array_reference (name))
6011 assign_array_element (name, t1, 0);
6012 else
6013 #endif /* ARRAY_VARS */
6014 bind_variable (name, t1, 0);
6015 #if 0
6016 if (STREQ (name, "IFS") == 0)
6017 #endif
6018 stupidly_hack_special_variables (name);
6019
6020 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6021 free (temp);
6022
6023 w->word = t1;
6024 return w;
6025 }
6026
6027 /* Deal with the right hand side of a ${name:?value} expansion in the case
6028 that NAME is null or not set. If VALUE is non-null it is expanded and
6029 used as the error message to print, otherwise a standard message is
6030 printed. */
6031 static void
6032 parameter_brace_expand_error (name, value)
6033 char *name, *value;
6034 {
6035 WORD_LIST *l;
6036 char *temp;
6037
6038 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
6039 if (value && *value)
6040 {
6041 l = expand_string (value, 0);
6042 temp = string_list (l);
6043 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6044 FREE (temp);
6045 dispose_words (l);
6046 }
6047 else
6048 report_error (_("%s: parameter null or not set"), name);
6049
6050 /* Free the data we have allocated during this expansion, since we
6051 are about to longjmp out. */
6052 free (name);
6053 FREE (value);
6054 }
6055
6056 /* Return 1 if NAME is something for which parameter_brace_expand_length is
6057 OK to do. */
6058 static int
6059 valid_length_expression (name)
6060 char *name;
6061 {
6062 return (name[1] == '\0' || /* ${#} */
6063 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6064 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
6065 #if defined (ARRAY_VARS)
6066 valid_array_reference (name + 1) || /* ${#a[7]} */
6067 #endif
6068 legal_identifier (name + 1)); /* ${#PS1} */
6069 }
6070
6071 /* Handle the parameter brace expansion that requires us to return the
6072 length of a parameter. */
6073 static intmax_t
6074 parameter_brace_expand_length (name)
6075 char *name;
6076 {
6077 char *t, *newname;
6078 intmax_t number, arg_index;
6079 WORD_LIST *list;
6080 #if defined (ARRAY_VARS)
6081 SHELL_VAR *var;
6082 #endif
6083
6084 if (name[1] == '\0') /* ${#} */
6085 number = number_of_args ();
6086 else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */
6087 number = number_of_args ();
6088 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
6089 {
6090 /* Take the lengths of some of the shell's special parameters. */
6091 switch (name[1])
6092 {
6093 case '-':
6094 t = which_set_flags ();
6095 break;
6096 case '?':
6097 t = itos (last_command_exit_value);
6098 break;
6099 case '$':
6100 t = itos (dollar_dollar_pid);
6101 break;
6102 case '!':
6103 if (last_asynchronous_pid == NO_PID)
6104 t = (char *)NULL; /* XXX - error if set -u set? */
6105 else
6106 t = itos (last_asynchronous_pid);
6107 break;
6108 case '#':
6109 t = itos (number_of_args ());
6110 break;
6111 }
6112 number = STRLEN (t);
6113 FREE (t);
6114 }
6115 #if defined (ARRAY_VARS)
6116 else if (valid_array_reference (name + 1))
6117 number = array_length_reference (name + 1);
6118 #endif /* ARRAY_VARS */
6119 else
6120 {
6121 number = 0;
6122
6123 if (legal_number (name + 1, &arg_index)) /* ${#1} */
6124 {
6125 t = get_dollar_var_value (arg_index);
6126 if (t == 0 && unbound_vars_is_error)
6127 return INTMAX_MIN;
6128 number = MB_STRLEN (t);
6129 FREE (t);
6130 }
6131 #if defined (ARRAY_VARS)
6132 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
6133 {
6134 if (assoc_p (var))
6135 t = assoc_reference (assoc_cell (var), "0");
6136 else
6137 t = array_reference (array_cell (var), 0);
6138 if (t == 0 && unbound_vars_is_error)
6139 return INTMAX_MIN;
6140 number = MB_STRLEN (t);
6141 }
6142 #endif
6143 else /* ${#PS1} */
6144 {
6145 newname = savestring (name);
6146 newname[0] = '$';
6147 list = expand_string (newname, Q_DOUBLE_QUOTES);
6148 t = list ? string_list (list) : (char *)NULL;
6149 free (newname);
6150 if (list)
6151 dispose_words (list);
6152
6153 number = t ? MB_STRLEN (t) : 0;
6154 FREE (t);
6155 }
6156 }
6157
6158 return (number);
6159 }
6160
6161 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6162 so we do some ad-hoc parsing of an arithmetic expression to find
6163 the first DELIM, instead of using strchr(3). Two rules:
6164 1. If the substring contains a `(', read until closing `)'.
6165 2. If the substring contains a `?', read past one `:' for each `?'.
6166 */
6167
6168 static char *
6169 skiparith (substr, delim)
6170 char *substr;
6171 int delim;
6172 {
6173 size_t sublen;
6174 int skipcol, pcount, i;
6175 DECLARE_MBSTATE;
6176
6177 sublen = strlen (substr);
6178 i = skipcol = pcount = 0;
6179 while (substr[i])
6180 {
6181 /* Balance parens */
6182 if (substr[i] == LPAREN)
6183 {
6184 pcount++;
6185 i++;
6186 continue;
6187 }
6188 if (substr[i] == RPAREN && pcount)
6189 {
6190 pcount--;
6191 i++;
6192 continue;
6193 }
6194 if (pcount)
6195 {
6196 ADVANCE_CHAR (substr, sublen, i);
6197 continue;
6198 }
6199
6200 /* Skip one `:' for each `?' */
6201 if (substr[i] == ':' && skipcol)
6202 {
6203 skipcol--;
6204 i++;
6205 continue;
6206 }
6207 if (substr[i] == delim)
6208 break;
6209 if (substr[i] == '?')
6210 {
6211 skipcol++;
6212 i++;
6213 continue;
6214 }
6215 ADVANCE_CHAR (substr, sublen, i);
6216 }
6217
6218 return (substr + i);
6219 }
6220
6221 /* Verify and limit the start and end of the desired substring. If
6222 VTYPE == 0, a regular shell variable is being used; if it is 1,
6223 then the positional parameters are being used; if it is 2, then
6224 VALUE is really a pointer to an array variable that should be used.
6225 Return value is 1 if both values were OK, 0 if there was a problem
6226 with an invalid expression, or -1 if the values were out of range. */
6227 static int
6228 verify_substring_values (v, value, substr, vtype, e1p, e2p)
6229 SHELL_VAR *v;
6230 char *value, *substr;
6231 int vtype;
6232 intmax_t *e1p, *e2p;
6233 {
6234 char *t, *temp1, *temp2;
6235 arrayind_t len;
6236 int expok;
6237 #if defined (ARRAY_VARS)
6238 ARRAY *a;
6239 HASH_TABLE *h;
6240 #endif
6241
6242 /* duplicate behavior of strchr(3) */
6243 t = skiparith (substr, ':');
6244 if (*t && *t == ':')
6245 *t = '\0';
6246 else
6247 t = (char *)0;
6248
6249 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
6250 *e1p = evalexp (temp1, &expok);
6251 free (temp1);
6252 if (expok == 0)
6253 return (0);
6254
6255 len = -1; /* paranoia */
6256 switch (vtype)
6257 {
6258 case VT_VARIABLE:
6259 case VT_ARRAYMEMBER:
6260 len = MB_STRLEN (value);
6261 break;
6262 case VT_POSPARMS:
6263 len = number_of_args () + 1;
6264 if (*e1p == 0)
6265 len++; /* add one arg if counting from $0 */
6266 break;
6267 #if defined (ARRAY_VARS)
6268 case VT_ARRAYVAR:
6269 /* For arrays, the first value deals with array indices. Negative
6270 offsets count from one past the array's maximum index. Associative
6271 arrays treat the number of elements as the maximum index. */
6272 if (assoc_p (v))
6273 {
6274 h = assoc_cell (v);
6275 len = assoc_num_elements (h) + (*e1p < 0);
6276 }
6277 else
6278 {
6279 a = (ARRAY *)value;
6280 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
6281 }
6282 break;
6283 #endif
6284 }
6285
6286 if (len == -1) /* paranoia */
6287 return -1;
6288
6289 if (*e1p < 0) /* negative offsets count from end */
6290 *e1p += len;
6291
6292 if (*e1p > len || *e1p < 0)
6293 return (-1);
6294
6295 #if defined (ARRAY_VARS)
6296 /* For arrays, the second offset deals with the number of elements. */
6297 if (vtype == VT_ARRAYVAR)
6298 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
6299 #endif
6300
6301 if (t)
6302 {
6303 t++;
6304 temp2 = savestring (t);
6305 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
6306 free (temp2);
6307 t[-1] = ':';
6308 *e2p = evalexp (temp1, &expok);
6309 free (temp1);
6310 if (expok == 0)
6311 return (0);
6312 #if 1
6313 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
6314 #else
6315 /* bash-4.3: allow positional parameter length < 0 to count backwards
6316 from end of positional parameters */
6317 if (vtype == VT_ARRAYVAR && *e2p < 0)
6318 #endif
6319 {
6320 internal_error (_("%s: substring expression < 0"), t);
6321 return (0);
6322 }
6323 #if defined (ARRAY_VARS)
6324 /* In order to deal with sparse arrays, push the intelligence about how
6325 to deal with the number of elements desired down to the array-
6326 specific functions. */
6327 if (vtype != VT_ARRAYVAR)
6328 #endif
6329 {
6330 if (*e2p < 0)
6331 {
6332 *e2p += len;
6333 if (*e2p < 0 || *e2p < *e1p)
6334 {
6335 internal_error (_("%s: substring expression < 0"), t);
6336 return (0);
6337 }
6338 }
6339 else
6340 *e2p += *e1p; /* want E2 chars starting at E1 */
6341 if (*e2p > len)
6342 *e2p = len;
6343 }
6344 }
6345 else
6346 *e2p = len;
6347
6348 return (1);
6349 }
6350
6351 /* Return the type of variable specified by VARNAME (simple variable,
6352 positional param, or array variable). Also return the value specified
6353 by VARNAME (value of a variable or a reference to an array element).
6354 QUOTED is the standard description of quoting state, using Q_* defines.
6355 FLAGS is currently a set of flags to pass to array_value. If IND is
6356 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
6357 passed to array_value so the array index is not computed again.
6358 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
6359 characters in the value are quoted with CTLESC and takes appropriate
6360 steps. For convenience, *VALP is set to the dequoted VALUE. */
6361 static int
6362 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
6363 char *varname, *value;
6364 arrayind_t ind;
6365 int quoted, flags;
6366 SHELL_VAR **varp;
6367 char **valp;
6368 {
6369 int vtype, want_indir;
6370 char *temp, *vname;
6371 WORD_DESC *wd;
6372 #if defined (ARRAY_VARS)
6373 SHELL_VAR *v;
6374 #endif
6375 arrayind_t lind;
6376
6377 want_indir = *varname == '!' &&
6378 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
6379 || VALID_INDIR_PARAM (varname[1]));
6380 if (want_indir)
6381 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
6382 else
6383 vname = varname;
6384
6385 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
6386 vtype = (vname[0] == '@' || vname[0] == '*') && vname[1] == '\0';
6387 if (vtype == VT_POSPARMS && vname[0] == '*')
6388 vtype |= VT_STARSUB;
6389 *varp = (SHELL_VAR *)NULL;
6390
6391 #if defined (ARRAY_VARS)
6392 if (valid_array_reference (vname))
6393 {
6394 v = array_variable_part (vname, &temp, (int *)0);
6395 /* If we want to signal array_value to use an already-computed index,
6396 set LIND to that index */
6397 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
6398 if (v && invisible_p (v))
6399 {
6400 vtype = VT_ARRAYMEMBER;
6401 *varp = (SHELL_VAR *)NULL;
6402 *valp = (char *)NULL;
6403 }
6404 if (v && (array_p (v) || assoc_p (v)))
6405 { /* [ */
6406 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
6407 {
6408 /* Callers have to differentiate between indexed and associative */
6409 vtype = VT_ARRAYVAR;
6410 if (temp[0] == '*')
6411 vtype |= VT_STARSUB;
6412 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
6413 }
6414 else
6415 {
6416 vtype = VT_ARRAYMEMBER;
6417 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6418 }
6419 *varp = v;
6420 }
6421 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
6422 {
6423 vtype = VT_VARIABLE;
6424 *varp = v;
6425 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6426 *valp = dequote_string (value);
6427 else
6428 *valp = dequote_escapes (value);
6429 }
6430 else
6431 {
6432 vtype = VT_ARRAYMEMBER;
6433 *varp = v;
6434 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6435 }
6436 }
6437 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
6438 {
6439 vtype = VT_ARRAYMEMBER;
6440 *varp = v;
6441 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
6442 }
6443 else
6444 #endif
6445 {
6446 if (value && vtype == VT_VARIABLE)
6447 {
6448 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6449 *valp = dequote_string (value);
6450 else
6451 *valp = dequote_escapes (value);
6452 }
6453 else
6454 *valp = value;
6455 }
6456
6457 if (want_indir)
6458 free (vname);
6459
6460 return vtype;
6461 }
6462
6463 /******************************************************/
6464 /* */
6465 /* Functions to extract substrings of variable values */
6466 /* */
6467 /******************************************************/
6468
6469 #if defined (HANDLE_MULTIBYTE)
6470 /* Character-oriented rather than strictly byte-oriented substrings. S and
6471 E, rather being strict indices into STRING, indicate character (possibly
6472 multibyte character) positions that require calculation.
6473 Used by the ${param:offset[:length]} expansion. */
6474 static char *
6475 mb_substring (string, s, e)
6476 char *string;
6477 int s, e;
6478 {
6479 char *tt;
6480 int start, stop, i, slen;
6481 DECLARE_MBSTATE;
6482
6483 start = 0;
6484 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
6485 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
6486
6487 i = s;
6488 while (string[start] && i--)
6489 ADVANCE_CHAR (string, slen, start);
6490 stop = start;
6491 i = e - s;
6492 while (string[stop] && i--)
6493 ADVANCE_CHAR (string, slen, stop);
6494 tt = substring (string, start, stop);
6495 return tt;
6496 }
6497 #endif
6498
6499 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
6500 is `@', use the positional parameters; otherwise, use the value of
6501 VARNAME. If VARNAME is an array variable, use the array elements. */
6502
6503 static char *
6504 parameter_brace_substring (varname, value, ind, substr, quoted, flags)
6505 char *varname, *value;
6506 int ind;
6507 char *substr;
6508 int quoted, flags;
6509 {
6510 intmax_t e1, e2;
6511 int vtype, r, starsub;
6512 char *temp, *val, *tt, *oname;
6513 SHELL_VAR *v;
6514
6515 if (value == 0)
6516 return ((char *)NULL);
6517
6518 oname = this_command_name;
6519 this_command_name = varname;
6520
6521 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6522 if (vtype == -1)
6523 {
6524 this_command_name = oname;
6525 return ((char *)NULL);
6526 }
6527
6528 starsub = vtype & VT_STARSUB;
6529 vtype &= ~VT_STARSUB;
6530
6531 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
6532 this_command_name = oname;
6533 if (r <= 0)
6534 {
6535 if (vtype == VT_VARIABLE)
6536 FREE (val);
6537 return ((r == 0) ? &expand_param_error : (char *)NULL);
6538 }
6539
6540 switch (vtype)
6541 {
6542 case VT_VARIABLE:
6543 case VT_ARRAYMEMBER:
6544 #if defined (HANDLE_MULTIBYTE)
6545 if (MB_CUR_MAX > 1)
6546 tt = mb_substring (val, e1, e2);
6547 else
6548 #endif
6549 tt = substring (val, e1, e2);
6550
6551 if (vtype == VT_VARIABLE)
6552 FREE (val);
6553 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6554 temp = quote_string (tt);
6555 else
6556 temp = tt ? quote_escapes (tt) : (char *)NULL;
6557 FREE (tt);
6558 break;
6559 case VT_POSPARMS:
6560 tt = pos_params (varname, e1, e2, quoted);
6561 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
6562 {
6563 temp = tt ? quote_escapes (tt) : (char *)NULL;
6564 FREE (tt);
6565 }
6566 else
6567 temp = tt;
6568 break;
6569 #if defined (ARRAY_VARS)
6570 case VT_ARRAYVAR:
6571 if (assoc_p (v))
6572 /* we convert to list and take first e2 elements starting at e1th
6573 element -- officially undefined for now */
6574 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
6575 else
6576 /* We want E2 to be the number of elements desired (arrays can be sparse,
6577 so verify_substring_values just returns the numbers specified and we
6578 rely on array_subrange to understand how to deal with them). */
6579 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
6580 /* array_subrange now calls array_quote_escapes as appropriate, so the
6581 caller no longer needs to. */
6582 break;
6583 #endif
6584 default:
6585 temp = (char *)NULL;
6586 }
6587
6588 return temp;
6589 }
6590
6591 /****************************************************************/
6592 /* */
6593 /* Functions to perform pattern substitution on variable values */
6594 /* */
6595 /****************************************************************/
6596
6597 static int
6598 shouldexp_replacement (s)
6599 char *s;
6600 {
6601 register char *p;
6602
6603 for (p = s; p && *p; p++)
6604 {
6605 if (*p == '\\')
6606 p++;
6607 else if (*p == '&')
6608 return 1;
6609 }
6610 return 0;
6611 }
6612
6613 char *
6614 pat_subst (string, pat, rep, mflags)
6615 char *string, *pat, *rep;
6616 int mflags;
6617 {
6618 char *ret, *s, *e, *str, *rstr, *mstr;
6619 int rsize, rptr, l, replen, mtype, rxpand, rslen, mlen;
6620
6621 if (string == 0)
6622 return (savestring (""));
6623
6624 mtype = mflags & MATCH_TYPEMASK;
6625
6626 #if 0 /* bash-4.2 ? */
6627 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
6628 #else
6629 rxpand = 0;
6630 #endif
6631
6632 /* Special cases:
6633 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
6634 * with REP and return the result.
6635 * 2. A null pattern with mtype == MATCH_END means to append REP to
6636 * STRING and return the result.
6637 * These don't understand or process `&' in the replacement string.
6638 */
6639 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
6640 {
6641 replen = STRLEN (rep);
6642 l = STRLEN (string);
6643 ret = (char *)xmalloc (replen + l + 2);
6644 if (replen == 0)
6645 strcpy (ret, string);
6646 else if (mtype == MATCH_BEG)
6647 {
6648 strcpy (ret, rep);
6649 strcpy (ret + replen, string);
6650 }
6651 else
6652 {
6653 strcpy (ret, string);
6654 strcpy (ret + l, rep);
6655 }
6656 return (ret);
6657 }
6658
6659 ret = (char *)xmalloc (rsize = 64);
6660 ret[0] = '\0';
6661
6662 for (replen = STRLEN (rep), rptr = 0, str = string;;)
6663 {
6664 if (match_pattern (str, pat, mtype, &s, &e) == 0)
6665 break;
6666 l = s - str;
6667
6668 if (rxpand)
6669 {
6670 int x;
6671 mlen = e - s;
6672 mstr = xmalloc (mlen + 1);
6673 for (x = 0; x < mlen; x++)
6674 mstr[x] = s[x];
6675 mstr[mlen] = '\0';
6676 rstr = strcreplace (rep, '&', mstr, 0);
6677 rslen = strlen (rstr);
6678 }
6679 else
6680 {
6681 rstr = rep;
6682 rslen = replen;
6683 }
6684
6685 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
6686
6687 /* OK, now copy the leading unmatched portion of the string (from
6688 str to s) to ret starting at rptr (the current offset). Then copy
6689 the replacement string at ret + rptr + (s - str). Increment
6690 rptr (if necessary) and str and go on. */
6691 if (l)
6692 {
6693 strncpy (ret + rptr, str, l);
6694 rptr += l;
6695 }
6696 if (replen)
6697 {
6698 strncpy (ret + rptr, rstr, rslen);
6699 rptr += rslen;
6700 }
6701 str = e; /* e == end of match */
6702
6703 if (rstr != rep)
6704 free (rstr);
6705
6706 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
6707 break;
6708
6709 if (s == e)
6710 {
6711 /* On a zero-length match, make sure we copy one character, since
6712 we increment one character to avoid infinite recursion. */
6713 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
6714 ret[rptr++] = *str++;
6715 e++; /* avoid infinite recursion on zero-length match */
6716 }
6717 }
6718
6719 /* Now copy the unmatched portion of the input string */
6720 if (str && *str)
6721 {
6722 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
6723 strcpy (ret + rptr, str);
6724 }
6725 else
6726 ret[rptr] = '\0';
6727
6728 return ret;
6729 }
6730
6731 /* Do pattern match and replacement on the positional parameters. */
6732 static char *
6733 pos_params_pat_subst (string, pat, rep, mflags)
6734 char *string, *pat, *rep;
6735 int mflags;
6736 {
6737 WORD_LIST *save, *params;
6738 WORD_DESC *w;
6739 char *ret;
6740 int pchar, qflags;
6741
6742 save = params = list_rest_of_args ();
6743 if (save == 0)
6744 return ((char *)NULL);
6745
6746 for ( ; params; params = params->next)
6747 {
6748 ret = pat_subst (params->word->word, pat, rep, mflags);
6749 w = alloc_word_desc ();
6750 w->word = ret ? ret : savestring ("");
6751 dispose_word (params->word);
6752 params->word = w;
6753 }
6754
6755 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6756 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6757
6758 ret = string_list_pos_params (pchar, save, qflags);
6759
6760 dispose_words (save);
6761
6762 return (ret);
6763 }
6764
6765 /* Perform pattern substitution on VALUE, which is the expansion of
6766 VARNAME. PATSUB is an expression supplying the pattern to match
6767 and the string to substitute. QUOTED is a flags word containing
6768 the type of quoting currently in effect. */
6769 static char *
6770 parameter_brace_patsub (varname, value, ind, patsub, quoted, flags)
6771 char *varname, *value;
6772 int ind;
6773 char *patsub;
6774 int quoted, flags;
6775 {
6776 int vtype, mflags, starsub, delim;
6777 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
6778 SHELL_VAR *v;
6779
6780 if (value == 0)
6781 return ((char *)NULL);
6782
6783 this_command_name = varname;
6784
6785 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6786 if (vtype == -1)
6787 return ((char *)NULL);
6788
6789 starsub = vtype & VT_STARSUB;
6790 vtype &= ~VT_STARSUB;
6791
6792 mflags = 0;
6793 /* PATSUB is never NULL when this is called. */
6794 if (*patsub == '/')
6795 {
6796 mflags |= MATCH_GLOBREP;
6797 patsub++;
6798 }
6799
6800 /* Malloc this because expand_string_if_necessary or one of the expansion
6801 functions in its call chain may free it on a substitution error. */
6802 lpatsub = savestring (patsub);
6803
6804 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6805 mflags |= MATCH_QUOTED;
6806
6807 if (starsub)
6808 mflags |= MATCH_STARSUB;
6809
6810 /* If the pattern starts with a `/', make sure we skip over it when looking
6811 for the replacement delimiter. */
6812 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
6813 if (lpatsub[delim] == '/')
6814 {
6815 lpatsub[delim] = 0;
6816 rep = lpatsub + delim + 1;
6817 }
6818 else
6819 rep = (char *)NULL;
6820
6821 if (rep && *rep == '\0')
6822 rep = (char *)NULL;
6823
6824 /* Perform the same expansions on the pattern as performed by the
6825 pattern removal expansions. */
6826 pat = getpattern (lpatsub, quoted, 1);
6827
6828 if (rep)
6829 {
6830 /* We want to perform quote removal on the expanded replacement even if
6831 the entire expansion is double-quoted because the parser and string
6832 extraction functions treated quotes in the replacement string as
6833 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
6834 if (shell_compatibility_level > 42)
6835 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
6836 /* This is the bash-4.2 code. */
6837 else if ((mflags & MATCH_QUOTED) == 0)
6838 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
6839 else
6840 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
6841 }
6842
6843 /* ksh93 doesn't allow the match specifier to be a part of the expanded
6844 pattern. This is an extension. Make sure we don't anchor the pattern
6845 at the beginning or end of the string if we're doing global replacement,
6846 though. */
6847 p = pat;
6848 if (mflags & MATCH_GLOBREP)
6849 mflags |= MATCH_ANY;
6850 else if (pat && pat[0] == '#')
6851 {
6852 mflags |= MATCH_BEG;
6853 p++;
6854 }
6855 else if (pat && pat[0] == '%')
6856 {
6857 mflags |= MATCH_END;
6858 p++;
6859 }
6860 else
6861 mflags |= MATCH_ANY;
6862
6863 /* OK, we now want to substitute REP for PAT in VAL. If
6864 flags & MATCH_GLOBREP is non-zero, the substitution is done
6865 everywhere, otherwise only the first occurrence of PAT is
6866 replaced. The pattern matching code doesn't understand
6867 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
6868 values passed in (VT_VARIABLE) so the pattern substitution
6869 code works right. We need to requote special chars after
6870 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
6871 other cases if QUOTED == 0, since the posparams and arrays
6872 indexed by * or @ do special things when QUOTED != 0. */
6873
6874 switch (vtype)
6875 {
6876 case VT_VARIABLE:
6877 case VT_ARRAYMEMBER:
6878 temp = pat_subst (val, p, rep, mflags);
6879 if (vtype == VT_VARIABLE)
6880 FREE (val);
6881 if (temp)
6882 {
6883 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
6884 free (temp);
6885 temp = tt;
6886 }
6887 break;
6888 case VT_POSPARMS:
6889 temp = pos_params_pat_subst (val, p, rep, mflags);
6890 if (temp && (mflags & MATCH_QUOTED) == 0)
6891 {
6892 tt = quote_escapes (temp);
6893 free (temp);
6894 temp = tt;
6895 }
6896 break;
6897 #if defined (ARRAY_VARS)
6898 case VT_ARRAYVAR:
6899 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
6900 : array_patsub (array_cell (v), p, rep, mflags);
6901 /* Don't call quote_escapes anymore; array_patsub calls
6902 array_quote_escapes as appropriate before adding the
6903 space separators; ditto for assoc_patsub. */
6904 break;
6905 #endif
6906 }
6907
6908 FREE (pat);
6909 FREE (rep);
6910 free (lpatsub);
6911
6912 return temp;
6913 }
6914
6915 /****************************************************************/
6916 /* */
6917 /* Functions to perform case modification on variable values */
6918 /* */
6919 /****************************************************************/
6920
6921 /* Do case modification on the positional parameters. */
6922
6923 static char *
6924 pos_params_modcase (string, pat, modop, mflags)
6925 char *string, *pat;
6926 int modop;
6927 int mflags;
6928 {
6929 WORD_LIST *save, *params;
6930 WORD_DESC *w;
6931 char *ret;
6932 int pchar, qflags;
6933
6934 save = params = list_rest_of_args ();
6935 if (save == 0)
6936 return ((char *)NULL);
6937
6938 for ( ; params; params = params->next)
6939 {
6940 ret = sh_modcase (params->word->word, pat, modop);
6941 w = alloc_word_desc ();
6942 w->word = ret ? ret : savestring ("");
6943 dispose_word (params->word);
6944 params->word = w;
6945 }
6946
6947 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6948 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6949
6950 ret = string_list_pos_params (pchar, save, qflags);
6951 dispose_words (save);
6952
6953 return (ret);
6954 }
6955
6956 /* Perform case modification on VALUE, which is the expansion of
6957 VARNAME. MODSPEC is an expression supplying the type of modification
6958 to perform. QUOTED is a flags word containing the type of quoting
6959 currently in effect. */
6960 static char *
6961 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
6962 char *varname, *value;
6963 int ind, modspec;
6964 char *patspec;
6965 int quoted, flags;
6966 {
6967 int vtype, starsub, modop, mflags, x;
6968 char *val, *temp, *pat, *p, *lpat, *tt;
6969 SHELL_VAR *v;
6970
6971 if (value == 0)
6972 return ((char *)NULL);
6973
6974 this_command_name = varname;
6975
6976 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6977 if (vtype == -1)
6978 return ((char *)NULL);
6979
6980 starsub = vtype & VT_STARSUB;
6981 vtype &= ~VT_STARSUB;
6982
6983 modop = 0;
6984 mflags = 0;
6985 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6986 mflags |= MATCH_QUOTED;
6987 if (starsub)
6988 mflags |= MATCH_STARSUB;
6989
6990 p = patspec;
6991 if (modspec == '^')
6992 {
6993 x = p && p[0] == modspec;
6994 modop = x ? CASE_UPPER : CASE_UPFIRST;
6995 p += x;
6996 }
6997 else if (modspec == ',')
6998 {
6999 x = p && p[0] == modspec;
7000 modop = x ? CASE_LOWER : CASE_LOWFIRST;
7001 p += x;
7002 }
7003 else if (modspec == '~')
7004 {
7005 x = p && p[0] == modspec;
7006 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7007 p += x;
7008 }
7009
7010 lpat = p ? savestring (p) : 0;
7011 /* Perform the same expansions on the pattern as performed by the
7012 pattern removal expansions. FOR LATER */
7013 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7014
7015 /* OK, now we do the case modification. */
7016 switch (vtype)
7017 {
7018 case VT_VARIABLE:
7019 case VT_ARRAYMEMBER:
7020 temp = sh_modcase (val, pat, modop);
7021 if (vtype == VT_VARIABLE)
7022 FREE (val);
7023 if (temp)
7024 {
7025 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7026 free (temp);
7027 temp = tt;
7028 }
7029 break;
7030
7031 case VT_POSPARMS:
7032 temp = pos_params_modcase (val, pat, modop, mflags);
7033 if (temp && (mflags & MATCH_QUOTED) == 0)
7034 {
7035 tt = quote_escapes (temp);
7036 free (temp);
7037 temp = tt;
7038 }
7039 break;
7040
7041 #if defined (ARRAY_VARS)
7042 case VT_ARRAYVAR:
7043 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7044 : array_modcase (array_cell (v), pat, modop, mflags);
7045 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7046 as appropriate before adding the space separators; ditto for
7047 assoc_modcase. */
7048 break;
7049 #endif
7050 }
7051
7052 FREE (pat);
7053 free (lpat);
7054
7055 return temp;
7056 }
7057
7058 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7059 any occur, this must be a nested command substitution, so return 0.
7060 Otherwise, return 1. A valid arithmetic expression must always have a
7061 ( before a matching ), so any cases where there are more right parens
7062 means that this must not be an arithmetic expression, though the parser
7063 will not accept it without a balanced total number of parens. */
7064 static int
7065 chk_arithsub (s, len)
7066 const char *s;
7067 int len;
7068 {
7069 int i, count;
7070 DECLARE_MBSTATE;
7071
7072 i = count = 0;
7073 while (i < len)
7074 {
7075 if (s[i] == LPAREN)
7076 count++;
7077 else if (s[i] == RPAREN)
7078 {
7079 count--;
7080 if (count < 0)
7081 return 0;
7082 }
7083
7084 switch (s[i])
7085 {
7086 default:
7087 ADVANCE_CHAR (s, len, i);
7088 break;
7089
7090 case '\\':
7091 i++;
7092 if (s[i])
7093 ADVANCE_CHAR (s, len, i);
7094 break;
7095
7096 case '\'':
7097 i = skip_single_quoted (s, len, ++i);
7098 break;
7099
7100 case '"':
7101 i = skip_double_quoted ((char *)s, len, ++i);
7102 break;
7103 }
7104 }
7105
7106 return (count == 0);
7107 }
7108
7109 /****************************************************************/
7110 /* */
7111 /* Functions to perform parameter expansion on a string */
7112 /* */
7113 /****************************************************************/
7114
7115 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
7116 static WORD_DESC *
7117 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
7118 char *string;
7119 int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at, pflags;
7120 {
7121 int check_nullness, var_is_set, var_is_null, var_is_special;
7122 int want_substring, want_indir, want_patsub, want_casemod;
7123 char *name, *value, *temp, *temp1;
7124 WORD_DESC *tdesc, *ret;
7125 int t_index, sindex, c, tflag, modspec;
7126 intmax_t number;
7127 arrayind_t ind;
7128
7129 temp = temp1 = value = (char *)NULL;
7130 var_is_set = var_is_null = var_is_special = check_nullness = 0;
7131 want_substring = want_indir = want_patsub = want_casemod = 0;
7132
7133 sindex = *indexp;
7134 t_index = ++sindex;
7135 /* ${#var} doesn't have any of the other parameter expansions on it. */
7136 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
7137 name = string_extract (string, &t_index, "}", SX_VARNAME);
7138 else
7139 #if defined (CASEMOD_EXPANSIONS)
7140 /* To enable case-toggling expansions using the `~' operator character
7141 change the 1 to 0. */
7142 # if defined (CASEMOD_CAPCASE)
7143 name = string_extract (string, &t_index, "#%^,~:-=?+/}", SX_VARNAME);
7144 # else
7145 name = string_extract (string, &t_index, "#%^,:-=?+/}", SX_VARNAME);
7146 # endif /* CASEMOD_CAPCASE */
7147 #else
7148 name = string_extract (string, &t_index, "#%:-=?+/}", SX_VARNAME);
7149 #endif /* CASEMOD_EXPANSIONS */
7150
7151 ret = 0;
7152 tflag = 0;
7153
7154 ind = INTMAX_MIN;
7155
7156 /* If the name really consists of a special variable, then make sure
7157 that we have the entire name. We don't allow indirect references
7158 to special variables except `#', `?', `@' and `*'. */
7159 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
7160 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
7161 {
7162 t_index++;
7163 temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0);
7164 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
7165 *name = string[sindex];
7166 if (string[sindex] == '!')
7167 {
7168 /* indirect reference of $#, $?, $@, or $* */
7169 name[1] = string[sindex + 1];
7170 strcpy (name + 2, temp1);
7171 }
7172 else
7173 strcpy (name + 1, temp1);
7174 free (temp1);
7175 }
7176 sindex = t_index;
7177
7178 /* Find out what character ended the variable name. Then
7179 do the appropriate thing. */
7180 if (c = string[sindex])
7181 sindex++;
7182
7183 /* If c is followed by one of the valid parameter expansion
7184 characters, move past it as normal. If not, assume that
7185 a substring specification is being given, and do not move
7186 past it. */
7187 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
7188 {
7189 check_nullness++;
7190 if (c = string[sindex])
7191 sindex++;
7192 }
7193 else if (c == ':' && string[sindex] != RBRACE)
7194 want_substring = 1;
7195 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
7196 want_patsub = 1;
7197 #if defined (CASEMOD_EXPANSIONS)
7198 else if (c == '^' || c == ',' || c == '~')
7199 {
7200 modspec = c;
7201 want_casemod = 1;
7202 }
7203 #endif
7204
7205 /* Catch the valid and invalid brace expressions that made it through the
7206 tests above. */
7207 /* ${#-} is a valid expansion and means to take the length of $-.
7208 Similarly for ${#?} and ${##}... */
7209 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7210 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
7211 {
7212 name = (char *)xrealloc (name, 3);
7213 name[1] = c;
7214 name[2] = '\0';
7215 c = string[sindex++];
7216 }
7217
7218 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
7219 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7220 member (c, "%:=+/") && string[sindex] == RBRACE)
7221 {
7222 temp = (char *)NULL;
7223 goto bad_substitution;
7224 }
7225
7226 /* Indirect expansion begins with a `!'. A valid indirect expansion is
7227 either a variable name, one of the positional parameters or a special
7228 variable that expands to one of the positional parameters. */
7229 want_indir = *name == '!' &&
7230 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7231 || VALID_INDIR_PARAM (name[1]));
7232
7233 /* Determine the value of this variable. */
7234
7235 /* Check for special variables, directly referenced. */
7236 if (SPECIAL_VAR (name, want_indir))
7237 var_is_special++;
7238
7239 /* Check for special expansion things, like the length of a parameter */
7240 if (*name == '#' && name[1])
7241 {
7242 /* If we are not pointing at the character just after the
7243 closing brace, then we haven't gotten all of the name.
7244 Since it begins with a special character, this is a bad
7245 substitution. Also check NAME for validity before trying
7246 to go on. */
7247 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
7248 {
7249 temp = (char *)NULL;
7250 goto bad_substitution;
7251 }
7252
7253 number = parameter_brace_expand_length (name);
7254 if (number == INTMAX_MIN && unbound_vars_is_error)
7255 {
7256 last_command_exit_value = EXECUTION_FAILURE;
7257 err_unboundvar (name+1);
7258 free (name);
7259 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7260 }
7261 free (name);
7262
7263 *indexp = sindex;
7264 if (number < 0)
7265 return (&expand_wdesc_error);
7266 else
7267 {
7268 ret = alloc_word_desc ();
7269 ret->word = itos (number);
7270 return ret;
7271 }
7272 }
7273
7274 /* ${@} is identical to $@. */
7275 if (name[0] == '@' && name[1] == '\0')
7276 {
7277 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7278 *quoted_dollar_atp = 1;
7279
7280 if (contains_dollar_at)
7281 *contains_dollar_at = 1;
7282
7283 tflag |= W_DOLLARAT;
7284 }
7285
7286 /* Process ${!PREFIX*} expansion. */
7287 if (want_indir && string[sindex - 1] == RBRACE &&
7288 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
7289 legal_variable_starter ((unsigned char) name[1]))
7290 {
7291 char **x;
7292 WORD_LIST *xlist;
7293
7294 temp1 = savestring (name + 1);
7295 number = strlen (temp1);
7296 temp1[number - 1] = '\0';
7297 x = all_variables_matching_prefix (temp1);
7298 xlist = strvec_to_word_list (x, 0, 0);
7299 if (string[sindex - 2] == '*')
7300 temp = string_list_dollar_star (xlist);
7301 else
7302 {
7303 temp = string_list_dollar_at (xlist, quoted);
7304 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7305 *quoted_dollar_atp = 1;
7306 if (contains_dollar_at)
7307 *contains_dollar_at = 1;
7308
7309 tflag |= W_DOLLARAT;
7310 }
7311 free (x);
7312 dispose_words (xlist);
7313 free (temp1);
7314 *indexp = sindex;
7315
7316 free (name);
7317
7318 ret = alloc_word_desc ();
7319 ret->word = temp;
7320 ret->flags = tflag; /* XXX */
7321 return ret;
7322 }
7323
7324 #if defined (ARRAY_VARS)
7325 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
7326 if (want_indir && string[sindex - 1] == RBRACE &&
7327 string[sindex - 2] == ']' && valid_array_reference (name+1))
7328 {
7329 char *x, *x1;
7330
7331 temp1 = savestring (name + 1);
7332 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
7333 FREE (x);
7334 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
7335 {
7336 temp = array_keys (temp1, quoted); /* handles assoc vars too */
7337 if (x1[0] == '@')
7338 {
7339 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7340 *quoted_dollar_atp = 1;
7341 if (contains_dollar_at)
7342 *contains_dollar_at = 1;
7343
7344 tflag |= W_DOLLARAT;
7345 }
7346
7347 free (temp1);
7348 *indexp = sindex;
7349
7350 ret = alloc_word_desc ();
7351 ret->word = temp;
7352 ret->flags = tflag; /* XXX */
7353 return ret;
7354 }
7355
7356 free (temp1);
7357 }
7358 #endif /* ARRAY_VARS */
7359
7360 /* Make sure that NAME is valid before trying to go on. */
7361 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
7362 var_is_special) == 0)
7363 {
7364 temp = (char *)NULL;
7365 goto bad_substitution;
7366 }
7367
7368 if (want_indir)
7369 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
7370 else
7371 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
7372
7373 if (tdesc)
7374 {
7375 temp = tdesc->word;
7376 tflag = tdesc->flags;
7377 dispose_word_desc (tdesc);
7378 }
7379 else
7380 temp = (char *)0;
7381
7382 if (temp == &expand_param_error || temp == &expand_param_fatal)
7383 {
7384 FREE (name);
7385 FREE (value);
7386 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
7387 }
7388
7389 #if defined (ARRAY_VARS)
7390 if (valid_array_reference (name))
7391 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at);
7392 #endif
7393
7394 var_is_set = temp != (char *)0;
7395 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
7396 /* XXX - this may not need to be restricted to special variables */
7397 if (check_nullness)
7398 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
7399
7400 /* Get the rest of the stuff inside the braces. */
7401 if (c && c != RBRACE)
7402 {
7403 /* Extract the contents of the ${ ... } expansion
7404 according to the Posix.2 rules. */
7405 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
7406 if (string[sindex] == RBRACE)
7407 sindex++;
7408 else
7409 goto bad_substitution;
7410 }
7411 else
7412 value = (char *)NULL;
7413
7414 *indexp = sindex;
7415
7416 /* All the cases where an expansion can possibly generate an unbound
7417 variable error. */
7418 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
7419 {
7420 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]))
7421 {
7422 last_command_exit_value = EXECUTION_FAILURE;
7423 err_unboundvar (name);
7424 FREE (value);
7425 FREE (temp);
7426 free (name);
7427 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7428 }
7429 }
7430
7431 /* If this is a substring spec, process it and add the result. */
7432 if (want_substring)
7433 {
7434 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7435 FREE (name);
7436 FREE (value);
7437 FREE (temp);
7438
7439 if (temp1 == &expand_param_error)
7440 return (&expand_wdesc_error);
7441 else if (temp1 == &expand_param_fatal)
7442 return (&expand_wdesc_fatal);
7443
7444 ret = alloc_word_desc ();
7445 ret->word = temp1;
7446 /* We test quoted_dollar_atp because we want variants with double-quoted
7447 "$@" to take a different code path. In fact, we make sure at the end
7448 of expand_word_internal that we're only looking at these flags if
7449 quoted_dollar_at == 0. */
7450 if (temp1 &&
7451 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7452 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7453 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7454 return ret;
7455 }
7456 else if (want_patsub)
7457 {
7458 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7459 FREE (name);
7460 FREE (value);
7461 FREE (temp);
7462
7463 if (temp1 == &expand_param_error)
7464 return (&expand_wdesc_error);
7465 else if (temp1 == &expand_param_fatal)
7466 return (&expand_wdesc_fatal);
7467
7468 ret = alloc_word_desc ();
7469 ret->word = temp1;
7470 if (temp1 &&
7471 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7472 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7473 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7474 return ret;
7475 }
7476 #if defined (CASEMOD_EXPANSIONS)
7477 else if (want_casemod)
7478 {
7479 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7480 FREE (name);
7481 FREE (value);
7482 FREE (temp);
7483
7484 if (temp1 == &expand_param_error)
7485 return (&expand_wdesc_error);
7486 else if (temp1 == &expand_param_fatal)
7487 return (&expand_wdesc_fatal);
7488
7489 ret = alloc_word_desc ();
7490 ret->word = temp1;
7491 if (temp1 &&
7492 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7493 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7494 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7495 return ret;
7496 }
7497 #endif
7498
7499 /* Do the right thing based on which character ended the variable name. */
7500 switch (c)
7501 {
7502 default:
7503 case '\0':
7504 bad_substitution:
7505 last_command_exit_value = EXECUTION_FAILURE;
7506 report_error (_("%s: bad substitution"), string ? string : "??");
7507 FREE (value);
7508 FREE (temp);
7509 free (name);
7510 return &expand_wdesc_error;
7511
7512 case RBRACE:
7513 break;
7514
7515 case '#': /* ${param#[#]pattern} */
7516 case '%': /* ${param%[%]pattern} */
7517 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
7518 {
7519 FREE (value);
7520 break;
7521 }
7522 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7523 free (temp);
7524 free (value);
7525 free (name);
7526
7527 ret = alloc_word_desc ();
7528 ret->word = temp1;
7529 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7530 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7531 return ret;
7532
7533 case '-':
7534 case '=':
7535 case '?':
7536 case '+':
7537 if (var_is_set && var_is_null == 0)
7538 {
7539 /* If the operator is `+', we don't want the value of the named
7540 variable for anything, just the value of the right hand side. */
7541 if (c == '+')
7542 {
7543 /* XXX -- if we're double-quoted and the named variable is "$@",
7544 we want to turn off any special handling of "$@" --
7545 we're not using it, so whatever is on the rhs applies. */
7546 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7547 *quoted_dollar_atp = 0;
7548 if (contains_dollar_at)
7549 *contains_dollar_at = 0;
7550
7551 FREE (temp);
7552 if (value)
7553 {
7554 /* From Posix discussion on austin-group list. Issue 221
7555 requires that backslashes escaping `}' inside
7556 double-quoted ${...} be removed. */
7557 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7558 quoted |= Q_DOLBRACE;
7559 ret = parameter_brace_expand_rhs (name, value, c,
7560 quoted,
7561 quoted_dollar_atp,
7562 contains_dollar_at);
7563 /* XXX - fix up later, esp. noting presence of
7564 W_HASQUOTEDNULL in ret->flags */
7565 free (value);
7566 }
7567 else
7568 temp = (char *)NULL;
7569 }
7570 else
7571 {
7572 FREE (value);
7573 }
7574 /* Otherwise do nothing; just use the value in TEMP. */
7575 }
7576 else /* VAR not set or VAR is NULL. */
7577 {
7578 FREE (temp);
7579 temp = (char *)NULL;
7580 if (c == '=' && var_is_special)
7581 {
7582 last_command_exit_value = EXECUTION_FAILURE;
7583 report_error (_("$%s: cannot assign in this way"), name);
7584 free (name);
7585 free (value);
7586 return &expand_wdesc_error;
7587 }
7588 else if (c == '?')
7589 {
7590 parameter_brace_expand_error (name, value);
7591 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7592 }
7593 else if (c != '+')
7594 {
7595 /* XXX -- if we're double-quoted and the named variable is "$@",
7596 we want to turn off any special handling of "$@" --
7597 we're not using it, so whatever is on the rhs applies. */
7598 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7599 *quoted_dollar_atp = 0;
7600 if (contains_dollar_at)
7601 *contains_dollar_at = 0;
7602
7603 /* From Posix discussion on austin-group list. Issue 221 requires
7604 that backslashes escaping `}' inside double-quoted ${...} be
7605 removed. */
7606 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7607 quoted |= Q_DOLBRACE;
7608 ret = parameter_brace_expand_rhs (name, value, c, quoted,
7609 quoted_dollar_atp,
7610 contains_dollar_at);
7611 /* XXX - fix up later, esp. noting presence of
7612 W_HASQUOTEDNULL in tdesc->flags */
7613 }
7614 free (value);
7615 }
7616
7617 break;
7618 }
7619 free (name);
7620
7621 if (ret == 0)
7622 {
7623 ret = alloc_word_desc ();
7624 ret->flags = tflag;
7625 ret->word = temp;
7626 }
7627 return (ret);
7628 }
7629
7630 /* Expand a single ${xxx} expansion. The braces are optional. When
7631 the braces are used, parameter_brace_expand() does the work,
7632 possibly calling param_expand recursively. */
7633 static WORD_DESC *
7634 param_expand (string, sindex, quoted, expanded_something,
7635 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
7636 pflags)
7637 char *string;
7638 int *sindex, quoted, *expanded_something, *contains_dollar_at;
7639 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
7640 {
7641 char *temp, *temp1, uerror[3];
7642 int zindex, t_index, expok;
7643 unsigned char c;
7644 intmax_t number;
7645 SHELL_VAR *var;
7646 WORD_LIST *list;
7647 WORD_DESC *tdesc, *ret;
7648 int tflag;
7649
7650 zindex = *sindex;
7651 c = string[++zindex];
7652
7653 temp = (char *)NULL;
7654 ret = tdesc = (WORD_DESC *)NULL;
7655 tflag = 0;
7656
7657 /* Do simple cases first. Switch on what follows '$'. */
7658 switch (c)
7659 {
7660 /* $0 .. $9? */
7661 case '0':
7662 case '1':
7663 case '2':
7664 case '3':
7665 case '4':
7666 case '5':
7667 case '6':
7668 case '7':
7669 case '8':
7670 case '9':
7671 temp1 = dollar_vars[TODIGIT (c)];
7672 if (unbound_vars_is_error && temp1 == (char *)NULL)
7673 {
7674 uerror[0] = '$';
7675 uerror[1] = c;
7676 uerror[2] = '\0';
7677 last_command_exit_value = EXECUTION_FAILURE;
7678 err_unboundvar (uerror);
7679 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7680 }
7681 if (temp1)
7682 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7683 ? quote_string (temp1)
7684 : quote_escapes (temp1);
7685 else
7686 temp = (char *)NULL;
7687
7688 break;
7689
7690 /* $$ -- pid of the invoking shell. */
7691 case '$':
7692 temp = itos (dollar_dollar_pid);
7693 break;
7694
7695 /* $# -- number of positional parameters. */
7696 case '#':
7697 temp = itos (number_of_args ());
7698 break;
7699
7700 /* $? -- return value of the last synchronous command. */
7701 case '?':
7702 temp = itos (last_command_exit_value);
7703 break;
7704
7705 /* $- -- flags supplied to the shell on invocation or by `set'. */
7706 case '-':
7707 temp = which_set_flags ();
7708 break;
7709
7710 /* $! -- Pid of the last asynchronous command. */
7711 case '!':
7712 /* If no asynchronous pids have been created, expand to nothing.
7713 If `set -u' has been executed, and no async processes have
7714 been created, this is an expansion error. */
7715 if (last_asynchronous_pid == NO_PID)
7716 {
7717 if (expanded_something)
7718 *expanded_something = 0;
7719 temp = (char *)NULL;
7720 if (unbound_vars_is_error)
7721 {
7722 uerror[0] = '$';
7723 uerror[1] = c;
7724 uerror[2] = '\0';
7725 last_command_exit_value = EXECUTION_FAILURE;
7726 err_unboundvar (uerror);
7727 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7728 }
7729 }
7730 else
7731 temp = itos (last_asynchronous_pid);
7732 break;
7733
7734 /* The only difference between this and $@ is when the arg is quoted. */
7735 case '*': /* `$*' */
7736 list = list_rest_of_args ();
7737
7738 #if 0
7739 /* According to austin-group posix proposal by Geoff Clare in
7740 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7741
7742 "The shell shall write a message to standard error and
7743 immediately exit when it tries to expand an unset parameter
7744 other than the '@' and '*' special parameters."
7745 */
7746
7747 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7748 {
7749 uerror[0] = '$';
7750 uerror[1] = '*';
7751 uerror[2] = '\0';
7752 last_command_exit_value = EXECUTION_FAILURE;
7753 err_unboundvar (uerror);
7754 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7755 }
7756 #endif
7757
7758 /* If there are no command-line arguments, this should just
7759 disappear if there are other characters in the expansion,
7760 even if it's quoted. */
7761 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
7762 temp = (char *)NULL;
7763 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
7764 {
7765 /* If we have "$*" we want to make a string of the positional
7766 parameters, separated by the first character of $IFS, and
7767 quote the whole string, including the separators. If IFS
7768 is unset, the parameters are separated by ' '; if $IFS is
7769 null, the parameters are concatenated. */
7770 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
7771 if (temp)
7772 {
7773 temp1 = quote_string (temp);
7774 if (*temp == 0)
7775 tflag |= W_HASQUOTEDNULL;
7776 free (temp);
7777 temp = temp1;
7778 }
7779 }
7780 else
7781 {
7782 /* We check whether or not we're eventually going to split $* here,
7783 for example when IFS is empty and we are processing the rhs of
7784 an assignment statement. In that case, we don't separate the
7785 arguments at all. Otherwise, if the $* is not quoted it is
7786 identical to $@ */
7787 # if defined (HANDLE_MULTIBYTE)
7788 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
7789 # else
7790 if (expand_no_split_dollar_star && ifs_firstc == 0)
7791 # endif
7792 temp = string_list_dollar_star (list);
7793 else
7794 {
7795 temp = string_list_dollar_at (list, quoted);
7796 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
7797 tflag |= W_SPLITSPACE;
7798 }
7799
7800 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
7801 *contains_dollar_at = 1;
7802 }
7803
7804 dispose_words (list);
7805 break;
7806
7807 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
7808 means that we have to turn quoting off after we split into
7809 the individually quoted arguments so that the final split
7810 on the first character of $IFS is still done. */
7811 case '@': /* `$@' */
7812 list = list_rest_of_args ();
7813
7814 #if 0
7815 /* According to austin-group posix proposal by Geoff Clare in
7816 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7817
7818 "The shell shall write a message to standard error and
7819 immediately exit when it tries to expand an unset parameter
7820 other than the '@' and '*' special parameters."
7821 */
7822
7823 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7824 {
7825 uerror[0] = '$';
7826 uerror[1] = '@';
7827 uerror[2] = '\0';
7828 last_command_exit_value = EXECUTION_FAILURE;
7829 err_unboundvar (uerror);
7830 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7831 }
7832 #endif
7833
7834 /* We want to flag the fact that we saw this. We can't turn
7835 off quoting entirely, because other characters in the
7836 string might need it (consider "\"$@\""), but we need some
7837 way to signal that the final split on the first character
7838 of $IFS should be done, even though QUOTED is 1. */
7839 /* XXX - should this test include Q_PATQUOTE? */
7840 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7841 *quoted_dollar_at_p = 1;
7842 if (contains_dollar_at)
7843 *contains_dollar_at = 1;
7844
7845 /* We want to separate the positional parameters with the first
7846 character of $IFS in case $IFS is something other than a space.
7847 We also want to make sure that splitting is done no matter what --
7848 according to POSIX.2, this expands to a list of the positional
7849 parameters no matter what IFS is set to. */
7850 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted);
7851
7852 tflag |= W_DOLLARAT;
7853 dispose_words (list);
7854 break;
7855
7856 case LBRACE:
7857 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
7858 quoted_dollar_at_p,
7859 contains_dollar_at);
7860
7861 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
7862 return (tdesc);
7863 temp = tdesc ? tdesc->word : (char *)0;
7864
7865 /* XXX */
7866 /* Quoted nulls should be removed if there is anything else
7867 in the string. */
7868 /* Note that we saw the quoted null so we can add one back at
7869 the end of this function if there are no other characters
7870 in the string, discard TEMP, and go on. The exception to
7871 this is when we have "${@}" and $1 is '', since $@ needs
7872 special handling. */
7873 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
7874 {
7875 if (had_quoted_null_p)
7876 *had_quoted_null_p = 1;
7877 if (*quoted_dollar_at_p == 0)
7878 {
7879 free (temp);
7880 tdesc->word = temp = (char *)NULL;
7881 }
7882
7883 }
7884
7885 ret = tdesc;
7886 goto return0;
7887
7888 /* Do command or arithmetic substitution. */
7889 case LPAREN:
7890 /* We have to extract the contents of this paren substitution. */
7891 t_index = zindex + 1;
7892 temp = extract_command_subst (string, &t_index, 0);
7893 zindex = t_index;
7894
7895 /* For Posix.2-style `$(( ))' arithmetic substitution,
7896 extract the expression and pass it to the evaluator. */
7897 if (temp && *temp == LPAREN)
7898 {
7899 char *temp2;
7900 temp1 = temp + 1;
7901 temp2 = savestring (temp1);
7902 t_index = strlen (temp2) - 1;
7903
7904 if (temp2[t_index] != RPAREN)
7905 {
7906 free (temp2);
7907 goto comsub;
7908 }
7909
7910 /* Cut off ending `)' */
7911 temp2[t_index] = '\0';
7912
7913 if (chk_arithsub (temp2, t_index) == 0)
7914 {
7915 free (temp2);
7916 #if 0
7917 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
7918 #endif
7919 goto comsub;
7920 }
7921
7922 /* Expand variables found inside the expression. */
7923 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7924 free (temp2);
7925
7926 arithsub:
7927 /* No error messages. */
7928 this_command_name = (char *)NULL;
7929 number = evalexp (temp1, &expok);
7930 free (temp);
7931 free (temp1);
7932 if (expok == 0)
7933 {
7934 if (interactive_shell == 0 && posixly_correct)
7935 {
7936 last_command_exit_value = EXECUTION_FAILURE;
7937 return (&expand_wdesc_fatal);
7938 }
7939 else
7940 return (&expand_wdesc_error);
7941 }
7942 temp = itos (number);
7943 break;
7944 }
7945
7946 comsub:
7947 if (pflags & PF_NOCOMSUB)
7948 /* we need zindex+1 because string[zindex] == RPAREN */
7949 temp1 = substring (string, *sindex, zindex+1);
7950 else
7951 {
7952 tdesc = command_substitute (temp, quoted);
7953 temp1 = tdesc ? tdesc->word : (char *)NULL;
7954 if (tdesc)
7955 dispose_word_desc (tdesc);
7956 }
7957 FREE (temp);
7958 temp = temp1;
7959 break;
7960
7961 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
7962 away in a future bash release. */
7963 case '[':
7964 /* Extract the contents of this arithmetic substitution. */
7965 t_index = zindex + 1;
7966 temp = extract_arithmetic_subst (string, &t_index);
7967 zindex = t_index;
7968 if (temp == 0)
7969 {
7970 temp = savestring (string);
7971 if (expanded_something)
7972 *expanded_something = 0;
7973 goto return0;
7974 }
7975
7976 /* Do initial variable expansion. */
7977 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES);
7978
7979 goto arithsub;
7980
7981 default:
7982 /* Find the variable in VARIABLE_LIST. */
7983 temp = (char *)NULL;
7984
7985 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
7986 ;
7987 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
7988
7989 /* If this isn't a variable name, then just output the `$'. */
7990 if (temp1 == 0 || *temp1 == '\0')
7991 {
7992 FREE (temp1);
7993 temp = (char *)xmalloc (2);
7994 temp[0] = '$';
7995 temp[1] = '\0';
7996 if (expanded_something)
7997 *expanded_something = 0;
7998 goto return0;
7999 }
8000
8001 /* If the variable exists, return its value cell. */
8002 var = find_variable (temp1);
8003
8004 if (var && invisible_p (var) == 0 && var_isset (var))
8005 {
8006 #if defined (ARRAY_VARS)
8007 if (assoc_p (var) || array_p (var))
8008 {
8009 temp = array_p (var) ? array_reference (array_cell (var), 0)
8010 : assoc_reference (assoc_cell (var), "0");
8011 if (temp)
8012 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8013 ? quote_string (temp)
8014 : quote_escapes (temp);
8015 else if (unbound_vars_is_error)
8016 goto unbound_variable;
8017 }
8018 else
8019 #endif
8020 {
8021 temp = value_cell (var);
8022
8023 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8024 ? quote_string (temp)
8025 : quote_escapes (temp);
8026 }
8027
8028 free (temp1);
8029
8030 goto return0;
8031 }
8032 else if (var = find_variable_last_nameref (temp1))
8033 {
8034 temp = nameref_cell (var);
8035 #if defined (ARRAY_VARS)
8036 if (temp && *temp && valid_array_reference (temp))
8037 {
8038 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8039 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8040 return (tdesc);
8041 ret = tdesc;
8042 goto return0;
8043 }
8044 else
8045 #endif
8046 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8047 if (temp && *temp && legal_identifier (temp) == 0)
8048 {
8049 last_command_exit_value = EXECUTION_FAILURE;
8050 report_error (_("%s: invalid variable name for name reference"), temp);
8051 return (&expand_wdesc_error); /* XXX */
8052 }
8053 else
8054 temp = (char *)NULL;
8055 }
8056
8057 temp = (char *)NULL;
8058
8059 unbound_variable:
8060 if (unbound_vars_is_error)
8061 {
8062 last_command_exit_value = EXECUTION_FAILURE;
8063 err_unboundvar (temp1);
8064 }
8065 else
8066 {
8067 free (temp1);
8068 goto return0;
8069 }
8070
8071 free (temp1);
8072 last_command_exit_value = EXECUTION_FAILURE;
8073 return ((unbound_vars_is_error && interactive_shell == 0)
8074 ? &expand_wdesc_fatal
8075 : &expand_wdesc_error);
8076 }
8077
8078 if (string[zindex])
8079 zindex++;
8080
8081 return0:
8082 *sindex = zindex;
8083
8084 if (ret == 0)
8085 {
8086 ret = alloc_word_desc ();
8087 ret->flags = tflag; /* XXX */
8088 ret->word = temp;
8089 }
8090 return ret;
8091 }
8092
8093 /* Make a word list which is the result of parameter and variable
8094 expansion, command substitution, arithmetic substitution, and
8095 quote removal of WORD. Return a pointer to a WORD_LIST which is
8096 the result of the expansion. If WORD contains a null word, the
8097 word list returned is also null.
8098
8099 QUOTED contains flag values defined in shell.h.
8100
8101 ISEXP is used to tell expand_word_internal that the word should be
8102 treated as the result of an expansion. This has implications for
8103 how IFS characters in the word are treated.
8104
8105 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
8106 they point to an integer value which receives information about expansion.
8107 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
8108 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
8109 else zero.
8110
8111 This only does word splitting in the case of $@ expansion. In that
8112 case, we split on ' '. */
8113
8114 /* Values for the local variable quoted_state. */
8115 #define UNQUOTED 0
8116 #define PARTIALLY_QUOTED 1
8117 #define WHOLLY_QUOTED 2
8118
8119 static WORD_LIST *
8120 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
8121 WORD_DESC *word;
8122 int quoted, isexp;
8123 int *contains_dollar_at;
8124 int *expanded_something;
8125 {
8126 WORD_LIST *list;
8127 WORD_DESC *tword;
8128
8129 /* The intermediate string that we build while expanding. */
8130 char *istring;
8131
8132 /* The current size of the above object. */
8133 int istring_size;
8134
8135 /* Index into ISTRING. */
8136 int istring_index;
8137
8138 /* Temporary string storage. */
8139 char *temp, *temp1;
8140
8141 /* The text of WORD. */
8142 register char *string;
8143
8144 /* The size of STRING. */
8145 size_t string_size;
8146
8147 /* The index into STRING. */
8148 int sindex;
8149
8150 /* This gets 1 if we see a $@ while quoted. */
8151 int quoted_dollar_at;
8152
8153 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
8154 whether WORD contains no quoting characters, a partially quoted
8155 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
8156 int quoted_state;
8157
8158 /* State flags */
8159 int had_quoted_null;
8160 int has_dollar_at, temp_has_dollar_at;
8161 int split_on_spaces;
8162 int tflag;
8163 int pflags; /* flags passed to param_expand */
8164
8165 int assignoff; /* If assignment, offset of `=' */
8166
8167 register unsigned char c; /* Current character. */
8168 int t_index; /* For calls to string_extract_xxx. */
8169
8170 char twochars[2];
8171
8172 DECLARE_MBSTATE;
8173
8174 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
8175 istring[istring_index = 0] = '\0';
8176 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
8177 split_on_spaces = 0;
8178 quoted_state = UNQUOTED;
8179
8180 string = word->word;
8181 if (string == 0)
8182 goto finished_with_string;
8183 /* Don't need the string length for the SADD... and COPY_ macros unless
8184 multibyte characters are possible. */
8185 string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1;
8186
8187 if (contains_dollar_at)
8188 *contains_dollar_at = 0;
8189
8190 assignoff = -1;
8191
8192 /* Begin the expansion. */
8193
8194 for (sindex = 0; ;)
8195 {
8196 c = string[sindex];
8197
8198 /* Case on top-level character. */
8199 switch (c)
8200 {
8201 case '\0':
8202 goto finished_with_string;
8203
8204 case CTLESC:
8205 sindex++;
8206 #if HANDLE_MULTIBYTE
8207 if (MB_CUR_MAX > 1 && string[sindex])
8208 {
8209 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8210 }
8211 else
8212 #endif
8213 {
8214 temp = (char *)xmalloc (3);
8215 temp[0] = CTLESC;
8216 temp[1] = c = string[sindex];
8217 temp[2] = '\0';
8218 }
8219
8220 dollar_add_string:
8221 if (string[sindex])
8222 sindex++;
8223
8224 add_string:
8225 if (temp)
8226 {
8227 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
8228 temp = (char *)0;
8229 }
8230
8231 break;
8232
8233 #if defined (PROCESS_SUBSTITUTION)
8234 /* Process substitution. */
8235 case '<':
8236 case '>':
8237 {
8238 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
8239 {
8240 sindex--; /* add_character: label increments sindex */
8241 goto add_character;
8242 }
8243 else
8244 t_index = sindex + 1; /* skip past both '<' and LPAREN */
8245
8246 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index); /*))*/
8247 sindex = t_index;
8248
8249 /* If the process substitution specification is `<()', we want to
8250 open the pipe for writing in the child and produce output; if
8251 it is `>()', we want to open the pipe for reading in the child
8252 and consume input. */
8253 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
8254
8255 FREE (temp1);
8256
8257 goto dollar_add_string;
8258 }
8259 #endif /* PROCESS_SUBSTITUTION */
8260
8261 case '=':
8262 /* Posix.2 section 3.6.1 says that tildes following `=' in words
8263 which are not assignment statements are not expanded. If the
8264 shell isn't in posix mode, though, we perform tilde expansion
8265 on `likely candidate' unquoted assignment statements (flags
8266 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
8267 contains an unquoted :~ or =~. Something to think about: we
8268 now have a flag that says to perform tilde expansion on arguments
8269 to `assignment builtins' like declare and export that look like
8270 assignment statements. We now do tilde expansion on such words
8271 even in POSIX mode. */
8272 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
8273 {
8274 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8275 goto add_ifs_character;
8276 else
8277 goto add_character;
8278 }
8279 /* If we're not in posix mode or forcing assignment-statement tilde
8280 expansion, note where the `=' appears in the word and prepare to
8281 do tilde expansion following the first `='. */
8282 if ((word->flags & W_ASSIGNMENT) &&
8283 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8284 assignoff == -1 && sindex > 0)
8285 assignoff = sindex;
8286 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
8287 word->flags |= W_ITILDE;
8288 #if 0
8289 else if ((word->flags & W_ASSIGNMENT) &&
8290 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8291 string[sindex+1] == '~')
8292 word->flags |= W_ITILDE;
8293 #endif
8294 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8295 goto add_ifs_character;
8296 else
8297 goto add_character;
8298
8299 case ':':
8300 if (word->flags & W_NOTILDE)
8301 {
8302 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8303 goto add_ifs_character;
8304 else
8305 goto add_character;
8306 }
8307
8308 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
8309 string[sindex+1] == '~')
8310 word->flags |= W_ITILDE;
8311
8312 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8313 goto add_ifs_character;
8314 else
8315 goto add_character;
8316
8317 case '~':
8318 /* If the word isn't supposed to be tilde expanded, or we're not
8319 at the start of a word or after an unquoted : or = in an
8320 assignment statement, we don't do tilde expansion. */
8321 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
8322 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
8323 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8324 {
8325 word->flags &= ~W_ITILDE;
8326 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
8327 goto add_ifs_character;
8328 else
8329 goto add_character;
8330 }
8331
8332 if (word->flags & W_ASSIGNRHS)
8333 tflag = 2;
8334 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
8335 tflag = 1;
8336 else
8337 tflag = 0;
8338
8339 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
8340
8341 word->flags &= ~W_ITILDE;
8342
8343 if (temp && *temp && t_index > 0)
8344 {
8345 temp1 = bash_tilde_expand (temp, tflag);
8346 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
8347 {
8348 FREE (temp);
8349 FREE (temp1);
8350 goto add_character; /* tilde expansion failed */
8351 }
8352 free (temp);
8353 temp = temp1;
8354 sindex += t_index;
8355 goto add_quoted_string; /* XXX was add_string */
8356 }
8357 else
8358 {
8359 FREE (temp);
8360 goto add_character;
8361 }
8362
8363 case '$':
8364 if (expanded_something)
8365 *expanded_something = 1;
8366
8367 temp_has_dollar_at = 0;
8368 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
8369 if (word->flags & W_NOSPLIT2)
8370 pflags |= PF_NOSPLIT2;
8371 if (word->flags & W_ASSIGNRHS)
8372 pflags |= PF_ASSIGNRHS;
8373 tword = param_expand (string, &sindex, quoted, expanded_something,
8374 &temp_has_dollar_at, &quoted_dollar_at,
8375 &had_quoted_null, pflags);
8376 has_dollar_at += temp_has_dollar_at;
8377 split_on_spaces += (tword->flags & W_SPLITSPACE);
8378
8379 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
8380 {
8381 free (string);
8382 free (istring);
8383 return ((tword == &expand_wdesc_error) ? &expand_word_error
8384 : &expand_word_fatal);
8385 }
8386 if (contains_dollar_at && has_dollar_at)
8387 *contains_dollar_at = 1;
8388
8389 if (tword && (tword->flags & W_HASQUOTEDNULL))
8390 had_quoted_null = 1;
8391
8392 temp = tword ? tword->word : (char *)NULL;
8393 dispose_word_desc (tword);
8394
8395 /* Kill quoted nulls; we will add them back at the end of
8396 expand_word_internal if nothing else in the string */
8397 if (had_quoted_null && temp && QUOTED_NULL (temp))
8398 {
8399 FREE (temp);
8400 temp = (char *)NULL;
8401 }
8402
8403 goto add_string;
8404 break;
8405
8406 case '`': /* Backquoted command substitution. */
8407 {
8408 t_index = sindex++;
8409
8410 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
8411 /* The test of sindex against t_index is to allow bare instances of
8412 ` to pass through, for backwards compatibility. */
8413 if (temp == &extract_string_error || temp == &extract_string_fatal)
8414 {
8415 if (sindex - 1 == t_index)
8416 {
8417 sindex = t_index;
8418 goto add_character;
8419 }
8420 last_command_exit_value = EXECUTION_FAILURE;
8421 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
8422 free (string);
8423 free (istring);
8424 return ((temp == &extract_string_error) ? &expand_word_error
8425 : &expand_word_fatal);
8426 }
8427
8428 if (expanded_something)
8429 *expanded_something = 1;
8430
8431 if (word->flags & W_NOCOMSUB)
8432 /* sindex + 1 because string[sindex] == '`' */
8433 temp1 = substring (string, t_index, sindex + 1);
8434 else
8435 {
8436 de_backslash (temp);
8437 tword = command_substitute (temp, quoted);
8438 temp1 = tword ? tword->word : (char *)NULL;
8439 if (tword)
8440 dispose_word_desc (tword);
8441 }
8442 FREE (temp);
8443 temp = temp1;
8444 goto dollar_add_string;
8445 }
8446
8447 case '\\':
8448 if (string[sindex + 1] == '\n')
8449 {
8450 sindex += 2;
8451 continue;
8452 }
8453
8454 c = string[++sindex];
8455
8456 if (quoted & Q_HERE_DOCUMENT)
8457 tflag = CBSHDOC;
8458 else if (quoted & Q_DOUBLE_QUOTES)
8459 tflag = CBSDQUOTE;
8460 else
8461 tflag = 0;
8462
8463 /* From Posix discussion on austin-group list: Backslash escaping
8464 a } in ${...} is removed. Issue 0000221 */
8465 if ((quoted & Q_DOLBRACE) && c == RBRACE)
8466 {
8467 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8468 }
8469 /* This is the fix for " $@\ " */
8470 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
8471 {
8472 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8473 DEFAULT_ARRAY_SIZE);
8474 istring[istring_index++] = CTLESC;
8475 istring[istring_index++] = '\\';
8476 istring[istring_index] = '\0';
8477
8478 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8479 }
8480 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
8481 {
8482 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
8483 }
8484 else if (c == 0)
8485 {
8486 c = CTLNUL;
8487 sindex--; /* add_character: label increments sindex */
8488 goto add_character;
8489 }
8490 else
8491 {
8492 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8493 }
8494
8495 sindex++;
8496 add_twochars:
8497 /* BEFORE jumping here, we need to increment sindex if appropriate */
8498 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8499 DEFAULT_ARRAY_SIZE);
8500 istring[istring_index++] = twochars[0];
8501 istring[istring_index++] = twochars[1];
8502 istring[istring_index] = '\0';
8503
8504 break;
8505
8506 case '"':
8507 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8508 goto add_character;
8509
8510 t_index = ++sindex;
8511 temp = string_extract_double_quoted (string, &sindex, 0);
8512
8513 /* If the quotes surrounded the entire string, then the
8514 whole word was quoted. */
8515 quoted_state = (t_index == 1 && string[sindex] == '\0')
8516 ? WHOLLY_QUOTED
8517 : PARTIALLY_QUOTED;
8518
8519 if (temp && *temp)
8520 {
8521 tword = alloc_word_desc ();
8522 tword->word = temp;
8523
8524 temp = (char *)NULL;
8525
8526 temp_has_dollar_at = 0; /* XXX */
8527 /* Need to get W_HASQUOTEDNULL flag through this function. */
8528 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
8529 has_dollar_at += temp_has_dollar_at;
8530
8531 if (list == &expand_word_error || list == &expand_word_fatal)
8532 {
8533 free (istring);
8534 free (string);
8535 /* expand_word_internal has already freed temp_word->word
8536 for us because of the way it prints error messages. */
8537 tword->word = (char *)NULL;
8538 dispose_word (tword);
8539 return list;
8540 }
8541
8542 dispose_word (tword);
8543
8544 /* "$@" (a double-quoted dollar-at) expands into nothing,
8545 not even a NULL word, when there are no positional
8546 parameters. */
8547 if (list == 0 && has_dollar_at)
8548 {
8549 quoted_dollar_at++;
8550 break;
8551 }
8552
8553 /* If we get "$@", we know we have expanded something, so we
8554 need to remember it for the final split on $IFS. This is
8555 a special case; it's the only case where a quoted string
8556 can expand into more than one word. It's going to come back
8557 from the above call to expand_word_internal as a list with
8558 a single word, in which all characters are quoted and
8559 separated by blanks. What we want to do is to turn it back
8560 into a list for the next piece of code. */
8561 if (list)
8562 dequote_list (list);
8563
8564 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
8565 had_quoted_null = 1; /* XXX */
8566
8567 if (has_dollar_at)
8568 {
8569 quoted_dollar_at++;
8570 if (contains_dollar_at)
8571 *contains_dollar_at = 1;
8572 if (expanded_something)
8573 *expanded_something = 1;
8574 }
8575 }
8576 else
8577 {
8578 /* What we have is "". This is a minor optimization. */
8579 FREE (temp);
8580 list = (WORD_LIST *)NULL;
8581 }
8582
8583 /* The code above *might* return a list (consider the case of "$@",
8584 where it returns "$1", "$2", etc.). We can't throw away the
8585 rest of the list, and we have to make sure each word gets added
8586 as quoted. We test on tresult->next: if it is non-NULL, we
8587 quote the whole list, save it to a string with string_list, and
8588 add that string. We don't need to quote the results of this
8589 (and it would be wrong, since that would quote the separators
8590 as well), so we go directly to add_string. */
8591 if (list)
8592 {
8593 if (list->next)
8594 {
8595 /* Testing quoted_dollar_at makes sure that "$@" is
8596 split correctly when $IFS does not contain a space. */
8597 temp = quoted_dollar_at
8598 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES)
8599 : string_list (quote_list (list));
8600 dispose_words (list);
8601 goto add_string;
8602 }
8603 else
8604 {
8605 temp = savestring (list->word->word);
8606 tflag = list->word->flags;
8607 dispose_words (list);
8608
8609 /* If the string is not a quoted null string, we want
8610 to remove any embedded unquoted CTLNUL characters.
8611 We do not want to turn quoted null strings back into
8612 the empty string, though. We do this because we
8613 want to remove any quoted nulls from expansions that
8614 contain other characters. For example, if we have
8615 x"$*"y or "x$*y" and there are no positional parameters,
8616 the $* should expand into nothing. */
8617 /* We use the W_HASQUOTEDNULL flag to differentiate the
8618 cases: a quoted null character as above and when
8619 CTLNUL is contained in the (non-null) expansion
8620 of some variable. We use the had_quoted_null flag to
8621 pass the value through this function to its caller. */
8622 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
8623 remove_quoted_nulls (temp); /* XXX */
8624 }
8625 }
8626 else
8627 temp = (char *)NULL;
8628
8629 /* We do not want to add quoted nulls to strings that are only
8630 partially quoted; we can throw them away. The exception to
8631 this is when we are going to be performing word splitting,
8632 since we have to preserve a null argument if the next character
8633 will cause word splitting. */
8634 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
8635 continue;
8636
8637 add_quoted_string:
8638
8639 if (temp)
8640 {
8641 temp1 = temp;
8642 temp = quote_string (temp);
8643 free (temp1);
8644 goto add_string;
8645 }
8646 else
8647 {
8648 /* Add NULL arg. */
8649 c = CTLNUL;
8650 sindex--; /* add_character: label increments sindex */
8651 goto add_character;
8652 }
8653
8654 /* break; */
8655
8656 case '\'':
8657 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8658 goto add_character;
8659
8660 t_index = ++sindex;
8661 temp = string_extract_single_quoted (string, &sindex);
8662
8663 /* If the entire STRING was surrounded by single quotes,
8664 then the string is wholly quoted. */
8665 quoted_state = (t_index == 1 && string[sindex] == '\0')
8666 ? WHOLLY_QUOTED
8667 : PARTIALLY_QUOTED;
8668
8669 /* If all we had was '', it is a null expansion. */
8670 if (*temp == '\0')
8671 {
8672 free (temp);
8673 temp = (char *)NULL;
8674 }
8675 else
8676 remove_quoted_escapes (temp); /* ??? */
8677
8678 /* We do not want to add quoted nulls to strings that are only
8679 partially quoted; such nulls are discarded. */
8680 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED))
8681 continue;
8682
8683 /* If we have a quoted null expansion, add a quoted NULL to istring. */
8684 if (temp == 0)
8685 {
8686 c = CTLNUL;
8687 sindex--; /* add_character: label increments sindex */
8688 goto add_character;
8689 }
8690 else
8691 goto add_quoted_string;
8692
8693 /* break; */
8694
8695 default:
8696 /* This is the fix for " $@ " */
8697 add_ifs_character:
8698 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c)))
8699 {
8700 if (string[sindex]) /* from old goto dollar_add_string */
8701 sindex++;
8702 if (c == 0)
8703 {
8704 c = CTLNUL;
8705 goto add_character;
8706 }
8707 else
8708 {
8709 #if HANDLE_MULTIBYTE
8710 if (MB_CUR_MAX > 1)
8711 sindex--;
8712
8713 if (MB_CUR_MAX > 1)
8714 {
8715 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8716 }
8717 else
8718 #endif
8719 {
8720 twochars[0] = CTLESC;
8721 twochars[1] = c;
8722 goto add_twochars;
8723 }
8724 }
8725 }
8726
8727 SADD_MBCHAR (temp, string, sindex, string_size);
8728
8729 add_character:
8730 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
8731 DEFAULT_ARRAY_SIZE);
8732 istring[istring_index++] = c;
8733 istring[istring_index] = '\0';
8734
8735 /* Next character. */
8736 sindex++;
8737 }
8738 }
8739
8740 finished_with_string:
8741 /* OK, we're ready to return. If we have a quoted string, and
8742 quoted_dollar_at is not set, we do no splitting at all; otherwise
8743 we split on ' '. The routines that call this will handle what to
8744 do if nothing has been expanded. */
8745
8746 /* Partially and wholly quoted strings which expand to the empty
8747 string are retained as an empty arguments. Unquoted strings
8748 which expand to the empty string are discarded. The single
8749 exception is the case of expanding "$@" when there are no
8750 positional parameters. In that case, we discard the expansion. */
8751
8752 /* Because of how the code that handles "" and '' in partially
8753 quoted strings works, we need to make ISTRING into a QUOTED_NULL
8754 if we saw quoting characters, but the expansion was empty.
8755 "" and '' are tossed away before we get to this point when
8756 processing partially quoted strings. This makes "" and $xxx""
8757 equivalent when xxx is unset. We also look to see whether we
8758 saw a quoted null from a ${} expansion and add one back if we
8759 need to. */
8760
8761 /* If we expand to nothing and there were no single or double quotes
8762 in the word, we throw it away. Otherwise, we return a NULL word.
8763 The single exception is for $@ surrounded by double quotes when
8764 there are no positional parameters. In that case, we also throw
8765 the word away. */
8766
8767 if (*istring == '\0')
8768 {
8769 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
8770 {
8771 istring[0] = CTLNUL;
8772 istring[1] = '\0';
8773 tword = make_bare_word (istring);
8774 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8775 list = make_word_list (tword, (WORD_LIST *)NULL);
8776 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8777 tword->flags |= W_QUOTED;
8778 }
8779 /* According to sh, ksh, and Posix.2, if a word expands into nothing
8780 and a double-quoted "$@" appears anywhere in it, then the entire
8781 word is removed. */
8782 else if (quoted_state == UNQUOTED || quoted_dollar_at)
8783 list = (WORD_LIST *)NULL;
8784 #if 0
8785 else
8786 {
8787 tword = make_bare_word (istring);
8788 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8789 tword->flags |= W_QUOTED;
8790 list = make_word_list (tword, (WORD_LIST *)NULL);
8791 }
8792 #else
8793 else
8794 list = (WORD_LIST *)NULL;
8795 #endif
8796 }
8797 else if (word->flags & W_NOSPLIT)
8798 {
8799 tword = make_bare_word (istring);
8800 if (word->flags & W_ASSIGNMENT)
8801 tword->flags |= W_ASSIGNMENT; /* XXX */
8802 if (word->flags & W_COMPASSIGN)
8803 tword->flags |= W_COMPASSIGN; /* XXX */
8804 if (word->flags & W_NOGLOB)
8805 tword->flags |= W_NOGLOB; /* XXX */
8806 if (word->flags & W_NOBRACE)
8807 tword->flags |= W_NOBRACE; /* XXX */
8808 if (word->flags & W_NOEXPAND)
8809 tword->flags |= W_NOEXPAND; /* XXX */
8810 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8811 tword->flags |= W_QUOTED;
8812 if (had_quoted_null && QUOTED_NULL (istring))
8813 tword->flags |= W_HASQUOTEDNULL;
8814 list = make_word_list (tword, (WORD_LIST *)NULL);
8815 }
8816 else
8817 {
8818 char *ifs_chars;
8819
8820 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
8821
8822 /* If we have $@, we need to split the results no matter what. If
8823 IFS is unset or NULL, string_list_dollar_at has separated the
8824 positional parameters with a space, so we split on space (we have
8825 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
8826 string_list_dollar_at has separated the positional parameters
8827 with the first character of $IFS, so we split on $IFS. If
8828 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
8829 unset or null, and we want to make sure that we split on spaces
8830 regardless of what else has happened to IFS since the expansion. */
8831 if (split_on_spaces)
8832 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
8833 else if (has_dollar_at && ifs_chars)
8834 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
8835 else
8836 {
8837 tword = make_bare_word (istring);
8838 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
8839 tword->flags |= W_QUOTED;
8840 if (word->flags & W_ASSIGNMENT)
8841 tword->flags |= W_ASSIGNMENT;
8842 if (word->flags & W_COMPASSIGN)
8843 tword->flags |= W_COMPASSIGN;
8844 if (word->flags & W_NOGLOB)
8845 tword->flags |= W_NOGLOB;
8846 if (word->flags & W_NOBRACE)
8847 tword->flags |= W_NOBRACE;
8848 if (word->flags & W_NOEXPAND)
8849 tword->flags |= W_NOEXPAND;
8850 if (had_quoted_null && QUOTED_NULL (istring))
8851 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8852 list = make_word_list (tword, (WORD_LIST *)NULL);
8853 }
8854 }
8855
8856 free (istring);
8857 return (list);
8858 }
8859
8860 /* **************************************************************** */
8861 /* */
8862 /* Functions for Quote Removal */
8863 /* */
8864 /* **************************************************************** */
8865
8866 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
8867 backslash quoting rules for within double quotes or a here document. */
8868 char *
8869 string_quote_removal (string, quoted)
8870 char *string;
8871 int quoted;
8872 {
8873 size_t slen;
8874 char *r, *result_string, *temp, *send;
8875 int sindex, tindex, dquote;
8876 unsigned char c;
8877 DECLARE_MBSTATE;
8878
8879 /* The result can be no longer than the original string. */
8880 slen = strlen (string);
8881 send = string + slen;
8882
8883 r = result_string = (char *)xmalloc (slen + 1);
8884
8885 for (dquote = sindex = 0; c = string[sindex];)
8886 {
8887 switch (c)
8888 {
8889 case '\\':
8890 c = string[++sindex];
8891 if (c == 0)
8892 {
8893 *r++ = '\\';
8894 break;
8895 }
8896 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
8897 *r++ = '\\';
8898 /* FALLTHROUGH */
8899
8900 default:
8901 SCOPY_CHAR_M (r, string, send, sindex);
8902 break;
8903
8904 case '\'':
8905 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
8906 {
8907 *r++ = c;
8908 sindex++;
8909 break;
8910 }
8911 tindex = sindex + 1;
8912 temp = string_extract_single_quoted (string, &tindex);
8913 if (temp)
8914 {
8915 strcpy (r, temp);
8916 r += strlen (r);
8917 free (temp);
8918 }
8919 sindex = tindex;
8920 break;
8921
8922 case '"':
8923 dquote = 1 - dquote;
8924 sindex++;
8925 break;
8926 }
8927 }
8928 *r = '\0';
8929 return (result_string);
8930 }
8931
8932 #if 0
8933 /* UNUSED */
8934 /* Perform quote removal on word WORD. This allocates and returns a new
8935 WORD_DESC *. */
8936 WORD_DESC *
8937 word_quote_removal (word, quoted)
8938 WORD_DESC *word;
8939 int quoted;
8940 {
8941 WORD_DESC *w;
8942 char *t;
8943
8944 t = string_quote_removal (word->word, quoted);
8945 w = alloc_word_desc ();
8946 w->word = t ? t : savestring ("");
8947 return (w);
8948 }
8949
8950 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
8951 the members of the list are treated as if they are surrounded by
8952 double quotes. Return a new list, or NULL if LIST is NULL. */
8953 WORD_LIST *
8954 word_list_quote_removal (list, quoted)
8955 WORD_LIST *list;
8956 int quoted;
8957 {
8958 WORD_LIST *result, *t, *tresult, *e;
8959
8960 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
8961 {
8962 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
8963 #if 0
8964 result = (WORD_LIST *) list_append (result, tresult);
8965 #else
8966 if (result == 0)
8967 result = e = tresult;
8968 else
8969 {
8970 e->next = tresult;
8971 while (e->next)
8972 e = e->next;
8973 }
8974 #endif
8975 }
8976 return (result);
8977 }
8978 #endif
8979
8980 /*******************************************
8981 * *
8982 * Functions to perform word splitting *
8983 * *
8984 *******************************************/
8985
8986 void
8987 setifs (v)
8988 SHELL_VAR *v;
8989 {
8990 char *t;
8991 unsigned char uc;
8992
8993 ifs_var = v;
8994 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
8995
8996 ifs_is_set = ifs_var != 0;
8997 ifs_is_null = ifs_is_set && (*ifs_value == 0);
8998
8999 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9000 handle multibyte chars in IFS */
9001 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9002 for (t = ifs_value ; t && *t; t++)
9003 {
9004 uc = *t;
9005 ifs_cmap[uc] = 1;
9006 }
9007
9008 #if defined (HANDLE_MULTIBYTE)
9009 if (ifs_value == 0)
9010 {
9011 ifs_firstc[0] = '\0';
9012 ifs_firstc_len = 1;
9013 }
9014 else
9015 {
9016 size_t ifs_len;
9017 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
9018 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
9019 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
9020 {
9021 ifs_firstc[0] = ifs_value[0];
9022 ifs_firstc[1] = '\0';
9023 ifs_firstc_len = 1;
9024 }
9025 else
9026 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
9027 }
9028 #else
9029 ifs_firstc = ifs_value ? *ifs_value : 0;
9030 #endif
9031 }
9032
9033 char *
9034 getifs ()
9035 {
9036 return ifs_value;
9037 }
9038
9039 /* This splits a single word into a WORD LIST on $IFS, but only if the word
9040 is not quoted. list_string () performs quote removal for us, even if we
9041 don't do any splitting. */
9042 WORD_LIST *
9043 word_split (w, ifs_chars)
9044 WORD_DESC *w;
9045 char *ifs_chars;
9046 {
9047 WORD_LIST *result;
9048
9049 if (w)
9050 {
9051 char *xifs;
9052
9053 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
9054 result = list_string (w->word, xifs, w->flags & W_QUOTED);
9055 }
9056 else
9057 result = (WORD_LIST *)NULL;
9058
9059 return (result);
9060 }
9061
9062 /* Perform word splitting on LIST and return the RESULT. It is possible
9063 to return (WORD_LIST *)NULL. */
9064 static WORD_LIST *
9065 word_list_split (list)
9066 WORD_LIST *list;
9067 {
9068 WORD_LIST *result, *t, *tresult, *e;
9069
9070 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
9071 {
9072 tresult = word_split (t->word, ifs_value);
9073 if (result == 0)
9074 result = e = tresult;
9075 else
9076 {
9077 e->next = tresult;
9078 while (e->next)
9079 e = e->next;
9080 }
9081 }
9082 return (result);
9083 }
9084
9085 /**************************************************
9086 * *
9087 * Functions to expand an entire WORD_LIST *
9088 * *
9089 **************************************************/
9090
9091 /* Do any word-expansion-specific cleanup and jump to top_level */
9092 static void
9093 exp_jump_to_top_level (v)
9094 int v;
9095 {
9096 set_pipestatus_from_exit (last_command_exit_value);
9097
9098 /* Cleanup code goes here. */
9099 expand_no_split_dollar_star = 0; /* XXX */
9100 expanding_redir = 0;
9101 assigning_in_environment = 0;
9102
9103 if (parse_and_execute_level == 0)
9104 top_level_cleanup (); /* from sig.c */
9105
9106 jump_to_top_level (v);
9107 }
9108
9109 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
9110 ELIST, and set ELIST to the new list. */
9111 #define PREPEND_LIST(nlist, elist) \
9112 do { nlist->next = elist; elist = nlist; } while (0)
9113
9114 /* Separate out any initial variable assignments from TLIST. If set -k has
9115 been executed, remove all assignment statements from TLIST. Initial
9116 variable assignments and other environment assignments are placed
9117 on SUBST_ASSIGN_VARLIST. */
9118 static WORD_LIST *
9119 separate_out_assignments (tlist)
9120 WORD_LIST *tlist;
9121 {
9122 register WORD_LIST *vp, *lp;
9123
9124 if (tlist == 0)
9125 return ((WORD_LIST *)NULL);
9126
9127 if (subst_assign_varlist)
9128 dispose_words (subst_assign_varlist); /* Clean up after previous error */
9129
9130 subst_assign_varlist = (WORD_LIST *)NULL;
9131 vp = lp = tlist;
9132
9133 /* Separate out variable assignments at the start of the command.
9134 Loop invariant: vp->next == lp
9135 Loop postcondition:
9136 lp = list of words left after assignment statements skipped
9137 tlist = original list of words
9138 */
9139 while (lp && (lp->word->flags & W_ASSIGNMENT))
9140 {
9141 vp = lp;
9142 lp = lp->next;
9143 }
9144
9145 /* If lp != tlist, we have some initial assignment statements.
9146 We make SUBST_ASSIGN_VARLIST point to the list of assignment
9147 words and TLIST point to the remaining words. */
9148 if (lp != tlist)
9149 {
9150 subst_assign_varlist = tlist;
9151 /* ASSERT(vp->next == lp); */
9152 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
9153 tlist = lp; /* remainder of word list */
9154 }
9155
9156 /* vp == end of variable list */
9157 /* tlist == remainder of original word list without variable assignments */
9158 if (!tlist)
9159 /* All the words in tlist were assignment statements */
9160 return ((WORD_LIST *)NULL);
9161
9162 /* ASSERT(tlist != NULL); */
9163 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
9164
9165 /* If the -k option is in effect, we need to go through the remaining
9166 words, separate out the assignment words, and place them on
9167 SUBST_ASSIGN_VARLIST. */
9168 if (place_keywords_in_env)
9169 {
9170 WORD_LIST *tp; /* tp == running pointer into tlist */
9171
9172 tp = tlist;
9173 lp = tlist->next;
9174
9175 /* Loop Invariant: tp->next == lp */
9176 /* Loop postcondition: tlist == word list without assignment statements */
9177 while (lp)
9178 {
9179 if (lp->word->flags & W_ASSIGNMENT)
9180 {
9181 /* Found an assignment statement, add this word to end of
9182 subst_assign_varlist (vp). */
9183 if (!subst_assign_varlist)
9184 subst_assign_varlist = vp = lp;
9185 else
9186 {
9187 vp->next = lp;
9188 vp = lp;
9189 }
9190
9191 /* Remove the word pointed to by LP from TLIST. */
9192 tp->next = lp->next;
9193 /* ASSERT(vp == lp); */
9194 lp->next = (WORD_LIST *)NULL;
9195 lp = tp->next;
9196 }
9197 else
9198 {
9199 tp = lp;
9200 lp = lp->next;
9201 }
9202 }
9203 }
9204 return (tlist);
9205 }
9206
9207 #define WEXP_VARASSIGN 0x001
9208 #define WEXP_BRACEEXP 0x002
9209 #define WEXP_TILDEEXP 0x004
9210 #define WEXP_PARAMEXP 0x008
9211 #define WEXP_PATHEXP 0x010
9212
9213 /* All of the expansions, including variable assignments at the start of
9214 the list. */
9215 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
9216
9217 /* All of the expansions except variable assignments at the start of
9218 the list. */
9219 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
9220
9221 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
9222 expansion, command substitution, arithmetic expansion, word splitting, and
9223 quote removal. */
9224 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
9225
9226 /* Take the list of words in LIST and do the various substitutions. Return
9227 a new list of words which is the expanded list, and without things like
9228 variable assignments. */
9229
9230 WORD_LIST *
9231 expand_words (list)
9232 WORD_LIST *list;
9233 {
9234 return (expand_word_list_internal (list, WEXP_ALL));
9235 }
9236
9237 /* Same as expand_words (), but doesn't hack variable or environment
9238 variables. */
9239 WORD_LIST *
9240 expand_words_no_vars (list)
9241 WORD_LIST *list;
9242 {
9243 return (expand_word_list_internal (list, WEXP_NOVARS));
9244 }
9245
9246 WORD_LIST *
9247 expand_words_shellexp (list)
9248 WORD_LIST *list;
9249 {
9250 return (expand_word_list_internal (list, WEXP_SHELLEXP));
9251 }
9252
9253 static WORD_LIST *
9254 glob_expand_word_list (tlist, eflags)
9255 WORD_LIST *tlist;
9256 int eflags;
9257 {
9258 char **glob_array, *temp_string;
9259 register int glob_index;
9260 WORD_LIST *glob_list, *output_list, *disposables, *next;
9261 WORD_DESC *tword;
9262
9263 output_list = disposables = (WORD_LIST *)NULL;
9264 glob_array = (char **)NULL;
9265 while (tlist)
9266 {
9267 /* For each word, either globbing is attempted or the word is
9268 added to orig_list. If globbing succeeds, the results are
9269 added to orig_list and the word (tlist) is added to the list
9270 of disposable words. If globbing fails and failed glob
9271 expansions are left unchanged (the shell default), the
9272 original word is added to orig_list. If globbing fails and
9273 failed glob expansions are removed, the original word is
9274 added to the list of disposable words. orig_list ends up
9275 in reverse order and requires a call to REVERSE_LIST to
9276 be set right. After all words are examined, the disposable
9277 words are freed. */
9278 next = tlist->next;
9279
9280 /* If the word isn't an assignment and contains an unquoted
9281 pattern matching character, then glob it. */
9282 if ((tlist->word->flags & W_NOGLOB) == 0 &&
9283 unquoted_glob_pattern_p (tlist->word->word))
9284 {
9285 glob_array = shell_glob_filename (tlist->word->word);
9286
9287 /* Handle error cases.
9288 I don't think we should report errors like "No such file
9289 or directory". However, I would like to report errors
9290 like "Read failed". */
9291
9292 if (glob_array == 0 || GLOB_FAILED (glob_array))
9293 {
9294 glob_array = (char **)xmalloc (sizeof (char *));
9295 glob_array[0] = (char *)NULL;
9296 }
9297
9298 /* Dequote the current word in case we have to use it. */
9299 if (glob_array[0] == NULL)
9300 {
9301 temp_string = dequote_string (tlist->word->word);
9302 free (tlist->word->word);
9303 tlist->word->word = temp_string;
9304 }
9305
9306 /* Make the array into a word list. */
9307 glob_list = (WORD_LIST *)NULL;
9308 for (glob_index = 0; glob_array[glob_index]; glob_index++)
9309 {
9310 tword = make_bare_word (glob_array[glob_index]);
9311 glob_list = make_word_list (tword, glob_list);
9312 }
9313
9314 if (glob_list)
9315 {
9316 output_list = (WORD_LIST *)list_append (glob_list, output_list);
9317 PREPEND_LIST (tlist, disposables);
9318 }
9319 else if (fail_glob_expansion != 0)
9320 {
9321 last_command_exit_value = EXECUTION_FAILURE;
9322 report_error (_("no match: %s"), tlist->word->word);
9323 exp_jump_to_top_level (DISCARD);
9324 }
9325 else if (allow_null_glob_expansion == 0)
9326 {
9327 /* Failed glob expressions are left unchanged. */
9328 PREPEND_LIST (tlist, output_list);
9329 }
9330 else
9331 {
9332 /* Failed glob expressions are removed. */
9333 PREPEND_LIST (tlist, disposables);
9334 }
9335 }
9336 else
9337 {
9338 /* Dequote the string. */
9339 temp_string = dequote_string (tlist->word->word);
9340 free (tlist->word->word);
9341 tlist->word->word = temp_string;
9342 PREPEND_LIST (tlist, output_list);
9343 }
9344
9345 strvec_dispose (glob_array);
9346 glob_array = (char **)NULL;
9347
9348 tlist = next;
9349 }
9350
9351 if (disposables)
9352 dispose_words (disposables);
9353
9354 if (output_list)
9355 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9356
9357 return (output_list);
9358 }
9359
9360 #if defined (BRACE_EXPANSION)
9361 static WORD_LIST *
9362 brace_expand_word_list (tlist, eflags)
9363 WORD_LIST *tlist;
9364 int eflags;
9365 {
9366 register char **expansions;
9367 char *temp_string;
9368 WORD_LIST *disposables, *output_list, *next;
9369 WORD_DESC *w;
9370 int eindex;
9371
9372 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
9373 {
9374 next = tlist->next;
9375
9376 if (tlist->word->flags & W_NOBRACE)
9377 {
9378 /*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
9379 PREPEND_LIST (tlist, output_list);
9380 continue;
9381 }
9382
9383 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9384 {
9385 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
9386 PREPEND_LIST (tlist, output_list);
9387 continue;
9388 }
9389
9390 /* Only do brace expansion if the word has a brace character. If
9391 not, just add the word list element to BRACES and continue. In
9392 the common case, at least when running shell scripts, this will
9393 degenerate to a bunch of calls to `mbschr', and then what is
9394 basically a reversal of TLIST into BRACES, which is corrected
9395 by a call to REVERSE_LIST () on BRACES when the end of TLIST
9396 is reached. */
9397 if (mbschr (tlist->word->word, LBRACE))
9398 {
9399 expansions = brace_expand (tlist->word->word);
9400
9401 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
9402 {
9403 w = alloc_word_desc ();
9404 w->word = temp_string;
9405
9406 /* If brace expansion didn't change the word, preserve
9407 the flags. We may want to preserve the flags
9408 unconditionally someday -- XXX */
9409 if (STREQ (temp_string, tlist->word->word))
9410 w->flags = tlist->word->flags;
9411 else
9412 w = make_word_flags (w, temp_string);
9413
9414 output_list = make_word_list (w, output_list);
9415 }
9416 free (expansions);
9417
9418 /* Add TLIST to the list of words to be freed after brace
9419 expansion has been performed. */
9420 PREPEND_LIST (tlist, disposables);
9421 }
9422 else
9423 PREPEND_LIST (tlist, output_list);
9424 }
9425
9426 if (disposables)
9427 dispose_words (disposables);
9428
9429 if (output_list)
9430 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9431
9432 return (output_list);
9433 }
9434 #endif
9435
9436 #if defined (ARRAY_VARS)
9437 /* Take WORD, a compound associative array assignment, and internally run
9438 'declare -A w', where W is the variable name portion of WORD. */
9439 static int
9440 make_internal_declare (word, option)
9441 char *word;
9442 char *option;
9443 {
9444 int t;
9445 WORD_LIST *wl;
9446 WORD_DESC *w;
9447
9448 w = make_word (word);
9449
9450 t = assignment (w->word, 0);
9451 w->word[t] = '\0';
9452
9453 wl = make_word_list (w, (WORD_LIST *)NULL);
9454 wl = make_word_list (make_word (option), wl);
9455
9456 return (declare_builtin (wl));
9457 }
9458 #endif
9459
9460 static WORD_LIST *
9461 shell_expand_word_list (tlist, eflags)
9462 WORD_LIST *tlist;
9463 int eflags;
9464 {
9465 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list;
9466 int expanded_something, has_dollar_at;
9467 char *temp_string;
9468
9469 /* We do tilde expansion all the time. This is what 1003.2 says. */
9470 new_list = (WORD_LIST *)NULL;
9471 for (orig_list = tlist; tlist; tlist = next)
9472 {
9473 temp_string = tlist->word->word;
9474
9475 next = tlist->next;
9476
9477 #if defined (ARRAY_VARS)
9478 /* If this is a compound array assignment to a builtin that accepts
9479 such assignments (e.g., `declare'), take the assignment and perform
9480 it separately, handling the semantics of declarations inside shell
9481 functions. This avoids the double-evaluation of such arguments,
9482 because `declare' does some evaluation of compound assignments on
9483 its own. */
9484 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9485 {
9486 int t;
9487 char opts[8], opti;
9488
9489 opti = 0;
9490 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
9491 opts[opti++] = '-';
9492
9493 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
9494 {
9495 opts[opti++] = 'g';
9496 opts[opti++] = 'A';
9497 }
9498 else if (tlist->word->flags & W_ASSIGNASSOC)
9499 opts[opti++] = 'A';
9500 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
9501 {
9502 opts[opti++] = 'g';
9503 opts[opti++] = 'a';
9504 }
9505 else if (tlist->word->flags & W_ASSIGNARRAY)
9506 opts[opti++] = 'a';
9507 else if (tlist->word->flags & W_ASSNGLOBAL)
9508 opts[opti++] = 'g';
9509
9510 #if 0
9511 /* If we have special handling note the integer attribute */
9512 if (opti > 0 && (tlist->word->flags & W_ASSIGNINT))
9513 opts[opti++] = 'i';
9514 #endif
9515
9516 opts[opti] = '\0';
9517 if (opti > 0)
9518 make_internal_declare (tlist->word->word, opts);
9519
9520 t = do_word_assignment (tlist->word, 0);
9521 if (t == 0)
9522 {
9523 last_command_exit_value = EXECUTION_FAILURE;
9524 exp_jump_to_top_level (DISCARD);
9525 }
9526
9527 /* Now transform the word as ksh93 appears to do and go on */
9528 t = assignment (tlist->word->word, 0);
9529 tlist->word->word[t] = '\0';
9530 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
9531 }
9532 #endif
9533
9534 expanded_something = 0;
9535 expanded = expand_word_internal
9536 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
9537
9538 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
9539 {
9540 /* By convention, each time this error is returned,
9541 tlist->word->word has already been freed. */
9542 tlist->word->word = (char *)NULL;
9543
9544 /* Dispose our copy of the original list. */
9545 dispose_words (orig_list);
9546 /* Dispose the new list we're building. */
9547 dispose_words (new_list);
9548
9549 last_command_exit_value = EXECUTION_FAILURE;
9550 if (expanded == &expand_word_error)
9551 exp_jump_to_top_level (DISCARD);
9552 else
9553 exp_jump_to_top_level (FORCE_EOF);
9554 }
9555
9556 /* Don't split words marked W_NOSPLIT. */
9557 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
9558 {
9559 temp_list = word_list_split (expanded);
9560 dispose_words (expanded);
9561 }
9562 else
9563 {
9564 /* If no parameter expansion, command substitution, process
9565 substitution, or arithmetic substitution took place, then
9566 do not do word splitting. We still have to remove quoted
9567 null characters from the result. */
9568 word_list_remove_quoted_nulls (expanded);
9569 temp_list = expanded;
9570 }
9571
9572 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
9573 new_list = (WORD_LIST *)list_append (expanded, new_list);
9574 }
9575
9576 if (orig_list)
9577 dispose_words (orig_list);
9578
9579 if (new_list)
9580 new_list = REVERSE_LIST (new_list, WORD_LIST *);
9581
9582 return (new_list);
9583 }
9584
9585 /* The workhorse for expand_words () and expand_words_no_vars ().
9586 First arg is LIST, a WORD_LIST of words.
9587 Second arg EFLAGS is a flags word controlling which expansions are
9588 performed.
9589
9590 This does all of the substitutions: brace expansion, tilde expansion,
9591 parameter expansion, command substitution, arithmetic expansion,
9592 process substitution, word splitting, and pathname expansion, according
9593 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
9594 set, or for which no expansion is done, do not undergo word splitting.
9595 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
9596 with W_NOBRACE set do not undergo brace expansion (see
9597 brace_expand_word_list above). */
9598 static WORD_LIST *
9599 expand_word_list_internal (list, eflags)
9600 WORD_LIST *list;
9601 int eflags;
9602 {
9603 WORD_LIST *new_list, *temp_list;
9604 int tint;
9605
9606 tempenv_assign_error = 0;
9607 if (list == 0)
9608 return ((WORD_LIST *)NULL);
9609
9610 garglist = new_list = copy_word_list (list);
9611 if (eflags & WEXP_VARASSIGN)
9612 {
9613 garglist = new_list = separate_out_assignments (new_list);
9614 if (new_list == 0)
9615 {
9616 if (subst_assign_varlist)
9617 {
9618 /* All the words were variable assignments, so they are placed
9619 into the shell's environment. */
9620 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9621 {
9622 this_command_name = (char *)NULL; /* no arithmetic errors */
9623 tint = do_word_assignment (temp_list->word, 0);
9624 /* Variable assignment errors in non-interactive shells
9625 running in Posix.2 mode cause the shell to exit. */
9626 if (tint == 0)
9627 {
9628 last_command_exit_value = EXECUTION_FAILURE;
9629 if (interactive_shell == 0 && posixly_correct)
9630 exp_jump_to_top_level (FORCE_EOF);
9631 else
9632 exp_jump_to_top_level (DISCARD);
9633 }
9634 }
9635 dispose_words (subst_assign_varlist);
9636 subst_assign_varlist = (WORD_LIST *)NULL;
9637 }
9638 return ((WORD_LIST *)NULL);
9639 }
9640 }
9641
9642 /* Begin expanding the words that remain. The expansions take place on
9643 things that aren't really variable assignments. */
9644
9645 #if defined (BRACE_EXPANSION)
9646 /* Do brace expansion on this word if there are any brace characters
9647 in the string. */
9648 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
9649 new_list = brace_expand_word_list (new_list, eflags);
9650 #endif /* BRACE_EXPANSION */
9651
9652 /* Perform the `normal' shell expansions: tilde expansion, parameter and
9653 variable substitution, command substitution, arithmetic expansion,
9654 and word splitting. */
9655 new_list = shell_expand_word_list (new_list, eflags);
9656
9657 /* Okay, we're almost done. Now let's just do some filename
9658 globbing. */
9659 if (new_list)
9660 {
9661 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
9662 /* Glob expand the word list unless globbing has been disabled. */
9663 new_list = glob_expand_word_list (new_list, eflags);
9664 else
9665 /* Dequote the words, because we're not performing globbing. */
9666 new_list = dequote_list (new_list);
9667 }
9668
9669 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
9670 {
9671 sh_wassign_func_t *assign_func;
9672 int is_special_builtin, is_builtin_or_func;
9673
9674 /* If the remainder of the words expand to nothing, Posix.2 requires
9675 that the variable and environment assignments affect the shell's
9676 environment. */
9677 assign_func = new_list ? assign_in_env : do_word_assignment;
9678 tempenv_assign_error = 0;
9679
9680 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
9681 /* Posix says that special builtins exit if a variable assignment error
9682 occurs in an assignment preceding it. */
9683 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
9684
9685 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9686 {
9687 this_command_name = (char *)NULL;
9688 assigning_in_environment = (assign_func == assign_in_env);
9689 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
9690 assigning_in_environment = 0;
9691 /* Variable assignment errors in non-interactive shells running
9692 in Posix.2 mode cause the shell to exit. */
9693 if (tint == 0)
9694 {
9695 if (assign_func == do_word_assignment)
9696 {
9697 last_command_exit_value = EXECUTION_FAILURE;
9698 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
9699 exp_jump_to_top_level (FORCE_EOF);
9700 else
9701 exp_jump_to_top_level (DISCARD);
9702 }
9703 else
9704 tempenv_assign_error++;
9705 }
9706 }
9707
9708 dispose_words (subst_assign_varlist);
9709 subst_assign_varlist = (WORD_LIST *)NULL;
9710 }
9711
9712 return (new_list);
9713 }