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