]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
Bash-4.3 patch 39
[thirdparty/bash.git] / subst.c
1 /* subst.c -- The part of the shell that does parameter, command, arithmetic,
2 and globbing substitutions. */
3
4 /* ``Have a little faith, there's magic in the night. You ain't a
5 beauty, but, hey, you're alright.'' */
6
7 /* Copyright (C) 1987-2013 Free Software Foundation, Inc.
8
9 This file is part of GNU Bash, the Bourne Again SHell.
10
11 Bash is free software: you can redistribute it and/or modify
12 it under the terms of the GNU General Public License as published by
13 the Free Software Foundation, either version 3 of the License, or
14 (at your option) any later version.
15
16 Bash is distributed in the hope that it will be useful,
17 but WITHOUT ANY WARRANTY; without even the implied warranty of
18 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 GNU General Public License for more details.
20
21 You should have received a copy of the GNU General Public License
22 along with Bash. If not, see <http://www.gnu.org/licenses/>.
23 */
24
25 #include "config.h"
26
27 #include "bashtypes.h"
28 #include <stdio.h>
29 #include "chartypes.h"
30 #if defined (HAVE_PWD_H)
31 # include <pwd.h>
32 #endif
33 #include <signal.h>
34 #include <errno.h>
35
36 #if defined (HAVE_UNISTD_H)
37 # include <unistd.h>
38 #endif
39
40 #include "bashansi.h"
41 #include "posixstat.h"
42 #include "bashintl.h"
43
44 #include "shell.h"
45 #include "parser.h"
46 #include "flags.h"
47 #include "jobs.h"
48 #include "execute_cmd.h"
49 #include "filecntl.h"
50 #include "trap.h"
51 #include "pathexp.h"
52 #include "mailcheck.h"
53
54 #include "shmbutil.h"
55 #include "typemax.h"
56
57 #include "builtins/getopt.h"
58 #include "builtins/common.h"
59
60 #include "builtins/builtext.h"
61
62 #include <tilde/tilde.h>
63 #include <glob/strmatch.h>
64
65 #if !defined (errno)
66 extern int errno;
67 #endif /* !errno */
68
69 /* The size that strings change by. */
70 #define DEFAULT_INITIAL_ARRAY_SIZE 112
71 #define DEFAULT_ARRAY_SIZE 128
72
73 /* Variable types. */
74 #define VT_VARIABLE 0
75 #define VT_POSPARMS 1
76 #define VT_ARRAYVAR 2
77 #define VT_ARRAYMEMBER 3
78 #define VT_ASSOCVAR 4
79
80 #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */
81
82 /* Flags for quoted_strchr */
83 #define ST_BACKSL 0x01
84 #define ST_CTLESC 0x02
85 #define ST_SQUOTE 0x04 /* unused yet */
86 #define ST_DQUOTE 0x08 /* unused yet */
87
88 /* Flags for the `pflags' argument to param_expand() */
89 #define PF_NOCOMSUB 0x01 /* Do not perform command substitution */
90 #define PF_IGNUNBOUND 0x02 /* ignore unbound vars even if -u set */
91 #define PF_NOSPLIT2 0x04 /* same as W_NOSPLIT2 */
92 #define PF_ASSIGNRHS 0x08 /* same as W_ASSIGNRHS */
93
94 /* These defs make it easier to use the editor. */
95 #define LBRACE '{'
96 #define RBRACE '}'
97 #define LPAREN '('
98 #define RPAREN ')'
99 #define LBRACK '['
100 #define RBRACK ']'
101
102 #if defined (HANDLE_MULTIBYTE)
103 #define WLPAREN L'('
104 #define WRPAREN L')'
105 #endif
106
107 /* Evaluates to 1 if C is one of the shell's special parameters whose length
108 can be taken, but is also one of the special expansion characters. */
109 #define VALID_SPECIAL_LENGTH_PARAM(c) \
110 ((c) == '-' || (c) == '?' || (c) == '#')
111
112 /* Evaluates to 1 if C is one of the shell's special parameters for which an
113 indirect variable reference may be made. */
114 #define VALID_INDIR_PARAM(c) \
115 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
116
117 /* Evaluates to 1 if C is one of the OP characters that follows the parameter
118 in ${parameter[:]OPword}. */
119 #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
120
121 /* Evaluates to 1 if this is one of the shell's special variables. */
122 #define SPECIAL_VAR(name, wi) \
123 ((DIGIT (*name) && all_digits (name)) || \
124 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
125 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1])))
126
127 /* An expansion function that takes a string and a quoted flag and returns
128 a WORD_LIST *. Used as the type of the third argument to
129 expand_string_if_necessary(). */
130 typedef WORD_LIST *EXPFUNC __P((char *, int));
131
132 /* Process ID of the last command executed within command substitution. */
133 pid_t last_command_subst_pid = NO_PID;
134 pid_t current_command_subst_pid = NO_PID;
135
136 /* Variables used to keep track of the characters in IFS. */
137 SHELL_VAR *ifs_var;
138 char *ifs_value;
139 unsigned char ifs_cmap[UCHAR_MAX + 1];
140 int ifs_is_set, ifs_is_null;
141
142 #if defined (HANDLE_MULTIBYTE)
143 unsigned char ifs_firstc[MB_LEN_MAX];
144 size_t ifs_firstc_len;
145 #else
146 unsigned char ifs_firstc;
147 #endif
148
149 /* Sentinel to tell when we are performing variable assignments preceding a
150 command name and putting them into the environment. Used to make sure
151 we use the temporary environment when looking up variable values. */
152 int assigning_in_environment;
153
154 /* Used to hold a list of variable assignments preceding a command. Global
155 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
156 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
157 WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
158
159 /* Extern functions and variables from different files. */
160 extern int last_command_exit_value, last_command_exit_signal;
161 extern int subshell_environment, line_number;
162 extern int subshell_level, parse_and_execute_level, sourcelevel;
163 extern int eof_encountered;
164 extern int return_catch_flag, return_catch_value;
165 extern pid_t dollar_dollar_pid;
166 extern int posixly_correct;
167 extern char *this_command_name;
168 extern struct fd_bitmap *current_fds_to_close;
169 extern int wordexp_only;
170 extern int expanding_redir;
171 extern int tempenv_assign_error;
172 extern int builtin_ignoring_errexit;
173
174 #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
175 extern wchar_t *wcsdup __P((const wchar_t *));
176 #endif
177
178 /* Non-zero means to allow unmatched globbed filenames to expand to
179 a null file. */
180 int allow_null_glob_expansion;
181
182 /* Non-zero means to throw an error when globbing fails to match anything. */
183 int fail_glob_expansion;
184
185 #if 0
186 /* Variables to keep track of which words in an expanded word list (the
187 output of expand_word_list_internal) are the result of globbing
188 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
189 (CURRENTLY UNUSED). */
190 char *glob_argv_flags;
191 static int glob_argv_flags_size;
192 #endif
193
194 static WORD_LIST expand_word_error, expand_word_fatal;
195 static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
196 static char expand_param_error, expand_param_fatal;
197 static char extract_string_error, extract_string_fatal;
198
199 /* Tell the expansion functions to not longjmp back to top_level on fatal
200 errors. Enabled when doing completion and prompt string expansion. */
201 static int no_longjmp_on_fatal_error = 0;
202
203 /* Set by expand_word_unsplit; used to inhibit splitting and re-joining
204 $* on $IFS, primarily when doing assignment statements. */
205 static int expand_no_split_dollar_star = 0;
206
207 /* A WORD_LIST of words to be expanded by expand_word_list_internal,
208 without any leading variable assignments. */
209 static WORD_LIST *garglist = (WORD_LIST *)NULL;
210
211 static char *quoted_substring __P((char *, int, int));
212 static int quoted_strlen __P((char *));
213 static char *quoted_strchr __P((char *, int, int));
214
215 static char *expand_string_if_necessary __P((char *, int, EXPFUNC *));
216 static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *));
217 static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
218 static WORD_LIST *expand_string_internal __P((char *, int));
219 static WORD_LIST *expand_string_leave_quoted __P((char *, int));
220 static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *));
221
222 static WORD_LIST *list_quote_escapes __P((WORD_LIST *));
223 static WORD_LIST *list_dequote_escapes __P((WORD_LIST *));
224
225 static char *make_quoted_char __P((int));
226 static WORD_LIST *quote_list __P((WORD_LIST *));
227
228 static int unquoted_substring __P((char *, char *));
229 static int unquoted_member __P((int, char *));
230
231 #if defined (ARRAY_VARS)
232 static SHELL_VAR *do_compound_assignment __P((char *, char *, int));
233 #endif
234 static int do_assignment_internal __P((const WORD_DESC *, int));
235
236 static char *string_extract_verbatim __P((char *, size_t, int *, char *, int));
237 static char *string_extract __P((char *, int *, char *, int));
238 static char *string_extract_double_quoted __P((char *, int *, int));
239 static inline char *string_extract_single_quoted __P((char *, int *));
240 static inline int skip_single_quoted __P((const char *, size_t, int));
241 static int skip_double_quoted __P((char *, size_t, int));
242 static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int));
243 static char *extract_dollar_brace_string __P((char *, int *, int, int));
244 static int skip_matched_pair __P((const char *, int, int, int, int));
245
246 static char *pos_params __P((char *, int, int, int));
247
248 static unsigned char *mb_getcharlens __P((char *, int));
249
250 static char *remove_upattern __P((char *, char *, int));
251 #if defined (HANDLE_MULTIBYTE)
252 static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int));
253 #endif
254 static char *remove_pattern __P((char *, char *, int));
255
256 static int match_upattern __P((char *, char *, int, char **, char **));
257 #if defined (HANDLE_MULTIBYTE)
258 static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
259 #endif
260 static int match_pattern __P((char *, char *, int, char **, char **));
261 static int getpatspec __P((int, char *));
262 static char *getpattern __P((char *, int, int));
263 static char *variable_remove_pattern __P((char *, char *, int, int));
264 static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int));
265 static char *parameter_list_remove_pattern __P((int, char *, int, int));
266 #ifdef ARRAY_VARS
267 static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int));
268 #endif
269 static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int));
270
271 static char *process_substitute __P((char *, int));
272
273 static char *read_comsub __P((int, int, int *));
274
275 #ifdef ARRAY_VARS
276 static arrayind_t array_length_reference __P((char *));
277 #endif
278
279 static int valid_brace_expansion_word __P((char *, int));
280 static int chk_atstar __P((char *, int, int *, int *));
281 static int chk_arithsub __P((const char *, int));
282
283 static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *));
284 static char *parameter_brace_find_indir __P((char *, int, int, int));
285 static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *));
286 static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *));
287 static void parameter_brace_expand_error __P((char *, char *));
288
289 static int valid_length_expression __P((char *));
290 static intmax_t parameter_brace_expand_length __P((char *));
291
292 static char *skiparith __P((char *, int));
293 static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
294 static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
295 static char *mb_substring __P((char *, int, int));
296 static char *parameter_brace_substring __P((char *, char *, int, char *, int, int));
297
298 static int shouldexp_replacement __P((char *));
299
300 static char *pos_params_pat_subst __P((char *, char *, char *, int));
301
302 static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int));
303
304 static char *pos_params_casemod __P((char *, char *, int, int));
305 static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int));
306
307 static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *));
308 static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int));
309
310 static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
311
312 static WORD_LIST *word_list_split __P((WORD_LIST *));
313
314 static void exp_jump_to_top_level __P((int));
315
316 static WORD_LIST *separate_out_assignments __P((WORD_LIST *));
317 static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int));
318 #ifdef BRACE_EXPANSION
319 static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int));
320 #endif
321 #if defined (ARRAY_VARS)
322 static int make_internal_declare __P((char *, char *));
323 #endif
324 static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int));
325 static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int));
326
327 /* **************************************************************** */
328 /* */
329 /* Utility Functions */
330 /* */
331 /* **************************************************************** */
332
333 #if defined (DEBUG)
334 void
335 dump_word_flags (flags)
336 int flags;
337 {
338 int f;
339
340 f = flags;
341 fprintf (stderr, "%d -> ", f);
342 if (f & W_ASSIGNASSOC)
343 {
344 f &= ~W_ASSIGNASSOC;
345 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
346 }
347 if (f & W_ASSIGNARRAY)
348 {
349 f &= ~W_ASSIGNARRAY;
350 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
351 }
352 if (f & W_HASCTLESC)
353 {
354 f &= ~W_HASCTLESC;
355 fprintf (stderr, "W_HASCTLESC%s", f ? "|" : "");
356 }
357 if (f & W_NOPROCSUB)
358 {
359 f &= ~W_NOPROCSUB;
360 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
361 }
362 if (f & W_DQUOTE)
363 {
364 f &= ~W_DQUOTE;
365 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
366 }
367 if (f & W_HASQUOTEDNULL)
368 {
369 f &= ~W_HASQUOTEDNULL;
370 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
371 }
372 if (f & W_ASSIGNARG)
373 {
374 f &= ~W_ASSIGNARG;
375 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
376 }
377 if (f & W_ASSNBLTIN)
378 {
379 f &= ~W_ASSNBLTIN;
380 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
381 }
382 if (f & W_ASSNGLOBAL)
383 {
384 f &= ~W_ASSNGLOBAL;
385 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
386 }
387 if (f & W_ASSIGNINT)
388 {
389 f &= ~W_ASSIGNINT;
390 fprintf (stderr, "W_ASSIGNINT%s", f ? "|" : "");
391 }
392 if (f & W_COMPASSIGN)
393 {
394 f &= ~W_COMPASSIGN;
395 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
396 }
397 if (f & W_NOEXPAND)
398 {
399 f &= ~W_NOEXPAND;
400 fprintf (stderr, "W_NOEXPAND%s", f ? "|" : "");
401 }
402 if (f & W_ITILDE)
403 {
404 f &= ~W_ITILDE;
405 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
406 }
407 if (f & W_NOTILDE)
408 {
409 f &= ~W_NOTILDE;
410 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
411 }
412 if (f & W_ASSIGNRHS)
413 {
414 f &= ~W_ASSIGNRHS;
415 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
416 }
417 if (f & W_NOCOMSUB)
418 {
419 f &= ~W_NOCOMSUB;
420 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
421 }
422 if (f & W_DOLLARSTAR)
423 {
424 f &= ~W_DOLLARSTAR;
425 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
426 }
427 if (f & W_DOLLARAT)
428 {
429 f &= ~W_DOLLARAT;
430 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
431 }
432 if (f & W_TILDEEXP)
433 {
434 f &= ~W_TILDEEXP;
435 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
436 }
437 if (f & W_NOSPLIT2)
438 {
439 f &= ~W_NOSPLIT2;
440 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
441 }
442 if (f & W_NOSPLIT)
443 {
444 f &= ~W_NOSPLIT;
445 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
446 }
447 if (f & W_NOBRACE)
448 {
449 f &= ~W_NOBRACE;
450 fprintf (stderr, "W_NOBRACE%s", f ? "|" : "");
451 }
452 if (f & W_NOGLOB)
453 {
454 f &= ~W_NOGLOB;
455 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
456 }
457 if (f & W_SPLITSPACE)
458 {
459 f &= ~W_SPLITSPACE;
460 fprintf (stderr, "W_SPLITSPACE%s", f ? "|" : "");
461 }
462 if (f & W_ASSIGNMENT)
463 {
464 f &= ~W_ASSIGNMENT;
465 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
466 }
467 if (f & W_QUOTED)
468 {
469 f &= ~W_QUOTED;
470 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
471 }
472 if (f & W_HASDOLLAR)
473 {
474 f &= ~W_HASDOLLAR;
475 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
476 }
477 fprintf (stderr, "\n");
478 fflush (stderr);
479 }
480 #endif
481
482 #ifdef INCLUDE_UNUSED
483 static char *
484 quoted_substring (string, start, end)
485 char *string;
486 int start, end;
487 {
488 register int len, l;
489 register char *result, *s, *r;
490
491 len = end - start;
492
493 /* Move to string[start], skipping quoted characters. */
494 for (s = string, l = 0; *s && l < start; )
495 {
496 if (*s == CTLESC)
497 {
498 s++;
499 continue;
500 }
501 l++;
502 if (*s == 0)
503 break;
504 }
505
506 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
507
508 /* Copy LEN characters, including quote characters. */
509 s = string + l;
510 for (l = 0; l < len; s++)
511 {
512 if (*s == CTLESC)
513 *r++ = *s++;
514 *r++ = *s;
515 l++;
516 if (*s == 0)
517 break;
518 }
519 *r = '\0';
520 return result;
521 }
522 #endif
523
524 #ifdef INCLUDE_UNUSED
525 /* Return the length of S, skipping over quoted characters */
526 static int
527 quoted_strlen (s)
528 char *s;
529 {
530 register char *p;
531 int i;
532
533 i = 0;
534 for (p = s; *p; p++)
535 {
536 if (*p == CTLESC)
537 {
538 p++;
539 if (*p == 0)
540 return (i + 1);
541 }
542 i++;
543 }
544
545 return i;
546 }
547 #endif
548
549 /* Find the first occurrence of character C in string S, obeying shell
550 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
551 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
552 escaped with CTLESC are skipped. */
553 static char *
554 quoted_strchr (s, c, flags)
555 char *s;
556 int c, flags;
557 {
558 register char *p;
559
560 for (p = s; *p; p++)
561 {
562 if (((flags & ST_BACKSL) && *p == '\\')
563 || ((flags & ST_CTLESC) && *p == CTLESC))
564 {
565 p++;
566 if (*p == '\0')
567 return ((char *)NULL);
568 continue;
569 }
570 else if (*p == c)
571 return p;
572 }
573 return ((char *)NULL);
574 }
575
576 /* Return 1 if CHARACTER appears in an unquoted portion of
577 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
578 static int
579 unquoted_member (character, string)
580 int character;
581 char *string;
582 {
583 size_t slen;
584 int sindex, c;
585 DECLARE_MBSTATE;
586
587 slen = strlen (string);
588 sindex = 0;
589 while (c = string[sindex])
590 {
591 if (c == character)
592 return (1);
593
594 switch (c)
595 {
596 default:
597 ADVANCE_CHAR (string, slen, sindex);
598 break;
599
600 case '\\':
601 sindex++;
602 if (string[sindex])
603 ADVANCE_CHAR (string, slen, sindex);
604 break;
605
606 case '\'':
607 sindex = skip_single_quoted (string, slen, ++sindex);
608 break;
609
610 case '"':
611 sindex = skip_double_quoted (string, slen, ++sindex);
612 break;
613 }
614 }
615 return (0);
616 }
617
618 /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
619 static int
620 unquoted_substring (substr, string)
621 char *substr, *string;
622 {
623 size_t slen;
624 int sindex, c, sublen;
625 DECLARE_MBSTATE;
626
627 if (substr == 0 || *substr == '\0')
628 return (0);
629
630 slen = strlen (string);
631 sublen = strlen (substr);
632 for (sindex = 0; c = string[sindex]; )
633 {
634 if (STREQN (string + sindex, substr, sublen))
635 return (1);
636
637 switch (c)
638 {
639 case '\\':
640 sindex++;
641 if (string[sindex])
642 ADVANCE_CHAR (string, slen, sindex);
643 break;
644
645 case '\'':
646 sindex = skip_single_quoted (string, slen, ++sindex);
647 break;
648
649 case '"':
650 sindex = skip_double_quoted (string, slen, ++sindex);
651 break;
652
653 default:
654 ADVANCE_CHAR (string, slen, sindex);
655 break;
656 }
657 }
658 return (0);
659 }
660
661 /* Most of the substitutions must be done in parallel. In order
662 to avoid using tons of unclear goto's, I have some functions
663 for manipulating malloc'ed strings. They all take INDX, a
664 pointer to an integer which is the offset into the string
665 where manipulation is taking place. They also take SIZE, a
666 pointer to an integer which is the current length of the
667 character array for this string. */
668
669 /* Append SOURCE to TARGET at INDEX. SIZE is the current amount
670 of space allocated to TARGET. SOURCE can be NULL, in which
671 case nothing happens. Gets rid of SOURCE by freeing it.
672 Returns TARGET in case the location has changed. */
673 INLINE char *
674 sub_append_string (source, target, indx, size)
675 char *source, *target;
676 int *indx, *size;
677 {
678 if (source)
679 {
680 int srclen, n;
681
682 srclen = STRLEN (source);
683 if (srclen >= (int)(*size - *indx))
684 {
685 n = srclen + *indx;
686 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
687 target = (char *)xrealloc (target, (*size = n));
688 }
689
690 FASTCOPY (source, target + *indx, srclen);
691 *indx += srclen;
692 target[*indx] = '\0';
693
694 free (source);
695 }
696 return (target);
697 }
698
699 #if 0
700 /* UNUSED */
701 /* Append the textual representation of NUMBER to TARGET.
702 INDX and SIZE are as in SUB_APPEND_STRING. */
703 char *
704 sub_append_number (number, target, indx, size)
705 intmax_t number;
706 int *indx, *size;
707 char *target;
708 {
709 char *temp;
710
711 temp = itos (number);
712 return (sub_append_string (temp, target, indx, size));
713 }
714 #endif
715
716 /* Extract a substring from STRING, starting at SINDEX and ending with
717 one of the characters in CHARLIST. Don't make the ending character
718 part of the string. Leave SINDEX pointing at the ending character.
719 Understand about backslashes in the string. If (flags & SX_VARNAME)
720 is non-zero, and array variables have been compiled into the shell,
721 everything between a `[' and a corresponding `]' is skipped over.
722 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
723 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
724 contain a closing character from CHARLIST. */
725 static char *
726 string_extract (string, sindex, charlist, flags)
727 char *string;
728 int *sindex;
729 char *charlist;
730 int flags;
731 {
732 register int c, i;
733 int found;
734 size_t slen;
735 char *temp;
736 DECLARE_MBSTATE;
737
738 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
739 i = *sindex;
740 found = 0;
741 while (c = string[i])
742 {
743 if (c == '\\')
744 {
745 if (string[i + 1])
746 i++;
747 else
748 break;
749 }
750 #if defined (ARRAY_VARS)
751 else if ((flags & SX_VARNAME) && c == '[')
752 {
753 int ni;
754 /* If this is an array subscript, skip over it and continue. */
755 ni = skipsubscript (string, i, 0);
756 if (string[ni] == ']')
757 i = ni;
758 }
759 #endif
760 else if (MEMBER (c, charlist))
761 {
762 found = 1;
763 break;
764 }
765
766 ADVANCE_CHAR (string, slen, i);
767 }
768
769 /* If we had to have a matching delimiter and didn't find one, return an
770 error and let the caller deal with it. */
771 if ((flags & SX_REQMATCH) && found == 0)
772 {
773 *sindex = i;
774 return (&extract_string_error);
775 }
776
777 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
778 *sindex = i;
779
780 return (temp);
781 }
782
783 /* Extract the contents of STRING as if it is enclosed in double quotes.
784 SINDEX, when passed in, is the offset of the character immediately
785 following the opening double quote; on exit, SINDEX is left pointing after
786 the closing double quote. If STRIPDQ is non-zero, unquoted double
787 quotes are stripped and the string is terminated by a null byte.
788 Backslashes between the embedded double quotes are processed. If STRIPDQ
789 is zero, an unquoted `"' terminates the string. */
790 static char *
791 string_extract_double_quoted (string, sindex, stripdq)
792 char *string;
793 int *sindex, stripdq;
794 {
795 size_t slen;
796 char *send;
797 int j, i, t;
798 unsigned char c;
799 char *temp, *ret; /* The new string we return. */
800 int pass_next, backquote, si; /* State variables for the machine. */
801 int dquote;
802 DECLARE_MBSTATE;
803
804 slen = strlen (string + *sindex) + *sindex;
805 send = string + slen;
806
807 pass_next = backquote = dquote = 0;
808 temp = (char *)xmalloc (1 + slen - *sindex);
809
810 j = 0;
811 i = *sindex;
812 while (c = string[i])
813 {
814 /* Process a character that was quoted by a backslash. */
815 if (pass_next)
816 {
817 /* XXX - take another look at this in light of Interp 221 */
818 /* Posix.2 sez:
819
820 ``The backslash shall retain its special meaning as an escape
821 character only when followed by one of the characters:
822 $ ` " \ <newline>''.
823
824 If STRIPDQ is zero, we handle the double quotes here and let
825 expand_word_internal handle the rest. If STRIPDQ is non-zero,
826 we have already been through one round of backslash stripping,
827 and want to strip these backslashes only if DQUOTE is non-zero,
828 indicating that we are inside an embedded double-quoted string. */
829
830 /* If we are in an embedded quoted string, then don't strip
831 backslashes before characters for which the backslash
832 retains its special meaning, but remove backslashes in
833 front of other characters. If we are not in an
834 embedded quoted string, don't strip backslashes at all.
835 This mess is necessary because the string was already
836 surrounded by double quotes (and sh has some really weird
837 quoting rules).
838 The returned string will be run through expansion as if
839 it were double-quoted. */
840 if ((stripdq == 0 && c != '"') ||
841 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
842 temp[j++] = '\\';
843 pass_next = 0;
844
845 add_one_character:
846 COPY_CHAR_I (temp, j, string, send, i);
847 continue;
848 }
849
850 /* A backslash protects the next character. The code just above
851 handles preserving the backslash in front of any character but
852 a double quote. */
853 if (c == '\\')
854 {
855 pass_next++;
856 i++;
857 continue;
858 }
859
860 /* Inside backquotes, ``the portion of the quoted string from the
861 initial backquote and the characters up to the next backquote
862 that is not preceded by a backslash, having escape characters
863 removed, defines that command''. */
864 if (backquote)
865 {
866 if (c == '`')
867 backquote = 0;
868 temp[j++] = c;
869 i++;
870 continue;
871 }
872
873 if (c == '`')
874 {
875 temp[j++] = c;
876 backquote++;
877 i++;
878 continue;
879 }
880
881 /* Pass everything between `$(' and the matching `)' or a quoted
882 ${ ... } pair through according to the Posix.2 specification. */
883 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
884 {
885 int free_ret = 1;
886
887 si = i + 2;
888 if (string[i + 1] == LPAREN)
889 ret = extract_command_subst (string, &si, 0);
890 else
891 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
892
893 temp[j++] = '$';
894 temp[j++] = string[i + 1];
895
896 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
897 is set. */
898 if (ret == 0 && no_longjmp_on_fatal_error)
899 {
900 free_ret = 0;
901 ret = string + i + 2;
902 }
903
904 for (t = 0; ret[t]; t++, j++)
905 temp[j] = ret[t];
906 temp[j] = string[si];
907
908 if (string[si])
909 {
910 j++;
911 i = si + 1;
912 }
913 else
914 i = si;
915
916 if (free_ret)
917 free (ret);
918 continue;
919 }
920
921 /* Add any character but a double quote to the quoted string we're
922 accumulating. */
923 if (c != '"')
924 goto add_one_character;
925
926 /* c == '"' */
927 if (stripdq)
928 {
929 dquote ^= 1;
930 i++;
931 continue;
932 }
933
934 break;
935 }
936 temp[j] = '\0';
937
938 /* Point to after the closing quote. */
939 if (c)
940 i++;
941 *sindex = i;
942
943 return (temp);
944 }
945
946 /* This should really be another option to string_extract_double_quoted. */
947 static int
948 skip_double_quoted (string, slen, sind)
949 char *string;
950 size_t slen;
951 int sind;
952 {
953 int c, i;
954 char *ret;
955 int pass_next, backquote, si;
956 DECLARE_MBSTATE;
957
958 pass_next = backquote = 0;
959 i = sind;
960 while (c = string[i])
961 {
962 if (pass_next)
963 {
964 pass_next = 0;
965 ADVANCE_CHAR (string, slen, i);
966 continue;
967 }
968 else if (c == '\\')
969 {
970 pass_next++;
971 i++;
972 continue;
973 }
974 else if (backquote)
975 {
976 if (c == '`')
977 backquote = 0;
978 ADVANCE_CHAR (string, slen, i);
979 continue;
980 }
981 else if (c == '`')
982 {
983 backquote++;
984 i++;
985 continue;
986 }
987 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
988 {
989 si = i + 2;
990 if (string[i + 1] == LPAREN)
991 ret = extract_command_subst (string, &si, SX_NOALLOC);
992 else
993 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
994
995 i = si + 1;
996 continue;
997 }
998 else if (c != '"')
999 {
1000 ADVANCE_CHAR (string, slen, i);
1001 continue;
1002 }
1003 else
1004 break;
1005 }
1006
1007 if (c)
1008 i++;
1009
1010 return (i);
1011 }
1012
1013 /* Extract the contents of STRING as if it is enclosed in single quotes.
1014 SINDEX, when passed in, is the offset of the character immediately
1015 following the opening single quote; on exit, SINDEX is left pointing after
1016 the closing single quote. */
1017 static inline char *
1018 string_extract_single_quoted (string, sindex)
1019 char *string;
1020 int *sindex;
1021 {
1022 register int i;
1023 size_t slen;
1024 char *t;
1025 DECLARE_MBSTATE;
1026
1027 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1028 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
1029 i = *sindex;
1030 while (string[i] && string[i] != '\'')
1031 ADVANCE_CHAR (string, slen, i);
1032
1033 t = substring (string, *sindex, i);
1034
1035 if (string[i])
1036 i++;
1037 *sindex = i;
1038
1039 return (t);
1040 }
1041
1042 static inline int
1043 skip_single_quoted (string, slen, sind)
1044 const char *string;
1045 size_t slen;
1046 int sind;
1047 {
1048 register int c;
1049 DECLARE_MBSTATE;
1050
1051 c = sind;
1052 while (string[c] && string[c] != '\'')
1053 ADVANCE_CHAR (string, slen, c);
1054
1055 if (string[c])
1056 c++;
1057 return c;
1058 }
1059
1060 /* Just like string_extract, but doesn't hack backslashes or any of
1061 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
1062 static char *
1063 string_extract_verbatim (string, slen, sindex, charlist, flags)
1064 char *string;
1065 size_t slen;
1066 int *sindex;
1067 char *charlist;
1068 int flags;
1069 {
1070 register int i;
1071 #if defined (HANDLE_MULTIBYTE)
1072 size_t clen;
1073 wchar_t *wcharlist;
1074 #endif
1075 int c;
1076 char *temp;
1077 DECLARE_MBSTATE;
1078
1079 if (charlist[0] == '\'' && charlist[1] == '\0')
1080 {
1081 temp = string_extract_single_quoted (string, sindex);
1082 --*sindex; /* leave *sindex at separator character */
1083 return temp;
1084 }
1085
1086 i = *sindex;
1087 #if 0
1088 /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need
1089 this only if MB_CUR_MAX > 1. */
1090 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1;
1091 #endif
1092 #if defined (HANDLE_MULTIBYTE)
1093 clen = strlen (charlist);
1094 wcharlist = 0;
1095 #endif
1096 while (c = string[i])
1097 {
1098 #if defined (HANDLE_MULTIBYTE)
1099 size_t mblength;
1100 #endif
1101 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1102 {
1103 i += 2;
1104 continue;
1105 }
1106 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1107 through, to protect the CTLNULs from later calls to
1108 remove_quoted_nulls. */
1109 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
1110 {
1111 i += 2;
1112 continue;
1113 }
1114
1115 #if defined (HANDLE_MULTIBYTE)
1116 mblength = MBLEN (string + i, slen - i);
1117 if (mblength > 1)
1118 {
1119 wchar_t wc;
1120 mblength = mbtowc (&wc, string + i, slen - i);
1121 if (MB_INVALIDCH (mblength))
1122 {
1123 if (MEMBER (c, charlist))
1124 break;
1125 }
1126 else
1127 {
1128 if (wcharlist == 0)
1129 {
1130 size_t len;
1131 len = mbstowcs (wcharlist, charlist, 0);
1132 if (len == -1)
1133 len = 0;
1134 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1135 mbstowcs (wcharlist, charlist, len + 1);
1136 }
1137
1138 if (wcschr (wcharlist, wc))
1139 break;
1140 }
1141 }
1142 else
1143 #endif
1144 if (MEMBER (c, charlist))
1145 break;
1146
1147 ADVANCE_CHAR (string, slen, i);
1148 }
1149
1150 #if defined (HANDLE_MULTIBYTE)
1151 FREE (wcharlist);
1152 #endif
1153
1154 temp = substring (string, *sindex, i);
1155 *sindex = i;
1156
1157 return (temp);
1158 }
1159
1160 /* Extract the $( construct in STRING, and return a new string.
1161 Start extracting at (SINDEX) as if we had just seen "$(".
1162 Make (SINDEX) get the position of the matching ")". )
1163 XFLAGS is additional flags to pass to other extraction functions. */
1164 char *
1165 extract_command_subst (string, sindex, xflags)
1166 char *string;
1167 int *sindex;
1168 int xflags;
1169 {
1170 if (string[*sindex] == LPAREN)
1171 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1172 else
1173 {
1174 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1175 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1176 }
1177 }
1178
1179 /* Extract the $[ construct in STRING, and return a new string. (])
1180 Start extracting at (SINDEX) as if we had just seen "$[".
1181 Make (SINDEX) get the position of the matching "]". */
1182 char *
1183 extract_arithmetic_subst (string, sindex)
1184 char *string;
1185 int *sindex;
1186 {
1187 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
1188 }
1189
1190 #if defined (PROCESS_SUBSTITUTION)
1191 /* Extract the <( or >( construct in STRING, and return a new string.
1192 Start extracting at (SINDEX) as if we had just seen "<(".
1193 Make (SINDEX) get the position of the matching ")". */ /*))*/
1194 char *
1195 extract_process_subst (string, starter, sindex, xflags)
1196 char *string;
1197 char *starter;
1198 int *sindex;
1199 int xflags;
1200 {
1201 #if 0
1202 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
1203 #else
1204 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1205 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1206 #endif
1207 }
1208 #endif /* PROCESS_SUBSTITUTION */
1209
1210 #if defined (ARRAY_VARS)
1211 /* This can be fooled by unquoted right parens in the passed string. If
1212 each caller verifies that the last character in STRING is a right paren,
1213 we don't even need to call extract_delimited_string. */
1214 char *
1215 extract_array_assignment_list (string, sindex)
1216 char *string;
1217 int *sindex;
1218 {
1219 int slen;
1220 char *ret;
1221
1222 slen = strlen (string); /* ( */
1223 if (string[slen - 1] == ')')
1224 {
1225 ret = substring (string, *sindex, slen - 1);
1226 *sindex = slen - 1;
1227 return ret;
1228 }
1229 return 0;
1230 }
1231 #endif
1232
1233 /* Extract and create a new string from the contents of STRING, a
1234 character string delimited with OPENER and CLOSER. SINDEX is
1235 the address of an int describing the current offset in STRING;
1236 it should point to just after the first OPENER found. On exit,
1237 SINDEX gets the position of the last character of the matching CLOSER.
1238 If OPENER is more than a single character, ALT_OPENER, if non-null,
1239 contains a character string that can also match CLOSER and thus
1240 needs to be skipped. */
1241 static char *
1242 extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
1243 char *string;
1244 int *sindex;
1245 char *opener, *alt_opener, *closer;
1246 int flags;
1247 {
1248 int i, c, si;
1249 size_t slen;
1250 char *t, *result;
1251 int pass_character, nesting_level, in_comment;
1252 int len_closer, len_opener, len_alt_opener;
1253 DECLARE_MBSTATE;
1254
1255 slen = strlen (string + *sindex) + *sindex;
1256 len_opener = STRLEN (opener);
1257 len_alt_opener = STRLEN (alt_opener);
1258 len_closer = STRLEN (closer);
1259
1260 pass_character = in_comment = 0;
1261
1262 nesting_level = 1;
1263 i = *sindex;
1264
1265 while (nesting_level)
1266 {
1267 c = string[i];
1268
1269 if (c == 0)
1270 break;
1271
1272 if (in_comment)
1273 {
1274 if (c == '\n')
1275 in_comment = 0;
1276 ADVANCE_CHAR (string, slen, i);
1277 continue;
1278 }
1279
1280 if (pass_character) /* previous char was backslash */
1281 {
1282 pass_character = 0;
1283 ADVANCE_CHAR (string, slen, i);
1284 continue;
1285 }
1286
1287 /* Not exactly right yet; should handle shell metacharacters and
1288 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
1289 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
1290 {
1291 in_comment = 1;
1292 ADVANCE_CHAR (string, slen, i);
1293 continue;
1294 }
1295
1296 if (c == CTLESC || c == '\\')
1297 {
1298 pass_character++;
1299 i++;
1300 continue;
1301 }
1302
1303 /* Process a nested command substitution, but only if we're parsing an
1304 arithmetic substitution. */
1305 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1306 {
1307 si = i + 2;
1308 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1309 i = si + 1;
1310 continue;
1311 }
1312
1313 /* Process a nested OPENER. */
1314 if (STREQN (string + i, opener, len_opener))
1315 {
1316 si = i + len_opener;
1317 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
1318 i = si + 1;
1319 continue;
1320 }
1321
1322 /* Process a nested ALT_OPENER */
1323 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
1324 {
1325 si = i + len_alt_opener;
1326 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
1327 i = si + 1;
1328 continue;
1329 }
1330
1331 /* If the current substring terminates the delimited string, decrement
1332 the nesting level. */
1333 if (STREQN (string + i, closer, len_closer))
1334 {
1335 i += len_closer - 1; /* move to last byte of the closer */
1336 nesting_level--;
1337 if (nesting_level == 0)
1338 break;
1339 }
1340
1341 /* Pass old-style command substitution through verbatim. */
1342 if (c == '`')
1343 {
1344 si = i + 1;
1345 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1346 i = si + 1;
1347 continue;
1348 }
1349
1350 /* Pass single-quoted and double-quoted strings through verbatim. */
1351 if (c == '\'' || c == '"')
1352 {
1353 si = i + 1;
1354 i = (c == '\'') ? skip_single_quoted (string, slen, si)
1355 : skip_double_quoted (string, slen, si);
1356 continue;
1357 }
1358
1359 /* move past this character, which was not special. */
1360 ADVANCE_CHAR (string, slen, i);
1361 }
1362
1363 if (c == 0 && nesting_level)
1364 {
1365 if (no_longjmp_on_fatal_error == 0)
1366 {
1367 last_command_exit_value = EXECUTION_FAILURE;
1368 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
1369 exp_jump_to_top_level (DISCARD);
1370 }
1371 else
1372 {
1373 *sindex = i;
1374 return (char *)NULL;
1375 }
1376 }
1377
1378 si = i - *sindex - len_closer + 1;
1379 if (flags & SX_NOALLOC)
1380 result = (char *)NULL;
1381 else
1382 {
1383 result = (char *)xmalloc (1 + si);
1384 strncpy (result, string + *sindex, si);
1385 result[si] = '\0';
1386 }
1387 *sindex = i;
1388
1389 return (result);
1390 }
1391
1392 /* Extract a parameter expansion expression within ${ and } from STRING.
1393 Obey the Posix.2 rules for finding the ending `}': count braces while
1394 skipping over enclosed quoted strings and command substitutions.
1395 SINDEX is the address of an int describing the current offset in STRING;
1396 it should point to just after the first `{' found. On exit, SINDEX
1397 gets the position of the matching `}'. QUOTED is non-zero if this
1398 occurs inside double quotes. */
1399 /* XXX -- this is very similar to extract_delimited_string -- XXX */
1400 static char *
1401 extract_dollar_brace_string (string, sindex, quoted, flags)
1402 char *string;
1403 int *sindex, quoted, flags;
1404 {
1405 register int i, c;
1406 size_t slen;
1407 int pass_character, nesting_level, si, dolbrace_state;
1408 char *result, *t;
1409 DECLARE_MBSTATE;
1410
1411 pass_character = 0;
1412 nesting_level = 1;
1413 slen = strlen (string + *sindex) + *sindex;
1414
1415 /* The handling of dolbrace_state needs to agree with the code in parse.y:
1416 parse_matched_pair(). The different initial value is to handle the
1417 case where this function is called to parse the word in
1418 ${param op word} (SX_WORD). */
1419 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1420 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1421 dolbrace_state = DOLBRACE_QUOTE;
1422
1423 i = *sindex;
1424 while (c = string[i])
1425 {
1426 if (pass_character)
1427 {
1428 pass_character = 0;
1429 ADVANCE_CHAR (string, slen, i);
1430 continue;
1431 }
1432
1433 /* CTLESCs and backslashes quote the next character. */
1434 if (c == CTLESC || c == '\\')
1435 {
1436 pass_character++;
1437 i++;
1438 continue;
1439 }
1440
1441 if (string[i] == '$' && string[i+1] == LBRACE)
1442 {
1443 nesting_level++;
1444 i += 2;
1445 continue;
1446 }
1447
1448 if (c == RBRACE)
1449 {
1450 nesting_level--;
1451 if (nesting_level == 0)
1452 break;
1453 i++;
1454 continue;
1455 }
1456
1457 /* Pass the contents of old-style command substitutions through
1458 verbatim. */
1459 if (c == '`')
1460 {
1461 si = i + 1;
1462 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1463 i = si + 1;
1464 continue;
1465 }
1466
1467 /* Pass the contents of new-style command substitutions and
1468 arithmetic substitutions through verbatim. */
1469 if (string[i] == '$' && string[i+1] == LPAREN)
1470 {
1471 si = i + 2;
1472 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1473 i = si + 1;
1474 continue;
1475 }
1476
1477 /* Pass the contents of double-quoted strings through verbatim. */
1478 if (c == '"')
1479 {
1480 si = i + 1;
1481 i = skip_double_quoted (string, slen, si);
1482 /* skip_XXX_quoted leaves index one past close quote */
1483 continue;
1484 }
1485
1486 if (c == '\'')
1487 {
1488 /*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
1489 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
1490 ADVANCE_CHAR (string, slen, i);
1491 else
1492 {
1493 si = i + 1;
1494 i = skip_single_quoted (string, slen, si);
1495 }
1496
1497 continue;
1498 }
1499
1500 /* move past this character, which was not special. */
1501 ADVANCE_CHAR (string, slen, i);
1502
1503 /* This logic must agree with parse.y:parse_matched_pair, since they
1504 share the same defines. */
1505 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1506 dolbrace_state = DOLBRACE_QUOTE;
1507 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1508 dolbrace_state = DOLBRACE_QUOTE;
1509 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
1510 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
1511 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1512 dolbrace_state = DOLBRACE_QUOTE;
1513 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1514 dolbrace_state = DOLBRACE_QUOTE;
1515 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1516 dolbrace_state = DOLBRACE_OP;
1517 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1518 dolbrace_state = DOLBRACE_WORD;
1519 }
1520
1521 if (c == 0 && nesting_level)
1522 {
1523 if (no_longjmp_on_fatal_error == 0)
1524 { /* { */
1525 last_command_exit_value = EXECUTION_FAILURE;
1526 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
1527 exp_jump_to_top_level (DISCARD);
1528 }
1529 else
1530 {
1531 *sindex = i;
1532 return ((char *)NULL);
1533 }
1534 }
1535
1536 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
1537 *sindex = i;
1538
1539 return (result);
1540 }
1541
1542 /* Remove backslashes which are quoting backquotes from STRING. Modifies
1543 STRING, and returns a pointer to it. */
1544 char *
1545 de_backslash (string)
1546 char *string;
1547 {
1548 register size_t slen;
1549 register int i, j, prev_i;
1550 DECLARE_MBSTATE;
1551
1552 slen = strlen (string);
1553 i = j = 0;
1554
1555 /* Loop copying string[i] to string[j], i >= j. */
1556 while (i < slen)
1557 {
1558 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
1559 string[i + 1] == '$'))
1560 i++;
1561 prev_i = i;
1562 ADVANCE_CHAR (string, slen, i);
1563 if (j < prev_i)
1564 do string[j++] = string[prev_i++]; while (prev_i < i);
1565 else
1566 j = i;
1567 }
1568 string[j] = '\0';
1569
1570 return (string);
1571 }
1572
1573 #if 0
1574 /*UNUSED*/
1575 /* Replace instances of \! in a string with !. */
1576 void
1577 unquote_bang (string)
1578 char *string;
1579 {
1580 register int i, j;
1581 register char *temp;
1582
1583 temp = (char *)xmalloc (1 + strlen (string));
1584
1585 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1586 {
1587 if (string[i] == '\\' && string[i + 1] == '!')
1588 {
1589 temp[j] = '!';
1590 i++;
1591 }
1592 }
1593 strcpy (string, temp);
1594 free (temp);
1595 }
1596 #endif
1597
1598 #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0)
1599
1600 /* This function assumes s[i] == open; returns with s[ret] == close; used to
1601 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1602 matched pairs of quotes or backquotes, or skip word expansions; it is
1603 intended to be used after expansion has been performed and during final
1604 assignment parsing (see arrayfunc.c:assign_compound_array_list()). */
1605 static int
1606 skip_matched_pair (string, start, open, close, flags)
1607 const char *string;
1608 int start, open, close, flags;
1609 {
1610 int i, pass_next, backq, si, c, count;
1611 size_t slen;
1612 char *temp, *ss;
1613 DECLARE_MBSTATE;
1614
1615 slen = strlen (string + start) + start;
1616 no_longjmp_on_fatal_error = 1;
1617
1618 i = start + 1; /* skip over leading bracket */
1619 count = 1;
1620 pass_next = backq = 0;
1621 ss = (char *)string;
1622 while (c = string[i])
1623 {
1624 if (pass_next)
1625 {
1626 pass_next = 0;
1627 if (c == 0)
1628 CQ_RETURN(i);
1629 ADVANCE_CHAR (string, slen, i);
1630 continue;
1631 }
1632 else if (c == '\\')
1633 {
1634 pass_next = 1;
1635 i++;
1636 continue;
1637 }
1638 else if (backq)
1639 {
1640 if (c == '`')
1641 backq = 0;
1642 ADVANCE_CHAR (string, slen, i);
1643 continue;
1644 }
1645 else if ((flags & 1) == 0 && c == '`')
1646 {
1647 backq = 1;
1648 i++;
1649 continue;
1650 }
1651 else if ((flags & 1) == 0 && c == open)
1652 {
1653 count++;
1654 i++;
1655 continue;
1656 }
1657 else if (c == close)
1658 {
1659 count--;
1660 if (count == 0)
1661 break;
1662 i++;
1663 continue;
1664 }
1665 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
1666 {
1667 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i)
1668 : skip_double_quoted (ss, slen, ++i);
1669 /* no increment, the skip functions increment past the closing quote. */
1670 }
1671 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
1672 {
1673 si = i + 2;
1674 if (string[si] == '\0')
1675 CQ_RETURN(si);
1676
1677 if (string[i+1] == LPAREN)
1678 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1679 else
1680 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
1681 i = si;
1682 if (string[i] == '\0') /* don't increment i past EOS in loop */
1683 break;
1684 i++;
1685 continue;
1686 }
1687 else
1688 ADVANCE_CHAR (string, slen, i);
1689 }
1690
1691 CQ_RETURN(i);
1692 }
1693
1694 #if defined (ARRAY_VARS)
1695 int
1696 skipsubscript (string, start, flags)
1697 const char *string;
1698 int start, flags;
1699 {
1700 return (skip_matched_pair (string, start, '[', ']', flags));
1701 }
1702 #endif
1703
1704 /* Skip characters in STRING until we find a character in DELIMS, and return
1705 the index of that character. START is the index into string at which we
1706 begin. This is similar in spirit to strpbrk, but it returns an index into
1707 STRING and takes a starting index. This little piece of code knows quite
1708 a lot of shell syntax. It's very similar to skip_double_quoted and other
1709 functions of that ilk. */
1710 int
1711 skip_to_delim (string, start, delims, flags)
1712 char *string;
1713 int start;
1714 char *delims;
1715 int flags;
1716 {
1717 int i, pass_next, backq, si, c, invert, skipquote, skipcmd;
1718 size_t slen;
1719 char *temp, open[3];
1720 DECLARE_MBSTATE;
1721
1722 slen = strlen (string + start) + start;
1723 if (flags & SD_NOJMP)
1724 no_longjmp_on_fatal_error = 1;
1725 invert = (flags & SD_INVERT);
1726 skipcmd = (flags & SD_NOSKIPCMD) == 0;
1727
1728 i = start;
1729 pass_next = backq = 0;
1730 while (c = string[i])
1731 {
1732 /* If this is non-zero, we should not let quote characters be delimiters
1733 and the current character is a single or double quote. We should not
1734 test whether or not it's a delimiter until after we skip single- or
1735 double-quoted strings. */
1736 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
1737 if (pass_next)
1738 {
1739 pass_next = 0;
1740 if (c == 0)
1741 CQ_RETURN(i);
1742 ADVANCE_CHAR (string, slen, i);
1743 continue;
1744 }
1745 else if (c == '\\')
1746 {
1747 pass_next = 1;
1748 i++;
1749 continue;
1750 }
1751 else if (backq)
1752 {
1753 if (c == '`')
1754 backq = 0;
1755 ADVANCE_CHAR (string, slen, i);
1756 continue;
1757 }
1758 else if (c == '`')
1759 {
1760 backq = 1;
1761 i++;
1762 continue;
1763 }
1764 else if (skipquote == 0 && invert == 0 && member (c, delims))
1765 break;
1766 else if (c == '\'' || c == '"')
1767 {
1768 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1769 : skip_double_quoted (string, slen, ++i);
1770 /* no increment, the skip functions increment past the closing quote. */
1771 }
1772 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
1773 {
1774 si = i + 2;
1775 if (string[si] == '\0')
1776 CQ_RETURN(si);
1777
1778 if (string[i+1] == LPAREN)
1779 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1780 else
1781 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1782 i = si;
1783 if (string[i] == '\0') /* don't increment i past EOS in loop */
1784 break;
1785 i++;
1786 continue;
1787 }
1788 #if defined (PROCESS_SUBSTITUTION)
1789 else if (skipcmd && (c == '<' || c == '>') && string[i+1] == LPAREN)
1790 {
1791 si = i + 2;
1792 if (string[si] == '\0')
1793 CQ_RETURN(si);
1794 temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si, 0);
1795 free (temp); /* no SX_ALLOC here */
1796 i = si;
1797 if (string[i] == '\0')
1798 break;
1799 i++;
1800 continue;
1801 }
1802 #endif /* PROCESS_SUBSTITUTION */
1803 #if defined (EXTENDED_GLOB)
1804 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1805 {
1806 si = i + 2;
1807 if (string[si] == '\0')
1808 CQ_RETURN(si);
1809
1810 open[0] = c;
1811 open[1] = LPAREN;
1812 open[2] = '\0';
1813 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
1814
1815 i = si;
1816 if (string[i] == '\0') /* don't increment i past EOS in loop */
1817 break;
1818 i++;
1819 continue;
1820 }
1821 #endif
1822 else if ((flags & SD_GLOB) && c == LBRACK)
1823 {
1824 si = i + 1;
1825 if (string[si] == '\0')
1826 CQ_RETURN(si);
1827
1828 temp = extract_delimited_string (string, &si, "[", "[", "]", SX_NOALLOC); /* ] */
1829
1830 i = si;
1831 if (string[i] == '\0') /* don't increment i past EOS in loop */
1832 break;
1833 i++;
1834 continue;
1835 }
1836 else if ((skipquote || invert) && (member (c, delims) == 0))
1837 break;
1838 else
1839 ADVANCE_CHAR (string, slen, i);
1840 }
1841
1842 CQ_RETURN(i);
1843 }
1844
1845 #if defined (READLINE)
1846 /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
1847 an unclosed quoted string), or if the character at EINDEX is quoted
1848 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
1849 single and double-quoted string parsing functions should not return an
1850 error if there are unclosed quotes or braces. The characters that this
1851 recognizes need to be the same as the contents of
1852 rl_completer_quote_characters. */
1853
1854 int
1855 char_is_quoted (string, eindex)
1856 char *string;
1857 int eindex;
1858 {
1859 int i, pass_next, c;
1860 size_t slen;
1861 DECLARE_MBSTATE;
1862
1863 slen = strlen (string);
1864 no_longjmp_on_fatal_error = 1;
1865 i = pass_next = 0;
1866 while (i <= eindex)
1867 {
1868 c = string[i];
1869
1870 if (pass_next)
1871 {
1872 pass_next = 0;
1873 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1874 CQ_RETURN(1);
1875 ADVANCE_CHAR (string, slen, i);
1876 continue;
1877 }
1878 else if (c == '\\')
1879 {
1880 pass_next = 1;
1881 i++;
1882 continue;
1883 }
1884 else if (c == '\'' || c == '"')
1885 {
1886 i = (c == '\'') ? skip_single_quoted (string, slen, ++i)
1887 : skip_double_quoted (string, slen, ++i);
1888 if (i > eindex)
1889 CQ_RETURN(1);
1890 /* no increment, the skip_xxx functions go one past end */
1891 }
1892 else
1893 ADVANCE_CHAR (string, slen, i);
1894 }
1895
1896 CQ_RETURN(0);
1897 }
1898
1899 int
1900 unclosed_pair (string, eindex, openstr)
1901 char *string;
1902 int eindex;
1903 char *openstr;
1904 {
1905 int i, pass_next, openc, olen;
1906 size_t slen;
1907 DECLARE_MBSTATE;
1908
1909 slen = strlen (string);
1910 olen = strlen (openstr);
1911 i = pass_next = openc = 0;
1912 while (i <= eindex)
1913 {
1914 if (pass_next)
1915 {
1916 pass_next = 0;
1917 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
1918 return 0;
1919 ADVANCE_CHAR (string, slen, i);
1920 continue;
1921 }
1922 else if (string[i] == '\\')
1923 {
1924 pass_next = 1;
1925 i++;
1926 continue;
1927 }
1928 else if (STREQN (string + i, openstr, olen))
1929 {
1930 openc = 1 - openc;
1931 i += olen;
1932 }
1933 else if (string[i] == '\'' || string[i] == '"')
1934 {
1935 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i)
1936 : skip_double_quoted (string, slen, i);
1937 if (i > eindex)
1938 return 0;
1939 }
1940 else
1941 ADVANCE_CHAR (string, slen, i);
1942 }
1943 return (openc);
1944 }
1945
1946 /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
1947 individual words. If DELIMS is NULL, the current value of $IFS is used
1948 to split the string, and the function follows the shell field splitting
1949 rules. SENTINEL is an index to look for. NWP, if non-NULL,
1950 gets the number of words in the returned list. CWP, if non-NULL, gets
1951 the index of the word containing SENTINEL. Non-whitespace chars in
1952 DELIMS delimit separate fields. */
1953 WORD_LIST *
1954 split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
1955 char *string;
1956 int slen;
1957 char *delims;
1958 int sentinel, flags;
1959 int *nwp, *cwp;
1960 {
1961 int ts, te, i, nw, cw, ifs_split, dflags;
1962 char *token, *d, *d2;
1963 WORD_LIST *ret, *tl;
1964
1965 if (string == 0 || *string == '\0')
1966 {
1967 if (nwp)
1968 *nwp = 0;
1969 if (cwp)
1970 *cwp = 0;
1971 return ((WORD_LIST *)NULL);
1972 }
1973
1974 d = (delims == 0) ? ifs_value : delims;
1975 ifs_split = delims == 0;
1976
1977 /* Make d2 the non-whitespace characters in delims */
1978 d2 = 0;
1979 if (delims)
1980 {
1981 size_t slength;
1982 #if defined (HANDLE_MULTIBYTE)
1983 size_t mblength = 1;
1984 #endif
1985 DECLARE_MBSTATE;
1986
1987 slength = strlen (delims);
1988 d2 = (char *)xmalloc (slength + 1);
1989 i = ts = 0;
1990 while (delims[i])
1991 {
1992 #if defined (HANDLE_MULTIBYTE)
1993 mbstate_t state_bak;
1994 state_bak = state;
1995 mblength = MBRLEN (delims + i, slength, &state);
1996 if (MB_INVALIDCH (mblength))
1997 state = state_bak;
1998 else if (mblength > 1)
1999 {
2000 memcpy (d2 + ts, delims + i, mblength);
2001 ts += mblength;
2002 i += mblength;
2003 slength -= mblength;
2004 continue;
2005 }
2006 #endif
2007 if (whitespace (delims[i]) == 0)
2008 d2[ts++] = delims[i];
2009
2010 i++;
2011 slength--;
2012 }
2013 d2[ts] = '\0';
2014 }
2015
2016 ret = (WORD_LIST *)NULL;
2017
2018 /* Remove sequences of whitespace characters at the start of the string, as
2019 long as those characters are delimiters. */
2020 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
2021 ;
2022 if (string[i] == '\0')
2023 return (ret);
2024
2025 ts = i;
2026 nw = 0;
2027 cw = -1;
2028 dflags = flags|SD_NOJMP;
2029 while (1)
2030 {
2031 te = skip_to_delim (string, ts, d, dflags);
2032
2033 /* If we have a non-whitespace delimiter character, use it to make a
2034 separate field. This is just about what $IFS splitting does and
2035 is closer to the behavior of the shell parser. */
2036 if (ts == te && d2 && member (string[ts], d2))
2037 {
2038 te = ts + 1;
2039 /* If we're using IFS splitting, the non-whitespace delimiter char
2040 and any additional IFS whitespace delimits a field. */
2041 if (ifs_split)
2042 while (member (string[te], d) && spctabnl (string[te]))
2043 te++;
2044 else
2045 while (member (string[te], d2))
2046 te++;
2047 }
2048
2049 token = substring (string, ts, te);
2050
2051 ret = add_string_to_list (token, ret);
2052 free (token);
2053 nw++;
2054
2055 if (sentinel >= ts && sentinel <= te)
2056 cw = nw;
2057
2058 /* If the cursor is at whitespace just before word start, set the
2059 sentinel word to the current word. */
2060 if (cwp && cw == -1 && sentinel == ts-1)
2061 cw = nw;
2062
2063 /* If the cursor is at whitespace between two words, make a new, empty
2064 word, add it before (well, after, since the list is in reverse order)
2065 the word we just added, and set the current word to that one. */
2066 if (cwp && cw == -1 && sentinel < ts)
2067 {
2068 tl = make_word_list (make_word (""), ret->next);
2069 ret->next = tl;
2070 cw = nw;
2071 nw++;
2072 }
2073
2074 if (string[te] == 0)
2075 break;
2076
2077 i = te;
2078 while (member (string[i], d) && (ifs_split || spctabnl(string[i])))
2079 i++;
2080
2081 if (string[i])
2082 ts = i;
2083 else
2084 break;
2085 }
2086
2087 /* Special case for SENTINEL at the end of STRING. If we haven't found
2088 the word containing SENTINEL yet, and the index we're looking for is at
2089 the end of STRING (or past the end of the previously-found token,
2090 possible if the end of the line is composed solely of IFS whitespace)
2091 add an additional null argument and set the current word pointer to that. */
2092 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
2093 {
2094 if (whitespace (string[sentinel - 1]))
2095 {
2096 token = "";
2097 ret = add_string_to_list (token, ret);
2098 nw++;
2099 }
2100 cw = nw;
2101 }
2102
2103 if (nwp)
2104 *nwp = nw;
2105 if (cwp)
2106 *cwp = cw;
2107
2108 FREE (d2);
2109
2110 return (REVERSE_LIST (ret, WORD_LIST *));
2111 }
2112 #endif /* READLINE */
2113
2114 #if 0
2115 /* UNUSED */
2116 /* Extract the name of the variable to bind to from the assignment string. */
2117 char *
2118 assignment_name (string)
2119 char *string;
2120 {
2121 int offset;
2122 char *temp;
2123
2124 offset = assignment (string, 0);
2125 if (offset == 0)
2126 return (char *)NULL;
2127 temp = substring (string, 0, offset);
2128 return (temp);
2129 }
2130 #endif
2131
2132 /* **************************************************************** */
2133 /* */
2134 /* Functions to convert strings to WORD_LISTs and vice versa */
2135 /* */
2136 /* **************************************************************** */
2137
2138 /* Return a single string of all the words in LIST. SEP is the separator
2139 to put between individual elements of LIST in the output string. */
2140 char *
2141 string_list_internal (list, sep)
2142 WORD_LIST *list;
2143 char *sep;
2144 {
2145 register WORD_LIST *t;
2146 char *result, *r;
2147 int word_len, sep_len, result_size;
2148
2149 if (list == 0)
2150 return ((char *)NULL);
2151
2152 /* Short-circuit quickly if we don't need to separate anything. */
2153 if (list->next == 0)
2154 return (savestring (list->word->word));
2155
2156 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2157 sep_len = STRLEN (sep);
2158 result_size = 0;
2159
2160 for (t = list; t; t = t->next)
2161 {
2162 if (t != list)
2163 result_size += sep_len;
2164 result_size += strlen (t->word->word);
2165 }
2166
2167 r = result = (char *)xmalloc (result_size + 1);
2168
2169 for (t = list; t; t = t->next)
2170 {
2171 if (t != list && sep_len)
2172 {
2173 if (sep_len > 1)
2174 {
2175 FASTCOPY (sep, r, sep_len);
2176 r += sep_len;
2177 }
2178 else
2179 *r++ = sep[0];
2180 }
2181
2182 word_len = strlen (t->word->word);
2183 FASTCOPY (t->word->word, r, word_len);
2184 r += word_len;
2185 }
2186
2187 *r = '\0';
2188 return (result);
2189 }
2190
2191 /* Return a single string of all the words present in LIST, separating
2192 each word with a space. */
2193 char *
2194 string_list (list)
2195 WORD_LIST *list;
2196 {
2197 return (string_list_internal (list, " "));
2198 }
2199
2200 /* An external interface that can be used by the rest of the shell to
2201 obtain a string containing the first character in $IFS. Handles all
2202 the multibyte complications. If LENP is non-null, it is set to the
2203 length of the returned string. */
2204 char *
2205 ifs_firstchar (lenp)
2206 int *lenp;
2207 {
2208 char *ret;
2209 int len;
2210
2211 ret = xmalloc (MB_LEN_MAX + 1);
2212 #if defined (HANDLE_MULTIBYTE)
2213 if (ifs_firstc_len == 1)
2214 {
2215 ret[0] = ifs_firstc[0];
2216 ret[1] = '\0';
2217 len = ret[0] ? 1 : 0;
2218 }
2219 else
2220 {
2221 memcpy (ret, ifs_firstc, ifs_firstc_len);
2222 ret[len = ifs_firstc_len] = '\0';
2223 }
2224 #else
2225 ret[0] = ifs_firstc;
2226 ret[1] = '\0';
2227 len = ret[0] ? 0 : 1;
2228 #endif
2229
2230 if (lenp)
2231 *lenp = len;
2232
2233 return ret;
2234 }
2235
2236 /* Return a single string of all the words present in LIST, obeying the
2237 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2238 expansion [of $*] appears within a double quoted string, it expands
2239 to a single field with the value of each parameter separated by the
2240 first character of the IFS variable, or by a <space> if IFS is unset." */
2241 char *
2242 string_list_dollar_star (list)
2243 WORD_LIST *list;
2244 {
2245 char *ret;
2246 #if defined (HANDLE_MULTIBYTE)
2247 # if defined (__GNUC__)
2248 char sep[MB_CUR_MAX + 1];
2249 # else
2250 char *sep = 0;
2251 # endif
2252 #else
2253 char sep[2];
2254 #endif
2255
2256 #if defined (HANDLE_MULTIBYTE)
2257 # if !defined (__GNUC__)
2258 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2259 # endif /* !__GNUC__ */
2260 if (ifs_firstc_len == 1)
2261 {
2262 sep[0] = ifs_firstc[0];
2263 sep[1] = '\0';
2264 }
2265 else
2266 {
2267 memcpy (sep, ifs_firstc, ifs_firstc_len);
2268 sep[ifs_firstc_len] = '\0';
2269 }
2270 #else
2271 sep[0] = ifs_firstc;
2272 sep[1] = '\0';
2273 #endif
2274
2275 ret = string_list_internal (list, sep);
2276 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2277 free (sep);
2278 #endif
2279 return ret;
2280 }
2281
2282 /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2283 is non-zero, the $@ appears within double quotes, and we should quote
2284 the list before converting it into a string. If IFS is unset, and the
2285 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2286 in the words in the list, because the default value of $IFS is
2287 <space><tab><newline>, IFS characters in the words in the list should
2288 also be split. If IFS is null, and the word is not quoted, we need
2289 to quote the words in the list to preserve the positional parameters
2290 exactly. */
2291 char *
2292 string_list_dollar_at (list, quoted)
2293 WORD_LIST *list;
2294 int quoted;
2295 {
2296 char *ifs, *ret;
2297 #if defined (HANDLE_MULTIBYTE)
2298 # if defined (__GNUC__)
2299 char sep[MB_CUR_MAX + 1];
2300 # else
2301 char *sep = 0;
2302 # endif /* !__GNUC__ */
2303 #else
2304 char sep[2];
2305 #endif
2306 WORD_LIST *tlist;
2307
2308 /* XXX this could just be ifs = ifs_value; */
2309 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
2310
2311 #if defined (HANDLE_MULTIBYTE)
2312 # if !defined (__GNUC__)
2313 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2314 # endif /* !__GNUC__ */
2315 if (ifs && *ifs)
2316 {
2317 if (ifs_firstc_len == 1)
2318 {
2319 sep[0] = ifs_firstc[0];
2320 sep[1] = '\0';
2321 }
2322 else
2323 {
2324 memcpy (sep, ifs_firstc, ifs_firstc_len);
2325 sep[ifs_firstc_len] = '\0';
2326 }
2327 }
2328 else
2329 {
2330 sep[0] = ' ';
2331 sep[1] = '\0';
2332 }
2333 #else
2334 sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs;
2335 sep[1] = '\0';
2336 #endif
2337
2338 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2339 it now that quote_escapes quotes spaces */
2340 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
2341 ? quote_list (list)
2342 : list_quote_escapes (list);
2343
2344 ret = string_list_internal (tlist, sep);
2345 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2346 free (sep);
2347 #endif
2348 return ret;
2349 }
2350
2351 /* Turn the positional parameters into a string, understanding quoting and
2352 the various subtleties of using the first character of $IFS as the
2353 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2354 string_list as appropriate. */
2355 char *
2356 string_list_pos_params (pchar, list, quoted)
2357 int pchar;
2358 WORD_LIST *list;
2359 int quoted;
2360 {
2361 char *ret;
2362 WORD_LIST *tlist;
2363
2364 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2365 {
2366 tlist = quote_list (list);
2367 word_list_remove_quoted_nulls (tlist);
2368 ret = string_list_dollar_star (tlist);
2369 }
2370 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2371 {
2372 tlist = quote_list (list);
2373 word_list_remove_quoted_nulls (tlist);
2374 ret = string_list (tlist);
2375 }
2376 else if (pchar == '*')
2377 {
2378 /* Even when unquoted, string_list_dollar_star does the right thing
2379 making sure that the first character of $IFS is used as the
2380 separator. */
2381 ret = string_list_dollar_star (list);
2382 }
2383 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2384 /* We use string_list_dollar_at, but only if the string is quoted, since
2385 that quotes the escapes if it's not, which we don't want. We could
2386 use string_list (the old code did), but that doesn't do the right
2387 thing if the first character of $IFS is not a space. We use
2388 string_list_dollar_star if the string is unquoted so we make sure that
2389 the elements of $@ are separated by the first character of $IFS for
2390 later splitting. */
2391 ret = string_list_dollar_at (list, quoted);
2392 else if (pchar == '@')
2393 ret = string_list_dollar_star (list);
2394 else
2395 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2396
2397 return ret;
2398 }
2399
2400 /* Return the list of words present in STRING. Separate the string into
2401 words at any of the characters found in SEPARATORS. If QUOTED is
2402 non-zero then word in the list will have its quoted flag set, otherwise
2403 the quoted flag is left as make_word () deemed fit.
2404
2405 This obeys the P1003.2 word splitting semantics. If `separators' is
2406 exactly <space><tab><newline>, then the splitting algorithm is that of
2407 the Bourne shell, which treats any sequence of characters from `separators'
2408 as a delimiter. If IFS is unset, which results in `separators' being set
2409 to "", no splitting occurs. If separators has some other value, the
2410 following rules are applied (`IFS white space' means zero or more
2411 occurrences of <space>, <tab>, or <newline>, as long as those characters
2412 are in `separators'):
2413
2414 1) IFS white space is ignored at the start and the end of the
2415 string.
2416 2) Each occurrence of a character in `separators' that is not
2417 IFS white space, along with any adjacent occurrences of
2418 IFS white space delimits a field.
2419 3) Any nonzero-length sequence of IFS white space delimits a field.
2420 */
2421
2422 /* BEWARE! list_string strips null arguments. Don't call it twice and
2423 expect to have "" preserved! */
2424
2425 /* This performs word splitting and quoted null character removal on
2426 STRING. */
2427 #define issep(c) \
2428 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2429 : (c) == (separators)[0]) \
2430 : 0)
2431
2432 WORD_LIST *
2433 list_string (string, separators, quoted)
2434 register char *string, *separators;
2435 int quoted;
2436 {
2437 WORD_LIST *result;
2438 WORD_DESC *t;
2439 char *current_word, *s;
2440 int sindex, sh_style_split, whitesep, xflags;
2441 size_t slen;
2442
2443 if (!string || !*string)
2444 return ((WORD_LIST *)NULL);
2445
2446 sh_style_split = separators && separators[0] == ' ' &&
2447 separators[1] == '\t' &&
2448 separators[2] == '\n' &&
2449 separators[3] == '\0';
2450 for (xflags = 0, s = ifs_value; s && *s; s++)
2451 {
2452 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2453 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2454 }
2455
2456 slen = 0;
2457 /* Remove sequences of whitespace at the beginning of STRING, as
2458 long as those characters appear in IFS. Do not do this if
2459 STRING is quoted or if there are no separator characters. */
2460 if (!quoted || !separators || !*separators)
2461 {
2462 for (s = string; *s && spctabnl (*s) && issep (*s); s++);
2463
2464 if (!*s)
2465 return ((WORD_LIST *)NULL);
2466
2467 string = s;
2468 }
2469
2470 /* OK, now STRING points to a word that does not begin with white space.
2471 The splitting algorithm is:
2472 extract a word, stopping at a separator
2473 skip sequences of spc, tab, or nl as long as they are separators
2474 This obeys the field splitting rules in Posix.2. */
2475 slen = (MB_CUR_MAX > 1) ? strlen (string) : 1;
2476 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
2477 {
2478 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2479 unless multibyte chars are possible. */
2480 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
2481 if (current_word == 0)
2482 break;
2483
2484 /* If we have a quoted empty string, add a quoted null argument. We
2485 want to preserve the quoted null character iff this is a quoted
2486 empty string; otherwise the quoted null characters are removed
2487 below. */
2488 if (QUOTED_NULL (current_word))
2489 {
2490 t = alloc_word_desc ();
2491 t->word = make_quoted_char ('\0');
2492 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2493 result = make_word_list (t, result);
2494 }
2495 else if (current_word[0] != '\0')
2496 {
2497 /* If we have something, then add it regardless. However,
2498 perform quoted null character removal on the current word. */
2499 remove_quoted_nulls (current_word);
2500 result = add_string_to_list (current_word, result);
2501 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
2502 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2503 result->word->flags |= W_QUOTED;
2504 }
2505
2506 /* If we're not doing sequences of separators in the traditional
2507 Bourne shell style, then add a quoted null argument. */
2508 else if (!sh_style_split && !spctabnl (string[sindex]))
2509 {
2510 t = alloc_word_desc ();
2511 t->word = make_quoted_char ('\0');
2512 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2513 result = make_word_list (t, result);
2514 }
2515
2516 free (current_word);
2517
2518 /* Note whether or not the separator is IFS whitespace, used later. */
2519 whitesep = string[sindex] && spctabnl (string[sindex]);
2520
2521 /* Move past the current separator character. */
2522 if (string[sindex])
2523 {
2524 DECLARE_MBSTATE;
2525 ADVANCE_CHAR (string, slen, sindex);
2526 }
2527
2528 /* Now skip sequences of space, tab, or newline characters if they are
2529 in the list of separators. */
2530 while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex]))
2531 sindex++;
2532
2533 /* If the first separator was IFS whitespace and the current character
2534 is a non-whitespace IFS character, it should be part of the current
2535 field delimiter, not a separate delimiter that would result in an
2536 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
2537 if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex]))
2538 {
2539 sindex++;
2540 /* An IFS character that is not IFS white space, along with any
2541 adjacent IFS white space, shall delimit a field. (SUSv3) */
2542 while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex]))
2543 sindex++;
2544 }
2545 }
2546 return (REVERSE_LIST (result, WORD_LIST *));
2547 }
2548
2549 /* Parse a single word from STRING, using SEPARATORS to separate fields.
2550 ENDPTR is set to the first character after the word. This is used by
2551 the `read' builtin. This is never called with SEPARATORS != $IFS;
2552 it should be simplified.
2553
2554 XXX - this function is very similar to list_string; they should be
2555 combined - XXX */
2556 char *
2557 get_word_from_string (stringp, separators, endptr)
2558 char **stringp, *separators, **endptr;
2559 {
2560 register char *s;
2561 char *current_word;
2562 int sindex, sh_style_split, whitesep, xflags;
2563 size_t slen;
2564
2565 if (!stringp || !*stringp || !**stringp)
2566 return ((char *)NULL);
2567
2568 sh_style_split = separators && separators[0] == ' ' &&
2569 separators[1] == '\t' &&
2570 separators[2] == '\n' &&
2571 separators[3] == '\0';
2572 for (xflags = 0, s = ifs_value; s && *s; s++)
2573 {
2574 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2575 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2576 }
2577
2578 s = *stringp;
2579 slen = 0;
2580
2581 /* Remove sequences of whitespace at the beginning of STRING, as
2582 long as those characters appear in IFS. */
2583 if (sh_style_split || !separators || !*separators)
2584 {
2585 for (; *s && spctabnl (*s) && isifs (*s); s++);
2586
2587 /* If the string is nothing but whitespace, update it and return. */
2588 if (!*s)
2589 {
2590 *stringp = s;
2591 if (endptr)
2592 *endptr = s;
2593 return ((char *)NULL);
2594 }
2595 }
2596
2597 /* OK, S points to a word that does not begin with white space.
2598 Now extract a word, stopping at a separator, save a pointer to
2599 the first character after the word, then skip sequences of spc,
2600 tab, or nl as long as they are separators.
2601
2602 This obeys the field splitting rules in Posix.2. */
2603 sindex = 0;
2604 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2605 unless multibyte chars are possible. */
2606 slen = (MB_CUR_MAX > 1) ? strlen (s) : 1;
2607 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
2608
2609 /* Set ENDPTR to the first character after the end of the word. */
2610 if (endptr)
2611 *endptr = s + sindex;
2612
2613 /* Note whether or not the separator is IFS whitespace, used later. */
2614 whitesep = s[sindex] && spctabnl (s[sindex]);
2615
2616 /* Move past the current separator character. */
2617 if (s[sindex])
2618 {
2619 DECLARE_MBSTATE;
2620 ADVANCE_CHAR (s, slen, sindex);
2621 }
2622
2623 /* Now skip sequences of space, tab, or newline characters if they are
2624 in the list of separators. */
2625 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2626 sindex++;
2627
2628 /* If the first separator was IFS whitespace and the current character is
2629 a non-whitespace IFS character, it should be part of the current field
2630 delimiter, not a separate delimiter that would result in an empty field.
2631 Look at POSIX.2, 3.6.5, (3)(b). */
2632 if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex]))
2633 {
2634 sindex++;
2635 /* An IFS character that is not IFS white space, along with any adjacent
2636 IFS white space, shall delimit a field. */
2637 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2638 sindex++;
2639 }
2640
2641 /* Update STRING to point to the next field. */
2642 *stringp = s + sindex;
2643 return (current_word);
2644 }
2645
2646 /* Remove IFS white space at the end of STRING. Start at the end
2647 of the string and walk backwards until the beginning of the string
2648 or we find a character that's not IFS white space and not CTLESC.
2649 Only let CTLESC escape a white space character if SAW_ESCAPE is
2650 non-zero. */
2651 char *
2652 strip_trailing_ifs_whitespace (string, separators, saw_escape)
2653 char *string, *separators;
2654 int saw_escape;
2655 {
2656 char *s;
2657
2658 s = string + STRLEN (string) - 1;
2659 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
2660 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2661 s--;
2662 *++s = '\0';
2663 return string;
2664 }
2665
2666 #if 0
2667 /* UNUSED */
2668 /* Split STRING into words at whitespace. Obeys shell-style quoting with
2669 backslashes, single and double quotes. */
2670 WORD_LIST *
2671 list_string_with_quotes (string)
2672 char *string;
2673 {
2674 WORD_LIST *list;
2675 char *token, *s;
2676 size_t s_len;
2677 int c, i, tokstart, len;
2678
2679 for (s = string; s && *s && spctabnl (*s); s++)
2680 ;
2681 if (s == 0 || *s == 0)
2682 return ((WORD_LIST *)NULL);
2683
2684 s_len = strlen (s);
2685 tokstart = i = 0;
2686 list = (WORD_LIST *)NULL;
2687 while (1)
2688 {
2689 c = s[i];
2690 if (c == '\\')
2691 {
2692 i++;
2693 if (s[i])
2694 i++;
2695 }
2696 else if (c == '\'')
2697 i = skip_single_quoted (s, s_len, ++i);
2698 else if (c == '"')
2699 i = skip_double_quoted (s, s_len, ++i);
2700 else if (c == 0 || spctabnl (c))
2701 {
2702 /* We have found the end of a token. Make a word out of it and
2703 add it to the word list. */
2704 token = substring (s, tokstart, i);
2705 list = add_string_to_list (token, list);
2706 free (token);
2707 while (spctabnl (s[i]))
2708 i++;
2709 if (s[i])
2710 tokstart = i;
2711 else
2712 break;
2713 }
2714 else
2715 i++; /* normal character */
2716 }
2717 return (REVERSE_LIST (list, WORD_LIST *));
2718 }
2719 #endif
2720
2721 /********************************************************/
2722 /* */
2723 /* Functions to perform assignment statements */
2724 /* */
2725 /********************************************************/
2726
2727 #if defined (ARRAY_VARS)
2728 static SHELL_VAR *
2729 do_compound_assignment (name, value, flags)
2730 char *name, *value;
2731 int flags;
2732 {
2733 SHELL_VAR *v;
2734 int mklocal, mkassoc, mkglobal;
2735 WORD_LIST *list;
2736
2737 mklocal = flags & ASS_MKLOCAL;
2738 mkassoc = flags & ASS_MKASSOC;
2739 mkglobal = flags & ASS_MKGLOBAL;
2740
2741 if (mklocal && variable_context)
2742 {
2743 v = find_variable (name);
2744 list = expand_compound_array_assignment (v, value, flags);
2745 if (mkassoc)
2746 v = make_local_assoc_variable (name);
2747 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
2748 v = make_local_array_variable (name, 0);
2749 if (v)
2750 assign_compound_array_list (v, list, flags);
2751 }
2752 /* In a function but forcing assignment in global context */
2753 else if (mkglobal && variable_context)
2754 {
2755 v = find_global_variable (name);
2756 list = expand_compound_array_assignment (v, value, flags);
2757 if (v == 0 && mkassoc)
2758 v = make_new_assoc_variable (name);
2759 else if (v && mkassoc && assoc_p (v) == 0)
2760 v = convert_var_to_assoc (v);
2761 else if (v == 0)
2762 v = make_new_array_variable (name);
2763 else if (v && mkassoc == 0 && array_p (v) == 0)
2764 v = convert_var_to_array (v);
2765 if (v)
2766 assign_compound_array_list (v, list, flags);
2767 }
2768 else
2769 v = assign_array_from_string (name, value, flags);
2770
2771 return (v);
2772 }
2773 #endif
2774
2775 /* Given STRING, an assignment string, get the value of the right side
2776 of the `=', and bind it to the left side. If EXPAND is true, then
2777 perform parameter expansion, command substitution, and arithmetic
2778 expansion on the right-hand side. Perform tilde expansion in any
2779 case. Do not perform word splitting on the result of expansion. */
2780 static int
2781 do_assignment_internal (word, expand)
2782 const WORD_DESC *word;
2783 int expand;
2784 {
2785 int offset, appendop, assign_list, aflags, retval;
2786 char *name, *value, *temp;
2787 SHELL_VAR *entry;
2788 #if defined (ARRAY_VARS)
2789 char *t;
2790 int ni;
2791 #endif
2792 const char *string;
2793
2794 if (word == 0 || word->word == 0)
2795 return 0;
2796
2797 appendop = assign_list = aflags = 0;
2798 string = word->word;
2799 offset = assignment (string, 0);
2800 name = savestring (string);
2801 value = (char *)NULL;
2802
2803 if (name[offset] == '=')
2804 {
2805 if (name[offset - 1] == '+')
2806 {
2807 appendop = 1;
2808 name[offset - 1] = '\0';
2809 }
2810
2811 name[offset] = 0; /* might need this set later */
2812 temp = name + offset + 1;
2813
2814 #if defined (ARRAY_VARS)
2815 if (expand && (word->flags & W_COMPASSIGN))
2816 {
2817 assign_list = ni = 1;
2818 value = extract_array_assignment_list (temp, &ni);
2819 }
2820 else
2821 #endif
2822 if (expand && temp[0])
2823 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
2824 else
2825 value = savestring (temp);
2826 }
2827
2828 if (value == 0)
2829 {
2830 value = (char *)xmalloc (1);
2831 value[0] = '\0';
2832 }
2833
2834 if (echo_command_at_execute)
2835 {
2836 if (appendop)
2837 name[offset - 1] = '+';
2838 xtrace_print_assignment (name, value, assign_list, 1);
2839 if (appendop)
2840 name[offset - 1] = '\0';
2841 }
2842
2843 #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
2844
2845 if (appendop)
2846 aflags |= ASS_APPEND;
2847
2848 #if defined (ARRAY_VARS)
2849 if (t = mbschr (name, '[')) /*]*/
2850 {
2851 if (assign_list)
2852 {
2853 report_error (_("%s: cannot assign list to array member"), name);
2854 ASSIGN_RETURN (0);
2855 }
2856 entry = assign_array_element (name, value, aflags);
2857 if (entry == 0)
2858 ASSIGN_RETURN (0);
2859 }
2860 else if (assign_list)
2861 {
2862 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
2863 aflags |= ASS_MKLOCAL;
2864 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
2865 aflags |= ASS_MKGLOBAL;
2866 if (word->flags & W_ASSIGNASSOC)
2867 aflags |= ASS_MKASSOC;
2868 entry = do_compound_assignment (name, value, aflags);
2869 }
2870 else
2871 #endif /* ARRAY_VARS */
2872 entry = bind_variable (name, value, aflags);
2873
2874 stupidly_hack_special_variables (name);
2875
2876 /* Return 1 if the assignment seems to have been performed correctly. */
2877 if (entry == 0 || readonly_p (entry))
2878 retval = 0; /* assignment failure */
2879 else if (noassign_p (entry))
2880 {
2881 last_command_exit_value = EXECUTION_FAILURE;
2882 retval = 1; /* error status, but not assignment failure */
2883 }
2884 else
2885 retval = 1;
2886
2887 if (entry && retval != 0 && noassign_p (entry) == 0)
2888 VUNSETATTR (entry, att_invisible);
2889
2890 ASSIGN_RETURN (retval);
2891 }
2892
2893 /* Perform the assignment statement in STRING, and expand the
2894 right side by doing tilde, command and parameter expansion. */
2895 int
2896 do_assignment (string)
2897 char *string;
2898 {
2899 WORD_DESC td;
2900
2901 td.flags = W_ASSIGNMENT;
2902 td.word = string;
2903
2904 return do_assignment_internal (&td, 1);
2905 }
2906
2907 int
2908 do_word_assignment (word, flags)
2909 WORD_DESC *word;
2910 int flags;
2911 {
2912 return do_assignment_internal (word, 1);
2913 }
2914
2915 /* Given STRING, an assignment string, get the value of the right side
2916 of the `=', and bind it to the left side. Do not perform any word
2917 expansions on the right hand side. */
2918 int
2919 do_assignment_no_expand (string)
2920 char *string;
2921 {
2922 WORD_DESC td;
2923
2924 td.flags = W_ASSIGNMENT;
2925 td.word = string;
2926
2927 return (do_assignment_internal (&td, 0));
2928 }
2929
2930 /***************************************************
2931 * *
2932 * Functions to manage the positional parameters *
2933 * *
2934 ***************************************************/
2935
2936 /* Return the word list that corresponds to `$*'. */
2937 WORD_LIST *
2938 list_rest_of_args ()
2939 {
2940 register WORD_LIST *list, *args;
2941 int i;
2942
2943 /* Break out of the loop as soon as one of the dollar variables is null. */
2944 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
2945 list = make_word_list (make_bare_word (dollar_vars[i]), list);
2946
2947 for (args = rest_of_args; args; args = args->next)
2948 list = make_word_list (make_bare_word (args->word->word), list);
2949
2950 return (REVERSE_LIST (list, WORD_LIST *));
2951 }
2952
2953 int
2954 number_of_args ()
2955 {
2956 register WORD_LIST *list;
2957 int n;
2958
2959 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
2960 ;
2961 for (list = rest_of_args; list; list = list->next)
2962 n++;
2963 return n;
2964 }
2965
2966 /* Return the value of a positional parameter. This handles values > 10. */
2967 char *
2968 get_dollar_var_value (ind)
2969 intmax_t ind;
2970 {
2971 char *temp;
2972 WORD_LIST *p;
2973
2974 if (ind < 10)
2975 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
2976 else /* We want something like ${11} */
2977 {
2978 ind -= 10;
2979 for (p = rest_of_args; p && ind--; p = p->next)
2980 ;
2981 temp = p ? savestring (p->word->word) : (char *)NULL;
2982 }
2983 return (temp);
2984 }
2985
2986 /* Make a single large string out of the dollar digit variables,
2987 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
2988 case of "$*" with respect to IFS. */
2989 char *
2990 string_rest_of_args (dollar_star)
2991 int dollar_star;
2992 {
2993 register WORD_LIST *list;
2994 char *string;
2995
2996 list = list_rest_of_args ();
2997 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
2998 dispose_words (list);
2999 return (string);
3000 }
3001
3002 /* Return a string containing the positional parameters from START to
3003 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
3004 which only makes a difference if QUOTED is non-zero. If QUOTED includes
3005 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3006 no quoting chars are added. */
3007 static char *
3008 pos_params (string, start, end, quoted)
3009 char *string;
3010 int start, end, quoted;
3011 {
3012 WORD_LIST *save, *params, *h, *t;
3013 char *ret;
3014 int i;
3015
3016 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3017 if (start == end)
3018 return ((char *)NULL);
3019
3020 save = params = list_rest_of_args ();
3021 if (save == 0)
3022 return ((char *)NULL);
3023
3024 if (start == 0) /* handle ${@:0[:x]} specially */
3025 {
3026 t = make_word_list (make_word (dollar_vars[0]), params);
3027 save = params = t;
3028 }
3029
3030 for (i = start ? 1 : 0; params && i < start; i++)
3031 params = params->next;
3032 if (params == 0)
3033 return ((char *)NULL);
3034 for (h = t = params; params && i < end; i++)
3035 {
3036 t = params;
3037 params = params->next;
3038 }
3039
3040 t->next = (WORD_LIST *)NULL;
3041
3042 ret = string_list_pos_params (string[0], h, quoted);
3043
3044 if (t != params)
3045 t->next = params;
3046
3047 dispose_words (save);
3048 return (ret);
3049 }
3050
3051 /******************************************************************/
3052 /* */
3053 /* Functions to expand strings to strings or WORD_LISTs */
3054 /* */
3055 /******************************************************************/
3056
3057 #if defined (PROCESS_SUBSTITUTION)
3058 #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
3059 #else
3060 #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
3061 #endif
3062
3063 /* If there are any characters in STRING that require full expansion,
3064 then call FUNC to expand STRING; otherwise just perform quote
3065 removal if necessary. This returns a new string. */
3066 static char *
3067 expand_string_if_necessary (string, quoted, func)
3068 char *string;
3069 int quoted;
3070 EXPFUNC *func;
3071 {
3072 WORD_LIST *list;
3073 size_t slen;
3074 int i, saw_quote;
3075 char *ret;
3076 DECLARE_MBSTATE;
3077
3078 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3079 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3080 i = saw_quote = 0;
3081 while (string[i])
3082 {
3083 if (EXP_CHAR (string[i]))
3084 break;
3085 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3086 saw_quote = 1;
3087 ADVANCE_CHAR (string, slen, i);
3088 }
3089
3090 if (string[i])
3091 {
3092 list = (*func) (string, quoted);
3093 if (list)
3094 {
3095 ret = string_list (list);
3096 dispose_words (list);
3097 }
3098 else
3099 ret = (char *)NULL;
3100 }
3101 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3102 ret = string_quote_removal (string, quoted);
3103 else
3104 ret = savestring (string);
3105
3106 return ret;
3107 }
3108
3109 static inline char *
3110 expand_string_to_string_internal (string, quoted, func)
3111 char *string;
3112 int quoted;
3113 EXPFUNC *func;
3114 {
3115 WORD_LIST *list;
3116 char *ret;
3117
3118 if (string == 0 || *string == '\0')
3119 return ((char *)NULL);
3120
3121 list = (*func) (string, quoted);
3122 if (list)
3123 {
3124 ret = string_list (list);
3125 dispose_words (list);
3126 }
3127 else
3128 ret = (char *)NULL;
3129
3130 return (ret);
3131 }
3132
3133 char *
3134 expand_string_to_string (string, quoted)
3135 char *string;
3136 int quoted;
3137 {
3138 return (expand_string_to_string_internal (string, quoted, expand_string));
3139 }
3140
3141 char *
3142 expand_string_unsplit_to_string (string, quoted)
3143 char *string;
3144 int quoted;
3145 {
3146 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3147 }
3148
3149 char *
3150 expand_assignment_string_to_string (string, quoted)
3151 char *string;
3152 int quoted;
3153 {
3154 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3155 }
3156
3157 char *
3158 expand_arith_string (string, quoted)
3159 char *string;
3160 int quoted;
3161 {
3162 WORD_DESC td;
3163 WORD_LIST *list, *tlist;
3164 size_t slen;
3165 int i, saw_quote;
3166 char *ret;
3167 DECLARE_MBSTATE;
3168
3169 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3170 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3171 i = saw_quote = 0;
3172 while (string[i])
3173 {
3174 if (EXP_CHAR (string[i]))
3175 break;
3176 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3177 saw_quote = 1;
3178 ADVANCE_CHAR (string, slen, i);
3179 }
3180
3181 if (string[i])
3182 {
3183 /* This is expanded version of expand_string_internal as it's called by
3184 expand_string_leave_quoted */
3185 td.flags = W_NOPROCSUB; /* don't want process substitution */
3186 td.word = savestring (string);
3187 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3188 /* This takes care of the calls from expand_string_leave_quoted and
3189 expand_string */
3190 if (list)
3191 {
3192 tlist = word_list_split (list);
3193 dispose_words (list);
3194 list = tlist;
3195 if (list)
3196 dequote_list (list);
3197 }
3198 /* This comes from expand_string_if_necessary */
3199 if (list)
3200 {
3201 ret = string_list (list);
3202 dispose_words (list);
3203 }
3204 else
3205 ret = (char *)NULL;
3206 FREE (td.word);
3207 }
3208 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3209 ret = string_quote_removal (string, quoted);
3210 else
3211 ret = savestring (string);
3212
3213 return ret;
3214 }
3215
3216 #if defined (COND_COMMAND)
3217 /* Just remove backslashes in STRING. Returns a new string. */
3218 char *
3219 remove_backslashes (string)
3220 char *string;
3221 {
3222 char *r, *ret, *s;
3223
3224 r = ret = (char *)xmalloc (strlen (string) + 1);
3225 for (s = string; s && *s; )
3226 {
3227 if (*s == '\\')
3228 s++;
3229 if (*s == 0)
3230 break;
3231 *r++ = *s++;
3232 }
3233 *r = '\0';
3234 return ret;
3235 }
3236
3237 /* This needs better error handling. */
3238 /* Expand W for use as an argument to a unary or binary operator in a
3239 [[...]] expression. If SPECIAL is 1, this is the rhs argument
3240 to the != or == operator, and should be treated as a pattern. In
3241 this case, we quote the string specially for the globbing code. If
3242 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3243 be quoted appropriately for regcomp/regexec. The caller is responsible
3244 for removing the backslashes if the unquoted word is needed later. */
3245 char *
3246 cond_expand_word (w, special)
3247 WORD_DESC *w;
3248 int special;
3249 {
3250 char *r, *p;
3251 WORD_LIST *l;
3252 int qflags;
3253
3254 if (w->word == 0 || w->word[0] == '\0')
3255 return ((char *)NULL);
3256
3257 expand_no_split_dollar_star = 1;
3258 w->flags |= W_NOSPLIT2;
3259 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3260 expand_no_split_dollar_star = 0;
3261 if (l)
3262 {
3263 if (special == 0) /* LHS */
3264 {
3265 dequote_list (l);
3266 r = string_list (l);
3267 }
3268 else
3269 {
3270 /* Need to figure out whether or not we should call dequote_escapes
3271 or a new dequote_ctlnul function here, and under what
3272 circumstances. */
3273 qflags = QGLOB_CVTNULL;
3274 if (special == 2)
3275 qflags |= QGLOB_REGEXP;
3276 p = string_list (l);
3277 r = quote_string_for_globbing (p, qflags);
3278 free (p);
3279 }
3280 dispose_words (l);
3281 }
3282 else
3283 r = (char *)NULL;
3284
3285 return r;
3286 }
3287 #endif
3288
3289 /* Call expand_word_internal to expand W and handle error returns.
3290 A convenience function for functions that don't want to handle
3291 any errors or free any memory before aborting. */
3292 static WORD_LIST *
3293 call_expand_word_internal (w, q, i, c, e)
3294 WORD_DESC *w;
3295 int q, i, *c, *e;
3296 {
3297 WORD_LIST *result;
3298
3299 result = expand_word_internal (w, q, i, c, e);
3300 if (result == &expand_word_error || result == &expand_word_fatal)
3301 {
3302 /* By convention, each time this error is returned, w->word has
3303 already been freed (it sometimes may not be in the fatal case,
3304 but that doesn't result in a memory leak because we're going
3305 to exit in most cases). */
3306 w->word = (char *)NULL;
3307 last_command_exit_value = EXECUTION_FAILURE;
3308 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
3309 /* NOTREACHED */
3310 return (NULL);
3311 }
3312 else
3313 return (result);
3314 }
3315
3316 /* Perform parameter expansion, command substitution, and arithmetic
3317 expansion on STRING, as if it were a word. Leave the result quoted.
3318 Since this does not perform word splitting, it leaves quoted nulls
3319 in the result. */
3320 static WORD_LIST *
3321 expand_string_internal (string, quoted)
3322 char *string;
3323 int quoted;
3324 {
3325 WORD_DESC td;
3326 WORD_LIST *tresult;
3327
3328 if (string == 0 || *string == 0)
3329 return ((WORD_LIST *)NULL);
3330
3331 td.flags = 0;
3332 td.word = savestring (string);
3333
3334 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3335
3336 FREE (td.word);
3337 return (tresult);
3338 }
3339
3340 /* Expand STRING by performing parameter expansion, command substitution,
3341 and arithmetic expansion. Dequote the resulting WORD_LIST before
3342 returning it, but do not perform word splitting. The call to
3343 remove_quoted_nulls () is in here because word splitting normally
3344 takes care of quote removal. */
3345 WORD_LIST *
3346 expand_string_unsplit (string, quoted)
3347 char *string;
3348 int quoted;
3349 {
3350 WORD_LIST *value;
3351
3352 if (string == 0 || *string == '\0')
3353 return ((WORD_LIST *)NULL);
3354
3355 expand_no_split_dollar_star = 1;
3356 value = expand_string_internal (string, quoted);
3357 expand_no_split_dollar_star = 0;
3358
3359 if (value)
3360 {
3361 if (value->word)
3362 {
3363 remove_quoted_nulls (value->word->word);
3364 value->word->flags &= ~W_HASQUOTEDNULL;
3365 }
3366 dequote_list (value);
3367 }
3368 return (value);
3369 }
3370
3371 /* Expand the rhs of an assignment statement */
3372 WORD_LIST *
3373 expand_string_assignment (string, quoted)
3374 char *string;
3375 int quoted;
3376 {
3377 WORD_DESC td;
3378 WORD_LIST *value;
3379
3380 if (string == 0 || *string == '\0')
3381 return ((WORD_LIST *)NULL);
3382
3383 expand_no_split_dollar_star = 1;
3384
3385 td.flags = W_ASSIGNRHS;
3386 td.word = savestring (string);
3387 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3388 FREE (td.word);
3389
3390 expand_no_split_dollar_star = 0;
3391
3392 if (value)
3393 {
3394 if (value->word)
3395 {
3396 remove_quoted_nulls (value->word->word);
3397 value->word->flags &= ~W_HASQUOTEDNULL;
3398 }
3399 dequote_list (value);
3400 }
3401 return (value);
3402 }
3403
3404
3405 /* Expand one of the PS? prompt strings. This is a sort of combination of
3406 expand_string_unsplit and expand_string_internal, but returns the
3407 passed string when an error occurs. Might want to trap other calls
3408 to jump_to_top_level here so we don't endlessly loop. */
3409 WORD_LIST *
3410 expand_prompt_string (string, quoted, wflags)
3411 char *string;
3412 int quoted;
3413 int wflags;
3414 {
3415 WORD_LIST *value;
3416 WORD_DESC td;
3417
3418 if (string == 0 || *string == 0)
3419 return ((WORD_LIST *)NULL);
3420
3421 td.flags = wflags;
3422 td.word = savestring (string);
3423
3424 no_longjmp_on_fatal_error = 1;
3425 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3426 no_longjmp_on_fatal_error = 0;
3427
3428 if (value == &expand_word_error || value == &expand_word_fatal)
3429 {
3430 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3431 return value;
3432 }
3433 FREE (td.word);
3434 if (value)
3435 {
3436 if (value->word)
3437 {
3438 remove_quoted_nulls (value->word->word);
3439 value->word->flags &= ~W_HASQUOTEDNULL;
3440 }
3441 dequote_list (value);
3442 }
3443 return (value);
3444 }
3445
3446 /* Expand STRING just as if you were expanding a word, but do not dequote
3447 the resultant WORD_LIST. This is called only from within this file,
3448 and is used to correctly preserve quoted characters when expanding
3449 things like ${1+"$@"}. This does parameter expansion, command
3450 substitution, arithmetic expansion, and word splitting. */
3451 static WORD_LIST *
3452 expand_string_leave_quoted (string, quoted)
3453 char *string;
3454 int quoted;
3455 {
3456 WORD_LIST *tlist;
3457 WORD_LIST *tresult;
3458
3459 if (string == 0 || *string == '\0')
3460 return ((WORD_LIST *)NULL);
3461
3462 tlist = expand_string_internal (string, quoted);
3463
3464 if (tlist)
3465 {
3466 tresult = word_list_split (tlist);
3467 dispose_words (tlist);
3468 return (tresult);
3469 }
3470 return ((WORD_LIST *)NULL);
3471 }
3472
3473 /* This does not perform word splitting or dequote the WORD_LIST
3474 it returns. */
3475 static WORD_LIST *
3476 expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at)
3477 char *string;
3478 int quoted, *dollar_at_p, *has_dollar_at;
3479 {
3480 WORD_DESC td;
3481 WORD_LIST *tresult;
3482
3483 if (string == 0 || *string == '\0')
3484 return (WORD_LIST *)NULL;
3485
3486 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
3487 td.word = string;
3488 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at);
3489 return (tresult);
3490 }
3491
3492 /* Expand STRING just as if you were expanding a word. This also returns
3493 a list of words. Note that filename globbing is *NOT* done for word
3494 or string expansion, just when the shell is expanding a command. This
3495 does parameter expansion, command substitution, arithmetic expansion,
3496 and word splitting. Dequote the resultant WORD_LIST before returning. */
3497 WORD_LIST *
3498 expand_string (string, quoted)
3499 char *string;
3500 int quoted;
3501 {
3502 WORD_LIST *result;
3503
3504 if (string == 0 || *string == '\0')
3505 return ((WORD_LIST *)NULL);
3506
3507 result = expand_string_leave_quoted (string, quoted);
3508 return (result ? dequote_list (result) : result);
3509 }
3510
3511 /***************************************************
3512 * *
3513 * Functions to handle quoting chars *
3514 * *
3515 ***************************************************/
3516
3517 /* Conventions:
3518
3519 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3520 The parser passes CTLNUL as CTLESC CTLNUL. */
3521
3522 /* Quote escape characters in string s, but no other characters. This is
3523 used to protect CTLESC and CTLNUL in variable values from the rest of
3524 the word expansion process after the variable is expanded (word splitting
3525 and filename generation). If IFS is null, we quote spaces as well, just
3526 in case we split on spaces later (in the case of unquoted $@, we will
3527 eventually attempt to split the entire word on spaces). Corresponding
3528 code exists in dequote_escapes. Even if we don't end up splitting on
3529 spaces, quoting spaces is not a problem. This should never be called on
3530 a string that is quoted with single or double quotes or part of a here
3531 document (effectively double-quoted). */
3532 char *
3533 quote_escapes (string)
3534 char *string;
3535 {
3536 register char *s, *t;
3537 size_t slen;
3538 char *result, *send;
3539 int quote_spaces, skip_ctlesc, skip_ctlnul;
3540 DECLARE_MBSTATE;
3541
3542 slen = strlen (string);
3543 send = string + slen;
3544
3545 quote_spaces = (ifs_value && *ifs_value == 0);
3546
3547 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3548 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3549
3550 t = result = (char *)xmalloc ((slen * 2) + 1);
3551 s = string;
3552
3553 while (*s)
3554 {
3555 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
3556 *t++ = CTLESC;
3557 COPY_CHAR_P (t, s, send);
3558 }
3559 *t = '\0';
3560
3561 return (result);
3562 }
3563
3564 static WORD_LIST *
3565 list_quote_escapes (list)
3566 WORD_LIST *list;
3567 {
3568 register WORD_LIST *w;
3569 char *t;
3570
3571 for (w = list; w; w = w->next)
3572 {
3573 t = w->word->word;
3574 w->word->word = quote_escapes (t);
3575 free (t);
3576 }
3577 return list;
3578 }
3579
3580 /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3581
3582 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3583 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3584 data stream pass through properly.
3585
3586 We need to remove doubled CTLESC characters inside quoted strings before
3587 quoting the entire string, so we do not double the number of CTLESC
3588 characters.
3589
3590 Also used by parts of the pattern substitution code. */
3591 char *
3592 dequote_escapes (string)
3593 char *string;
3594 {
3595 register char *s, *t, *s1;
3596 size_t slen;
3597 char *result, *send;
3598 int quote_spaces;
3599 DECLARE_MBSTATE;
3600
3601 if (string == 0)
3602 return string;
3603
3604 slen = strlen (string);
3605 send = string + slen;
3606
3607 t = result = (char *)xmalloc (slen + 1);
3608
3609 if (strchr (string, CTLESC) == 0)
3610 return (strcpy (result, string));
3611
3612 quote_spaces = (ifs_value && *ifs_value == 0);
3613
3614 s = string;
3615 while (*s)
3616 {
3617 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
3618 {
3619 s++;
3620 if (*s == '\0')
3621 break;
3622 }
3623 COPY_CHAR_P (t, s, send);
3624 }
3625 *t = '\0';
3626
3627 return result;
3628 }
3629
3630 static WORD_LIST *
3631 list_dequote_escapes (list)
3632 WORD_LIST *list;
3633 {
3634 register WORD_LIST *w;
3635 char *t;
3636
3637 for (w = list; w; w = w->next)
3638 {
3639 t = w->word->word;
3640 w->word->word = dequote_escapes (t);
3641 free (t);
3642 }
3643 return list;
3644 }
3645
3646 /* Return a new string with the quoted representation of character C.
3647 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3648 set in any resultant WORD_DESC where this value is the word. */
3649 static char *
3650 make_quoted_char (c)
3651 int c;
3652 {
3653 char *temp;
3654
3655 temp = (char *)xmalloc (3);
3656 if (c == 0)
3657 {
3658 temp[0] = CTLNUL;
3659 temp[1] = '\0';
3660 }
3661 else
3662 {
3663 temp[0] = CTLESC;
3664 temp[1] = c;
3665 temp[2] = '\0';
3666 }
3667 return (temp);
3668 }
3669
3670 /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3671 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3672 this value is the word. */
3673 char *
3674 quote_string (string)
3675 char *string;
3676 {
3677 register char *t;
3678 size_t slen;
3679 char *result, *send;
3680
3681 if (*string == 0)
3682 {
3683 result = (char *)xmalloc (2);
3684 result[0] = CTLNUL;
3685 result[1] = '\0';
3686 }
3687 else
3688 {
3689 DECLARE_MBSTATE;
3690
3691 slen = strlen (string);
3692 send = string + slen;
3693
3694 result = (char *)xmalloc ((slen * 2) + 1);
3695
3696 for (t = result; string < send; )
3697 {
3698 *t++ = CTLESC;
3699 COPY_CHAR_P (t, string, send);
3700 }
3701 *t = '\0';
3702 }
3703 return (result);
3704 }
3705
3706 /* De-quote quoted characters in STRING. */
3707 char *
3708 dequote_string (string)
3709 char *string;
3710 {
3711 register char *s, *t;
3712 size_t slen;
3713 char *result, *send;
3714 DECLARE_MBSTATE;
3715
3716 slen = strlen (string);
3717
3718 t = result = (char *)xmalloc (slen + 1);
3719
3720 if (QUOTED_NULL (string))
3721 {
3722 result[0] = '\0';
3723 return (result);
3724 }
3725
3726 /* If no character in the string can be quoted, don't bother examining
3727 each character. Just return a copy of the string passed to us. */
3728 if (strchr (string, CTLESC) == NULL)
3729 return (strcpy (result, string));
3730
3731 send = string + slen;
3732 s = string;
3733 while (*s)
3734 {
3735 if (*s == CTLESC)
3736 {
3737 s++;
3738 if (*s == '\0')
3739 break;
3740 }
3741 COPY_CHAR_P (t, s, send);
3742 }
3743
3744 *t = '\0';
3745 return (result);
3746 }
3747
3748 /* Quote the entire WORD_LIST list. */
3749 static WORD_LIST *
3750 quote_list (list)
3751 WORD_LIST *list;
3752 {
3753 register WORD_LIST *w;
3754 char *t;
3755
3756 for (w = list; w; w = w->next)
3757 {
3758 t = w->word->word;
3759 w->word->word = quote_string (t);
3760 if (*t == 0)
3761 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
3762 w->word->flags |= W_QUOTED;
3763 free (t);
3764 }
3765 return list;
3766 }
3767
3768 /* De-quote quoted characters in each word in LIST. */
3769 WORD_LIST *
3770 dequote_list (list)
3771 WORD_LIST *list;
3772 {
3773 register char *s;
3774 register WORD_LIST *tlist;
3775
3776 for (tlist = list; tlist; tlist = tlist->next)
3777 {
3778 s = dequote_string (tlist->word->word);
3779 if (QUOTED_NULL (tlist->word->word))
3780 tlist->word->flags &= ~W_HASQUOTEDNULL;
3781 free (tlist->word->word);
3782 tlist->word->word = s;
3783 }
3784 return list;
3785 }
3786
3787 /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
3788 string. */
3789 char *
3790 remove_quoted_escapes (string)
3791 char *string;
3792 {
3793 char *t;
3794
3795 if (string)
3796 {
3797 t = dequote_escapes (string);
3798 strcpy (string, t);
3799 free (t);
3800 }
3801
3802 return (string);
3803 }
3804
3805 /* Perform quoted null character removal on STRING. We don't allow any
3806 quoted null characters in the middle or at the ends of strings because
3807 of how expand_word_internal works. remove_quoted_nulls () turns
3808 STRING into an empty string iff it only consists of a quoted null,
3809 and removes all unquoted CTLNUL characters. */
3810 char *
3811 remove_quoted_nulls (string)
3812 char *string;
3813 {
3814 register size_t slen;
3815 register int i, j, prev_i;
3816 DECLARE_MBSTATE;
3817
3818 if (strchr (string, CTLNUL) == 0) /* XXX */
3819 return string; /* XXX */
3820
3821 slen = strlen (string);
3822 i = j = 0;
3823
3824 while (i < slen)
3825 {
3826 if (string[i] == CTLESC)
3827 {
3828 /* Old code had j++, but we cannot assume that i == j at this
3829 point -- what if a CTLNUL has already been removed from the
3830 string? We don't want to drop the CTLESC or recopy characters
3831 that we've already copied down. */
3832 i++; string[j++] = CTLESC;
3833 if (i == slen)
3834 break;
3835 }
3836 else if (string[i] == CTLNUL)
3837 {
3838 i++;
3839 continue;
3840 }
3841
3842 prev_i = i;
3843 ADVANCE_CHAR (string, slen, i);
3844 if (j < prev_i)
3845 {
3846 do string[j++] = string[prev_i++]; while (prev_i < i);
3847 }
3848 else
3849 j = i;
3850 }
3851 string[j] = '\0';
3852
3853 return (string);
3854 }
3855
3856 /* Perform quoted null character removal on each element of LIST.
3857 This modifies LIST. */
3858 void
3859 word_list_remove_quoted_nulls (list)
3860 WORD_LIST *list;
3861 {
3862 register WORD_LIST *t;
3863
3864 for (t = list; t; t = t->next)
3865 {
3866 remove_quoted_nulls (t->word->word);
3867 t->word->flags &= ~W_HASQUOTEDNULL;
3868 }
3869 }
3870
3871 /* **************************************************************** */
3872 /* */
3873 /* Functions for Matching and Removing Patterns */
3874 /* */
3875 /* **************************************************************** */
3876
3877 #if defined (HANDLE_MULTIBYTE)
3878 #if 0 /* Currently unused */
3879 static unsigned char *
3880 mb_getcharlens (string, len)
3881 char *string;
3882 int len;
3883 {
3884 int i, offset, last;
3885 unsigned char *ret;
3886 char *p;
3887 DECLARE_MBSTATE;
3888
3889 i = offset = 0;
3890 last = 0;
3891 ret = (unsigned char *)xmalloc (len);
3892 memset (ret, 0, len);
3893 while (string[last])
3894 {
3895 ADVANCE_CHAR (string, len, offset);
3896 ret[last] = offset - last;
3897 last = offset;
3898 }
3899 return ret;
3900 }
3901 #endif
3902 #endif
3903
3904 /* Remove the portion of PARAM matched by PATTERN according to OP, where OP
3905 can have one of 4 values:
3906 RP_LONG_LEFT remove longest matching portion at start of PARAM
3907 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
3908 RP_LONG_RIGHT remove longest matching portion at end of PARAM
3909 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
3910 */
3911
3912 #define RP_LONG_LEFT 1
3913 #define RP_SHORT_LEFT 2
3914 #define RP_LONG_RIGHT 3
3915 #define RP_SHORT_RIGHT 4
3916
3917 /* Returns its first argument if nothing matched; new memory otherwise */
3918 static char *
3919 remove_upattern (param, pattern, op)
3920 char *param, *pattern;
3921 int op;
3922 {
3923 register int len;
3924 register char *end;
3925 register char *p, *ret, c;
3926
3927 len = STRLEN (param);
3928 end = param + len;
3929
3930 switch (op)
3931 {
3932 case RP_LONG_LEFT: /* remove longest match at start */
3933 for (p = end; p >= param; p--)
3934 {
3935 c = *p; *p = '\0';
3936 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3937 {
3938 *p = c;
3939 return (savestring (p));
3940 }
3941 *p = c;
3942
3943 }
3944 break;
3945
3946 case RP_SHORT_LEFT: /* remove shortest match at start */
3947 for (p = param; p <= end; p++)
3948 {
3949 c = *p; *p = '\0';
3950 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3951 {
3952 *p = c;
3953 return (savestring (p));
3954 }
3955 *p = c;
3956 }
3957 break;
3958
3959 case RP_LONG_RIGHT: /* remove longest match at end */
3960 for (p = param; p <= end; p++)
3961 {
3962 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3963 {
3964 c = *p; *p = '\0';
3965 ret = savestring (param);
3966 *p = c;
3967 return (ret);
3968 }
3969 }
3970 break;
3971
3972 case RP_SHORT_RIGHT: /* remove shortest match at end */
3973 for (p = end; p >= param; p--)
3974 {
3975 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
3976 {
3977 c = *p; *p = '\0';
3978 ret = savestring (param);
3979 *p = c;
3980 return (ret);
3981 }
3982 }
3983 break;
3984 }
3985
3986 return (param); /* no match, return original string */
3987 }
3988
3989 #if defined (HANDLE_MULTIBYTE)
3990 /* Returns its first argument if nothing matched; new memory otherwise */
3991 static wchar_t *
3992 remove_wpattern (wparam, wstrlen, wpattern, op)
3993 wchar_t *wparam;
3994 size_t wstrlen;
3995 wchar_t *wpattern;
3996 int op;
3997 {
3998 wchar_t wc, *ret;
3999 int n;
4000
4001 switch (op)
4002 {
4003 case RP_LONG_LEFT: /* remove longest match at start */
4004 for (n = wstrlen; n >= 0; n--)
4005 {
4006 wc = wparam[n]; wparam[n] = L'\0';
4007 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4008 {
4009 wparam[n] = wc;
4010 return (wcsdup (wparam + n));
4011 }
4012 wparam[n] = wc;
4013 }
4014 break;
4015
4016 case RP_SHORT_LEFT: /* remove shortest match at start */
4017 for (n = 0; n <= wstrlen; n++)
4018 {
4019 wc = wparam[n]; wparam[n] = L'\0';
4020 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4021 {
4022 wparam[n] = wc;
4023 return (wcsdup (wparam + n));
4024 }
4025 wparam[n] = wc;
4026 }
4027 break;
4028
4029 case RP_LONG_RIGHT: /* remove longest match at end */
4030 for (n = 0; n <= wstrlen; n++)
4031 {
4032 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4033 {
4034 wc = wparam[n]; wparam[n] = L'\0';
4035 ret = wcsdup (wparam);
4036 wparam[n] = wc;
4037 return (ret);
4038 }
4039 }
4040 break;
4041
4042 case RP_SHORT_RIGHT: /* remove shortest match at end */
4043 for (n = wstrlen; n >= 0; n--)
4044 {
4045 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4046 {
4047 wc = wparam[n]; wparam[n] = L'\0';
4048 ret = wcsdup (wparam);
4049 wparam[n] = wc;
4050 return (ret);
4051 }
4052 }
4053 break;
4054 }
4055
4056 return (wparam); /* no match, return original string */
4057 }
4058 #endif /* HANDLE_MULTIBYTE */
4059
4060 static char *
4061 remove_pattern (param, pattern, op)
4062 char *param, *pattern;
4063 int op;
4064 {
4065 char *xret;
4066
4067 if (param == NULL)
4068 return (param);
4069 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4070 return (savestring (param));
4071
4072 #if defined (HANDLE_MULTIBYTE)
4073 if (MB_CUR_MAX > 1)
4074 {
4075 wchar_t *ret, *oret;
4076 size_t n;
4077 wchar_t *wparam, *wpattern;
4078 mbstate_t ps;
4079
4080 n = xdupmbstowcs (&wpattern, NULL, pattern);
4081 if (n == (size_t)-1)
4082 {
4083 xret = remove_upattern (param, pattern, op);
4084 return ((xret == param) ? savestring (param) : xret);
4085 }
4086 n = xdupmbstowcs (&wparam, NULL, param);
4087
4088 if (n == (size_t)-1)
4089 {
4090 free (wpattern);
4091 xret = remove_upattern (param, pattern, op);
4092 return ((xret == param) ? savestring (param) : xret);
4093 }
4094 oret = ret = remove_wpattern (wparam, n, wpattern, op);
4095 /* Don't bother to convert wparam back to multibyte string if nothing
4096 matched; just return copy of original string */
4097 if (ret == wparam)
4098 {
4099 free (wparam);
4100 free (wpattern);
4101 return (savestring (param));
4102 }
4103
4104 free (wparam);
4105 free (wpattern);
4106
4107 n = strlen (param);
4108 xret = (char *)xmalloc (n + 1);
4109 memset (&ps, '\0', sizeof (mbstate_t));
4110 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4111 xret[n] = '\0'; /* just to make sure */
4112 free (oret);
4113 return xret;
4114 }
4115 else
4116 #endif
4117 {
4118 xret = remove_upattern (param, pattern, op);
4119 return ((xret == param) ? savestring (param) : xret);
4120 }
4121 }
4122
4123 /* Match PAT anywhere in STRING and return the match boundaries.
4124 This returns 1 in case of a successful match, 0 otherwise. SP
4125 and EP are pointers into the string where the match begins and
4126 ends, respectively. MTYPE controls what kind of match is attempted.
4127 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4128 of the string, respectively. The longest match is returned. */
4129 static int
4130 match_upattern (string, pat, mtype, sp, ep)
4131 char *string, *pat;
4132 int mtype;
4133 char **sp, **ep;
4134 {
4135 int c, len, mlen;
4136 register char *p, *p1, *npat;
4137 char *end;
4138 int n1;
4139
4140 /* If the pattern doesn't match anywhere in the string, go ahead and
4141 short-circuit right away. A minor optimization, saves a bunch of
4142 unnecessary calls to strmatch (up to N calls for a string of N
4143 characters) if the match is unsuccessful. To preserve the semantics
4144 of the substring matches below, we make sure that the pattern has
4145 `*' as first and last character, making a new pattern if necessary. */
4146 /* XXX - check this later if I ever implement `**' with special meaning,
4147 since this will potentially result in `**' at the beginning or end */
4148 len = STRLEN (pat);
4149 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
4150 {
4151 p = npat = (char *)xmalloc (len + 3);
4152 p1 = pat;
4153 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
4154 *p++ = '*';
4155 while (*p1)
4156 *p++ = *p1++;
4157 if (p1[-1] != '*' || p[-2] == '\\')
4158 *p++ = '*';
4159 *p = '\0';
4160 }
4161 else
4162 npat = pat;
4163 c = strmatch (npat, string, FNMATCH_EXTFLAG);
4164 if (npat != pat)
4165 free (npat);
4166 if (c == FNM_NOMATCH)
4167 return (0);
4168
4169 len = STRLEN (string);
4170 end = string + len;
4171
4172 mlen = umatchlen (pat, len);
4173
4174 switch (mtype)
4175 {
4176 case MATCH_ANY:
4177 for (p = string; p <= end; p++)
4178 {
4179 if (match_pattern_char (pat, p))
4180 {
4181 p1 = (mlen == -1) ? end : p + mlen;
4182 /* p1 - p = length of portion of string to be considered
4183 p = current position in string
4184 mlen = number of characters consumed by match (-1 for entire string)
4185 end = end of string
4186 we want to break immediately if the potential match len
4187 is greater than the number of characters remaining in the
4188 string
4189 */
4190 if (p1 > end)
4191 break;
4192 for ( ; p1 >= p; p1--)
4193 {
4194 c = *p1; *p1 = '\0';
4195 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4196 {
4197 *p1 = c;
4198 *sp = p;
4199 *ep = p1;
4200 return 1;
4201 }
4202 *p1 = c;
4203 #if 1
4204 /* If MLEN != -1, we have a fixed length pattern. */
4205 if (mlen != -1)
4206 break;
4207 #endif
4208 }
4209 }
4210 }
4211
4212 return (0);
4213
4214 case MATCH_BEG:
4215 if (match_pattern_char (pat, string) == 0)
4216 return (0);
4217
4218 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
4219 {
4220 c = *p; *p = '\0';
4221 if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0)
4222 {
4223 *p = c;
4224 *sp = string;
4225 *ep = p;
4226 return 1;
4227 }
4228 *p = c;
4229 /* If MLEN != -1, we have a fixed length pattern. */
4230 if (mlen != -1)
4231 break;
4232 }
4233
4234 return (0);
4235
4236 case MATCH_END:
4237 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
4238 {
4239 if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0)
4240 {
4241 *sp = p;
4242 *ep = end;
4243 return 1;
4244 }
4245 /* If MLEN != -1, we have a fixed length pattern. */
4246 if (mlen != -1)
4247 break;
4248 }
4249
4250 return (0);
4251 }
4252
4253 return (0);
4254 }
4255
4256 #if defined (HANDLE_MULTIBYTE)
4257 /* Match WPAT anywhere in WSTRING and return the match boundaries.
4258 This returns 1 in case of a successful match, 0 otherwise. Wide
4259 character version. */
4260 static int
4261 match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4262 wchar_t *wstring;
4263 char **indices;
4264 size_t wstrlen;
4265 wchar_t *wpat;
4266 int mtype;
4267 char **sp, **ep;
4268 {
4269 wchar_t wc, *wp, *nwpat, *wp1;
4270 size_t len;
4271 int mlen;
4272 int n, n1, n2, simple;
4273
4274 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4275 #if defined (EXTENDED_GLOB)
4276 if (extended_glob)
4277 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
4278 #endif
4279
4280 /* If the pattern doesn't match anywhere in the string, go ahead and
4281 short-circuit right away. A minor optimization, saves a bunch of
4282 unnecessary calls to strmatch (up to N calls for a string of N
4283 characters) if the match is unsuccessful. To preserve the semantics
4284 of the substring matches below, we make sure that the pattern has
4285 `*' as first and last character, making a new pattern if necessary. */
4286 len = wcslen (wpat);
4287 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
4288 {
4289 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
4290 wp1 = wpat;
4291 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
4292 *wp++ = L'*';
4293 while (*wp1 != L'\0')
4294 *wp++ = *wp1++;
4295 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4296 *wp++ = L'*';
4297 *wp = '\0';
4298 }
4299 else
4300 nwpat = wpat;
4301 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG);
4302 if (nwpat != wpat)
4303 free (nwpat);
4304 if (len == FNM_NOMATCH)
4305 return (0);
4306
4307 mlen = wmatchlen (wpat, wstrlen);
4308
4309 /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
4310 switch (mtype)
4311 {
4312 case MATCH_ANY:
4313 for (n = 0; n <= wstrlen; n++)
4314 {
4315 n2 = simple ? (*wpat == wstring[n]) : match_pattern_wchar (wpat, wstring + n);
4316 if (n2)
4317 {
4318 n1 = (mlen == -1) ? wstrlen : n + mlen;
4319 if (n1 > wstrlen)
4320 break;
4321
4322 for ( ; n1 >= n; n1--)
4323 {
4324 wc = wstring[n1]; wstring[n1] = L'\0';
4325 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4326 {
4327 wstring[n1] = wc;
4328 *sp = indices[n];
4329 *ep = indices[n1];
4330 return 1;
4331 }
4332 wstring[n1] = wc;
4333 /* If MLEN != -1, we have a fixed length pattern. */
4334 if (mlen != -1)
4335 break;
4336 }
4337 }
4338 }
4339
4340 return (0);
4341
4342 case MATCH_BEG:
4343 if (match_pattern_wchar (wpat, wstring) == 0)
4344 return (0);
4345
4346 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
4347 {
4348 wc = wstring[n]; wstring[n] = L'\0';
4349 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0)
4350 {
4351 wstring[n] = wc;
4352 *sp = indices[0];
4353 *ep = indices[n];
4354 return 1;
4355 }
4356 wstring[n] = wc;
4357 /* If MLEN != -1, we have a fixed length pattern. */
4358 if (mlen != -1)
4359 break;
4360 }
4361
4362 return (0);
4363
4364 case MATCH_END:
4365 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
4366 {
4367 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0)
4368 {
4369 *sp = indices[n];
4370 *ep = indices[wstrlen];
4371 return 1;
4372 }
4373 /* If MLEN != -1, we have a fixed length pattern. */
4374 if (mlen != -1)
4375 break;
4376 }
4377
4378 return (0);
4379 }
4380
4381 return (0);
4382 }
4383 #endif /* HANDLE_MULTIBYTE */
4384
4385 static int
4386 match_pattern (string, pat, mtype, sp, ep)
4387 char *string, *pat;
4388 int mtype;
4389 char **sp, **ep;
4390 {
4391 #if defined (HANDLE_MULTIBYTE)
4392 int ret;
4393 size_t n;
4394 wchar_t *wstring, *wpat;
4395 char **indices;
4396 size_t slen, plen, mslen, mplen;
4397 #endif
4398
4399 if (string == 0 || *string == 0 || pat == 0 || *pat == 0)
4400 return (0);
4401
4402 #if defined (HANDLE_MULTIBYTE)
4403 if (MB_CUR_MAX > 1)
4404 {
4405 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
4406 return (match_upattern (string, pat, mtype, sp, ep));
4407
4408 n = xdupmbstowcs (&wpat, NULL, pat);
4409 if (n == (size_t)-1)
4410 return (match_upattern (string, pat, mtype, sp, ep));
4411 n = xdupmbstowcs (&wstring, &indices, string);
4412 if (n == (size_t)-1)
4413 {
4414 free (wpat);
4415 return (match_upattern (string, pat, mtype, sp, ep));
4416 }
4417 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4418
4419 free (wpat);
4420 free (wstring);
4421 free (indices);
4422
4423 return (ret);
4424 }
4425 else
4426 #endif
4427 return (match_upattern (string, pat, mtype, sp, ep));
4428 }
4429
4430 static int
4431 getpatspec (c, value)
4432 int c;
4433 char *value;
4434 {
4435 if (c == '#')
4436 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4437 else /* c == '%' */
4438 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4439 }
4440
4441 /* Posix.2 says that the WORD should be run through tilde expansion,
4442 parameter expansion, command substitution and arithmetic expansion.
4443 This leaves the result quoted, so quote_string_for_globbing () has
4444 to be called to fix it up for strmatch (). If QUOTED is non-zero,
4445 it means that the entire expression was enclosed in double quotes.
4446 This means that quoting characters in the pattern do not make any
4447 special pattern characters quoted. For example, the `*' in the
4448 following retains its special meaning: "${foo#'*'}". */
4449 static char *
4450 getpattern (value, quoted, expandpat)
4451 char *value;
4452 int quoted, expandpat;
4453 {
4454 char *pat, *tword;
4455 WORD_LIST *l;
4456 #if 0
4457 int i;
4458 #endif
4459 /* There is a problem here: how to handle single or double quotes in the
4460 pattern string when the whole expression is between double quotes?
4461 POSIX.2 says that enclosing double quotes do not cause the pattern to
4462 be quoted, but does that leave us a problem with @ and array[@] and their
4463 expansions inside a pattern? */
4464 #if 0
4465 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4466 {
4467 i = 0;
4468 pat = string_extract_double_quoted (tword, &i, 1);
4469 free (tword);
4470 tword = pat;
4471 }
4472 #endif
4473
4474 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4475 word splitting. */
4476 l = *value ? expand_string_for_rhs (value,
4477 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
4478 (int *)NULL, (int *)NULL)
4479 : (WORD_LIST *)0;
4480 pat = string_list (l);
4481 dispose_words (l);
4482 if (pat)
4483 {
4484 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4485 free (pat);
4486 pat = tword;
4487 }
4488 return (pat);
4489 }
4490
4491 #if 0
4492 /* Handle removing a pattern from a string as a result of ${name%[%]value}
4493 or ${name#[#]value}. */
4494 static char *
4495 variable_remove_pattern (value, pattern, patspec, quoted)
4496 char *value, *pattern;
4497 int patspec, quoted;
4498 {
4499 char *tword;
4500
4501 tword = remove_pattern (value, pattern, patspec);
4502
4503 return (tword);
4504 }
4505 #endif
4506
4507 static char *
4508 list_remove_pattern (list, pattern, patspec, itype, quoted)
4509 WORD_LIST *list;
4510 char *pattern;
4511 int patspec, itype, quoted;
4512 {
4513 WORD_LIST *new, *l;
4514 WORD_DESC *w;
4515 char *tword;
4516
4517 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4518 {
4519 tword = remove_pattern (l->word->word, pattern, patspec);
4520 w = alloc_word_desc ();
4521 w->word = tword ? tword : savestring ("");
4522 new = make_word_list (w, new);
4523 }
4524
4525 l = REVERSE_LIST (new, WORD_LIST *);
4526 tword = string_list_pos_params (itype, l, quoted);
4527 dispose_words (l);
4528
4529 return (tword);
4530 }
4531
4532 static char *
4533 parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4534 int itype;
4535 char *pattern;
4536 int patspec, quoted;
4537 {
4538 char *ret;
4539 WORD_LIST *list;
4540
4541 list = list_rest_of_args ();
4542 if (list == 0)
4543 return ((char *)NULL);
4544 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4545 dispose_words (list);
4546 return (ret);
4547 }
4548
4549 #if defined (ARRAY_VARS)
4550 static char *
4551 array_remove_pattern (var, pattern, patspec, varname, quoted)
4552 SHELL_VAR *var;
4553 char *pattern;
4554 int patspec;
4555 char *varname; /* so we can figure out how it's indexed */
4556 int quoted;
4557 {
4558 ARRAY *a;
4559 HASH_TABLE *h;
4560 int itype;
4561 char *ret;
4562 WORD_LIST *list;
4563 SHELL_VAR *v;
4564
4565 /* compute itype from varname here */
4566 v = array_variable_part (varname, &ret, 0);
4567
4568 /* XXX */
4569 if (v && invisible_p (var))
4570 return ((char *)NULL);
4571
4572 itype = ret[0];
4573
4574 a = (v && array_p (v)) ? array_cell (v) : 0;
4575 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4576
4577 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4578 if (list == 0)
4579 return ((char *)NULL);
4580 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4581 dispose_words (list);
4582
4583 return ret;
4584 }
4585 #endif /* ARRAY_VARS */
4586
4587 static char *
4588 parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4589 char *varname, *value;
4590 int ind;
4591 char *patstr;
4592 int rtype, quoted, flags;
4593 {
4594 int vtype, patspec, starsub;
4595 char *temp1, *val, *pattern;
4596 SHELL_VAR *v;
4597
4598 if (value == 0)
4599 return ((char *)NULL);
4600
4601 this_command_name = varname;
4602
4603 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4604 if (vtype == -1)
4605 return ((char *)NULL);
4606
4607 starsub = vtype & VT_STARSUB;
4608 vtype &= ~VT_STARSUB;
4609
4610 patspec = getpatspec (rtype, patstr);
4611 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4612 patstr++;
4613
4614 /* Need to pass getpattern newly-allocated memory in case of expansion --
4615 the expansion code will free the passed string on an error. */
4616 temp1 = savestring (patstr);
4617 pattern = getpattern (temp1, quoted, 1);
4618 free (temp1);
4619
4620 temp1 = (char *)NULL; /* shut up gcc */
4621 switch (vtype)
4622 {
4623 case VT_VARIABLE:
4624 case VT_ARRAYMEMBER:
4625 temp1 = remove_pattern (val, pattern, patspec);
4626 if (vtype == VT_VARIABLE)
4627 FREE (val);
4628 if (temp1)
4629 {
4630 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
4631 ? quote_string (temp1)
4632 : quote_escapes (temp1);
4633 free (temp1);
4634 temp1 = val;
4635 }
4636 break;
4637 #if defined (ARRAY_VARS)
4638 case VT_ARRAYVAR:
4639 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
4640 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4641 {
4642 val = quote_escapes (temp1);
4643 free (temp1);
4644 temp1 = val;
4645 }
4646 break;
4647 #endif
4648 case VT_POSPARMS:
4649 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
4650 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
4651 {
4652 val = quote_escapes (temp1);
4653 free (temp1);
4654 temp1 = val;
4655 }
4656 break;
4657 }
4658
4659 FREE (pattern);
4660 return temp1;
4661 }
4662
4663 /*******************************************
4664 * *
4665 * Functions to expand WORD_DESCs *
4666 * *
4667 *******************************************/
4668
4669 /* Expand WORD, performing word splitting on the result. This does
4670 parameter expansion, command substitution, arithmetic expansion,
4671 word splitting, and quote removal. */
4672
4673 WORD_LIST *
4674 expand_word (word, quoted)
4675 WORD_DESC *word;
4676 int quoted;
4677 {
4678 WORD_LIST *result, *tresult;
4679
4680 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4681 result = word_list_split (tresult);
4682 dispose_words (tresult);
4683 return (result ? dequote_list (result) : result);
4684 }
4685
4686 /* Expand WORD, but do not perform word splitting on the result. This
4687 does parameter expansion, command substitution, arithmetic expansion,
4688 and quote removal. */
4689 WORD_LIST *
4690 expand_word_unsplit (word, quoted)
4691 WORD_DESC *word;
4692 int quoted;
4693 {
4694 WORD_LIST *result;
4695
4696 expand_no_split_dollar_star = 1;
4697 #if defined (HANDLE_MULTIBYTE)
4698 if (ifs_firstc[0] == 0)
4699 #else
4700 if (ifs_firstc == 0)
4701 #endif
4702 word->flags |= W_NOSPLIT;
4703 word->flags |= W_NOSPLIT2;
4704 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4705 expand_no_split_dollar_star = 0;
4706
4707 return (result ? dequote_list (result) : result);
4708 }
4709
4710 /* Perform shell expansions on WORD, but do not perform word splitting or
4711 quote removal on the result. Virtually identical to expand_word_unsplit;
4712 could be combined if implementations don't diverge. */
4713 WORD_LIST *
4714 expand_word_leave_quoted (word, quoted)
4715 WORD_DESC *word;
4716 int quoted;
4717 {
4718 WORD_LIST *result;
4719
4720 expand_no_split_dollar_star = 1;
4721 #if defined (HANDLE_MULTIBYTE)
4722 if (ifs_firstc[0] == 0)
4723 #else
4724 if (ifs_firstc == 0)
4725 #endif
4726 word->flags |= W_NOSPLIT;
4727 word->flags |= W_NOSPLIT2;
4728 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
4729 expand_no_split_dollar_star = 0;
4730
4731 return result;
4732 }
4733
4734 #if defined (PROCESS_SUBSTITUTION)
4735
4736 /*****************************************************************/
4737 /* */
4738 /* Hacking Process Substitution */
4739 /* */
4740 /*****************************************************************/
4741
4742 #if !defined (HAVE_DEV_FD)
4743 /* Named pipes must be removed explicitly with `unlink'. This keeps a list
4744 of FIFOs the shell has open. unlink_fifo_list will walk the list and
4745 unlink all of them. add_fifo_list adds the name of an open FIFO to the
4746 list. NFIFO is a count of the number of FIFOs in the list. */
4747 #define FIFO_INCR 20
4748
4749 struct temp_fifo {
4750 char *file;
4751 pid_t proc;
4752 };
4753
4754 static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
4755 static int nfifo;
4756 static int fifo_list_size;
4757
4758 char *
4759 copy_fifo_list (sizep)
4760 int *sizep;
4761 {
4762 if (sizep)
4763 *sizep = 0;
4764 return (char *)NULL;
4765 }
4766
4767 static void
4768 add_fifo_list (pathname)
4769 char *pathname;
4770 {
4771 if (nfifo >= fifo_list_size - 1)
4772 {
4773 fifo_list_size += FIFO_INCR;
4774 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
4775 fifo_list_size * sizeof (struct temp_fifo));
4776 }
4777
4778 fifo_list[nfifo].file = savestring (pathname);
4779 nfifo++;
4780 }
4781
4782 void
4783 unlink_fifo (i)
4784 int i;
4785 {
4786 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4787 {
4788 unlink (fifo_list[i].file);
4789 free (fifo_list[i].file);
4790 fifo_list[i].file = (char *)NULL;
4791 fifo_list[i].proc = -1;
4792 }
4793 }
4794
4795 void
4796 unlink_fifo_list ()
4797 {
4798 int saved, i, j;
4799
4800 if (nfifo == 0)
4801 return;
4802
4803 for (i = saved = 0; i < nfifo; i++)
4804 {
4805 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
4806 {
4807 unlink (fifo_list[i].file);
4808 free (fifo_list[i].file);
4809 fifo_list[i].file = (char *)NULL;
4810 fifo_list[i].proc = -1;
4811 }
4812 else
4813 saved++;
4814 }
4815
4816 /* If we didn't remove some of the FIFOs, compact the list. */
4817 if (saved)
4818 {
4819 for (i = j = 0; i < nfifo; i++)
4820 if (fifo_list[i].file)
4821 {
4822 fifo_list[j].file = fifo_list[i].file;
4823 fifo_list[j].proc = fifo_list[i].proc;
4824 j++;
4825 }
4826 nfifo = j;
4827 }
4828 else
4829 nfifo = 0;
4830 }
4831
4832 /* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
4833 from some point in the past, and close all open FIFOs in fifo_list
4834 that are not marked as active in LIST. If LIST is NULL, close
4835 everything in fifo_list. LSIZE is the number of elements in LIST, in
4836 case it's larger than fifo_list_size (size of fifo_list). */
4837 void
4838 close_new_fifos (list, lsize)
4839 char *list;
4840 int lsize;
4841 {
4842 int i;
4843
4844 if (list == 0)
4845 {
4846 unlink_fifo_list ();
4847 return;
4848 }
4849
4850 for (i = 0; i < lsize; i++)
4851 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
4852 unlink_fifo (i);
4853
4854 for (i = lsize; i < fifo_list_size; i++)
4855 unlink_fifo (i);
4856 }
4857
4858 int
4859 fifos_pending ()
4860 {
4861 return nfifo;
4862 }
4863
4864 int
4865 num_fifos ()
4866 {
4867 return nfifo;
4868 }
4869
4870 static char *
4871 make_named_pipe ()
4872 {
4873 char *tname;
4874
4875 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
4876 if (mkfifo (tname, 0600) < 0)
4877 {
4878 free (tname);
4879 return ((char *)NULL);
4880 }
4881
4882 add_fifo_list (tname);
4883 return (tname);
4884 }
4885
4886 #else /* HAVE_DEV_FD */
4887
4888 /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
4889 has open to children. NFDS is a count of the number of bits currently
4890 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
4891 of open files. */
4892 static char *dev_fd_list = (char *)NULL;
4893 static int nfds;
4894 static int totfds; /* The highest possible number of open files. */
4895
4896 char *
4897 copy_fifo_list (sizep)
4898 int *sizep;
4899 {
4900 char *ret;
4901
4902 if (nfds == 0 || totfds == 0)
4903 {
4904 if (sizep)
4905 *sizep = 0;
4906 return (char *)NULL;
4907 }
4908
4909 if (sizep)
4910 *sizep = totfds;
4911 ret = (char *)xmalloc (totfds);
4912 return (memcpy (ret, dev_fd_list, totfds));
4913 }
4914
4915 static void
4916 add_fifo_list (fd)
4917 int fd;
4918 {
4919 if (dev_fd_list == 0 || fd >= totfds)
4920 {
4921 int ofds;
4922
4923 ofds = totfds;
4924 totfds = getdtablesize ();
4925 if (totfds < 0 || totfds > 256)
4926 totfds = 256;
4927 if (fd >= totfds)
4928 totfds = fd + 2;
4929
4930 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
4931 memset (dev_fd_list + ofds, '\0', totfds - ofds);
4932 }
4933
4934 dev_fd_list[fd] = 1;
4935 nfds++;
4936 }
4937
4938 int
4939 fifos_pending ()
4940 {
4941 return 0; /* used for cleanup; not needed with /dev/fd */
4942 }
4943
4944 int
4945 num_fifos ()
4946 {
4947 return nfds;
4948 }
4949
4950 void
4951 unlink_fifo (fd)
4952 int fd;
4953 {
4954 if (dev_fd_list[fd])
4955 {
4956 close (fd);
4957 dev_fd_list[fd] = 0;
4958 nfds--;
4959 }
4960 }
4961
4962 void
4963 unlink_fifo_list ()
4964 {
4965 register int i;
4966
4967 if (nfds == 0)
4968 return;
4969
4970 for (i = 0; nfds && i < totfds; i++)
4971 unlink_fifo (i);
4972
4973 nfds = 0;
4974 }
4975
4976 /* Take LIST, which is a snapshot copy of dev_fd_list from some point in
4977 the past, and close all open fds in dev_fd_list that are not marked
4978 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
4979 LSIZE is the number of elements in LIST, in case it's larger than
4980 totfds (size of dev_fd_list). */
4981 void
4982 close_new_fifos (list, lsize)
4983 char *list;
4984 int lsize;
4985 {
4986 int i;
4987
4988 if (list == 0)
4989 {
4990 unlink_fifo_list ();
4991 return;
4992 }
4993
4994 for (i = 0; i < lsize; i++)
4995 if (list[i] == 0 && i < totfds && dev_fd_list[i])
4996 unlink_fifo (i);
4997
4998 for (i = lsize; i < totfds; i++)
4999 unlink_fifo (i);
5000 }
5001
5002 #if defined (NOTDEF)
5003 print_dev_fd_list ()
5004 {
5005 register int i;
5006
5007 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
5008 fflush (stderr);
5009
5010 for (i = 0; i < totfds; i++)
5011 {
5012 if (dev_fd_list[i])
5013 fprintf (stderr, " %d", i);
5014 }
5015 fprintf (stderr, "\n");
5016 }
5017 #endif /* NOTDEF */
5018
5019 static char *
5020 make_dev_fd_filename (fd)
5021 int fd;
5022 {
5023 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
5024
5025 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
5026
5027 strcpy (ret, DEV_FD_PREFIX);
5028 p = inttostr (fd, intbuf, sizeof (intbuf));
5029 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5030
5031 add_fifo_list (fd);
5032 return (ret);
5033 }
5034
5035 #endif /* HAVE_DEV_FD */
5036
5037 /* Return a filename that will open a connection to the process defined by
5038 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5039 a filename in /dev/fd corresponding to a descriptor that is one of the
5040 ends of the pipe. If not defined, we use named pipes on systems that have
5041 them. Systems without /dev/fd and named pipes are out of luck.
5042
5043 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5044 use the read end of the pipe and dup that file descriptor to fd 0 in
5045 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5046 writing or use the write end of the pipe in the child, and dup that
5047 file descriptor to fd 1 in the child. The parent does the opposite. */
5048
5049 static char *
5050 process_substitute (string, open_for_read_in_child)
5051 char *string;
5052 int open_for_read_in_child;
5053 {
5054 char *pathname;
5055 int fd, result;
5056 pid_t old_pid, pid;
5057 #if defined (HAVE_DEV_FD)
5058 int parent_pipe_fd, child_pipe_fd;
5059 int fildes[2];
5060 #endif /* HAVE_DEV_FD */
5061 #if defined (JOB_CONTROL)
5062 pid_t old_pipeline_pgrp;
5063 #endif
5064
5065 if (!string || !*string || wordexp_only)
5066 return ((char *)NULL);
5067
5068 #if !defined (HAVE_DEV_FD)
5069 pathname = make_named_pipe ();
5070 #else /* HAVE_DEV_FD */
5071 if (pipe (fildes) < 0)
5072 {
5073 sys_error (_("cannot make pipe for process substitution"));
5074 return ((char *)NULL);
5075 }
5076 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5077 the pipe in the parent, otherwise the read end. */
5078 parent_pipe_fd = fildes[open_for_read_in_child];
5079 child_pipe_fd = fildes[1 - open_for_read_in_child];
5080 /* Move the parent end of the pipe to some high file descriptor, to
5081 avoid clashes with FDs used by the script. */
5082 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5083
5084 pathname = make_dev_fd_filename (parent_pipe_fd);
5085 #endif /* HAVE_DEV_FD */
5086
5087 if (pathname == 0)
5088 {
5089 sys_error (_("cannot make pipe for process substitution"));
5090 return ((char *)NULL);
5091 }
5092
5093 old_pid = last_made_pid;
5094
5095 #if defined (JOB_CONTROL)
5096 old_pipeline_pgrp = pipeline_pgrp;
5097 pipeline_pgrp = shell_pgrp;
5098 save_pipeline (1);
5099 #endif /* JOB_CONTROL */
5100
5101 pid = make_child ((char *)NULL, 1);
5102 if (pid == 0)
5103 {
5104 reset_terminating_signals (); /* XXX */
5105 free_pushed_string_input ();
5106 /* Cancel traps, in trap.c. */
5107 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
5108 setup_async_signals ();
5109 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
5110 }
5111
5112 #if defined (JOB_CONTROL)
5113 set_sigchld_handler ();
5114 stop_making_children ();
5115 /* XXX - should we only do this in the parent? (as in command subst) */
5116 pipeline_pgrp = old_pipeline_pgrp;
5117 #endif /* JOB_CONTROL */
5118
5119 if (pid < 0)
5120 {
5121 sys_error (_("cannot make child for process substitution"));
5122 free (pathname);
5123 #if defined (HAVE_DEV_FD)
5124 close (parent_pipe_fd);
5125 close (child_pipe_fd);
5126 #endif /* HAVE_DEV_FD */
5127 return ((char *)NULL);
5128 }
5129
5130 if (pid > 0)
5131 {
5132 #if defined (JOB_CONTROL)
5133 restore_pipeline (1);
5134 #endif
5135
5136 #if !defined (HAVE_DEV_FD)
5137 fifo_list[nfifo-1].proc = pid;
5138 #endif
5139
5140 last_made_pid = old_pid;
5141
5142 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5143 close_pgrp_pipe ();
5144 #endif /* JOB_CONTROL && PGRP_PIPE */
5145
5146 #if defined (HAVE_DEV_FD)
5147 close (child_pipe_fd);
5148 #endif /* HAVE_DEV_FD */
5149
5150 return (pathname);
5151 }
5152
5153 set_sigint_handler ();
5154
5155 #if defined (JOB_CONTROL)
5156 set_job_control (0);
5157 #endif /* JOB_CONTROL */
5158
5159 #if !defined (HAVE_DEV_FD)
5160 /* Open the named pipe in the child. */
5161 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
5162 if (fd < 0)
5163 {
5164 /* Two separate strings for ease of translation. */
5165 if (open_for_read_in_child)
5166 sys_error (_("cannot open named pipe %s for reading"), pathname);
5167 else
5168 sys_error (_("cannot open named pipe %s for writing"), pathname);
5169
5170 exit (127);
5171 }
5172 if (open_for_read_in_child)
5173 {
5174 if (sh_unset_nodelay_mode (fd) < 0)
5175 {
5176 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
5177 exit (127);
5178 }
5179 }
5180 #else /* HAVE_DEV_FD */
5181 fd = child_pipe_fd;
5182 #endif /* HAVE_DEV_FD */
5183
5184 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5185 {
5186 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
5187 open_for_read_in_child ? 0 : 1);
5188 exit (127);
5189 }
5190
5191 if (fd != (open_for_read_in_child ? 0 : 1))
5192 close (fd);
5193
5194 /* Need to close any files that this process has open to pipes inherited
5195 from its parent. */
5196 if (current_fds_to_close)
5197 {
5198 close_fd_bitmap (current_fds_to_close);
5199 current_fds_to_close = (struct fd_bitmap *)NULL;
5200 }
5201
5202 #if defined (HAVE_DEV_FD)
5203 /* Make sure we close the parent's end of the pipe and clear the slot
5204 in the fd list so it is not closed later, if reallocated by, for
5205 instance, pipe(2). */
5206 close (parent_pipe_fd);
5207 dev_fd_list[parent_pipe_fd] = 0;
5208 #endif /* HAVE_DEV_FD */
5209
5210 /* subshells shouldn't have this flag, which controls using the temporary
5211 environment for variable lookups. */
5212 expanding_redir = 0;
5213
5214 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
5215
5216 #if !defined (HAVE_DEV_FD)
5217 /* Make sure we close the named pipe in the child before we exit. */
5218 close (open_for_read_in_child ? 0 : 1);
5219 #endif /* !HAVE_DEV_FD */
5220
5221 last_command_exit_value = result;
5222 result = run_exit_trap ();
5223 exit (result);
5224 /*NOTREACHED*/
5225 }
5226 #endif /* PROCESS_SUBSTITUTION */
5227
5228 /***********************************/
5229 /* */
5230 /* Command Substitution */
5231 /* */
5232 /***********************************/
5233
5234 static char *
5235 read_comsub (fd, quoted, rflag)
5236 int fd, quoted;
5237 int *rflag;
5238 {
5239 char *istring, buf[128], *bufp, *s;
5240 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
5241 ssize_t bufn;
5242
5243 istring = (char *)NULL;
5244 istring_index = istring_size = bufn = tflag = 0;
5245
5246 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5247 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
5248
5249 /* Read the output of the command through the pipe. This may need to be
5250 changed to understand multibyte characters in the future. */
5251 while (1)
5252 {
5253 if (fd < 0)
5254 break;
5255 if (--bufn <= 0)
5256 {
5257 bufn = zread (fd, buf, sizeof (buf));
5258 if (bufn <= 0)
5259 break;
5260 bufp = buf;
5261 }
5262 c = *bufp++;
5263
5264 if (c == 0)
5265 {
5266 #if 0
5267 internal_warning ("read_comsub: ignored null byte in input");
5268 #endif
5269 continue;
5270 }
5271
5272 /* Add the character to ISTRING, possibly after resizing it. */
5273 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5274
5275 /* This is essentially quote_string inline */
5276 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5277 istring[istring_index++] = CTLESC;
5278 /* Escape CTLESC and CTLNUL in the output to protect those characters
5279 from the rest of the word expansions (word splitting and globbing.)
5280 This is essentially quote_escapes inline. */
5281 else if (skip_ctlesc == 0 && c == CTLESC)
5282 {
5283 tflag |= W_HASCTLESC;
5284 istring[istring_index++] = CTLESC;
5285 }
5286 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
5287 istring[istring_index++] = CTLESC;
5288
5289 istring[istring_index++] = c;
5290
5291 #if 0
5292 #if defined (__CYGWIN__)
5293 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5294 {
5295 istring_index--;
5296 istring[istring_index - 1] = '\n';
5297 }
5298 #endif
5299 #endif
5300 }
5301
5302 if (istring)
5303 istring[istring_index] = '\0';
5304
5305 /* If we read no output, just return now and save ourselves some
5306 trouble. */
5307 if (istring_index == 0)
5308 {
5309 FREE (istring);
5310 if (rflag)
5311 *rflag = tflag;
5312 return (char *)NULL;
5313 }
5314
5315 /* Strip trailing newlines from the output of the command. */
5316 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5317 {
5318 while (istring_index > 0)
5319 {
5320 if (istring[istring_index - 1] == '\n')
5321 {
5322 --istring_index;
5323
5324 /* If the newline was quoted, remove the quoting char. */
5325 if (istring[istring_index - 1] == CTLESC)
5326 --istring_index;
5327 }
5328 else
5329 break;
5330 }
5331 istring[istring_index] = '\0';
5332 }
5333 else
5334 strip_trailing (istring, istring_index - 1, 1);
5335
5336 if (rflag)
5337 *rflag = tflag;
5338 return istring;
5339 }
5340
5341 /* Perform command substitution on STRING. This returns a WORD_DESC * with the
5342 contained string possibly quoted. */
5343 WORD_DESC *
5344 command_substitute (string, quoted)
5345 char *string;
5346 int quoted;
5347 {
5348 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
5349 char *istring;
5350 int result, fildes[2], function_value, pflags, rc, tflag;
5351 WORD_DESC *ret;
5352
5353 istring = (char *)NULL;
5354
5355 /* Don't fork () if there is no need to. In the case of no command to
5356 run, just return NULL. */
5357 if (!string || !*string || (string[0] == '\n' && !string[1]))
5358 return ((WORD_DESC *)NULL);
5359
5360 if (wordexp_only && read_but_dont_execute)
5361 {
5362 last_command_exit_value = EX_WEXPCOMSUB;
5363 jump_to_top_level (EXITPROG);
5364 }
5365
5366 /* We're making the assumption here that the command substitution will
5367 eventually run a command from the file system. Since we'll run
5368 maybe_make_export_env in this subshell before executing that command,
5369 the parent shell and any other shells it starts will have to remake
5370 the environment. If we make it before we fork, other shells won't
5371 have to. Don't bother if we have any temporary variable assignments,
5372 though, because the export environment will be remade after this
5373 command completes anyway, but do it if all the words to be expanded
5374 are variable assignments. */
5375 if (subst_assign_varlist == 0 || garglist == 0)
5376 maybe_make_export_env (); /* XXX */
5377
5378 /* Flags to pass to parse_and_execute() */
5379 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
5380
5381 /* Pipe the output of executing STRING into the current shell. */
5382 if (pipe (fildes) < 0)
5383 {
5384 sys_error (_("cannot make pipe for command substitution"));
5385 goto error_exit;
5386 }
5387
5388 old_pid = last_made_pid;
5389 #if defined (JOB_CONTROL)
5390 old_pipeline_pgrp = pipeline_pgrp;
5391 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
5392 if ((subshell_environment & SUBSHELL_PIPE) == 0)
5393 pipeline_pgrp = shell_pgrp;
5394 cleanup_the_pipeline ();
5395 #endif /* JOB_CONTROL */
5396
5397 old_async_pid = last_asynchronous_pid;
5398 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
5399 last_asynchronous_pid = old_async_pid;
5400
5401 if (pid == 0)
5402 {
5403 /* Reset the signal handlers in the child, but don't free the
5404 trap strings. Set a flag noting that we have to free the
5405 trap strings if we run trap to change a signal disposition. */
5406 reset_signal_handlers ();
5407 subshell_environment |= SUBSHELL_RESETTRAP;
5408 }
5409
5410 #if defined (JOB_CONTROL)
5411 /* XXX DO THIS ONLY IN PARENT ? XXX */
5412 set_sigchld_handler ();
5413 stop_making_children ();
5414 if (pid != 0)
5415 pipeline_pgrp = old_pipeline_pgrp;
5416 #else
5417 stop_making_children ();
5418 #endif /* JOB_CONTROL */
5419
5420 if (pid < 0)
5421 {
5422 sys_error (_("cannot make child for command substitution"));
5423 error_exit:
5424
5425 last_made_pid = old_pid;
5426
5427 FREE (istring);
5428 close (fildes[0]);
5429 close (fildes[1]);
5430 return ((WORD_DESC *)NULL);
5431 }
5432
5433 if (pid == 0)
5434 {
5435 set_sigint_handler (); /* XXX */
5436
5437 free_pushed_string_input ();
5438
5439 if (dup2 (fildes[1], 1) < 0)
5440 {
5441 sys_error (_("command_substitute: cannot duplicate pipe as fd 1"));
5442 exit (EXECUTION_FAILURE);
5443 }
5444
5445 /* If standard output is closed in the parent shell
5446 (such as after `exec >&-'), file descriptor 1 will be
5447 the lowest available file descriptor, and end up in
5448 fildes[0]. This can happen for stdin and stderr as well,
5449 but stdout is more important -- it will cause no output
5450 to be generated from this command. */
5451 if ((fildes[1] != fileno (stdin)) &&
5452 (fildes[1] != fileno (stdout)) &&
5453 (fildes[1] != fileno (stderr)))
5454 close (fildes[1]);
5455
5456 if ((fildes[0] != fileno (stdin)) &&
5457 (fildes[0] != fileno (stdout)) &&
5458 (fildes[0] != fileno (stderr)))
5459 close (fildes[0]);
5460
5461 #ifdef __CYGWIN__
5462 /* Let stdio know the fd may have changed from text to binary mode, and
5463 make sure to preserve stdout line buffering. */
5464 freopen (NULL, "w", stdout);
5465 sh_setlinebuf (stdout);
5466 #endif /* __CYGWIN__ */
5467
5468 /* The currently executing shell is not interactive. */
5469 interactive = 0;
5470
5471 /* This is a subshell environment. */
5472 subshell_environment |= SUBSHELL_COMSUB;
5473
5474 /* When not in POSIX mode, command substitution does not inherit
5475 the -e flag. */
5476 if (posixly_correct == 0)
5477 {
5478 builtin_ignoring_errexit = 0;
5479 change_flag ('e', FLAG_OFF);
5480 set_shellopts ();
5481 }
5482
5483 remove_quoted_escapes (string);
5484
5485 startup_state = 2; /* see if we can avoid a fork */
5486 /* Give command substitution a place to jump back to on failure,
5487 so we don't go back up to main (). */
5488 result = setjmp_nosigs (top_level);
5489
5490 /* If we're running a command substitution inside a shell function,
5491 trap `return' so we don't return from the function in the subshell
5492 and go off to never-never land. */
5493 if (result == 0 && return_catch_flag)
5494 function_value = setjmp_nosigs (return_catch);
5495 else
5496 function_value = 0;
5497
5498 if (result == ERREXIT)
5499 rc = last_command_exit_value;
5500 else if (result == EXITPROG)
5501 rc = last_command_exit_value;
5502 else if (result)
5503 rc = EXECUTION_FAILURE;
5504 else if (function_value)
5505 rc = return_catch_value;
5506 else
5507 {
5508 subshell_level++;
5509 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
5510 subshell_level--;
5511 }
5512
5513 last_command_exit_value = rc;
5514 rc = run_exit_trap ();
5515 #if defined (PROCESS_SUBSTITUTION)
5516 unlink_fifo_list ();
5517 #endif
5518 exit (rc);
5519 }
5520 else
5521 {
5522 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5523 close_pgrp_pipe ();
5524 #endif /* JOB_CONTROL && PGRP_PIPE */
5525
5526 close (fildes[1]);
5527
5528 tflag = 0;
5529 istring = read_comsub (fildes[0], quoted, &tflag);
5530
5531 close (fildes[0]);
5532
5533 current_command_subst_pid = pid;
5534 last_command_exit_value = wait_for (pid);
5535 last_command_subst_pid = pid;
5536 last_made_pid = old_pid;
5537
5538 #if defined (JOB_CONTROL)
5539 /* If last_command_exit_value > 128, then the substituted command
5540 was terminated by a signal. If that signal was SIGINT, then send
5541 SIGINT to ourselves. This will break out of loops, for instance. */
5542 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
5543 kill (getpid (), SIGINT);
5544
5545 /* wait_for gives the terminal back to shell_pgrp. If some other
5546 process group should have it, give it away to that group here.
5547 pipeline_pgrp is non-zero only while we are constructing a
5548 pipeline, so what we are concerned about is whether or not that
5549 pipeline was started in the background. A pipeline started in
5550 the background should never get the tty back here. */
5551 if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0)
5552 give_terminal_to (pipeline_pgrp, 0);
5553 #endif /* JOB_CONTROL */
5554
5555 ret = alloc_word_desc ();
5556 ret->word = istring;
5557 ret->flags = tflag;
5558
5559 return ret;
5560 }
5561 }
5562
5563 /********************************************************
5564 * *
5565 * Utility functions for parameter expansion *
5566 * *
5567 ********************************************************/
5568
5569 #if defined (ARRAY_VARS)
5570
5571 static arrayind_t
5572 array_length_reference (s)
5573 char *s;
5574 {
5575 int len;
5576 arrayind_t ind;
5577 char *akey;
5578 char *t, c;
5579 ARRAY *array;
5580 HASH_TABLE *h;
5581 SHELL_VAR *var;
5582
5583 var = array_variable_part (s, &t, &len);
5584
5585 /* If unbound variables should generate an error, report one and return
5586 failure. */
5587 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
5588 {
5589 c = *--t;
5590 *t = '\0';
5591 last_command_exit_value = EXECUTION_FAILURE;
5592 err_unboundvar (s);
5593 *t = c;
5594 return (-1);
5595 }
5596 else if (var == 0 || invisible_p (var))
5597 return 0;
5598
5599 /* We support a couple of expansions for variables that are not arrays.
5600 We'll return the length of the value for v[0], and 1 for v[@] or
5601 v[*]. Return 0 for everything else. */
5602
5603 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
5604 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
5605
5606 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
5607 {
5608 if (assoc_p (var))
5609 return (h ? assoc_num_elements (h) : 0);
5610 else if (array_p (var))
5611 return (array ? array_num_elements (array) : 0);
5612 else
5613 return (var_isset (var) ? 1 : 0);
5614 }
5615
5616 if (assoc_p (var))
5617 {
5618 t[len - 1] = '\0';
5619 akey = expand_assignment_string_to_string (t, 0); /* [ */
5620 t[len - 1] = ']';
5621 if (akey == 0 || *akey == 0)
5622 {
5623 err_badarraysub (t);
5624 FREE (akey);
5625 return (-1);
5626 }
5627 t = assoc_reference (assoc_cell (var), akey);
5628 free (akey);
5629 }
5630 else
5631 {
5632 ind = array_expand_index (var, t, len);
5633 /* negative subscripts to indexed arrays count back from end */
5634 if (var && array_p (var) && ind < 0)
5635 ind = array_max_index (array_cell (var)) + 1 + ind;
5636 if (ind < 0)
5637 {
5638 err_badarraysub (t);
5639 return (-1);
5640 }
5641 if (array_p (var))
5642 t = array_reference (array, ind);
5643 else
5644 t = (ind == 0) ? value_cell (var) : (char *)NULL;
5645 }
5646
5647 len = MB_STRLEN (t);
5648 return (len);
5649 }
5650 #endif /* ARRAY_VARS */
5651
5652 static int
5653 valid_brace_expansion_word (name, var_is_special)
5654 char *name;
5655 int var_is_special;
5656 {
5657 if (DIGIT (*name) && all_digits (name))
5658 return 1;
5659 else if (var_is_special)
5660 return 1;
5661 #if defined (ARRAY_VARS)
5662 else if (valid_array_reference (name))
5663 return 1;
5664 #endif /* ARRAY_VARS */
5665 else if (legal_identifier (name))
5666 return 1;
5667 else
5668 return 0;
5669 }
5670
5671 static int
5672 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
5673 char *name;
5674 int quoted;
5675 int *quoted_dollar_atp, *contains_dollar_at;
5676 {
5677 char *temp1;
5678
5679 if (name == 0)
5680 {
5681 if (quoted_dollar_atp)
5682 *quoted_dollar_atp = 0;
5683 if (contains_dollar_at)
5684 *contains_dollar_at = 0;
5685 return 0;
5686 }
5687
5688 /* check for $@ and $* */
5689 if (name[0] == '@' && name[1] == 0)
5690 {
5691 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5692 *quoted_dollar_atp = 1;
5693 if (contains_dollar_at)
5694 *contains_dollar_at = 1;
5695 return 1;
5696 }
5697 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
5698 {
5699 if (contains_dollar_at)
5700 *contains_dollar_at = 1;
5701 return 1;
5702 }
5703
5704 /* Now check for ${array[@]} and ${array[*]} */
5705 #if defined (ARRAY_VARS)
5706 else if (valid_array_reference (name))
5707 {
5708 temp1 = mbschr (name, '[');
5709 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
5710 {
5711 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
5712 *quoted_dollar_atp = 1;
5713 if (contains_dollar_at)
5714 *contains_dollar_at = 1;
5715 return 1;
5716 } /* [ */
5717 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
5718 which should result in separate words even when IFS is unset. */
5719 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
5720 {
5721 if (contains_dollar_at)
5722 *contains_dollar_at = 1;
5723 return 1;
5724 }
5725 }
5726 #endif
5727 return 0;
5728 }
5729
5730 /* Parameter expand NAME, and return a new string which is the expansion,
5731 or NULL if there was no expansion.
5732 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
5733 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
5734 NAME was found inside of a double-quoted expression. */
5735 static WORD_DESC *
5736 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
5737 char *name;
5738 int var_is_special, quoted, pflags;
5739 arrayind_t *indp;
5740 {
5741 WORD_DESC *ret;
5742 char *temp, *tt;
5743 intmax_t arg_index;
5744 SHELL_VAR *var;
5745 int atype, rflags;
5746 arrayind_t ind;
5747
5748 ret = 0;
5749 temp = 0;
5750 rflags = 0;
5751
5752 if (indp)
5753 *indp = INTMAX_MIN;
5754
5755 /* Handle multiple digit arguments, as in ${11}. */
5756 if (legal_number (name, &arg_index))
5757 {
5758 tt = get_dollar_var_value (arg_index);
5759 if (tt)
5760 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5761 ? quote_string (tt)
5762 : quote_escapes (tt);
5763 else
5764 temp = (char *)NULL;
5765 FREE (tt);
5766 }
5767 else if (var_is_special) /* ${@} */
5768 {
5769 int sindex;
5770 tt = (char *)xmalloc (2 + strlen (name));
5771 tt[sindex = 0] = '$';
5772 strcpy (tt + 1, name);
5773
5774 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
5775 (int *)NULL, (int *)NULL, pflags);
5776 free (tt);
5777 }
5778 #if defined (ARRAY_VARS)
5779 else if (valid_array_reference (name))
5780 {
5781 expand_arrayref:
5782 /* XXX - does this leak if name[@] or name[*]? */
5783 if (pflags & PF_ASSIGNRHS)
5784 {
5785 temp = array_variable_name (name, &tt, (int *)0);
5786 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
5787 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, 0, &atype, &ind);
5788 else
5789 temp = array_value (name, quoted, 0, &atype, &ind);
5790 }
5791 else
5792 temp = array_value (name, quoted, 0, &atype, &ind);
5793 if (atype == 0 && temp)
5794 {
5795 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5796 ? quote_string (temp)
5797 : quote_escapes (temp);
5798 rflags |= W_ARRAYIND;
5799 if (indp)
5800 *indp = ind;
5801 }
5802 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5803 rflags |= W_HASQUOTEDNULL;
5804 }
5805 #endif
5806 else if (var = find_variable (name))
5807 {
5808 if (var_isset (var) && invisible_p (var) == 0)
5809 {
5810 #if defined (ARRAY_VARS)
5811 if (assoc_p (var))
5812 temp = assoc_reference (assoc_cell (var), "0");
5813 else if (array_p (var))
5814 temp = array_reference (array_cell (var), 0);
5815 else
5816 temp = value_cell (var);
5817 #else
5818 temp = value_cell (var);
5819 #endif
5820
5821 if (temp)
5822 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
5823 ? quote_string (temp)
5824 : quote_escapes (temp);
5825 }
5826 else
5827 temp = (char *)NULL;
5828 }
5829 else if (var = find_variable_last_nameref (name))
5830 {
5831 temp = nameref_cell (var);
5832 #if defined (ARRAY_VARS)
5833 /* Handle expanding nameref whose value is x[n] */
5834 if (temp && *temp && valid_array_reference (temp))
5835 {
5836 name = temp;
5837 goto expand_arrayref;
5838 }
5839 else
5840 #endif
5841 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
5842 if (temp && *temp && legal_identifier (temp) == 0)
5843 {
5844 last_command_exit_value = EXECUTION_FAILURE;
5845 report_error (_("%s: invalid variable name for name reference"), temp);
5846 temp = &expand_param_error;
5847 }
5848 else
5849 temp = (char *)NULL;
5850 }
5851 else
5852 temp = (char *)NULL;
5853
5854 if (ret == 0)
5855 {
5856 ret = alloc_word_desc ();
5857 ret->word = temp;
5858 ret->flags |= rflags;
5859 }
5860 return ret;
5861 }
5862
5863 static char *
5864 parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
5865 char *name;
5866 int var_is_special, quoted, find_nameref;
5867 {
5868 char *temp, *t;
5869 WORD_DESC *w;
5870 SHELL_VAR *v;
5871
5872 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name)) &&
5873 nameref_p (v) && (t = nameref_cell (v)) && *t)
5874 return (savestring (t));
5875
5876 /* If var_is_special == 0, and name is not an array reference, this does
5877 more expansion than necessary. It should really look up the variable's
5878 value and not try to expand it. */
5879 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
5880 t = w->word;
5881 /* Have to dequote here if necessary */
5882 if (t)
5883 {
5884 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
5885 ? dequote_string (t)
5886 : dequote_escapes (t);
5887 free (t);
5888 t = temp;
5889 }
5890 dispose_word_desc (w);
5891
5892 return t;
5893 }
5894
5895 /* Expand an indirect reference to a variable: ${!NAME} expands to the
5896 value of the variable whose name is the value of NAME. */
5897 static WORD_DESC *
5898 parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
5899 char *name;
5900 int var_is_special, quoted;
5901 int *quoted_dollar_atp, *contains_dollar_at;
5902 {
5903 char *temp, *t;
5904 WORD_DESC *w;
5905 SHELL_VAR *v;
5906
5907 /* See if it's a nameref first, behave in ksh93-compatible fashion.
5908 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
5909 bash performs an indirect lookup on foo[0] and expands the result;
5910 ksh93 expands bar[0]. We could do that here -- there are enough usable
5911 primitives to do that -- but do not at this point. */
5912 if (var_is_special == 0 && (v = find_variable_last_nameref (name)))
5913 {
5914 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
5915 {
5916 w = alloc_word_desc ();
5917 w->word = savestring (t);
5918 w->flags = 0;
5919 return w;
5920 }
5921 }
5922
5923 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
5924
5925 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
5926 if (t == 0)
5927 return (WORD_DESC *)NULL;
5928
5929 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
5930 free (t);
5931
5932 return w;
5933 }
5934
5935 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
5936 depending on the value of C, the separating character. C can be one of
5937 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
5938 between double quotes. */
5939 static WORD_DESC *
5940 parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat)
5941 char *name, *value;
5942 int c, quoted, *qdollaratp, *hasdollarat;
5943 {
5944 WORD_DESC *w;
5945 WORD_LIST *l;
5946 char *t, *t1, *temp;
5947 int hasdol;
5948
5949 /* If the entire expression is between double quotes, we want to treat
5950 the value as a double-quoted string, with the exception that we strip
5951 embedded unescaped double quotes (for sh backwards compatibility). */
5952 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
5953 {
5954 hasdol = 0;
5955 temp = string_extract_double_quoted (value, &hasdol, 1);
5956 }
5957 else
5958 temp = value;
5959
5960 w = alloc_word_desc ();
5961 hasdol = 0;
5962 /* XXX was 0 not quoted */
5963 l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL)
5964 : (WORD_LIST *)0;
5965 if (hasdollarat)
5966 *hasdollarat = hasdol || (l && l->next);
5967 if (temp != value)
5968 free (temp);
5969 if (l)
5970 {
5971 /* The expansion of TEMP returned something. We need to treat things
5972 slightly differently if HASDOL is non-zero. If we have "$@", the
5973 individual words have already been quoted. We need to turn them
5974 into a string with the words separated by the first character of
5975 $IFS without any additional quoting, so string_list_dollar_at won't
5976 do the right thing. We use string_list_dollar_star instead. */
5977 temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l);
5978
5979 /* If l->next is not null, we know that TEMP contained "$@", since that
5980 is the only expansion that creates more than one word. */
5981 if (qdollaratp && ((hasdol && quoted) || l->next))
5982 *qdollaratp = 1;
5983 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
5984 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
5985 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
5986 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5987 (which is more paranoia than anything else), we need to return the
5988 quoted null string and set the flags to indicate it. */
5989 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
5990 {
5991 w->flags |= W_HASQUOTEDNULL;
5992 }
5993 dispose_words (l);
5994 }
5995 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol)
5996 {
5997 /* The brace expansion occurred between double quotes and there was
5998 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
5999 it does not expand to anything. In this case, we want to return
6000 a quoted empty string. */
6001 temp = make_quoted_char ('\0');
6002 w->flags |= W_HASQUOTEDNULL;
6003 }
6004 else
6005 temp = (char *)NULL;
6006
6007 if (c == '-' || c == '+')
6008 {
6009 w->word = temp;
6010 return w;
6011 }
6012
6013 /* c == '=' */
6014 t = temp ? savestring (temp) : savestring ("");
6015 t1 = dequote_string (t);
6016 free (t);
6017 #if defined (ARRAY_VARS)
6018 if (valid_array_reference (name))
6019 assign_array_element (name, t1, 0);
6020 else
6021 #endif /* ARRAY_VARS */
6022 bind_variable (name, t1, 0);
6023 #if 0
6024 if (STREQ (name, "IFS") == 0)
6025 #endif
6026 stupidly_hack_special_variables (name);
6027
6028 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6029 free (temp);
6030
6031 w->word = t1;
6032 return w;
6033 }
6034
6035 /* Deal with the right hand side of a ${name:?value} expansion in the case
6036 that NAME is null or not set. If VALUE is non-null it is expanded and
6037 used as the error message to print, otherwise a standard message is
6038 printed. */
6039 static void
6040 parameter_brace_expand_error (name, value)
6041 char *name, *value;
6042 {
6043 WORD_LIST *l;
6044 char *temp;
6045
6046 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
6047 if (value && *value)
6048 {
6049 l = expand_string (value, 0);
6050 temp = string_list (l);
6051 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6052 FREE (temp);
6053 dispose_words (l);
6054 }
6055 else
6056 report_error (_("%s: parameter null or not set"), name);
6057
6058 /* Free the data we have allocated during this expansion, since we
6059 are about to longjmp out. */
6060 free (name);
6061 FREE (value);
6062 }
6063
6064 /* Return 1 if NAME is something for which parameter_brace_expand_length is
6065 OK to do. */
6066 static int
6067 valid_length_expression (name)
6068 char *name;
6069 {
6070 return (name[1] == '\0' || /* ${#} */
6071 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6072 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
6073 #if defined (ARRAY_VARS)
6074 valid_array_reference (name + 1) || /* ${#a[7]} */
6075 #endif
6076 legal_identifier (name + 1)); /* ${#PS1} */
6077 }
6078
6079 /* Handle the parameter brace expansion that requires us to return the
6080 length of a parameter. */
6081 static intmax_t
6082 parameter_brace_expand_length (name)
6083 char *name;
6084 {
6085 char *t, *newname;
6086 intmax_t number, arg_index;
6087 WORD_LIST *list;
6088 #if defined (ARRAY_VARS)
6089 SHELL_VAR *var;
6090 #endif
6091
6092 if (name[1] == '\0') /* ${#} */
6093 number = number_of_args ();
6094 else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */
6095 number = number_of_args ();
6096 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
6097 {
6098 /* Take the lengths of some of the shell's special parameters. */
6099 switch (name[1])
6100 {
6101 case '-':
6102 t = which_set_flags ();
6103 break;
6104 case '?':
6105 t = itos (last_command_exit_value);
6106 break;
6107 case '$':
6108 t = itos (dollar_dollar_pid);
6109 break;
6110 case '!':
6111 if (last_asynchronous_pid == NO_PID)
6112 t = (char *)NULL; /* XXX - error if set -u set? */
6113 else
6114 t = itos (last_asynchronous_pid);
6115 break;
6116 case '#':
6117 t = itos (number_of_args ());
6118 break;
6119 }
6120 number = STRLEN (t);
6121 FREE (t);
6122 }
6123 #if defined (ARRAY_VARS)
6124 else if (valid_array_reference (name + 1))
6125 number = array_length_reference (name + 1);
6126 #endif /* ARRAY_VARS */
6127 else
6128 {
6129 number = 0;
6130
6131 if (legal_number (name + 1, &arg_index)) /* ${#1} */
6132 {
6133 t = get_dollar_var_value (arg_index);
6134 if (t == 0 && unbound_vars_is_error)
6135 return INTMAX_MIN;
6136 number = MB_STRLEN (t);
6137 FREE (t);
6138 }
6139 #if defined (ARRAY_VARS)
6140 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
6141 {
6142 if (assoc_p (var))
6143 t = assoc_reference (assoc_cell (var), "0");
6144 else
6145 t = array_reference (array_cell (var), 0);
6146 if (t == 0 && unbound_vars_is_error)
6147 return INTMAX_MIN;
6148 number = MB_STRLEN (t);
6149 }
6150 #endif
6151 else /* ${#PS1} */
6152 {
6153 newname = savestring (name);
6154 newname[0] = '$';
6155 list = expand_string (newname, Q_DOUBLE_QUOTES);
6156 t = list ? string_list (list) : (char *)NULL;
6157 free (newname);
6158 if (list)
6159 dispose_words (list);
6160
6161 number = t ? MB_STRLEN (t) : 0;
6162 FREE (t);
6163 }
6164 }
6165
6166 return (number);
6167 }
6168
6169 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6170 so we do some ad-hoc parsing of an arithmetic expression to find
6171 the first DELIM, instead of using strchr(3). Two rules:
6172 1. If the substring contains a `(', read until closing `)'.
6173 2. If the substring contains a `?', read past one `:' for each `?'.
6174 */
6175
6176 static char *
6177 skiparith (substr, delim)
6178 char *substr;
6179 int delim;
6180 {
6181 size_t sublen;
6182 int skipcol, pcount, i;
6183 DECLARE_MBSTATE;
6184
6185 sublen = strlen (substr);
6186 i = skipcol = pcount = 0;
6187 while (substr[i])
6188 {
6189 /* Balance parens */
6190 if (substr[i] == LPAREN)
6191 {
6192 pcount++;
6193 i++;
6194 continue;
6195 }
6196 if (substr[i] == RPAREN && pcount)
6197 {
6198 pcount--;
6199 i++;
6200 continue;
6201 }
6202 if (pcount)
6203 {
6204 ADVANCE_CHAR (substr, sublen, i);
6205 continue;
6206 }
6207
6208 /* Skip one `:' for each `?' */
6209 if (substr[i] == ':' && skipcol)
6210 {
6211 skipcol--;
6212 i++;
6213 continue;
6214 }
6215 if (substr[i] == delim)
6216 break;
6217 if (substr[i] == '?')
6218 {
6219 skipcol++;
6220 i++;
6221 continue;
6222 }
6223 ADVANCE_CHAR (substr, sublen, i);
6224 }
6225
6226 return (substr + i);
6227 }
6228
6229 /* Verify and limit the start and end of the desired substring. If
6230 VTYPE == 0, a regular shell variable is being used; if it is 1,
6231 then the positional parameters are being used; if it is 2, then
6232 VALUE is really a pointer to an array variable that should be used.
6233 Return value is 1 if both values were OK, 0 if there was a problem
6234 with an invalid expression, or -1 if the values were out of range. */
6235 static int
6236 verify_substring_values (v, value, substr, vtype, e1p, e2p)
6237 SHELL_VAR *v;
6238 char *value, *substr;
6239 int vtype;
6240 intmax_t *e1p, *e2p;
6241 {
6242 char *t, *temp1, *temp2;
6243 arrayind_t len;
6244 int expok;
6245 #if defined (ARRAY_VARS)
6246 ARRAY *a;
6247 HASH_TABLE *h;
6248 #endif
6249
6250 /* duplicate behavior of strchr(3) */
6251 t = skiparith (substr, ':');
6252 if (*t && *t == ':')
6253 *t = '\0';
6254 else
6255 t = (char *)0;
6256
6257 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
6258 *e1p = evalexp (temp1, &expok);
6259 free (temp1);
6260 if (expok == 0)
6261 return (0);
6262
6263 len = -1; /* paranoia */
6264 switch (vtype)
6265 {
6266 case VT_VARIABLE:
6267 case VT_ARRAYMEMBER:
6268 len = MB_STRLEN (value);
6269 break;
6270 case VT_POSPARMS:
6271 len = number_of_args () + 1;
6272 if (*e1p == 0)
6273 len++; /* add one arg if counting from $0 */
6274 break;
6275 #if defined (ARRAY_VARS)
6276 case VT_ARRAYVAR:
6277 /* For arrays, the first value deals with array indices. Negative
6278 offsets count from one past the array's maximum index. Associative
6279 arrays treat the number of elements as the maximum index. */
6280 if (assoc_p (v))
6281 {
6282 h = assoc_cell (v);
6283 len = assoc_num_elements (h) + (*e1p < 0);
6284 }
6285 else
6286 {
6287 a = (ARRAY *)value;
6288 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
6289 }
6290 break;
6291 #endif
6292 }
6293
6294 if (len == -1) /* paranoia */
6295 return -1;
6296
6297 if (*e1p < 0) /* negative offsets count from end */
6298 *e1p += len;
6299
6300 if (*e1p > len || *e1p < 0)
6301 return (-1);
6302
6303 #if defined (ARRAY_VARS)
6304 /* For arrays, the second offset deals with the number of elements. */
6305 if (vtype == VT_ARRAYVAR)
6306 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
6307 #endif
6308
6309 if (t)
6310 {
6311 t++;
6312 temp2 = savestring (t);
6313 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
6314 free (temp2);
6315 t[-1] = ':';
6316 *e2p = evalexp (temp1, &expok);
6317 free (temp1);
6318 if (expok == 0)
6319 return (0);
6320 #if 1
6321 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
6322 #else
6323 /* bash-4.3: allow positional parameter length < 0 to count backwards
6324 from end of positional parameters */
6325 if (vtype == VT_ARRAYVAR && *e2p < 0)
6326 #endif
6327 {
6328 internal_error (_("%s: substring expression < 0"), t);
6329 return (0);
6330 }
6331 #if defined (ARRAY_VARS)
6332 /* In order to deal with sparse arrays, push the intelligence about how
6333 to deal with the number of elements desired down to the array-
6334 specific functions. */
6335 if (vtype != VT_ARRAYVAR)
6336 #endif
6337 {
6338 if (*e2p < 0)
6339 {
6340 *e2p += len;
6341 if (*e2p < 0 || *e2p < *e1p)
6342 {
6343 internal_error (_("%s: substring expression < 0"), t);
6344 return (0);
6345 }
6346 }
6347 else
6348 *e2p += *e1p; /* want E2 chars starting at E1 */
6349 if (*e2p > len)
6350 *e2p = len;
6351 }
6352 }
6353 else
6354 *e2p = len;
6355
6356 return (1);
6357 }
6358
6359 /* Return the type of variable specified by VARNAME (simple variable,
6360 positional param, or array variable). Also return the value specified
6361 by VARNAME (value of a variable or a reference to an array element).
6362 QUOTED is the standard description of quoting state, using Q_* defines.
6363 FLAGS is currently a set of flags to pass to array_value. If IND is
6364 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
6365 passed to array_value so the array index is not computed again.
6366 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
6367 characters in the value are quoted with CTLESC and takes appropriate
6368 steps. For convenience, *VALP is set to the dequoted VALUE. */
6369 static int
6370 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
6371 char *varname, *value;
6372 arrayind_t ind;
6373 int quoted, flags;
6374 SHELL_VAR **varp;
6375 char **valp;
6376 {
6377 int vtype, want_indir;
6378 char *temp, *vname;
6379 WORD_DESC *wd;
6380 #if defined (ARRAY_VARS)
6381 SHELL_VAR *v;
6382 #endif
6383 arrayind_t lind;
6384
6385 want_indir = *varname == '!' &&
6386 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
6387 || VALID_INDIR_PARAM (varname[1]));
6388 if (want_indir)
6389 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
6390 else
6391 vname = varname;
6392
6393 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
6394 vtype = (vname[0] == '@' || vname[0] == '*') && vname[1] == '\0';
6395 if (vtype == VT_POSPARMS && vname[0] == '*')
6396 vtype |= VT_STARSUB;
6397 *varp = (SHELL_VAR *)NULL;
6398
6399 #if defined (ARRAY_VARS)
6400 if (valid_array_reference (vname))
6401 {
6402 v = array_variable_part (vname, &temp, (int *)0);
6403 /* If we want to signal array_value to use an already-computed index,
6404 set LIND to that index */
6405 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
6406 if (v && invisible_p (v))
6407 {
6408 vtype = VT_ARRAYMEMBER;
6409 *varp = (SHELL_VAR *)NULL;
6410 *valp = (char *)NULL;
6411 }
6412 if (v && (array_p (v) || assoc_p (v)))
6413 { /* [ */
6414 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
6415 {
6416 /* Callers have to differentiate between indexed and associative */
6417 vtype = VT_ARRAYVAR;
6418 if (temp[0] == '*')
6419 vtype |= VT_STARSUB;
6420 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
6421 }
6422 else
6423 {
6424 vtype = VT_ARRAYMEMBER;
6425 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6426 }
6427 *varp = v;
6428 }
6429 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
6430 {
6431 vtype = VT_VARIABLE;
6432 *varp = v;
6433 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6434 *valp = dequote_string (value);
6435 else
6436 *valp = dequote_escapes (value);
6437 }
6438 else
6439 {
6440 vtype = VT_ARRAYMEMBER;
6441 *varp = v;
6442 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
6443 }
6444 }
6445 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
6446 {
6447 vtype = VT_ARRAYMEMBER;
6448 *varp = v;
6449 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
6450 }
6451 else
6452 #endif
6453 {
6454 if (value && vtype == VT_VARIABLE)
6455 {
6456 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6457 *valp = dequote_string (value);
6458 else
6459 *valp = dequote_escapes (value);
6460 }
6461 else
6462 *valp = value;
6463 }
6464
6465 if (want_indir)
6466 free (vname);
6467
6468 return vtype;
6469 }
6470
6471 /******************************************************/
6472 /* */
6473 /* Functions to extract substrings of variable values */
6474 /* */
6475 /******************************************************/
6476
6477 #if defined (HANDLE_MULTIBYTE)
6478 /* Character-oriented rather than strictly byte-oriented substrings. S and
6479 E, rather being strict indices into STRING, indicate character (possibly
6480 multibyte character) positions that require calculation.
6481 Used by the ${param:offset[:length]} expansion. */
6482 static char *
6483 mb_substring (string, s, e)
6484 char *string;
6485 int s, e;
6486 {
6487 char *tt;
6488 int start, stop, i, slen;
6489 DECLARE_MBSTATE;
6490
6491 start = 0;
6492 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
6493 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
6494
6495 i = s;
6496 while (string[start] && i--)
6497 ADVANCE_CHAR (string, slen, start);
6498 stop = start;
6499 i = e - s;
6500 while (string[stop] && i--)
6501 ADVANCE_CHAR (string, slen, stop);
6502 tt = substring (string, start, stop);
6503 return tt;
6504 }
6505 #endif
6506
6507 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
6508 is `@', use the positional parameters; otherwise, use the value of
6509 VARNAME. If VARNAME is an array variable, use the array elements. */
6510
6511 static char *
6512 parameter_brace_substring (varname, value, ind, substr, quoted, flags)
6513 char *varname, *value;
6514 int ind;
6515 char *substr;
6516 int quoted, flags;
6517 {
6518 intmax_t e1, e2;
6519 int vtype, r, starsub;
6520 char *temp, *val, *tt, *oname;
6521 SHELL_VAR *v;
6522
6523 if (value == 0)
6524 return ((char *)NULL);
6525
6526 oname = this_command_name;
6527 this_command_name = varname;
6528
6529 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6530 if (vtype == -1)
6531 {
6532 this_command_name = oname;
6533 return ((char *)NULL);
6534 }
6535
6536 starsub = vtype & VT_STARSUB;
6537 vtype &= ~VT_STARSUB;
6538
6539 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
6540 this_command_name = oname;
6541 if (r <= 0)
6542 {
6543 if (vtype == VT_VARIABLE)
6544 FREE (val);
6545 return ((r == 0) ? &expand_param_error : (char *)NULL);
6546 }
6547
6548 switch (vtype)
6549 {
6550 case VT_VARIABLE:
6551 case VT_ARRAYMEMBER:
6552 #if defined (HANDLE_MULTIBYTE)
6553 if (MB_CUR_MAX > 1)
6554 tt = mb_substring (val, e1, e2);
6555 else
6556 #endif
6557 tt = substring (val, e1, e2);
6558
6559 if (vtype == VT_VARIABLE)
6560 FREE (val);
6561 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6562 temp = quote_string (tt);
6563 else
6564 temp = tt ? quote_escapes (tt) : (char *)NULL;
6565 FREE (tt);
6566 break;
6567 case VT_POSPARMS:
6568 tt = pos_params (varname, e1, e2, quoted);
6569 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
6570 {
6571 temp = tt ? quote_escapes (tt) : (char *)NULL;
6572 FREE (tt);
6573 }
6574 else
6575 temp = tt;
6576 break;
6577 #if defined (ARRAY_VARS)
6578 case VT_ARRAYVAR:
6579 if (assoc_p (v))
6580 /* we convert to list and take first e2 elements starting at e1th
6581 element -- officially undefined for now */
6582 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
6583 else
6584 /* We want E2 to be the number of elements desired (arrays can be sparse,
6585 so verify_substring_values just returns the numbers specified and we
6586 rely on array_subrange to understand how to deal with them). */
6587 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
6588 /* array_subrange now calls array_quote_escapes as appropriate, so the
6589 caller no longer needs to. */
6590 break;
6591 #endif
6592 default:
6593 temp = (char *)NULL;
6594 }
6595
6596 return temp;
6597 }
6598
6599 /****************************************************************/
6600 /* */
6601 /* Functions to perform pattern substitution on variable values */
6602 /* */
6603 /****************************************************************/
6604
6605 static int
6606 shouldexp_replacement (s)
6607 char *s;
6608 {
6609 register char *p;
6610
6611 for (p = s; p && *p; p++)
6612 {
6613 if (*p == '\\')
6614 p++;
6615 else if (*p == '&')
6616 return 1;
6617 }
6618 return 0;
6619 }
6620
6621 char *
6622 pat_subst (string, pat, rep, mflags)
6623 char *string, *pat, *rep;
6624 int mflags;
6625 {
6626 char *ret, *s, *e, *str, *rstr, *mstr;
6627 int rsize, rptr, l, replen, mtype, rxpand, rslen, mlen;
6628
6629 if (string == 0)
6630 return (savestring (""));
6631
6632 mtype = mflags & MATCH_TYPEMASK;
6633
6634 #if 0 /* bash-4.2 ? */
6635 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
6636 #else
6637 rxpand = 0;
6638 #endif
6639
6640 /* Special cases:
6641 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
6642 * with REP and return the result.
6643 * 2. A null pattern with mtype == MATCH_END means to append REP to
6644 * STRING and return the result.
6645 * These don't understand or process `&' in the replacement string.
6646 */
6647 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
6648 {
6649 replen = STRLEN (rep);
6650 l = STRLEN (string);
6651 ret = (char *)xmalloc (replen + l + 2);
6652 if (replen == 0)
6653 strcpy (ret, string);
6654 else if (mtype == MATCH_BEG)
6655 {
6656 strcpy (ret, rep);
6657 strcpy (ret + replen, string);
6658 }
6659 else
6660 {
6661 strcpy (ret, string);
6662 strcpy (ret + l, rep);
6663 }
6664 return (ret);
6665 }
6666
6667 ret = (char *)xmalloc (rsize = 64);
6668 ret[0] = '\0';
6669
6670 for (replen = STRLEN (rep), rptr = 0, str = string;;)
6671 {
6672 if (match_pattern (str, pat, mtype, &s, &e) == 0)
6673 break;
6674 l = s - str;
6675
6676 if (rxpand)
6677 {
6678 int x;
6679 mlen = e - s;
6680 mstr = xmalloc (mlen + 1);
6681 for (x = 0; x < mlen; x++)
6682 mstr[x] = s[x];
6683 mstr[mlen] = '\0';
6684 rstr = strcreplace (rep, '&', mstr, 0);
6685 rslen = strlen (rstr);
6686 }
6687 else
6688 {
6689 rstr = rep;
6690 rslen = replen;
6691 }
6692
6693 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
6694
6695 /* OK, now copy the leading unmatched portion of the string (from
6696 str to s) to ret starting at rptr (the current offset). Then copy
6697 the replacement string at ret + rptr + (s - str). Increment
6698 rptr (if necessary) and str and go on. */
6699 if (l)
6700 {
6701 strncpy (ret + rptr, str, l);
6702 rptr += l;
6703 }
6704 if (replen)
6705 {
6706 strncpy (ret + rptr, rstr, rslen);
6707 rptr += rslen;
6708 }
6709 str = e; /* e == end of match */
6710
6711 if (rstr != rep)
6712 free (rstr);
6713
6714 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
6715 break;
6716
6717 if (s == e)
6718 {
6719 /* On a zero-length match, make sure we copy one character, since
6720 we increment one character to avoid infinite recursion. */
6721 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
6722 ret[rptr++] = *str++;
6723 e++; /* avoid infinite recursion on zero-length match */
6724 }
6725 }
6726
6727 /* Now copy the unmatched portion of the input string */
6728 if (str && *str)
6729 {
6730 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
6731 strcpy (ret + rptr, str);
6732 }
6733 else
6734 ret[rptr] = '\0';
6735
6736 return ret;
6737 }
6738
6739 /* Do pattern match and replacement on the positional parameters. */
6740 static char *
6741 pos_params_pat_subst (string, pat, rep, mflags)
6742 char *string, *pat, *rep;
6743 int mflags;
6744 {
6745 WORD_LIST *save, *params;
6746 WORD_DESC *w;
6747 char *ret;
6748 int pchar, qflags;
6749
6750 save = params = list_rest_of_args ();
6751 if (save == 0)
6752 return ((char *)NULL);
6753
6754 for ( ; params; params = params->next)
6755 {
6756 ret = pat_subst (params->word->word, pat, rep, mflags);
6757 w = alloc_word_desc ();
6758 w->word = ret ? ret : savestring ("");
6759 dispose_word (params->word);
6760 params->word = w;
6761 }
6762
6763 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6764 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6765
6766 ret = string_list_pos_params (pchar, save, qflags);
6767
6768 dispose_words (save);
6769
6770 return (ret);
6771 }
6772
6773 /* Perform pattern substitution on VALUE, which is the expansion of
6774 VARNAME. PATSUB is an expression supplying the pattern to match
6775 and the string to substitute. QUOTED is a flags word containing
6776 the type of quoting currently in effect. */
6777 static char *
6778 parameter_brace_patsub (varname, value, ind, patsub, quoted, flags)
6779 char *varname, *value;
6780 int ind;
6781 char *patsub;
6782 int quoted, flags;
6783 {
6784 int vtype, mflags, starsub, delim;
6785 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
6786 SHELL_VAR *v;
6787
6788 if (value == 0)
6789 return ((char *)NULL);
6790
6791 this_command_name = varname;
6792
6793 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6794 if (vtype == -1)
6795 return ((char *)NULL);
6796
6797 starsub = vtype & VT_STARSUB;
6798 vtype &= ~VT_STARSUB;
6799
6800 mflags = 0;
6801 /* PATSUB is never NULL when this is called. */
6802 if (*patsub == '/')
6803 {
6804 mflags |= MATCH_GLOBREP;
6805 patsub++;
6806 }
6807
6808 /* Malloc this because expand_string_if_necessary or one of the expansion
6809 functions in its call chain may free it on a substitution error. */
6810 lpatsub = savestring (patsub);
6811
6812 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6813 mflags |= MATCH_QUOTED;
6814
6815 if (starsub)
6816 mflags |= MATCH_STARSUB;
6817
6818 /* If the pattern starts with a `/', make sure we skip over it when looking
6819 for the replacement delimiter. */
6820 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
6821 if (lpatsub[delim] == '/')
6822 {
6823 lpatsub[delim] = 0;
6824 rep = lpatsub + delim + 1;
6825 }
6826 else
6827 rep = (char *)NULL;
6828
6829 if (rep && *rep == '\0')
6830 rep = (char *)NULL;
6831
6832 /* Perform the same expansions on the pattern as performed by the
6833 pattern removal expansions. */
6834 pat = getpattern (lpatsub, quoted, 1);
6835
6836 if (rep)
6837 {
6838 /* We want to perform quote removal on the expanded replacement even if
6839 the entire expansion is double-quoted because the parser and string
6840 extraction functions treated quotes in the replacement string as
6841 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
6842 if (shell_compatibility_level > 42)
6843 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
6844 /* This is the bash-4.2 code. */
6845 else if ((mflags & MATCH_QUOTED) == 0)
6846 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
6847 else
6848 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
6849 }
6850
6851 /* ksh93 doesn't allow the match specifier to be a part of the expanded
6852 pattern. This is an extension. Make sure we don't anchor the pattern
6853 at the beginning or end of the string if we're doing global replacement,
6854 though. */
6855 p = pat;
6856 if (mflags & MATCH_GLOBREP)
6857 mflags |= MATCH_ANY;
6858 else if (pat && pat[0] == '#')
6859 {
6860 mflags |= MATCH_BEG;
6861 p++;
6862 }
6863 else if (pat && pat[0] == '%')
6864 {
6865 mflags |= MATCH_END;
6866 p++;
6867 }
6868 else
6869 mflags |= MATCH_ANY;
6870
6871 /* OK, we now want to substitute REP for PAT in VAL. If
6872 flags & MATCH_GLOBREP is non-zero, the substitution is done
6873 everywhere, otherwise only the first occurrence of PAT is
6874 replaced. The pattern matching code doesn't understand
6875 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
6876 values passed in (VT_VARIABLE) so the pattern substitution
6877 code works right. We need to requote special chars after
6878 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
6879 other cases if QUOTED == 0, since the posparams and arrays
6880 indexed by * or @ do special things when QUOTED != 0. */
6881
6882 switch (vtype)
6883 {
6884 case VT_VARIABLE:
6885 case VT_ARRAYMEMBER:
6886 temp = pat_subst (val, p, rep, mflags);
6887 if (vtype == VT_VARIABLE)
6888 FREE (val);
6889 if (temp)
6890 {
6891 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
6892 free (temp);
6893 temp = tt;
6894 }
6895 break;
6896 case VT_POSPARMS:
6897 temp = pos_params_pat_subst (val, p, rep, mflags);
6898 if (temp && (mflags & MATCH_QUOTED) == 0)
6899 {
6900 tt = quote_escapes (temp);
6901 free (temp);
6902 temp = tt;
6903 }
6904 break;
6905 #if defined (ARRAY_VARS)
6906 case VT_ARRAYVAR:
6907 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
6908 : array_patsub (array_cell (v), p, rep, mflags);
6909 /* Don't call quote_escapes anymore; array_patsub calls
6910 array_quote_escapes as appropriate before adding the
6911 space separators; ditto for assoc_patsub. */
6912 break;
6913 #endif
6914 }
6915
6916 FREE (pat);
6917 FREE (rep);
6918 free (lpatsub);
6919
6920 return temp;
6921 }
6922
6923 /****************************************************************/
6924 /* */
6925 /* Functions to perform case modification on variable values */
6926 /* */
6927 /****************************************************************/
6928
6929 /* Do case modification on the positional parameters. */
6930
6931 static char *
6932 pos_params_modcase (string, pat, modop, mflags)
6933 char *string, *pat;
6934 int modop;
6935 int mflags;
6936 {
6937 WORD_LIST *save, *params;
6938 WORD_DESC *w;
6939 char *ret;
6940 int pchar, qflags;
6941
6942 save = params = list_rest_of_args ();
6943 if (save == 0)
6944 return ((char *)NULL);
6945
6946 for ( ; params; params = params->next)
6947 {
6948 ret = sh_modcase (params->word->word, pat, modop);
6949 w = alloc_word_desc ();
6950 w->word = ret ? ret : savestring ("");
6951 dispose_word (params->word);
6952 params->word = w;
6953 }
6954
6955 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
6956 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
6957
6958 ret = string_list_pos_params (pchar, save, qflags);
6959 dispose_words (save);
6960
6961 return (ret);
6962 }
6963
6964 /* Perform case modification on VALUE, which is the expansion of
6965 VARNAME. MODSPEC is an expression supplying the type of modification
6966 to perform. QUOTED is a flags word containing the type of quoting
6967 currently in effect. */
6968 static char *
6969 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
6970 char *varname, *value;
6971 int ind, modspec;
6972 char *patspec;
6973 int quoted, flags;
6974 {
6975 int vtype, starsub, modop, mflags, x;
6976 char *val, *temp, *pat, *p, *lpat, *tt;
6977 SHELL_VAR *v;
6978
6979 if (value == 0)
6980 return ((char *)NULL);
6981
6982 this_command_name = varname;
6983
6984 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
6985 if (vtype == -1)
6986 return ((char *)NULL);
6987
6988 starsub = vtype & VT_STARSUB;
6989 vtype &= ~VT_STARSUB;
6990
6991 modop = 0;
6992 mflags = 0;
6993 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6994 mflags |= MATCH_QUOTED;
6995 if (starsub)
6996 mflags |= MATCH_STARSUB;
6997
6998 p = patspec;
6999 if (modspec == '^')
7000 {
7001 x = p && p[0] == modspec;
7002 modop = x ? CASE_UPPER : CASE_UPFIRST;
7003 p += x;
7004 }
7005 else if (modspec == ',')
7006 {
7007 x = p && p[0] == modspec;
7008 modop = x ? CASE_LOWER : CASE_LOWFIRST;
7009 p += x;
7010 }
7011 else if (modspec == '~')
7012 {
7013 x = p && p[0] == modspec;
7014 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7015 p += x;
7016 }
7017
7018 lpat = p ? savestring (p) : 0;
7019 /* Perform the same expansions on the pattern as performed by the
7020 pattern removal expansions. FOR LATER */
7021 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7022
7023 /* OK, now we do the case modification. */
7024 switch (vtype)
7025 {
7026 case VT_VARIABLE:
7027 case VT_ARRAYMEMBER:
7028 temp = sh_modcase (val, pat, modop);
7029 if (vtype == VT_VARIABLE)
7030 FREE (val);
7031 if (temp)
7032 {
7033 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7034 free (temp);
7035 temp = tt;
7036 }
7037 break;
7038
7039 case VT_POSPARMS:
7040 temp = pos_params_modcase (val, pat, modop, mflags);
7041 if (temp && (mflags & MATCH_QUOTED) == 0)
7042 {
7043 tt = quote_escapes (temp);
7044 free (temp);
7045 temp = tt;
7046 }
7047 break;
7048
7049 #if defined (ARRAY_VARS)
7050 case VT_ARRAYVAR:
7051 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7052 : array_modcase (array_cell (v), pat, modop, mflags);
7053 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7054 as appropriate before adding the space separators; ditto for
7055 assoc_modcase. */
7056 break;
7057 #endif
7058 }
7059
7060 FREE (pat);
7061 free (lpat);
7062
7063 return temp;
7064 }
7065
7066 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7067 any occur, this must be a nested command substitution, so return 0.
7068 Otherwise, return 1. A valid arithmetic expression must always have a
7069 ( before a matching ), so any cases where there are more right parens
7070 means that this must not be an arithmetic expression, though the parser
7071 will not accept it without a balanced total number of parens. */
7072 static int
7073 chk_arithsub (s, len)
7074 const char *s;
7075 int len;
7076 {
7077 int i, count;
7078 DECLARE_MBSTATE;
7079
7080 i = count = 0;
7081 while (i < len)
7082 {
7083 if (s[i] == LPAREN)
7084 count++;
7085 else if (s[i] == RPAREN)
7086 {
7087 count--;
7088 if (count < 0)
7089 return 0;
7090 }
7091
7092 switch (s[i])
7093 {
7094 default:
7095 ADVANCE_CHAR (s, len, i);
7096 break;
7097
7098 case '\\':
7099 i++;
7100 if (s[i])
7101 ADVANCE_CHAR (s, len, i);
7102 break;
7103
7104 case '\'':
7105 i = skip_single_quoted (s, len, ++i);
7106 break;
7107
7108 case '"':
7109 i = skip_double_quoted ((char *)s, len, ++i);
7110 break;
7111 }
7112 }
7113
7114 return (count == 0);
7115 }
7116
7117 /****************************************************************/
7118 /* */
7119 /* Functions to perform parameter expansion on a string */
7120 /* */
7121 /****************************************************************/
7122
7123 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
7124 static WORD_DESC *
7125 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
7126 char *string;
7127 int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at, pflags;
7128 {
7129 int check_nullness, var_is_set, var_is_null, var_is_special;
7130 int want_substring, want_indir, want_patsub, want_casemod;
7131 char *name, *value, *temp, *temp1;
7132 WORD_DESC *tdesc, *ret;
7133 int t_index, sindex, c, tflag, modspec;
7134 intmax_t number;
7135 arrayind_t ind;
7136
7137 temp = temp1 = value = (char *)NULL;
7138 var_is_set = var_is_null = var_is_special = check_nullness = 0;
7139 want_substring = want_indir = want_patsub = want_casemod = 0;
7140
7141 sindex = *indexp;
7142 t_index = ++sindex;
7143 /* ${#var} doesn't have any of the other parameter expansions on it. */
7144 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
7145 name = string_extract (string, &t_index, "}", SX_VARNAME);
7146 else
7147 #if defined (CASEMOD_EXPANSIONS)
7148 /* To enable case-toggling expansions using the `~' operator character
7149 change the 1 to 0. */
7150 # if defined (CASEMOD_CAPCASE)
7151 name = string_extract (string, &t_index, "#%^,~:-=?+/}", SX_VARNAME);
7152 # else
7153 name = string_extract (string, &t_index, "#%^,:-=?+/}", SX_VARNAME);
7154 # endif /* CASEMOD_CAPCASE */
7155 #else
7156 name = string_extract (string, &t_index, "#%:-=?+/}", SX_VARNAME);
7157 #endif /* CASEMOD_EXPANSIONS */
7158
7159 ret = 0;
7160 tflag = 0;
7161
7162 ind = INTMAX_MIN;
7163
7164 /* If the name really consists of a special variable, then make sure
7165 that we have the entire name. We don't allow indirect references
7166 to special variables except `#', `?', `@' and `*'. */
7167 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
7168 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
7169 {
7170 t_index++;
7171 temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0);
7172 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
7173 *name = string[sindex];
7174 if (string[sindex] == '!')
7175 {
7176 /* indirect reference of $#, $?, $@, or $* */
7177 name[1] = string[sindex + 1];
7178 strcpy (name + 2, temp1);
7179 }
7180 else
7181 strcpy (name + 1, temp1);
7182 free (temp1);
7183 }
7184 sindex = t_index;
7185
7186 /* Find out what character ended the variable name. Then
7187 do the appropriate thing. */
7188 if (c = string[sindex])
7189 sindex++;
7190
7191 /* If c is followed by one of the valid parameter expansion
7192 characters, move past it as normal. If not, assume that
7193 a substring specification is being given, and do not move
7194 past it. */
7195 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
7196 {
7197 check_nullness++;
7198 if (c = string[sindex])
7199 sindex++;
7200 }
7201 else if (c == ':' && string[sindex] != RBRACE)
7202 want_substring = 1;
7203 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
7204 want_patsub = 1;
7205 #if defined (CASEMOD_EXPANSIONS)
7206 else if (c == '^' || c == ',' || c == '~')
7207 {
7208 modspec = c;
7209 want_casemod = 1;
7210 }
7211 #endif
7212
7213 /* Catch the valid and invalid brace expressions that made it through the
7214 tests above. */
7215 /* ${#-} is a valid expansion and means to take the length of $-.
7216 Similarly for ${#?} and ${##}... */
7217 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7218 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
7219 {
7220 name = (char *)xrealloc (name, 3);
7221 name[1] = c;
7222 name[2] = '\0';
7223 c = string[sindex++];
7224 }
7225
7226 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
7227 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
7228 member (c, "%:=+/") && string[sindex] == RBRACE)
7229 {
7230 temp = (char *)NULL;
7231 goto bad_substitution;
7232 }
7233
7234 /* Indirect expansion begins with a `!'. A valid indirect expansion is
7235 either a variable name, one of the positional parameters or a special
7236 variable that expands to one of the positional parameters. */
7237 want_indir = *name == '!' &&
7238 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
7239 || VALID_INDIR_PARAM (name[1]));
7240
7241 /* Determine the value of this variable. */
7242
7243 /* Check for special variables, directly referenced. */
7244 if (SPECIAL_VAR (name, want_indir))
7245 var_is_special++;
7246
7247 /* Check for special expansion things, like the length of a parameter */
7248 if (*name == '#' && name[1])
7249 {
7250 /* If we are not pointing at the character just after the
7251 closing brace, then we haven't gotten all of the name.
7252 Since it begins with a special character, this is a bad
7253 substitution. Also check NAME for validity before trying
7254 to go on. */
7255 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
7256 {
7257 temp = (char *)NULL;
7258 goto bad_substitution;
7259 }
7260
7261 number = parameter_brace_expand_length (name);
7262 if (number == INTMAX_MIN && unbound_vars_is_error)
7263 {
7264 last_command_exit_value = EXECUTION_FAILURE;
7265 err_unboundvar (name+1);
7266 free (name);
7267 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7268 }
7269 free (name);
7270
7271 *indexp = sindex;
7272 if (number < 0)
7273 return (&expand_wdesc_error);
7274 else
7275 {
7276 ret = alloc_word_desc ();
7277 ret->word = itos (number);
7278 return ret;
7279 }
7280 }
7281
7282 /* ${@} is identical to $@. */
7283 if (name[0] == '@' && name[1] == '\0')
7284 {
7285 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7286 *quoted_dollar_atp = 1;
7287
7288 if (contains_dollar_at)
7289 *contains_dollar_at = 1;
7290
7291 tflag |= W_DOLLARAT;
7292 }
7293
7294 /* Process ${!PREFIX*} expansion. */
7295 if (want_indir && string[sindex - 1] == RBRACE &&
7296 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
7297 legal_variable_starter ((unsigned char) name[1]))
7298 {
7299 char **x;
7300 WORD_LIST *xlist;
7301
7302 temp1 = savestring (name + 1);
7303 number = strlen (temp1);
7304 temp1[number - 1] = '\0';
7305 x = all_variables_matching_prefix (temp1);
7306 xlist = strvec_to_word_list (x, 0, 0);
7307 if (string[sindex - 2] == '*')
7308 temp = string_list_dollar_star (xlist);
7309 else
7310 {
7311 temp = string_list_dollar_at (xlist, quoted);
7312 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7313 *quoted_dollar_atp = 1;
7314 if (contains_dollar_at)
7315 *contains_dollar_at = 1;
7316
7317 tflag |= W_DOLLARAT;
7318 }
7319 free (x);
7320 dispose_words (xlist);
7321 free (temp1);
7322 *indexp = sindex;
7323
7324 free (name);
7325
7326 ret = alloc_word_desc ();
7327 ret->word = temp;
7328 ret->flags = tflag; /* XXX */
7329 return ret;
7330 }
7331
7332 #if defined (ARRAY_VARS)
7333 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
7334 if (want_indir && string[sindex - 1] == RBRACE &&
7335 string[sindex - 2] == ']' && valid_array_reference (name+1))
7336 {
7337 char *x, *x1;
7338
7339 temp1 = savestring (name + 1);
7340 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
7341 FREE (x);
7342 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
7343 {
7344 temp = array_keys (temp1, quoted); /* handles assoc vars too */
7345 if (x1[0] == '@')
7346 {
7347 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7348 *quoted_dollar_atp = 1;
7349 if (contains_dollar_at)
7350 *contains_dollar_at = 1;
7351
7352 tflag |= W_DOLLARAT;
7353 }
7354
7355 free (temp1);
7356 *indexp = sindex;
7357
7358 ret = alloc_word_desc ();
7359 ret->word = temp;
7360 ret->flags = tflag; /* XXX */
7361 return ret;
7362 }
7363
7364 free (temp1);
7365 }
7366 #endif /* ARRAY_VARS */
7367
7368 /* Make sure that NAME is valid before trying to go on. */
7369 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
7370 var_is_special) == 0)
7371 {
7372 temp = (char *)NULL;
7373 goto bad_substitution;
7374 }
7375
7376 if (want_indir)
7377 {
7378 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
7379 /* Turn off the W_ARRAYIND flag because there is no way for this function
7380 to return the index we're supposed to be using. */
7381 if (tdesc && tdesc->flags)
7382 tdesc->flags &= ~W_ARRAYIND;
7383 }
7384 else
7385 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
7386
7387 if (tdesc)
7388 {
7389 temp = tdesc->word;
7390 tflag = tdesc->flags;
7391 dispose_word_desc (tdesc);
7392 }
7393 else
7394 temp = (char *)0;
7395
7396 if (temp == &expand_param_error || temp == &expand_param_fatal)
7397 {
7398 FREE (name);
7399 FREE (value);
7400 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
7401 }
7402
7403 #if defined (ARRAY_VARS)
7404 if (valid_array_reference (name))
7405 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at);
7406 #endif
7407
7408 var_is_set = temp != (char *)0;
7409 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
7410 /* XXX - this may not need to be restricted to special variables */
7411 if (check_nullness)
7412 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
7413
7414 /* Get the rest of the stuff inside the braces. */
7415 if (c && c != RBRACE)
7416 {
7417 /* Extract the contents of the ${ ... } expansion
7418 according to the Posix.2 rules. */
7419 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
7420 if (string[sindex] == RBRACE)
7421 sindex++;
7422 else
7423 goto bad_substitution;
7424 }
7425 else
7426 value = (char *)NULL;
7427
7428 *indexp = sindex;
7429
7430 /* All the cases where an expansion can possibly generate an unbound
7431 variable error. */
7432 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
7433 {
7434 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]))
7435 {
7436 last_command_exit_value = EXECUTION_FAILURE;
7437 err_unboundvar (name);
7438 FREE (value);
7439 FREE (temp);
7440 free (name);
7441 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7442 }
7443 }
7444
7445 /* If this is a substring spec, process it and add the result. */
7446 if (want_substring)
7447 {
7448 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7449 FREE (name);
7450 FREE (value);
7451 FREE (temp);
7452
7453 if (temp1 == &expand_param_error)
7454 return (&expand_wdesc_error);
7455 else if (temp1 == &expand_param_fatal)
7456 return (&expand_wdesc_fatal);
7457
7458 ret = alloc_word_desc ();
7459 ret->word = temp1;
7460 /* We test quoted_dollar_atp because we want variants with double-quoted
7461 "$@" to take a different code path. In fact, we make sure at the end
7462 of expand_word_internal that we're only looking at these flags if
7463 quoted_dollar_at == 0. */
7464 if (temp1 &&
7465 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7466 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7467 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7468 return ret;
7469 }
7470 else if (want_patsub)
7471 {
7472 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7473 FREE (name);
7474 FREE (value);
7475 FREE (temp);
7476
7477 if (temp1 == &expand_param_error)
7478 return (&expand_wdesc_error);
7479 else if (temp1 == &expand_param_fatal)
7480 return (&expand_wdesc_fatal);
7481
7482 ret = alloc_word_desc ();
7483 ret->word = temp1;
7484 if (temp1 &&
7485 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7486 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7487 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7488 return ret;
7489 }
7490 #if defined (CASEMOD_EXPANSIONS)
7491 else if (want_casemod)
7492 {
7493 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7494 FREE (name);
7495 FREE (value);
7496 FREE (temp);
7497
7498 if (temp1 == &expand_param_error)
7499 return (&expand_wdesc_error);
7500 else if (temp1 == &expand_param_fatal)
7501 return (&expand_wdesc_fatal);
7502
7503 ret = alloc_word_desc ();
7504 ret->word = temp1;
7505 if (temp1 &&
7506 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
7507 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7508 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7509 return ret;
7510 }
7511 #endif
7512
7513 /* Do the right thing based on which character ended the variable name. */
7514 switch (c)
7515 {
7516 default:
7517 case '\0':
7518 bad_substitution:
7519 last_command_exit_value = EXECUTION_FAILURE;
7520 report_error (_("%s: bad substitution"), string ? string : "??");
7521 FREE (value);
7522 FREE (temp);
7523 free (name);
7524 return &expand_wdesc_error;
7525
7526 case RBRACE:
7527 break;
7528
7529 case '#': /* ${param#[#]pattern} */
7530 case '%': /* ${param%[%]pattern} */
7531 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
7532 {
7533 FREE (value);
7534 break;
7535 }
7536 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
7537 free (temp);
7538 free (value);
7539 free (name);
7540
7541 ret = alloc_word_desc ();
7542 ret->word = temp1;
7543 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7544 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
7545 return ret;
7546
7547 case '-':
7548 case '=':
7549 case '?':
7550 case '+':
7551 if (var_is_set && var_is_null == 0)
7552 {
7553 /* If the operator is `+', we don't want the value of the named
7554 variable for anything, just the value of the right hand side. */
7555 if (c == '+')
7556 {
7557 /* XXX -- if we're double-quoted and the named variable is "$@",
7558 we want to turn off any special handling of "$@" --
7559 we're not using it, so whatever is on the rhs applies. */
7560 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7561 *quoted_dollar_atp = 0;
7562 if (contains_dollar_at)
7563 *contains_dollar_at = 0;
7564
7565 FREE (temp);
7566 if (value)
7567 {
7568 /* From Posix discussion on austin-group list. Issue 221
7569 requires that backslashes escaping `}' inside
7570 double-quoted ${...} be removed. */
7571 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7572 quoted |= Q_DOLBRACE;
7573 ret = parameter_brace_expand_rhs (name, value, c,
7574 quoted,
7575 quoted_dollar_atp,
7576 contains_dollar_at);
7577 /* XXX - fix up later, esp. noting presence of
7578 W_HASQUOTEDNULL in ret->flags */
7579 free (value);
7580 }
7581 else
7582 temp = (char *)NULL;
7583 }
7584 else
7585 {
7586 FREE (value);
7587 }
7588 /* Otherwise do nothing; just use the value in TEMP. */
7589 }
7590 else /* VAR not set or VAR is NULL. */
7591 {
7592 FREE (temp);
7593 temp = (char *)NULL;
7594 if (c == '=' && var_is_special)
7595 {
7596 last_command_exit_value = EXECUTION_FAILURE;
7597 report_error (_("$%s: cannot assign in this way"), name);
7598 free (name);
7599 free (value);
7600 return &expand_wdesc_error;
7601 }
7602 else if (c == '?')
7603 {
7604 parameter_brace_expand_error (name, value);
7605 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7606 }
7607 else if (c != '+')
7608 {
7609 /* XXX -- if we're double-quoted and the named variable is "$@",
7610 we want to turn off any special handling of "$@" --
7611 we're not using it, so whatever is on the rhs applies. */
7612 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
7613 *quoted_dollar_atp = 0;
7614 if (contains_dollar_at)
7615 *contains_dollar_at = 0;
7616
7617 /* From Posix discussion on austin-group list. Issue 221 requires
7618 that backslashes escaping `}' inside double-quoted ${...} be
7619 removed. */
7620 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7621 quoted |= Q_DOLBRACE;
7622 ret = parameter_brace_expand_rhs (name, value, c, quoted,
7623 quoted_dollar_atp,
7624 contains_dollar_at);
7625 /* XXX - fix up later, esp. noting presence of
7626 W_HASQUOTEDNULL in tdesc->flags */
7627 }
7628 free (value);
7629 }
7630
7631 break;
7632 }
7633 free (name);
7634
7635 if (ret == 0)
7636 {
7637 ret = alloc_word_desc ();
7638 ret->flags = tflag;
7639 ret->word = temp;
7640 }
7641 return (ret);
7642 }
7643
7644 /* Expand a single ${xxx} expansion. The braces are optional. When
7645 the braces are used, parameter_brace_expand() does the work,
7646 possibly calling param_expand recursively. */
7647 static WORD_DESC *
7648 param_expand (string, sindex, quoted, expanded_something,
7649 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
7650 pflags)
7651 char *string;
7652 int *sindex, quoted, *expanded_something, *contains_dollar_at;
7653 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
7654 {
7655 char *temp, *temp1, uerror[3];
7656 int zindex, t_index, expok;
7657 unsigned char c;
7658 intmax_t number;
7659 SHELL_VAR *var;
7660 WORD_LIST *list;
7661 WORD_DESC *tdesc, *ret;
7662 int tflag;
7663
7664 zindex = *sindex;
7665 c = string[++zindex];
7666
7667 temp = (char *)NULL;
7668 ret = tdesc = (WORD_DESC *)NULL;
7669 tflag = 0;
7670
7671 /* Do simple cases first. Switch on what follows '$'. */
7672 switch (c)
7673 {
7674 /* $0 .. $9? */
7675 case '0':
7676 case '1':
7677 case '2':
7678 case '3':
7679 case '4':
7680 case '5':
7681 case '6':
7682 case '7':
7683 case '8':
7684 case '9':
7685 temp1 = dollar_vars[TODIGIT (c)];
7686 if (unbound_vars_is_error && temp1 == (char *)NULL)
7687 {
7688 uerror[0] = '$';
7689 uerror[1] = c;
7690 uerror[2] = '\0';
7691 last_command_exit_value = EXECUTION_FAILURE;
7692 err_unboundvar (uerror);
7693 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7694 }
7695 if (temp1)
7696 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7697 ? quote_string (temp1)
7698 : quote_escapes (temp1);
7699 else
7700 temp = (char *)NULL;
7701
7702 break;
7703
7704 /* $$ -- pid of the invoking shell. */
7705 case '$':
7706 temp = itos (dollar_dollar_pid);
7707 break;
7708
7709 /* $# -- number of positional parameters. */
7710 case '#':
7711 temp = itos (number_of_args ());
7712 break;
7713
7714 /* $? -- return value of the last synchronous command. */
7715 case '?':
7716 temp = itos (last_command_exit_value);
7717 break;
7718
7719 /* $- -- flags supplied to the shell on invocation or by `set'. */
7720 case '-':
7721 temp = which_set_flags ();
7722 break;
7723
7724 /* $! -- Pid of the last asynchronous command. */
7725 case '!':
7726 /* If no asynchronous pids have been created, expand to nothing.
7727 If `set -u' has been executed, and no async processes have
7728 been created, this is an expansion error. */
7729 if (last_asynchronous_pid == NO_PID)
7730 {
7731 if (expanded_something)
7732 *expanded_something = 0;
7733 temp = (char *)NULL;
7734 if (unbound_vars_is_error)
7735 {
7736 uerror[0] = '$';
7737 uerror[1] = c;
7738 uerror[2] = '\0';
7739 last_command_exit_value = EXECUTION_FAILURE;
7740 err_unboundvar (uerror);
7741 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7742 }
7743 }
7744 else
7745 temp = itos (last_asynchronous_pid);
7746 break;
7747
7748 /* The only difference between this and $@ is when the arg is quoted. */
7749 case '*': /* `$*' */
7750 list = list_rest_of_args ();
7751
7752 #if 0
7753 /* According to austin-group posix proposal by Geoff Clare in
7754 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7755
7756 "The shell shall write a message to standard error and
7757 immediately exit when it tries to expand an unset parameter
7758 other than the '@' and '*' special parameters."
7759 */
7760
7761 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7762 {
7763 uerror[0] = '$';
7764 uerror[1] = '*';
7765 uerror[2] = '\0';
7766 last_command_exit_value = EXECUTION_FAILURE;
7767 err_unboundvar (uerror);
7768 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7769 }
7770 #endif
7771
7772 /* If there are no command-line arguments, this should just
7773 disappear if there are other characters in the expansion,
7774 even if it's quoted. */
7775 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
7776 temp = (char *)NULL;
7777 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
7778 {
7779 /* If we have "$*" we want to make a string of the positional
7780 parameters, separated by the first character of $IFS, and
7781 quote the whole string, including the separators. If IFS
7782 is unset, the parameters are separated by ' '; if $IFS is
7783 null, the parameters are concatenated. */
7784 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
7785 if (temp)
7786 {
7787 temp1 = quote_string (temp);
7788 if (*temp == 0)
7789 tflag |= W_HASQUOTEDNULL;
7790 free (temp);
7791 temp = temp1;
7792 }
7793 }
7794 else
7795 {
7796 /* We check whether or not we're eventually going to split $* here,
7797 for example when IFS is empty and we are processing the rhs of
7798 an assignment statement. In that case, we don't separate the
7799 arguments at all. Otherwise, if the $* is not quoted it is
7800 identical to $@ */
7801 # if defined (HANDLE_MULTIBYTE)
7802 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
7803 # else
7804 if (expand_no_split_dollar_star && ifs_firstc == 0)
7805 # endif
7806 temp = string_list_dollar_star (list);
7807 else
7808 {
7809 temp = string_list_dollar_at (list, quoted);
7810 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
7811 tflag |= W_SPLITSPACE;
7812 }
7813
7814 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
7815 *contains_dollar_at = 1;
7816 }
7817
7818 dispose_words (list);
7819 break;
7820
7821 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
7822 means that we have to turn quoting off after we split into
7823 the individually quoted arguments so that the final split
7824 on the first character of $IFS is still done. */
7825 case '@': /* `$@' */
7826 list = list_rest_of_args ();
7827
7828 #if 0
7829 /* According to austin-group posix proposal by Geoff Clare in
7830 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
7831
7832 "The shell shall write a message to standard error and
7833 immediately exit when it tries to expand an unset parameter
7834 other than the '@' and '*' special parameters."
7835 */
7836
7837 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
7838 {
7839 uerror[0] = '$';
7840 uerror[1] = '@';
7841 uerror[2] = '\0';
7842 last_command_exit_value = EXECUTION_FAILURE;
7843 err_unboundvar (uerror);
7844 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
7845 }
7846 #endif
7847
7848 /* We want to flag the fact that we saw this. We can't turn
7849 off quoting entirely, because other characters in the
7850 string might need it (consider "\"$@\""), but we need some
7851 way to signal that the final split on the first character
7852 of $IFS should be done, even though QUOTED is 1. */
7853 /* XXX - should this test include Q_PATQUOTE? */
7854 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
7855 *quoted_dollar_at_p = 1;
7856 if (contains_dollar_at)
7857 *contains_dollar_at = 1;
7858
7859 /* We want to separate the positional parameters with the first
7860 character of $IFS in case $IFS is something other than a space.
7861 We also want to make sure that splitting is done no matter what --
7862 according to POSIX.2, this expands to a list of the positional
7863 parameters no matter what IFS is set to. */
7864 /* XXX - what to do when in a context where word splitting is not
7865 performed? Even when IFS is not the default, posix seems to imply
7866 that we behave like unquoted $* ? Maybe we should use PF_NOSPLIT2
7867 here. */
7868 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted);
7869
7870 tflag |= W_DOLLARAT;
7871 dispose_words (list);
7872 break;
7873
7874 case LBRACE:
7875 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
7876 quoted_dollar_at_p,
7877 contains_dollar_at);
7878
7879 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
7880 return (tdesc);
7881 temp = tdesc ? tdesc->word : (char *)0;
7882
7883 /* XXX */
7884 /* Quoted nulls should be removed if there is anything else
7885 in the string. */
7886 /* Note that we saw the quoted null so we can add one back at
7887 the end of this function if there are no other characters
7888 in the string, discard TEMP, and go on. The exception to
7889 this is when we have "${@}" and $1 is '', since $@ needs
7890 special handling. */
7891 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
7892 {
7893 if (had_quoted_null_p)
7894 *had_quoted_null_p = 1;
7895 if (*quoted_dollar_at_p == 0)
7896 {
7897 free (temp);
7898 tdesc->word = temp = (char *)NULL;
7899 }
7900
7901 }
7902
7903 ret = tdesc;
7904 goto return0;
7905
7906 /* Do command or arithmetic substitution. */
7907 case LPAREN:
7908 /* We have to extract the contents of this paren substitution. */
7909 t_index = zindex + 1;
7910 temp = extract_command_subst (string, &t_index, 0);
7911 zindex = t_index;
7912
7913 /* For Posix.2-style `$(( ))' arithmetic substitution,
7914 extract the expression and pass it to the evaluator. */
7915 if (temp && *temp == LPAREN)
7916 {
7917 char *temp2;
7918 temp1 = temp + 1;
7919 temp2 = savestring (temp1);
7920 t_index = strlen (temp2) - 1;
7921
7922 if (temp2[t_index] != RPAREN)
7923 {
7924 free (temp2);
7925 goto comsub;
7926 }
7927
7928 /* Cut off ending `)' */
7929 temp2[t_index] = '\0';
7930
7931 if (chk_arithsub (temp2, t_index) == 0)
7932 {
7933 free (temp2);
7934 #if 0
7935 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
7936 #endif
7937 goto comsub;
7938 }
7939
7940 /* Expand variables found inside the expression. */
7941 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7942 free (temp2);
7943
7944 arithsub:
7945 /* No error messages. */
7946 this_command_name = (char *)NULL;
7947 number = evalexp (temp1, &expok);
7948 free (temp);
7949 free (temp1);
7950 if (expok == 0)
7951 {
7952 if (interactive_shell == 0 && posixly_correct)
7953 {
7954 last_command_exit_value = EXECUTION_FAILURE;
7955 return (&expand_wdesc_fatal);
7956 }
7957 else
7958 return (&expand_wdesc_error);
7959 }
7960 temp = itos (number);
7961 break;
7962 }
7963
7964 comsub:
7965 if (pflags & PF_NOCOMSUB)
7966 /* we need zindex+1 because string[zindex] == RPAREN */
7967 temp1 = substring (string, *sindex, zindex+1);
7968 else
7969 {
7970 tdesc = command_substitute (temp, quoted);
7971 temp1 = tdesc ? tdesc->word : (char *)NULL;
7972 if (tdesc)
7973 dispose_word_desc (tdesc);
7974 }
7975 FREE (temp);
7976 temp = temp1;
7977 break;
7978
7979 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
7980 away in a future bash release. */
7981 case '[':
7982 /* Extract the contents of this arithmetic substitution. */
7983 t_index = zindex + 1;
7984 temp = extract_arithmetic_subst (string, &t_index);
7985 zindex = t_index;
7986 if (temp == 0)
7987 {
7988 temp = savestring (string);
7989 if (expanded_something)
7990 *expanded_something = 0;
7991 goto return0;
7992 }
7993
7994 /* Do initial variable expansion. */
7995 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES);
7996
7997 goto arithsub;
7998
7999 default:
8000 /* Find the variable in VARIABLE_LIST. */
8001 temp = (char *)NULL;
8002
8003 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
8004 ;
8005 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
8006
8007 /* If this isn't a variable name, then just output the `$'. */
8008 if (temp1 == 0 || *temp1 == '\0')
8009 {
8010 FREE (temp1);
8011 temp = (char *)xmalloc (2);
8012 temp[0] = '$';
8013 temp[1] = '\0';
8014 if (expanded_something)
8015 *expanded_something = 0;
8016 goto return0;
8017 }
8018
8019 /* If the variable exists, return its value cell. */
8020 var = find_variable (temp1);
8021
8022 if (var && invisible_p (var) == 0 && var_isset (var))
8023 {
8024 #if defined (ARRAY_VARS)
8025 if (assoc_p (var) || array_p (var))
8026 {
8027 temp = array_p (var) ? array_reference (array_cell (var), 0)
8028 : assoc_reference (assoc_cell (var), "0");
8029 if (temp)
8030 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8031 ? quote_string (temp)
8032 : quote_escapes (temp);
8033 else if (unbound_vars_is_error)
8034 goto unbound_variable;
8035 }
8036 else
8037 #endif
8038 {
8039 temp = value_cell (var);
8040
8041 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8042 ? quote_string (temp)
8043 : quote_escapes (temp);
8044 }
8045
8046 free (temp1);
8047
8048 goto return0;
8049 }
8050 else if (var && (invisible_p (var) || var_isset (var) == 0))
8051 temp = (char *)NULL;
8052 else if ((var = find_variable_last_nameref (temp1)) && var_isset (var) && invisible_p (var) == 0)
8053 {
8054 temp = nameref_cell (var);
8055 #if defined (ARRAY_VARS)
8056 if (temp && *temp && valid_array_reference (temp))
8057 {
8058 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8059 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8060 return (tdesc);
8061 ret = tdesc;
8062 goto return0;
8063 }
8064 else
8065 #endif
8066 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8067 if (temp && *temp && legal_identifier (temp) == 0)
8068 {
8069 last_command_exit_value = EXECUTION_FAILURE;
8070 report_error (_("%s: invalid variable name for name reference"), temp);
8071 return (&expand_wdesc_error); /* XXX */
8072 }
8073 else
8074 temp = (char *)NULL;
8075 }
8076
8077 temp = (char *)NULL;
8078
8079 unbound_variable:
8080 if (unbound_vars_is_error)
8081 {
8082 last_command_exit_value = EXECUTION_FAILURE;
8083 err_unboundvar (temp1);
8084 }
8085 else
8086 {
8087 free (temp1);
8088 goto return0;
8089 }
8090
8091 free (temp1);
8092 last_command_exit_value = EXECUTION_FAILURE;
8093 return ((unbound_vars_is_error && interactive_shell == 0)
8094 ? &expand_wdesc_fatal
8095 : &expand_wdesc_error);
8096 }
8097
8098 if (string[zindex])
8099 zindex++;
8100
8101 return0:
8102 *sindex = zindex;
8103
8104 if (ret == 0)
8105 {
8106 ret = alloc_word_desc ();
8107 ret->flags = tflag; /* XXX */
8108 ret->word = temp;
8109 }
8110 return ret;
8111 }
8112
8113 /* Make a word list which is the result of parameter and variable
8114 expansion, command substitution, arithmetic substitution, and
8115 quote removal of WORD. Return a pointer to a WORD_LIST which is
8116 the result of the expansion. If WORD contains a null word, the
8117 word list returned is also null.
8118
8119 QUOTED contains flag values defined in shell.h.
8120
8121 ISEXP is used to tell expand_word_internal that the word should be
8122 treated as the result of an expansion. This has implications for
8123 how IFS characters in the word are treated.
8124
8125 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
8126 they point to an integer value which receives information about expansion.
8127 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
8128 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
8129 else zero.
8130
8131 This only does word splitting in the case of $@ expansion. In that
8132 case, we split on ' '. */
8133
8134 /* Values for the local variable quoted_state. */
8135 #define UNQUOTED 0
8136 #define PARTIALLY_QUOTED 1
8137 #define WHOLLY_QUOTED 2
8138
8139 static WORD_LIST *
8140 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
8141 WORD_DESC *word;
8142 int quoted, isexp;
8143 int *contains_dollar_at;
8144 int *expanded_something;
8145 {
8146 WORD_LIST *list;
8147 WORD_DESC *tword;
8148
8149 /* The intermediate string that we build while expanding. */
8150 char *istring;
8151
8152 /* The current size of the above object. */
8153 int istring_size;
8154
8155 /* Index into ISTRING. */
8156 int istring_index;
8157
8158 /* Temporary string storage. */
8159 char *temp, *temp1;
8160
8161 /* The text of WORD. */
8162 register char *string;
8163
8164 /* The size of STRING. */
8165 size_t string_size;
8166
8167 /* The index into STRING. */
8168 int sindex;
8169
8170 /* This gets 1 if we see a $@ while quoted. */
8171 int quoted_dollar_at;
8172
8173 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
8174 whether WORD contains no quoting characters, a partially quoted
8175 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
8176 int quoted_state;
8177
8178 /* State flags */
8179 int had_quoted_null;
8180 int has_dollar_at, temp_has_dollar_at;
8181 int split_on_spaces;
8182 int tflag;
8183 int pflags; /* flags passed to param_expand */
8184
8185 int assignoff; /* If assignment, offset of `=' */
8186
8187 register unsigned char c; /* Current character. */
8188 int t_index; /* For calls to string_extract_xxx. */
8189
8190 char twochars[2];
8191
8192 DECLARE_MBSTATE;
8193
8194 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
8195 istring[istring_index = 0] = '\0';
8196 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
8197 split_on_spaces = 0;
8198 quoted_state = UNQUOTED;
8199
8200 string = word->word;
8201 if (string == 0)
8202 goto finished_with_string;
8203 /* Don't need the string length for the SADD... and COPY_ macros unless
8204 multibyte characters are possible. */
8205 string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1;
8206
8207 if (contains_dollar_at)
8208 *contains_dollar_at = 0;
8209
8210 assignoff = -1;
8211
8212 /* Begin the expansion. */
8213
8214 for (sindex = 0; ;)
8215 {
8216 c = string[sindex];
8217
8218 /* Case on top-level character. */
8219 switch (c)
8220 {
8221 case '\0':
8222 goto finished_with_string;
8223
8224 case CTLESC:
8225 sindex++;
8226 #if HANDLE_MULTIBYTE
8227 if (MB_CUR_MAX > 1 && string[sindex])
8228 {
8229 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8230 }
8231 else
8232 #endif
8233 {
8234 temp = (char *)xmalloc (3);
8235 temp[0] = CTLESC;
8236 temp[1] = c = string[sindex];
8237 temp[2] = '\0';
8238 }
8239
8240 dollar_add_string:
8241 if (string[sindex])
8242 sindex++;
8243
8244 add_string:
8245 if (temp)
8246 {
8247 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
8248 temp = (char *)0;
8249 }
8250
8251 break;
8252
8253 #if defined (PROCESS_SUBSTITUTION)
8254 /* Process substitution. */
8255 case '<':
8256 case '>':
8257 {
8258 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
8259 {
8260 sindex--; /* add_character: label increments sindex */
8261 goto add_character;
8262 }
8263 else
8264 t_index = sindex + 1; /* skip past both '<' and LPAREN */
8265
8266 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
8267 sindex = t_index;
8268
8269 /* If the process substitution specification is `<()', we want to
8270 open the pipe for writing in the child and produce output; if
8271 it is `>()', we want to open the pipe for reading in the child
8272 and consume input. */
8273 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
8274
8275 FREE (temp1);
8276
8277 goto dollar_add_string;
8278 }
8279 #endif /* PROCESS_SUBSTITUTION */
8280
8281 case '=':
8282 /* Posix.2 section 3.6.1 says that tildes following `=' in words
8283 which are not assignment statements are not expanded. If the
8284 shell isn't in posix mode, though, we perform tilde expansion
8285 on `likely candidate' unquoted assignment statements (flags
8286 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
8287 contains an unquoted :~ or =~. Something to think about: we
8288 now have a flag that says to perform tilde expansion on arguments
8289 to `assignment builtins' like declare and export that look like
8290 assignment statements. We now do tilde expansion on such words
8291 even in POSIX mode. */
8292 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
8293 {
8294 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8295 goto add_ifs_character;
8296 else
8297 goto add_character;
8298 }
8299 /* If we're not in posix mode or forcing assignment-statement tilde
8300 expansion, note where the `=' appears in the word and prepare to
8301 do tilde expansion following the first `='. */
8302 if ((word->flags & W_ASSIGNMENT) &&
8303 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8304 assignoff == -1 && sindex > 0)
8305 assignoff = sindex;
8306 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
8307 word->flags |= W_ITILDE;
8308 #if 0
8309 else if ((word->flags & W_ASSIGNMENT) &&
8310 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
8311 string[sindex+1] == '~')
8312 word->flags |= W_ITILDE;
8313 #endif
8314 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8315 goto add_ifs_character;
8316 else
8317 goto add_character;
8318
8319 case ':':
8320 if (word->flags & W_NOTILDE)
8321 {
8322 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8323 goto add_ifs_character;
8324 else
8325 goto add_character;
8326 }
8327
8328 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
8329 string[sindex+1] == '~')
8330 word->flags |= W_ITILDE;
8331
8332 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
8333 goto add_ifs_character;
8334 else
8335 goto add_character;
8336
8337 case '~':
8338 /* If the word isn't supposed to be tilde expanded, or we're not
8339 at the start of a word or after an unquoted : or = in an
8340 assignment statement, we don't do tilde expansion. */
8341 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
8342 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
8343 (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8344 {
8345 word->flags &= ~W_ITILDE;
8346 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
8347 goto add_ifs_character;
8348 else
8349 goto add_character;
8350 }
8351
8352 if (word->flags & W_ASSIGNRHS)
8353 tflag = 2;
8354 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
8355 tflag = 1;
8356 else
8357 tflag = 0;
8358
8359 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
8360
8361 word->flags &= ~W_ITILDE;
8362
8363 if (temp && *temp && t_index > 0)
8364 {
8365 temp1 = bash_tilde_expand (temp, tflag);
8366 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
8367 {
8368 FREE (temp);
8369 FREE (temp1);
8370 goto add_character; /* tilde expansion failed */
8371 }
8372 free (temp);
8373 temp = temp1;
8374 sindex += t_index;
8375 goto add_quoted_string; /* XXX was add_string */
8376 }
8377 else
8378 {
8379 FREE (temp);
8380 goto add_character;
8381 }
8382
8383 case '$':
8384 if (expanded_something)
8385 *expanded_something = 1;
8386
8387 temp_has_dollar_at = 0;
8388 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
8389 if (word->flags & W_NOSPLIT2)
8390 pflags |= PF_NOSPLIT2;
8391 if (word->flags & W_ASSIGNRHS)
8392 pflags |= PF_ASSIGNRHS;
8393 tword = param_expand (string, &sindex, quoted, expanded_something,
8394 &temp_has_dollar_at, &quoted_dollar_at,
8395 &had_quoted_null, pflags);
8396 has_dollar_at += temp_has_dollar_at;
8397 split_on_spaces += (tword->flags & W_SPLITSPACE);
8398
8399 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
8400 {
8401 free (string);
8402 free (istring);
8403 return ((tword == &expand_wdesc_error) ? &expand_word_error
8404 : &expand_word_fatal);
8405 }
8406 if (contains_dollar_at && has_dollar_at)
8407 *contains_dollar_at = 1;
8408
8409 if (tword && (tword->flags & W_HASQUOTEDNULL))
8410 had_quoted_null = 1;
8411
8412 temp = tword ? tword->word : (char *)NULL;
8413 dispose_word_desc (tword);
8414
8415 /* Kill quoted nulls; we will add them back at the end of
8416 expand_word_internal if nothing else in the string */
8417 if (had_quoted_null && temp && QUOTED_NULL (temp))
8418 {
8419 FREE (temp);
8420 temp = (char *)NULL;
8421 }
8422
8423 goto add_string;
8424 break;
8425
8426 case '`': /* Backquoted command substitution. */
8427 {
8428 t_index = sindex++;
8429
8430 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
8431 /* The test of sindex against t_index is to allow bare instances of
8432 ` to pass through, for backwards compatibility. */
8433 if (temp == &extract_string_error || temp == &extract_string_fatal)
8434 {
8435 if (sindex - 1 == t_index)
8436 {
8437 sindex = t_index;
8438 goto add_character;
8439 }
8440 last_command_exit_value = EXECUTION_FAILURE;
8441 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
8442 free (string);
8443 free (istring);
8444 return ((temp == &extract_string_error) ? &expand_word_error
8445 : &expand_word_fatal);
8446 }
8447
8448 if (expanded_something)
8449 *expanded_something = 1;
8450
8451 if (word->flags & W_NOCOMSUB)
8452 /* sindex + 1 because string[sindex] == '`' */
8453 temp1 = substring (string, t_index, sindex + 1);
8454 else
8455 {
8456 de_backslash (temp);
8457 tword = command_substitute (temp, quoted);
8458 temp1 = tword ? tword->word : (char *)NULL;
8459 if (tword)
8460 dispose_word_desc (tword);
8461 }
8462 FREE (temp);
8463 temp = temp1;
8464 goto dollar_add_string;
8465 }
8466
8467 case '\\':
8468 if (string[sindex + 1] == '\n')
8469 {
8470 sindex += 2;
8471 continue;
8472 }
8473
8474 c = string[++sindex];
8475
8476 if (quoted & Q_HERE_DOCUMENT)
8477 tflag = CBSHDOC;
8478 else if (quoted & Q_DOUBLE_QUOTES)
8479 tflag = CBSDQUOTE;
8480 else
8481 tflag = 0;
8482
8483 /* From Posix discussion on austin-group list: Backslash escaping
8484 a } in ${...} is removed. Issue 0000221 */
8485 if ((quoted & Q_DOLBRACE) && c == RBRACE)
8486 {
8487 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8488 }
8489 /* This is the fix for " $@\ " */
8490 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
8491 {
8492 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8493 DEFAULT_ARRAY_SIZE);
8494 istring[istring_index++] = CTLESC;
8495 istring[istring_index++] = '\\';
8496 istring[istring_index] = '\0';
8497
8498 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8499 }
8500 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
8501 {
8502 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
8503 }
8504 else if (c == 0)
8505 {
8506 c = CTLNUL;
8507 sindex--; /* add_character: label increments sindex */
8508 goto add_character;
8509 }
8510 else
8511 {
8512 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
8513 }
8514
8515 sindex++;
8516 add_twochars:
8517 /* BEFORE jumping here, we need to increment sindex if appropriate */
8518 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
8519 DEFAULT_ARRAY_SIZE);
8520 istring[istring_index++] = twochars[0];
8521 istring[istring_index++] = twochars[1];
8522 istring[istring_index] = '\0';
8523
8524 break;
8525
8526 case '"':
8527 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8528 goto add_character;
8529
8530 t_index = ++sindex;
8531 temp = string_extract_double_quoted (string, &sindex, 0);
8532
8533 /* If the quotes surrounded the entire string, then the
8534 whole word was quoted. */
8535 quoted_state = (t_index == 1 && string[sindex] == '\0')
8536 ? WHOLLY_QUOTED
8537 : PARTIALLY_QUOTED;
8538
8539 if (temp && *temp)
8540 {
8541 tword = alloc_word_desc ();
8542 tword->word = temp;
8543
8544 temp = (char *)NULL;
8545
8546 temp_has_dollar_at = 0; /* XXX */
8547 /* Need to get W_HASQUOTEDNULL flag through this function. */
8548 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
8549 has_dollar_at += temp_has_dollar_at;
8550
8551 if (list == &expand_word_error || list == &expand_word_fatal)
8552 {
8553 free (istring);
8554 free (string);
8555 /* expand_word_internal has already freed temp_word->word
8556 for us because of the way it prints error messages. */
8557 tword->word = (char *)NULL;
8558 dispose_word (tword);
8559 return list;
8560 }
8561
8562 dispose_word (tword);
8563
8564 /* "$@" (a double-quoted dollar-at) expands into nothing,
8565 not even a NULL word, when there are no positional
8566 parameters. */
8567 if (list == 0 && has_dollar_at)
8568 {
8569 quoted_dollar_at++;
8570 break;
8571 }
8572
8573 /* If we get "$@", we know we have expanded something, so we
8574 need to remember it for the final split on $IFS. This is
8575 a special case; it's the only case where a quoted string
8576 can expand into more than one word. It's going to come back
8577 from the above call to expand_word_internal as a list with
8578 a single word, in which all characters are quoted and
8579 separated by blanks. What we want to do is to turn it back
8580 into a list for the next piece of code. */
8581 if (list)
8582 dequote_list (list);
8583
8584 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
8585 had_quoted_null = 1; /* XXX */
8586
8587 if (has_dollar_at)
8588 {
8589 quoted_dollar_at++;
8590 if (contains_dollar_at)
8591 *contains_dollar_at = 1;
8592 if (expanded_something)
8593 *expanded_something = 1;
8594 }
8595 }
8596 else
8597 {
8598 /* What we have is "". This is a minor optimization. */
8599 FREE (temp);
8600 list = (WORD_LIST *)NULL;
8601 }
8602
8603 /* The code above *might* return a list (consider the case of "$@",
8604 where it returns "$1", "$2", etc.). We can't throw away the
8605 rest of the list, and we have to make sure each word gets added
8606 as quoted. We test on tresult->next: if it is non-NULL, we
8607 quote the whole list, save it to a string with string_list, and
8608 add that string. We don't need to quote the results of this
8609 (and it would be wrong, since that would quote the separators
8610 as well), so we go directly to add_string. */
8611 if (list)
8612 {
8613 if (list->next)
8614 {
8615 /* Testing quoted_dollar_at makes sure that "$@" is
8616 split correctly when $IFS does not contain a space. */
8617 temp = quoted_dollar_at
8618 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES)
8619 : string_list (quote_list (list));
8620 dispose_words (list);
8621 goto add_string;
8622 }
8623 else
8624 {
8625 temp = savestring (list->word->word);
8626 tflag = list->word->flags;
8627 dispose_words (list);
8628
8629 /* If the string is not a quoted null string, we want
8630 to remove any embedded unquoted CTLNUL characters.
8631 We do not want to turn quoted null strings back into
8632 the empty string, though. We do this because we
8633 want to remove any quoted nulls from expansions that
8634 contain other characters. For example, if we have
8635 x"$*"y or "x$*y" and there are no positional parameters,
8636 the $* should expand into nothing. */
8637 /* We use the W_HASQUOTEDNULL flag to differentiate the
8638 cases: a quoted null character as above and when
8639 CTLNUL is contained in the (non-null) expansion
8640 of some variable. We use the had_quoted_null flag to
8641 pass the value through this function to its caller. */
8642 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
8643 remove_quoted_nulls (temp); /* XXX */
8644 }
8645 }
8646 else
8647 temp = (char *)NULL;
8648
8649 /* We do not want to add quoted nulls to strings that are only
8650 partially quoted; we can throw them away. The exception to
8651 this is when we are going to be performing word splitting,
8652 since we have to preserve a null argument if the next character
8653 will cause word splitting. */
8654 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
8655 continue;
8656
8657 add_quoted_string:
8658
8659 if (temp)
8660 {
8661 temp1 = temp;
8662 temp = quote_string (temp);
8663 free (temp1);
8664 goto add_string;
8665 }
8666 else
8667 {
8668 /* Add NULL arg. */
8669 c = CTLNUL;
8670 sindex--; /* add_character: label increments sindex */
8671 goto add_character;
8672 }
8673
8674 /* break; */
8675
8676 case '\'':
8677 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
8678 goto add_character;
8679
8680 t_index = ++sindex;
8681 temp = string_extract_single_quoted (string, &sindex);
8682
8683 /* If the entire STRING was surrounded by single quotes,
8684 then the string is wholly quoted. */
8685 quoted_state = (t_index == 1 && string[sindex] == '\0')
8686 ? WHOLLY_QUOTED
8687 : PARTIALLY_QUOTED;
8688
8689 /* If all we had was '', it is a null expansion. */
8690 if (*temp == '\0')
8691 {
8692 free (temp);
8693 temp = (char *)NULL;
8694 }
8695 else
8696 remove_quoted_escapes (temp); /* ??? */
8697
8698 /* We do not want to add quoted nulls to strings that are only
8699 partially quoted; such nulls are discarded. */
8700 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED))
8701 continue;
8702
8703 /* If we have a quoted null expansion, add a quoted NULL to istring. */
8704 if (temp == 0)
8705 {
8706 c = CTLNUL;
8707 sindex--; /* add_character: label increments sindex */
8708 goto add_character;
8709 }
8710 else
8711 goto add_quoted_string;
8712
8713 /* break; */
8714
8715 default:
8716 /* This is the fix for " $@ " */
8717 add_ifs_character:
8718 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c)))
8719 {
8720 if (string[sindex]) /* from old goto dollar_add_string */
8721 sindex++;
8722 if (c == 0)
8723 {
8724 c = CTLNUL;
8725 goto add_character;
8726 }
8727 else
8728 {
8729 #if HANDLE_MULTIBYTE
8730 if (MB_CUR_MAX > 1)
8731 sindex--;
8732
8733 if (MB_CUR_MAX > 1)
8734 {
8735 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
8736 }
8737 else
8738 #endif
8739 {
8740 twochars[0] = CTLESC;
8741 twochars[1] = c;
8742 goto add_twochars;
8743 }
8744 }
8745 }
8746
8747 SADD_MBCHAR (temp, string, sindex, string_size);
8748
8749 add_character:
8750 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
8751 DEFAULT_ARRAY_SIZE);
8752 istring[istring_index++] = c;
8753 istring[istring_index] = '\0';
8754
8755 /* Next character. */
8756 sindex++;
8757 }
8758 }
8759
8760 finished_with_string:
8761 /* OK, we're ready to return. If we have a quoted string, and
8762 quoted_dollar_at is not set, we do no splitting at all; otherwise
8763 we split on ' '. The routines that call this will handle what to
8764 do if nothing has been expanded. */
8765
8766 /* Partially and wholly quoted strings which expand to the empty
8767 string are retained as an empty arguments. Unquoted strings
8768 which expand to the empty string are discarded. The single
8769 exception is the case of expanding "$@" when there are no
8770 positional parameters. In that case, we discard the expansion. */
8771
8772 /* Because of how the code that handles "" and '' in partially
8773 quoted strings works, we need to make ISTRING into a QUOTED_NULL
8774 if we saw quoting characters, but the expansion was empty.
8775 "" and '' are tossed away before we get to this point when
8776 processing partially quoted strings. This makes "" and $xxx""
8777 equivalent when xxx is unset. We also look to see whether we
8778 saw a quoted null from a ${} expansion and add one back if we
8779 need to. */
8780
8781 /* If we expand to nothing and there were no single or double quotes
8782 in the word, we throw it away. Otherwise, we return a NULL word.
8783 The single exception is for $@ surrounded by double quotes when
8784 there are no positional parameters. In that case, we also throw
8785 the word away. */
8786
8787 if (*istring == '\0')
8788 {
8789 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
8790 {
8791 istring[0] = CTLNUL;
8792 istring[1] = '\0';
8793 tword = make_bare_word (istring);
8794 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8795 list = make_word_list (tword, (WORD_LIST *)NULL);
8796 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8797 tword->flags |= W_QUOTED;
8798 }
8799 /* According to sh, ksh, and Posix.2, if a word expands into nothing
8800 and a double-quoted "$@" appears anywhere in it, then the entire
8801 word is removed. */
8802 else if (quoted_state == UNQUOTED || quoted_dollar_at)
8803 list = (WORD_LIST *)NULL;
8804 #if 0
8805 else
8806 {
8807 tword = make_bare_word (istring);
8808 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8809 tword->flags |= W_QUOTED;
8810 list = make_word_list (tword, (WORD_LIST *)NULL);
8811 }
8812 #else
8813 else
8814 list = (WORD_LIST *)NULL;
8815 #endif
8816 }
8817 else if (word->flags & W_NOSPLIT)
8818 {
8819 tword = make_bare_word (istring);
8820 if (word->flags & W_ASSIGNMENT)
8821 tword->flags |= W_ASSIGNMENT; /* XXX */
8822 if (word->flags & W_COMPASSIGN)
8823 tword->flags |= W_COMPASSIGN; /* XXX */
8824 if (word->flags & W_NOGLOB)
8825 tword->flags |= W_NOGLOB; /* XXX */
8826 if (word->flags & W_NOBRACE)
8827 tword->flags |= W_NOBRACE; /* XXX */
8828 if (word->flags & W_NOEXPAND)
8829 tword->flags |= W_NOEXPAND; /* XXX */
8830 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8831 tword->flags |= W_QUOTED;
8832 if (had_quoted_null && QUOTED_NULL (istring))
8833 tword->flags |= W_HASQUOTEDNULL;
8834 list = make_word_list (tword, (WORD_LIST *)NULL);
8835 }
8836 else
8837 {
8838 char *ifs_chars;
8839 char *tstring;
8840
8841 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
8842
8843 /* If we have $@, we need to split the results no matter what. If
8844 IFS is unset or NULL, string_list_dollar_at has separated the
8845 positional parameters with a space, so we split on space (we have
8846 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
8847 string_list_dollar_at has separated the positional parameters
8848 with the first character of $IFS, so we split on $IFS. If
8849 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
8850 unset or null, and we want to make sure that we split on spaces
8851 regardless of what else has happened to IFS since the expansion. */
8852 if (split_on_spaces)
8853 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
8854 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
8855 don't want to split the result (W_NOSPLIT2), and we are not quoted,
8856 we have already separated the arguments with the first character of
8857 $IFS. In this case, we want to return a list with a single word
8858 with the separator possibly replaced with a space (it's what other
8859 shells seem to do).
8860 quoted_dollar_at is internal to this function and is set if we are
8861 passed an argument that is unquoted (quoted == 0) but we encounter a
8862 double-quoted $@ while expanding it. */
8863 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
8864 {
8865 /* Only split and rejoin if we have to */
8866 if (*ifs_chars && *ifs_chars != ' ')
8867 {
8868 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
8869 tstring = string_list (list);
8870 }
8871 else
8872 tstring = istring;
8873 tword = make_bare_word (tstring);
8874 if (tstring != istring)
8875 free (tstring);
8876 goto set_word_flags;
8877 }
8878 else if (has_dollar_at && ifs_chars)
8879 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
8880 else
8881 {
8882 tword = make_bare_word (istring);
8883 set_word_flags:
8884 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
8885 tword->flags |= W_QUOTED;
8886 if (word->flags & W_ASSIGNMENT)
8887 tword->flags |= W_ASSIGNMENT;
8888 if (word->flags & W_COMPASSIGN)
8889 tword->flags |= W_COMPASSIGN;
8890 if (word->flags & W_NOGLOB)
8891 tword->flags |= W_NOGLOB;
8892 if (word->flags & W_NOBRACE)
8893 tword->flags |= W_NOBRACE;
8894 if (word->flags & W_NOEXPAND)
8895 tword->flags |= W_NOEXPAND;
8896 if (had_quoted_null && QUOTED_NULL (istring))
8897 tword->flags |= W_HASQUOTEDNULL; /* XXX */
8898 list = make_word_list (tword, (WORD_LIST *)NULL);
8899 }
8900 }
8901
8902 free (istring);
8903 return (list);
8904 }
8905
8906 /* **************************************************************** */
8907 /* */
8908 /* Functions for Quote Removal */
8909 /* */
8910 /* **************************************************************** */
8911
8912 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
8913 backslash quoting rules for within double quotes or a here document. */
8914 char *
8915 string_quote_removal (string, quoted)
8916 char *string;
8917 int quoted;
8918 {
8919 size_t slen;
8920 char *r, *result_string, *temp, *send;
8921 int sindex, tindex, dquote;
8922 unsigned char c;
8923 DECLARE_MBSTATE;
8924
8925 /* The result can be no longer than the original string. */
8926 slen = strlen (string);
8927 send = string + slen;
8928
8929 r = result_string = (char *)xmalloc (slen + 1);
8930
8931 for (dquote = sindex = 0; c = string[sindex];)
8932 {
8933 switch (c)
8934 {
8935 case '\\':
8936 c = string[++sindex];
8937 if (c == 0)
8938 {
8939 *r++ = '\\';
8940 break;
8941 }
8942 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
8943 *r++ = '\\';
8944 /* FALLTHROUGH */
8945
8946 default:
8947 SCOPY_CHAR_M (r, string, send, sindex);
8948 break;
8949
8950 case '\'':
8951 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
8952 {
8953 *r++ = c;
8954 sindex++;
8955 break;
8956 }
8957 tindex = sindex + 1;
8958 temp = string_extract_single_quoted (string, &tindex);
8959 if (temp)
8960 {
8961 strcpy (r, temp);
8962 r += strlen (r);
8963 free (temp);
8964 }
8965 sindex = tindex;
8966 break;
8967
8968 case '"':
8969 dquote = 1 - dquote;
8970 sindex++;
8971 break;
8972 }
8973 }
8974 *r = '\0';
8975 return (result_string);
8976 }
8977
8978 #if 0
8979 /* UNUSED */
8980 /* Perform quote removal on word WORD. This allocates and returns a new
8981 WORD_DESC *. */
8982 WORD_DESC *
8983 word_quote_removal (word, quoted)
8984 WORD_DESC *word;
8985 int quoted;
8986 {
8987 WORD_DESC *w;
8988 char *t;
8989
8990 t = string_quote_removal (word->word, quoted);
8991 w = alloc_word_desc ();
8992 w->word = t ? t : savestring ("");
8993 return (w);
8994 }
8995
8996 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
8997 the members of the list are treated as if they are surrounded by
8998 double quotes. Return a new list, or NULL if LIST is NULL. */
8999 WORD_LIST *
9000 word_list_quote_removal (list, quoted)
9001 WORD_LIST *list;
9002 int quoted;
9003 {
9004 WORD_LIST *result, *t, *tresult, *e;
9005
9006 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
9007 {
9008 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
9009 #if 0
9010 result = (WORD_LIST *) list_append (result, tresult);
9011 #else
9012 if (result == 0)
9013 result = e = tresult;
9014 else
9015 {
9016 e->next = tresult;
9017 while (e->next)
9018 e = e->next;
9019 }
9020 #endif
9021 }
9022 return (result);
9023 }
9024 #endif
9025
9026 /*******************************************
9027 * *
9028 * Functions to perform word splitting *
9029 * *
9030 *******************************************/
9031
9032 void
9033 setifs (v)
9034 SHELL_VAR *v;
9035 {
9036 char *t;
9037 unsigned char uc;
9038
9039 ifs_var = v;
9040 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
9041
9042 ifs_is_set = ifs_var != 0;
9043 ifs_is_null = ifs_is_set && (*ifs_value == 0);
9044
9045 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9046 handle multibyte chars in IFS */
9047 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9048 for (t = ifs_value ; t && *t; t++)
9049 {
9050 uc = *t;
9051 ifs_cmap[uc] = 1;
9052 }
9053
9054 #if defined (HANDLE_MULTIBYTE)
9055 if (ifs_value == 0)
9056 {
9057 ifs_firstc[0] = '\0';
9058 ifs_firstc_len = 1;
9059 }
9060 else
9061 {
9062 size_t ifs_len;
9063 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
9064 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
9065 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
9066 {
9067 ifs_firstc[0] = ifs_value[0];
9068 ifs_firstc[1] = '\0';
9069 ifs_firstc_len = 1;
9070 }
9071 else
9072 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
9073 }
9074 #else
9075 ifs_firstc = ifs_value ? *ifs_value : 0;
9076 #endif
9077 }
9078
9079 char *
9080 getifs ()
9081 {
9082 return ifs_value;
9083 }
9084
9085 /* This splits a single word into a WORD LIST on $IFS, but only if the word
9086 is not quoted. list_string () performs quote removal for us, even if we
9087 don't do any splitting. */
9088 WORD_LIST *
9089 word_split (w, ifs_chars)
9090 WORD_DESC *w;
9091 char *ifs_chars;
9092 {
9093 WORD_LIST *result;
9094
9095 if (w)
9096 {
9097 char *xifs;
9098
9099 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
9100 result = list_string (w->word, xifs, w->flags & W_QUOTED);
9101 }
9102 else
9103 result = (WORD_LIST *)NULL;
9104
9105 return (result);
9106 }
9107
9108 /* Perform word splitting on LIST and return the RESULT. It is possible
9109 to return (WORD_LIST *)NULL. */
9110 static WORD_LIST *
9111 word_list_split (list)
9112 WORD_LIST *list;
9113 {
9114 WORD_LIST *result, *t, *tresult, *e;
9115
9116 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
9117 {
9118 tresult = word_split (t->word, ifs_value);
9119 if (result == 0)
9120 result = e = tresult;
9121 else
9122 {
9123 e->next = tresult;
9124 while (e->next)
9125 e = e->next;
9126 }
9127 }
9128 return (result);
9129 }
9130
9131 /**************************************************
9132 * *
9133 * Functions to expand an entire WORD_LIST *
9134 * *
9135 **************************************************/
9136
9137 /* Do any word-expansion-specific cleanup and jump to top_level */
9138 static void
9139 exp_jump_to_top_level (v)
9140 int v;
9141 {
9142 set_pipestatus_from_exit (last_command_exit_value);
9143
9144 /* Cleanup code goes here. */
9145 expand_no_split_dollar_star = 0; /* XXX */
9146 expanding_redir = 0;
9147 assigning_in_environment = 0;
9148
9149 if (parse_and_execute_level == 0)
9150 top_level_cleanup (); /* from sig.c */
9151
9152 jump_to_top_level (v);
9153 }
9154
9155 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
9156 ELIST, and set ELIST to the new list. */
9157 #define PREPEND_LIST(nlist, elist) \
9158 do { nlist->next = elist; elist = nlist; } while (0)
9159
9160 /* Separate out any initial variable assignments from TLIST. If set -k has
9161 been executed, remove all assignment statements from TLIST. Initial
9162 variable assignments and other environment assignments are placed
9163 on SUBST_ASSIGN_VARLIST. */
9164 static WORD_LIST *
9165 separate_out_assignments (tlist)
9166 WORD_LIST *tlist;
9167 {
9168 register WORD_LIST *vp, *lp;
9169
9170 if (tlist == 0)
9171 return ((WORD_LIST *)NULL);
9172
9173 if (subst_assign_varlist)
9174 dispose_words (subst_assign_varlist); /* Clean up after previous error */
9175
9176 subst_assign_varlist = (WORD_LIST *)NULL;
9177 vp = lp = tlist;
9178
9179 /* Separate out variable assignments at the start of the command.
9180 Loop invariant: vp->next == lp
9181 Loop postcondition:
9182 lp = list of words left after assignment statements skipped
9183 tlist = original list of words
9184 */
9185 while (lp && (lp->word->flags & W_ASSIGNMENT))
9186 {
9187 vp = lp;
9188 lp = lp->next;
9189 }
9190
9191 /* If lp != tlist, we have some initial assignment statements.
9192 We make SUBST_ASSIGN_VARLIST point to the list of assignment
9193 words and TLIST point to the remaining words. */
9194 if (lp != tlist)
9195 {
9196 subst_assign_varlist = tlist;
9197 /* ASSERT(vp->next == lp); */
9198 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
9199 tlist = lp; /* remainder of word list */
9200 }
9201
9202 /* vp == end of variable list */
9203 /* tlist == remainder of original word list without variable assignments */
9204 if (!tlist)
9205 /* All the words in tlist were assignment statements */
9206 return ((WORD_LIST *)NULL);
9207
9208 /* ASSERT(tlist != NULL); */
9209 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
9210
9211 /* If the -k option is in effect, we need to go through the remaining
9212 words, separate out the assignment words, and place them on
9213 SUBST_ASSIGN_VARLIST. */
9214 if (place_keywords_in_env)
9215 {
9216 WORD_LIST *tp; /* tp == running pointer into tlist */
9217
9218 tp = tlist;
9219 lp = tlist->next;
9220
9221 /* Loop Invariant: tp->next == lp */
9222 /* Loop postcondition: tlist == word list without assignment statements */
9223 while (lp)
9224 {
9225 if (lp->word->flags & W_ASSIGNMENT)
9226 {
9227 /* Found an assignment statement, add this word to end of
9228 subst_assign_varlist (vp). */
9229 if (!subst_assign_varlist)
9230 subst_assign_varlist = vp = lp;
9231 else
9232 {
9233 vp->next = lp;
9234 vp = lp;
9235 }
9236
9237 /* Remove the word pointed to by LP from TLIST. */
9238 tp->next = lp->next;
9239 /* ASSERT(vp == lp); */
9240 lp->next = (WORD_LIST *)NULL;
9241 lp = tp->next;
9242 }
9243 else
9244 {
9245 tp = lp;
9246 lp = lp->next;
9247 }
9248 }
9249 }
9250 return (tlist);
9251 }
9252
9253 #define WEXP_VARASSIGN 0x001
9254 #define WEXP_BRACEEXP 0x002
9255 #define WEXP_TILDEEXP 0x004
9256 #define WEXP_PARAMEXP 0x008
9257 #define WEXP_PATHEXP 0x010
9258
9259 /* All of the expansions, including variable assignments at the start of
9260 the list. */
9261 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
9262
9263 /* All of the expansions except variable assignments at the start of
9264 the list. */
9265 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
9266
9267 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
9268 expansion, command substitution, arithmetic expansion, word splitting, and
9269 quote removal. */
9270 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
9271
9272 /* Take the list of words in LIST and do the various substitutions. Return
9273 a new list of words which is the expanded list, and without things like
9274 variable assignments. */
9275
9276 WORD_LIST *
9277 expand_words (list)
9278 WORD_LIST *list;
9279 {
9280 return (expand_word_list_internal (list, WEXP_ALL));
9281 }
9282
9283 /* Same as expand_words (), but doesn't hack variable or environment
9284 variables. */
9285 WORD_LIST *
9286 expand_words_no_vars (list)
9287 WORD_LIST *list;
9288 {
9289 return (expand_word_list_internal (list, WEXP_NOVARS));
9290 }
9291
9292 WORD_LIST *
9293 expand_words_shellexp (list)
9294 WORD_LIST *list;
9295 {
9296 return (expand_word_list_internal (list, WEXP_SHELLEXP));
9297 }
9298
9299 static WORD_LIST *
9300 glob_expand_word_list (tlist, eflags)
9301 WORD_LIST *tlist;
9302 int eflags;
9303 {
9304 char **glob_array, *temp_string;
9305 register int glob_index;
9306 WORD_LIST *glob_list, *output_list, *disposables, *next;
9307 WORD_DESC *tword;
9308
9309 output_list = disposables = (WORD_LIST *)NULL;
9310 glob_array = (char **)NULL;
9311 while (tlist)
9312 {
9313 /* For each word, either globbing is attempted or the word is
9314 added to orig_list. If globbing succeeds, the results are
9315 added to orig_list and the word (tlist) is added to the list
9316 of disposable words. If globbing fails and failed glob
9317 expansions are left unchanged (the shell default), the
9318 original word is added to orig_list. If globbing fails and
9319 failed glob expansions are removed, the original word is
9320 added to the list of disposable words. orig_list ends up
9321 in reverse order and requires a call to REVERSE_LIST to
9322 be set right. After all words are examined, the disposable
9323 words are freed. */
9324 next = tlist->next;
9325
9326 /* If the word isn't an assignment and contains an unquoted
9327 pattern matching character, then glob it. */
9328 if ((tlist->word->flags & W_NOGLOB) == 0 &&
9329 unquoted_glob_pattern_p (tlist->word->word))
9330 {
9331 glob_array = shell_glob_filename (tlist->word->word);
9332
9333 /* Handle error cases.
9334 I don't think we should report errors like "No such file
9335 or directory". However, I would like to report errors
9336 like "Read failed". */
9337
9338 if (glob_array == 0 || GLOB_FAILED (glob_array))
9339 {
9340 glob_array = (char **)xmalloc (sizeof (char *));
9341 glob_array[0] = (char *)NULL;
9342 }
9343
9344 /* Dequote the current word in case we have to use it. */
9345 if (glob_array[0] == NULL)
9346 {
9347 temp_string = dequote_string (tlist->word->word);
9348 free (tlist->word->word);
9349 tlist->word->word = temp_string;
9350 }
9351
9352 /* Make the array into a word list. */
9353 glob_list = (WORD_LIST *)NULL;
9354 for (glob_index = 0; glob_array[glob_index]; glob_index++)
9355 {
9356 tword = make_bare_word (glob_array[glob_index]);
9357 glob_list = make_word_list (tword, glob_list);
9358 }
9359
9360 if (glob_list)
9361 {
9362 output_list = (WORD_LIST *)list_append (glob_list, output_list);
9363 PREPEND_LIST (tlist, disposables);
9364 }
9365 else if (fail_glob_expansion != 0)
9366 {
9367 last_command_exit_value = EXECUTION_FAILURE;
9368 report_error (_("no match: %s"), tlist->word->word);
9369 exp_jump_to_top_level (DISCARD);
9370 }
9371 else if (allow_null_glob_expansion == 0)
9372 {
9373 /* Failed glob expressions are left unchanged. */
9374 PREPEND_LIST (tlist, output_list);
9375 }
9376 else
9377 {
9378 /* Failed glob expressions are removed. */
9379 PREPEND_LIST (tlist, disposables);
9380 }
9381 }
9382 else
9383 {
9384 /* Dequote the string. */
9385 temp_string = dequote_string (tlist->word->word);
9386 free (tlist->word->word);
9387 tlist->word->word = temp_string;
9388 PREPEND_LIST (tlist, output_list);
9389 }
9390
9391 strvec_dispose (glob_array);
9392 glob_array = (char **)NULL;
9393
9394 tlist = next;
9395 }
9396
9397 if (disposables)
9398 dispose_words (disposables);
9399
9400 if (output_list)
9401 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9402
9403 return (output_list);
9404 }
9405
9406 #if defined (BRACE_EXPANSION)
9407 static WORD_LIST *
9408 brace_expand_word_list (tlist, eflags)
9409 WORD_LIST *tlist;
9410 int eflags;
9411 {
9412 register char **expansions;
9413 char *temp_string;
9414 WORD_LIST *disposables, *output_list, *next;
9415 WORD_DESC *w;
9416 int eindex;
9417
9418 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
9419 {
9420 next = tlist->next;
9421
9422 if (tlist->word->flags & W_NOBRACE)
9423 {
9424 /*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
9425 PREPEND_LIST (tlist, output_list);
9426 continue;
9427 }
9428
9429 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9430 {
9431 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
9432 PREPEND_LIST (tlist, output_list);
9433 continue;
9434 }
9435
9436 /* Only do brace expansion if the word has a brace character. If
9437 not, just add the word list element to BRACES and continue. In
9438 the common case, at least when running shell scripts, this will
9439 degenerate to a bunch of calls to `mbschr', and then what is
9440 basically a reversal of TLIST into BRACES, which is corrected
9441 by a call to REVERSE_LIST () on BRACES when the end of TLIST
9442 is reached. */
9443 if (mbschr (tlist->word->word, LBRACE))
9444 {
9445 expansions = brace_expand (tlist->word->word);
9446
9447 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
9448 {
9449 w = alloc_word_desc ();
9450 w->word = temp_string;
9451
9452 /* If brace expansion didn't change the word, preserve
9453 the flags. We may want to preserve the flags
9454 unconditionally someday -- XXX */
9455 if (STREQ (temp_string, tlist->word->word))
9456 w->flags = tlist->word->flags;
9457 else
9458 w = make_word_flags (w, temp_string);
9459
9460 output_list = make_word_list (w, output_list);
9461 }
9462 free (expansions);
9463
9464 /* Add TLIST to the list of words to be freed after brace
9465 expansion has been performed. */
9466 PREPEND_LIST (tlist, disposables);
9467 }
9468 else
9469 PREPEND_LIST (tlist, output_list);
9470 }
9471
9472 if (disposables)
9473 dispose_words (disposables);
9474
9475 if (output_list)
9476 output_list = REVERSE_LIST (output_list, WORD_LIST *);
9477
9478 return (output_list);
9479 }
9480 #endif
9481
9482 #if defined (ARRAY_VARS)
9483 /* Take WORD, a compound associative array assignment, and internally run
9484 'declare -A w', where W is the variable name portion of WORD. */
9485 static int
9486 make_internal_declare (word, option)
9487 char *word;
9488 char *option;
9489 {
9490 int t;
9491 WORD_LIST *wl;
9492 WORD_DESC *w;
9493
9494 w = make_word (word);
9495
9496 t = assignment (w->word, 0);
9497 w->word[t] = '\0';
9498
9499 wl = make_word_list (w, (WORD_LIST *)NULL);
9500 wl = make_word_list (make_word (option), wl);
9501
9502 return (declare_builtin (wl));
9503 }
9504 #endif
9505
9506 static WORD_LIST *
9507 shell_expand_word_list (tlist, eflags)
9508 WORD_LIST *tlist;
9509 int eflags;
9510 {
9511 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list;
9512 int expanded_something, has_dollar_at;
9513 char *temp_string;
9514
9515 /* We do tilde expansion all the time. This is what 1003.2 says. */
9516 new_list = (WORD_LIST *)NULL;
9517 for (orig_list = tlist; tlist; tlist = next)
9518 {
9519 temp_string = tlist->word->word;
9520
9521 next = tlist->next;
9522
9523 #if defined (ARRAY_VARS)
9524 /* If this is a compound array assignment to a builtin that accepts
9525 such assignments (e.g., `declare'), take the assignment and perform
9526 it separately, handling the semantics of declarations inside shell
9527 functions. This avoids the double-evaluation of such arguments,
9528 because `declare' does some evaluation of compound assignments on
9529 its own. */
9530 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
9531 {
9532 int t;
9533 char opts[8], opti;
9534
9535 opti = 0;
9536 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
9537 opts[opti++] = '-';
9538
9539 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
9540 {
9541 opts[opti++] = 'g';
9542 opts[opti++] = 'A';
9543 }
9544 else if (tlist->word->flags & W_ASSIGNASSOC)
9545 opts[opti++] = 'A';
9546 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
9547 {
9548 opts[opti++] = 'g';
9549 opts[opti++] = 'a';
9550 }
9551 else if (tlist->word->flags & W_ASSIGNARRAY)
9552 opts[opti++] = 'a';
9553 else if (tlist->word->flags & W_ASSNGLOBAL)
9554 opts[opti++] = 'g';
9555
9556 #if 0
9557 /* If we have special handling note the integer attribute */
9558 if (opti > 0 && (tlist->word->flags & W_ASSIGNINT))
9559 opts[opti++] = 'i';
9560 #endif
9561
9562 opts[opti] = '\0';
9563 if (opti > 0)
9564 make_internal_declare (tlist->word->word, opts);
9565
9566 t = do_word_assignment (tlist->word, 0);
9567 if (t == 0)
9568 {
9569 last_command_exit_value = EXECUTION_FAILURE;
9570 exp_jump_to_top_level (DISCARD);
9571 }
9572
9573 /* Now transform the word as ksh93 appears to do and go on */
9574 t = assignment (tlist->word->word, 0);
9575 tlist->word->word[t] = '\0';
9576 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
9577 }
9578 #endif
9579
9580 expanded_something = 0;
9581 expanded = expand_word_internal
9582 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
9583
9584 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
9585 {
9586 /* By convention, each time this error is returned,
9587 tlist->word->word has already been freed. */
9588 tlist->word->word = (char *)NULL;
9589
9590 /* Dispose our copy of the original list. */
9591 dispose_words (orig_list);
9592 /* Dispose the new list we're building. */
9593 dispose_words (new_list);
9594
9595 last_command_exit_value = EXECUTION_FAILURE;
9596 if (expanded == &expand_word_error)
9597 exp_jump_to_top_level (DISCARD);
9598 else
9599 exp_jump_to_top_level (FORCE_EOF);
9600 }
9601
9602 /* Don't split words marked W_NOSPLIT. */
9603 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
9604 {
9605 temp_list = word_list_split (expanded);
9606 dispose_words (expanded);
9607 }
9608 else
9609 {
9610 /* If no parameter expansion, command substitution, process
9611 substitution, or arithmetic substitution took place, then
9612 do not do word splitting. We still have to remove quoted
9613 null characters from the result. */
9614 word_list_remove_quoted_nulls (expanded);
9615 temp_list = expanded;
9616 }
9617
9618 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
9619 new_list = (WORD_LIST *)list_append (expanded, new_list);
9620 }
9621
9622 if (orig_list)
9623 dispose_words (orig_list);
9624
9625 if (new_list)
9626 new_list = REVERSE_LIST (new_list, WORD_LIST *);
9627
9628 return (new_list);
9629 }
9630
9631 /* The workhorse for expand_words () and expand_words_no_vars ().
9632 First arg is LIST, a WORD_LIST of words.
9633 Second arg EFLAGS is a flags word controlling which expansions are
9634 performed.
9635
9636 This does all of the substitutions: brace expansion, tilde expansion,
9637 parameter expansion, command substitution, arithmetic expansion,
9638 process substitution, word splitting, and pathname expansion, according
9639 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
9640 set, or for which no expansion is done, do not undergo word splitting.
9641 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
9642 with W_NOBRACE set do not undergo brace expansion (see
9643 brace_expand_word_list above). */
9644 static WORD_LIST *
9645 expand_word_list_internal (list, eflags)
9646 WORD_LIST *list;
9647 int eflags;
9648 {
9649 WORD_LIST *new_list, *temp_list;
9650 int tint;
9651
9652 tempenv_assign_error = 0;
9653 if (list == 0)
9654 return ((WORD_LIST *)NULL);
9655
9656 garglist = new_list = copy_word_list (list);
9657 if (eflags & WEXP_VARASSIGN)
9658 {
9659 garglist = new_list = separate_out_assignments (new_list);
9660 if (new_list == 0)
9661 {
9662 if (subst_assign_varlist)
9663 {
9664 /* All the words were variable assignments, so they are placed
9665 into the shell's environment. */
9666 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9667 {
9668 this_command_name = (char *)NULL; /* no arithmetic errors */
9669 tint = do_word_assignment (temp_list->word, 0);
9670 /* Variable assignment errors in non-interactive shells
9671 running in Posix.2 mode cause the shell to exit. */
9672 if (tint == 0)
9673 {
9674 last_command_exit_value = EXECUTION_FAILURE;
9675 if (interactive_shell == 0 && posixly_correct)
9676 exp_jump_to_top_level (FORCE_EOF);
9677 else
9678 exp_jump_to_top_level (DISCARD);
9679 }
9680 }
9681 dispose_words (subst_assign_varlist);
9682 subst_assign_varlist = (WORD_LIST *)NULL;
9683 }
9684 return ((WORD_LIST *)NULL);
9685 }
9686 }
9687
9688 /* Begin expanding the words that remain. The expansions take place on
9689 things that aren't really variable assignments. */
9690
9691 #if defined (BRACE_EXPANSION)
9692 /* Do brace expansion on this word if there are any brace characters
9693 in the string. */
9694 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
9695 new_list = brace_expand_word_list (new_list, eflags);
9696 #endif /* BRACE_EXPANSION */
9697
9698 /* Perform the `normal' shell expansions: tilde expansion, parameter and
9699 variable substitution, command substitution, arithmetic expansion,
9700 and word splitting. */
9701 new_list = shell_expand_word_list (new_list, eflags);
9702
9703 /* Okay, we're almost done. Now let's just do some filename
9704 globbing. */
9705 if (new_list)
9706 {
9707 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
9708 /* Glob expand the word list unless globbing has been disabled. */
9709 new_list = glob_expand_word_list (new_list, eflags);
9710 else
9711 /* Dequote the words, because we're not performing globbing. */
9712 new_list = dequote_list (new_list);
9713 }
9714
9715 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
9716 {
9717 sh_wassign_func_t *assign_func;
9718 int is_special_builtin, is_builtin_or_func;
9719
9720 /* If the remainder of the words expand to nothing, Posix.2 requires
9721 that the variable and environment assignments affect the shell's
9722 environment. */
9723 assign_func = new_list ? assign_in_env : do_word_assignment;
9724 tempenv_assign_error = 0;
9725
9726 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
9727 /* Posix says that special builtins exit if a variable assignment error
9728 occurs in an assignment preceding it. */
9729 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
9730
9731 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
9732 {
9733 this_command_name = (char *)NULL;
9734 assigning_in_environment = (assign_func == assign_in_env);
9735 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
9736 assigning_in_environment = 0;
9737 /* Variable assignment errors in non-interactive shells running
9738 in Posix.2 mode cause the shell to exit. */
9739 if (tint == 0)
9740 {
9741 if (assign_func == do_word_assignment)
9742 {
9743 last_command_exit_value = EXECUTION_FAILURE;
9744 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
9745 exp_jump_to_top_level (FORCE_EOF);
9746 else
9747 exp_jump_to_top_level (DISCARD);
9748 }
9749 else
9750 tempenv_assign_error++;
9751 }
9752 }
9753
9754 dispose_words (subst_assign_varlist);
9755 subst_assign_varlist = (WORD_LIST *)NULL;
9756 }
9757
9758 return (new_list);
9759 }