]> git.ipfire.org Git - thirdparty/bash.git/blob - subst.c
Bash-4.4 patch 4
[thirdparty/bash.git] / subst.c
1 /* subst.c -- The part of the shell that does parameter, command, arithmetic,
2 and globbing substitutions. */
3
4 /* ``Have a little faith, there's magic in the night. You ain't a
5 beauty, but, hey, you're alright.'' */
6
7 /* Copyright (C) 1987-2016 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 #define NEED_FPURGE_DECL
41
42 #include "bashansi.h"
43 #include "posixstat.h"
44 #include "bashintl.h"
45
46 #include "shell.h"
47 #include "parser.h"
48 #include "flags.h"
49 #include "jobs.h"
50 #include "execute_cmd.h"
51 #include "filecntl.h"
52 #include "trap.h"
53 #include "pathexp.h"
54 #include "mailcheck.h"
55
56 #include "shmbutil.h"
57 #if defined (HAVE_MBSTR_H) && defined (HAVE_MBSCHR)
58 # include <mbstr.h> /* mbschr */
59 #endif
60 #include "typemax.h"
61
62 #include "builtins/getopt.h"
63 #include "builtins/common.h"
64
65 #include "builtins/builtext.h"
66
67 #include <tilde/tilde.h>
68 #include <glob/strmatch.h>
69
70 #if !defined (errno)
71 extern int errno;
72 #endif /* !errno */
73
74 /* The size that strings change by. */
75 #define DEFAULT_INITIAL_ARRAY_SIZE 112
76 #define DEFAULT_ARRAY_SIZE 128
77
78 /* Variable types. */
79 #define VT_VARIABLE 0
80 #define VT_POSPARMS 1
81 #define VT_ARRAYVAR 2
82 #define VT_ARRAYMEMBER 3
83 #define VT_ASSOCVAR 4
84
85 #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */
86
87 /* Flags for quoted_strchr */
88 #define ST_BACKSL 0x01
89 #define ST_CTLESC 0x02
90 #define ST_SQUOTE 0x04 /* unused yet */
91 #define ST_DQUOTE 0x08 /* unused yet */
92
93 /* These defs make it easier to use the editor. */
94 #define LBRACE '{'
95 #define RBRACE '}'
96 #define LPAREN '('
97 #define RPAREN ')'
98 #define LBRACK '['
99 #define RBRACK ']'
100
101 #if defined (HANDLE_MULTIBYTE)
102 #define WLPAREN L'('
103 #define WRPAREN L')'
104 #endif
105
106 #define DOLLAR_AT_STAR(c) ((c) == '@' || (c) == '*')
107 #define STR_DOLLAR_AT_STAR(s) (DOLLAR_AT_STAR ((s)[0]) && (s)[1] == '\0')
108
109 /* Evaluates to 1 if C is one of the shell's special parameters whose length
110 can be taken, but is also one of the special expansion characters. */
111 #define VALID_SPECIAL_LENGTH_PARAM(c) \
112 ((c) == '-' || (c) == '?' || (c) == '#' || (c) == '@')
113
114 /* Evaluates to 1 if C is one of the shell's special parameters for which an
115 indirect variable reference may be made. */
116 #define VALID_INDIR_PARAM(c) \
117 ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*')
118
119 /* Evaluates to 1 if C is one of the OP characters that follows the parameter
120 in ${parameter[:]OPword}. */
121 #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP)
122
123 /* Evaluates to 1 if this is one of the shell's special variables. */
124 #define SPECIAL_VAR(name, wi) \
125 (*name && ((DIGIT (*name) && all_digits (name)) || \
126 (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \
127 (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1]))))
128
129 /* This can be used by all of the *_extract_* functions that have a similar
130 structure. It can't just be wrapped in a do...while(0) loop because of
131 the embedded `break'. The dangling else accommodates a trailing semicolon;
132 we could also put in a do ; while (0) */
133
134
135 #define CHECK_STRING_OVERRUN(oind, ind, len, ch) \
136 if (ind >= len) \
137 { \
138 oind = len; \
139 ch = 0; \
140 break; \
141 } \
142 else \
143
144 /* An expansion function that takes a string and a quoted flag and returns
145 a WORD_LIST *. Used as the type of the third argument to
146 expand_string_if_necessary(). */
147 typedef WORD_LIST *EXPFUNC __P((char *, int));
148
149 /* Process ID of the last command executed within command substitution. */
150 pid_t last_command_subst_pid = NO_PID;
151 pid_t current_command_subst_pid = NO_PID;
152
153 /* Variables used to keep track of the characters in IFS. */
154 SHELL_VAR *ifs_var;
155 char *ifs_value;
156 unsigned char ifs_cmap[UCHAR_MAX + 1];
157 int ifs_is_set, ifs_is_null;
158
159 #if defined (HANDLE_MULTIBYTE)
160 unsigned char ifs_firstc[MB_LEN_MAX];
161 size_t ifs_firstc_len;
162 #else
163 unsigned char ifs_firstc;
164 #endif
165
166 /* If non-zero, command substitution inherits the value of errexit option */
167 int inherit_errexit = 0;
168
169 /* Sentinel to tell when we are performing variable assignments preceding a
170 command name and putting them into the environment. Used to make sure
171 we use the temporary environment when looking up variable values. */
172 int assigning_in_environment;
173
174 /* Used to hold a list of variable assignments preceding a command. Global
175 so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a
176 SIGCHLD trap and so it can be saved and restored by the trap handlers. */
177 WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL;
178
179 /* Tell the expansion functions to not longjmp back to top_level on fatal
180 errors. Enabled when doing completion and prompt string expansion. */
181 int no_longjmp_on_fatal_error = 0;
182
183 /* Extern functions and variables from different files. */
184 extern int last_command_exit_value, last_command_exit_signal;
185 extern int subshell_environment, running_in_background;
186 extern int subshell_level, parse_and_execute_level, sourcelevel;
187 extern int eof_encountered, line_number;
188 extern int return_catch_flag, return_catch_value;
189 extern pid_t dollar_dollar_pid;
190 extern int posixly_correct;
191 extern char *this_command_name;
192 extern struct fd_bitmap *current_fds_to_close;
193 extern int wordexp_only;
194 extern int expanding_redir;
195 extern int tempenv_assign_error;
196 extern int builtin_ignoring_errexit;
197
198 #if defined (JOB_CONTROL) && defined (PROCESS_SUBSTITUTION)
199 extern PROCESS *last_procsub_child;
200 #endif
201
202 #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE)
203 extern wchar_t *wcsdup __P((const wchar_t *));
204 #endif
205
206 /* Non-zero means to allow unmatched globbed filenames to expand to
207 a null file. */
208 int allow_null_glob_expansion;
209
210 /* Non-zero means to throw an error when globbing fails to match anything. */
211 int fail_glob_expansion;
212
213 #if 0
214 /* Variables to keep track of which words in an expanded word list (the
215 output of expand_word_list_internal) are the result of globbing
216 expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c.
217 (CURRENTLY UNUSED). */
218 char *glob_argv_flags;
219 static int glob_argv_flags_size;
220 #endif
221
222 static WORD_LIST *cached_quoted_dollar_at = 0;
223
224 static WORD_LIST expand_word_error, expand_word_fatal;
225 static WORD_DESC expand_wdesc_error, expand_wdesc_fatal;
226 static char expand_param_error, expand_param_fatal;
227 static char extract_string_error, extract_string_fatal;
228
229 /* Set by expand_word_unsplit; used to inhibit splitting and re-joining
230 $* on $IFS, primarily when doing assignment statements. */
231 static int expand_no_split_dollar_star = 0;
232
233 /* A WORD_LIST of words to be expanded by expand_word_list_internal,
234 without any leading variable assignments. */
235 static WORD_LIST *garglist = (WORD_LIST *)NULL;
236
237 static char *quoted_substring __P((char *, int, int));
238 static int quoted_strlen __P((char *));
239 static char *quoted_strchr __P((char *, int, int));
240
241 static char *expand_string_if_necessary __P((char *, int, EXPFUNC *));
242 static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *));
243 static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
244 static WORD_LIST *expand_string_internal __P((char *, int));
245 static WORD_LIST *expand_string_leave_quoted __P((char *, int));
246 static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *));
247
248 static WORD_LIST *list_quote_escapes __P((WORD_LIST *));
249 static WORD_LIST *list_dequote_escapes __P((WORD_LIST *));
250
251 static char *make_quoted_char __P((int));
252 static WORD_LIST *quote_list __P((WORD_LIST *));
253
254 static int unquoted_substring __P((char *, char *));
255 static int unquoted_member __P((int, char *));
256
257 #if defined (ARRAY_VARS)
258 static SHELL_VAR *do_compound_assignment __P((char *, char *, int));
259 #endif
260 static int do_assignment_internal __P((const WORD_DESC *, int));
261
262 static char *string_extract_verbatim __P((char *, size_t, int *, char *, int));
263 static char *string_extract __P((char *, int *, char *, int));
264 static char *string_extract_double_quoted __P((char *, int *, int));
265 static inline char *string_extract_single_quoted __P((char *, int *));
266 static inline int skip_single_quoted __P((const char *, size_t, int, int));
267 static int skip_double_quoted __P((char *, size_t, int, int));
268 static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int));
269 static char *extract_dollar_brace_string __P((char *, int *, int, int));
270 static int skip_matched_pair __P((const char *, int, int, int, int));
271
272 static char *pos_params __P((char *, int, int, int));
273
274 static unsigned char *mb_getcharlens __P((char *, int));
275
276 static char *remove_upattern __P((char *, char *, int));
277 #if defined (HANDLE_MULTIBYTE)
278 static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int));
279 #endif
280 static char *remove_pattern __P((char *, char *, int));
281
282 static int match_upattern __P((char *, char *, int, char **, char **));
283 #if defined (HANDLE_MULTIBYTE)
284 static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **));
285 #endif
286 static int match_pattern __P((char *, char *, int, char **, char **));
287 static int getpatspec __P((int, char *));
288 static char *getpattern __P((char *, int, int));
289 static char *variable_remove_pattern __P((char *, char *, int, int));
290 static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int));
291 static char *parameter_list_remove_pattern __P((int, char *, int, int));
292 #ifdef ARRAY_VARS
293 static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int));
294 #endif
295 static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int));
296
297 static char *string_var_assignment __P((SHELL_VAR *, char *));
298 #if defined (ARRAY_VARS)
299 static char *array_var_assignment __P((SHELL_VAR *, int, int));
300 #endif
301 static char *pos_params_assignment __P((WORD_LIST *, int, int));
302 static char *string_transform __P((int, SHELL_VAR *, char *));
303 static char *list_transform __P((int, SHELL_VAR *, WORD_LIST *, int, int));
304 static char *parameter_list_transform __P((int, int, int));
305 #if defined ARRAY_VARS
306 static char *array_transform __P((int, SHELL_VAR *, char *, int));
307 #endif
308 static char *parameter_brace_transform __P((char *, char *, int, char *, int, int, int));
309
310 static char *process_substitute __P((char *, int));
311
312 static char *read_comsub __P((int, int, int *));
313
314 #ifdef ARRAY_VARS
315 static arrayind_t array_length_reference __P((char *));
316 #endif
317
318 static int valid_brace_expansion_word __P((char *, int));
319 static int chk_atstar __P((char *, int, int *, int *));
320 static int chk_arithsub __P((const char *, int));
321
322 static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *));
323 static char *parameter_brace_find_indir __P((char *, int, int, int));
324 static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *));
325 static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int, int *, int *));
326 static void parameter_brace_expand_error __P((char *, char *));
327
328 static int valid_length_expression __P((char *));
329 static intmax_t parameter_brace_expand_length __P((char *));
330
331 static char *skiparith __P((char *, int));
332 static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *));
333 static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **));
334 static char *mb_substring __P((char *, int, int));
335 static char *parameter_brace_substring __P((char *, char *, int, char *, int, int));
336
337 static int shouldexp_replacement __P((char *));
338
339 static char *pos_params_pat_subst __P((char *, char *, char *, int));
340
341 static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int, int));
342
343 static char *pos_params_casemod __P((char *, char *, int, int));
344 static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int));
345
346 static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *));
347 static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int));
348
349 static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *));
350
351 static WORD_LIST *word_list_split __P((WORD_LIST *));
352
353 static void exp_jump_to_top_level __P((int));
354
355 static WORD_LIST *separate_out_assignments __P((WORD_LIST *));
356 static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int));
357 #ifdef BRACE_EXPANSION
358 static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int));
359 #endif
360 #if defined (ARRAY_VARS)
361 static int make_internal_declare __P((char *, char *, char *));
362 #endif
363 static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int));
364 static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int));
365
366 /* **************************************************************** */
367 /* */
368 /* Utility Functions */
369 /* */
370 /* **************************************************************** */
371
372 #if defined (DEBUG)
373 void
374 dump_word_flags (flags)
375 int flags;
376 {
377 int f;
378
379 f = flags;
380 fprintf (stderr, "%d -> ", f);
381 if (f & W_ARRAYIND)
382 {
383 f &= ~W_ARRAYIND;
384 fprintf (stderr, "W_ARRAYIND%s", f ? "|" : "");
385 }
386 if (f & W_ASSIGNASSOC)
387 {
388 f &= ~W_ASSIGNASSOC;
389 fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : "");
390 }
391 if (f & W_ASSIGNARRAY)
392 {
393 f &= ~W_ASSIGNARRAY;
394 fprintf (stderr, "W_ASSIGNARRAY%s", f ? "|" : "");
395 }
396 if (f & W_HASCTLESC)
397 {
398 f &= ~W_HASCTLESC;
399 fprintf (stderr, "W_HASCTLESC%s", f ? "|" : "");
400 }
401 if (f & W_NOPROCSUB)
402 {
403 f &= ~W_NOPROCSUB;
404 fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : "");
405 }
406 if (f & W_DQUOTE)
407 {
408 f &= ~W_DQUOTE;
409 fprintf (stderr, "W_DQUOTE%s", f ? "|" : "");
410 }
411 if (f & W_HASQUOTEDNULL)
412 {
413 f &= ~W_HASQUOTEDNULL;
414 fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : "");
415 }
416 if (f & W_ASSIGNARG)
417 {
418 f &= ~W_ASSIGNARG;
419 fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : "");
420 }
421 if (f & W_ASSNBLTIN)
422 {
423 f &= ~W_ASSNBLTIN;
424 fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : "");
425 }
426 if (f & W_ASSNGLOBAL)
427 {
428 f &= ~W_ASSNGLOBAL;
429 fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : "");
430 }
431 if (f & W_COMPASSIGN)
432 {
433 f &= ~W_COMPASSIGN;
434 fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : "");
435 }
436 if (f & W_NOEXPAND)
437 {
438 f &= ~W_NOEXPAND;
439 fprintf (stderr, "W_NOEXPAND%s", f ? "|" : "");
440 }
441 if (f & W_ITILDE)
442 {
443 f &= ~W_ITILDE;
444 fprintf (stderr, "W_ITILDE%s", f ? "|" : "");
445 }
446 if (f & W_NOTILDE)
447 {
448 f &= ~W_NOTILDE;
449 fprintf (stderr, "W_NOTILDE%s", f ? "|" : "");
450 }
451 if (f & W_ASSIGNRHS)
452 {
453 f &= ~W_ASSIGNRHS;
454 fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : "");
455 }
456 if (f & W_NOCOMSUB)
457 {
458 f &= ~W_NOCOMSUB;
459 fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : "");
460 }
461 if (f & W_DOLLARSTAR)
462 {
463 f &= ~W_DOLLARSTAR;
464 fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : "");
465 }
466 if (f & W_DOLLARAT)
467 {
468 f &= ~W_DOLLARAT;
469 fprintf (stderr, "W_DOLLARAT%s", f ? "|" : "");
470 }
471 if (f & W_TILDEEXP)
472 {
473 f &= ~W_TILDEEXP;
474 fprintf (stderr, "W_TILDEEXP%s", f ? "|" : "");
475 }
476 if (f & W_NOSPLIT2)
477 {
478 f &= ~W_NOSPLIT2;
479 fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : "");
480 }
481 if (f & W_NOSPLIT)
482 {
483 f &= ~W_NOSPLIT;
484 fprintf (stderr, "W_NOSPLIT%s", f ? "|" : "");
485 }
486 if (f & W_NOBRACE)
487 {
488 f &= ~W_NOBRACE;
489 fprintf (stderr, "W_NOBRACE%s", f ? "|" : "");
490 }
491 if (f & W_NOGLOB)
492 {
493 f &= ~W_NOGLOB;
494 fprintf (stderr, "W_NOGLOB%s", f ? "|" : "");
495 }
496 if (f & W_SPLITSPACE)
497 {
498 f &= ~W_SPLITSPACE;
499 fprintf (stderr, "W_SPLITSPACE%s", f ? "|" : "");
500 }
501 if (f & W_ASSIGNMENT)
502 {
503 f &= ~W_ASSIGNMENT;
504 fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : "");
505 }
506 if (f & W_QUOTED)
507 {
508 f &= ~W_QUOTED;
509 fprintf (stderr, "W_QUOTED%s", f ? "|" : "");
510 }
511 if (f & W_HASDOLLAR)
512 {
513 f &= ~W_HASDOLLAR;
514 fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : "");
515 }
516 if (f & W_COMPLETE)
517 {
518 f &= ~W_COMPLETE;
519 fprintf (stderr, "W_COMPLETE%s", f ? "|" : "");
520 }
521
522 fprintf (stderr, "\n");
523 fflush (stderr);
524 }
525 #endif
526
527 #ifdef INCLUDE_UNUSED
528 static char *
529 quoted_substring (string, start, end)
530 char *string;
531 int start, end;
532 {
533 register int len, l;
534 register char *result, *s, *r;
535
536 len = end - start;
537
538 /* Move to string[start], skipping quoted characters. */
539 for (s = string, l = 0; *s && l < start; )
540 {
541 if (*s == CTLESC)
542 {
543 s++;
544 continue;
545 }
546 l++;
547 if (*s == 0)
548 break;
549 }
550
551 r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */
552
553 /* Copy LEN characters, including quote characters. */
554 s = string + l;
555 for (l = 0; l < len; s++)
556 {
557 if (*s == CTLESC)
558 *r++ = *s++;
559 *r++ = *s;
560 l++;
561 if (*s == 0)
562 break;
563 }
564 *r = '\0';
565 return result;
566 }
567 #endif
568
569 #ifdef INCLUDE_UNUSED
570 /* Return the length of S, skipping over quoted characters */
571 static int
572 quoted_strlen (s)
573 char *s;
574 {
575 register char *p;
576 int i;
577
578 i = 0;
579 for (p = s; *p; p++)
580 {
581 if (*p == CTLESC)
582 {
583 p++;
584 if (*p == 0)
585 return (i + 1);
586 }
587 i++;
588 }
589
590 return i;
591 }
592 #endif
593
594 /* Find the first occurrence of character C in string S, obeying shell
595 quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped
596 characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters
597 escaped with CTLESC are skipped. */
598 static char *
599 quoted_strchr (s, c, flags)
600 char *s;
601 int c, flags;
602 {
603 register char *p;
604
605 for (p = s; *p; p++)
606 {
607 if (((flags & ST_BACKSL) && *p == '\\')
608 || ((flags & ST_CTLESC) && *p == CTLESC))
609 {
610 p++;
611 if (*p == '\0')
612 return ((char *)NULL);
613 continue;
614 }
615 else if (*p == c)
616 return p;
617 }
618 return ((char *)NULL);
619 }
620
621 #if defined (INCLUDE_UNUSED)
622 /* Return 1 if CHARACTER appears in an unquoted portion of
623 STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */
624 static int
625 unquoted_member (character, string)
626 int character;
627 char *string;
628 {
629 size_t slen;
630 int sindex, c;
631 DECLARE_MBSTATE;
632
633 slen = strlen (string);
634 sindex = 0;
635 while (c = string[sindex])
636 {
637 if (c == character)
638 return (1);
639
640 switch (c)
641 {
642 default:
643 ADVANCE_CHAR (string, slen, sindex);
644 break;
645
646 case '\\':
647 sindex++;
648 if (string[sindex])
649 ADVANCE_CHAR (string, slen, sindex);
650 break;
651
652 case '\'':
653 sindex = skip_single_quoted (string, slen, ++sindex, 0);
654 break;
655
656 case '"':
657 sindex = skip_double_quoted (string, slen, ++sindex, 0);
658 break;
659 }
660 }
661 return (0);
662 }
663
664 /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */
665 static int
666 unquoted_substring (substr, string)
667 char *substr, *string;
668 {
669 size_t slen;
670 int sindex, c, sublen;
671 DECLARE_MBSTATE;
672
673 if (substr == 0 || *substr == '\0')
674 return (0);
675
676 slen = strlen (string);
677 sublen = strlen (substr);
678 for (sindex = 0; c = string[sindex]; )
679 {
680 if (STREQN (string + sindex, substr, sublen))
681 return (1);
682
683 switch (c)
684 {
685 case '\\':
686 sindex++;
687 if (string[sindex])
688 ADVANCE_CHAR (string, slen, sindex);
689 break;
690
691 case '\'':
692 sindex = skip_single_quoted (string, slen, ++sindex, 0);
693 break;
694
695 case '"':
696 sindex = skip_double_quoted (string, slen, ++sindex, 0);
697 break;
698
699 default:
700 ADVANCE_CHAR (string, slen, sindex);
701 break;
702 }
703 }
704 return (0);
705 }
706 #endif
707
708 /* Most of the substitutions must be done in parallel. In order
709 to avoid using tons of unclear goto's, I have some functions
710 for manipulating malloc'ed strings. They all take INDX, a
711 pointer to an integer which is the offset into the string
712 where manipulation is taking place. They also take SIZE, a
713 pointer to an integer which is the current length of the
714 character array for this string. */
715
716 /* Append SOURCE to TARGET at INDEX. SIZE is the current amount
717 of space allocated to TARGET. SOURCE can be NULL, in which
718 case nothing happens. Gets rid of SOURCE by freeing it.
719 Returns TARGET in case the location has changed. */
720 INLINE char *
721 sub_append_string (source, target, indx, size)
722 char *source, *target;
723 int *indx;
724 size_t *size;
725 {
726 if (source)
727 {
728 int n;
729 size_t srclen;
730
731 srclen = STRLEN (source);
732 if (srclen >= (int)(*size - *indx))
733 {
734 n = srclen + *indx;
735 n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE);
736 target = (char *)xrealloc (target, (*size = n));
737 }
738
739 FASTCOPY (source, target + *indx, srclen);
740 *indx += srclen;
741 target[*indx] = '\0';
742
743 free (source);
744 }
745 return (target);
746 }
747
748 #if 0
749 /* UNUSED */
750 /* Append the textual representation of NUMBER to TARGET.
751 INDX and SIZE are as in SUB_APPEND_STRING. */
752 char *
753 sub_append_number (number, target, indx, size)
754 intmax_t number;
755 char *target;
756 int *indx;
757 size_t *size;
758 {
759 char *temp;
760
761 temp = itos (number);
762 return (sub_append_string (temp, target, indx, size));
763 }
764 #endif
765
766 /* Extract a substring from STRING, starting at SINDEX and ending with
767 one of the characters in CHARLIST. Don't make the ending character
768 part of the string. Leave SINDEX pointing at the ending character.
769 Understand about backslashes in the string. If (flags & SX_VARNAME)
770 is non-zero, and array variables have been compiled into the shell,
771 everything between a `[' and a corresponding `]' is skipped over.
772 If (flags & SX_NOALLOC) is non-zero, don't return the substring, just
773 update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must
774 contain a closing character from CHARLIST. */
775 static char *
776 string_extract (string, sindex, charlist, flags)
777 char *string;
778 int *sindex;
779 char *charlist;
780 int flags;
781 {
782 register int c, i;
783 int found;
784 size_t slen;
785 char *temp;
786 DECLARE_MBSTATE;
787
788 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
789 i = *sindex;
790 found = 0;
791 while (c = string[i])
792 {
793 if (c == '\\')
794 {
795 if (string[i + 1])
796 i++;
797 else
798 break;
799 }
800 #if defined (ARRAY_VARS)
801 else if ((flags & SX_VARNAME) && c == '[')
802 {
803 int ni;
804 /* If this is an array subscript, skip over it and continue. */
805 ni = skipsubscript (string, i, 0);
806 if (string[ni] == ']')
807 i = ni;
808 }
809 #endif
810 else if (MEMBER (c, charlist))
811 {
812 found = 1;
813 break;
814 }
815
816 ADVANCE_CHAR (string, slen, i);
817 }
818
819 /* If we had to have a matching delimiter and didn't find one, return an
820 error and let the caller deal with it. */
821 if ((flags & SX_REQMATCH) && found == 0)
822 {
823 *sindex = i;
824 return (&extract_string_error);
825 }
826
827 temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
828 *sindex = i;
829
830 return (temp);
831 }
832
833 /* Extract the contents of STRING as if it is enclosed in double quotes.
834 SINDEX, when passed in, is the offset of the character immediately
835 following the opening double quote; on exit, SINDEX is left pointing after
836 the closing double quote. If STRIPDQ is non-zero, unquoted double
837 quotes are stripped and the string is terminated by a null byte.
838 Backslashes between the embedded double quotes are processed. If STRIPDQ
839 is zero, an unquoted `"' terminates the string. */
840 static char *
841 string_extract_double_quoted (string, sindex, flags)
842 char *string;
843 int *sindex, flags;
844 {
845 size_t slen;
846 char *send;
847 int j, i, t;
848 unsigned char c;
849 char *temp, *ret; /* The new string we return. */
850 int pass_next, backquote, si; /* State variables for the machine. */
851 int dquote;
852 int stripdq;
853 DECLARE_MBSTATE;
854
855 slen = strlen (string + *sindex) + *sindex;
856 send = string + slen;
857
858 stripdq = (flags & SX_STRIPDQ);
859
860 pass_next = backquote = dquote = 0;
861 temp = (char *)xmalloc (1 + slen - *sindex);
862
863 j = 0;
864 i = *sindex;
865 while (c = string[i])
866 {
867 /* Process a character that was quoted by a backslash. */
868 if (pass_next)
869 {
870 /* XXX - take another look at this in light of Interp 221 */
871 /* Posix.2 sez:
872
873 ``The backslash shall retain its special meaning as an escape
874 character only when followed by one of the characters:
875 $ ` " \ <newline>''.
876
877 If STRIPDQ is zero, we handle the double quotes here and let
878 expand_word_internal handle the rest. If STRIPDQ is non-zero,
879 we have already been through one round of backslash stripping,
880 and want to strip these backslashes only if DQUOTE is non-zero,
881 indicating that we are inside an embedded double-quoted string. */
882
883 /* If we are in an embedded quoted string, then don't strip
884 backslashes before characters for which the backslash
885 retains its special meaning, but remove backslashes in
886 front of other characters. If we are not in an
887 embedded quoted string, don't strip backslashes at all.
888 This mess is necessary because the string was already
889 surrounded by double quotes (and sh has some really weird
890 quoting rules).
891 The returned string will be run through expansion as if
892 it were double-quoted. */
893 if ((stripdq == 0 && c != '"') ||
894 (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0)))
895 temp[j++] = '\\';
896 pass_next = 0;
897
898 add_one_character:
899 COPY_CHAR_I (temp, j, string, send, i);
900 continue;
901 }
902
903 /* A backslash protects the next character. The code just above
904 handles preserving the backslash in front of any character but
905 a double quote. */
906 if (c == '\\')
907 {
908 pass_next++;
909 i++;
910 continue;
911 }
912
913 /* Inside backquotes, ``the portion of the quoted string from the
914 initial backquote and the characters up to the next backquote
915 that is not preceded by a backslash, having escape characters
916 removed, defines that command''. */
917 if (backquote)
918 {
919 if (c == '`')
920 backquote = 0;
921 temp[j++] = c;
922 i++;
923 continue;
924 }
925
926 if (c == '`')
927 {
928 temp[j++] = c;
929 backquote++;
930 i++;
931 continue;
932 }
933
934 /* Pass everything between `$(' and the matching `)' or a quoted
935 ${ ... } pair through according to the Posix.2 specification. */
936 if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
937 {
938 int free_ret = 1;
939
940 si = i + 2;
941 if (string[i + 1] == LPAREN)
942 ret = extract_command_subst (string, &si, (flags & SX_COMPLETE));
943 else
944 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0);
945
946 temp[j++] = '$';
947 temp[j++] = string[i + 1];
948
949 /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error
950 is set. */
951 if (ret == 0 && no_longjmp_on_fatal_error)
952 {
953 free_ret = 0;
954 ret = string + i + 2;
955 }
956
957 for (t = 0; ret[t]; t++, j++)
958 temp[j] = ret[t];
959 temp[j] = string[si];
960
961 if (string[si])
962 {
963 j++;
964 i = si + 1;
965 }
966 else
967 i = si;
968
969 if (free_ret)
970 free (ret);
971 continue;
972 }
973
974 /* Add any character but a double quote to the quoted string we're
975 accumulating. */
976 if (c != '"')
977 goto add_one_character;
978
979 /* c == '"' */
980 if (stripdq)
981 {
982 dquote ^= 1;
983 i++;
984 continue;
985 }
986
987 break;
988 }
989 temp[j] = '\0';
990
991 /* Point to after the closing quote. */
992 if (c)
993 i++;
994 *sindex = i;
995
996 return (temp);
997 }
998
999 /* This should really be another option to string_extract_double_quoted. */
1000 static int
1001 skip_double_quoted (string, slen, sind, flags)
1002 char *string;
1003 size_t slen;
1004 int sind;
1005 int flags;
1006 {
1007 int c, i;
1008 char *ret;
1009 int pass_next, backquote, si;
1010 DECLARE_MBSTATE;
1011
1012 pass_next = backquote = 0;
1013 i = sind;
1014 while (c = string[i])
1015 {
1016 if (pass_next)
1017 {
1018 pass_next = 0;
1019 ADVANCE_CHAR (string, slen, i);
1020 continue;
1021 }
1022 else if (c == '\\')
1023 {
1024 pass_next++;
1025 i++;
1026 continue;
1027 }
1028 else if (backquote)
1029 {
1030 if (c == '`')
1031 backquote = 0;
1032 ADVANCE_CHAR (string, slen, i);
1033 continue;
1034 }
1035 else if (c == '`')
1036 {
1037 backquote++;
1038 i++;
1039 continue;
1040 }
1041 else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE)))
1042 {
1043 si = i + 2;
1044 if (string[i + 1] == LPAREN)
1045 ret = extract_command_subst (string, &si, SX_NOALLOC|(flags&SX_COMPLETE));
1046 else
1047 ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC);
1048
1049 /* These can consume the entire string if they are unterminated */
1050 CHECK_STRING_OVERRUN (i, si, slen, c);
1051
1052 i = si + 1;
1053 continue;
1054 }
1055 else if (c != '"')
1056 {
1057 ADVANCE_CHAR (string, slen, i);
1058 continue;
1059 }
1060 else
1061 break;
1062 }
1063
1064 if (c)
1065 i++;
1066
1067 return (i);
1068 }
1069
1070 /* Extract the contents of STRING as if it is enclosed in single quotes.
1071 SINDEX, when passed in, is the offset of the character immediately
1072 following the opening single quote; on exit, SINDEX is left pointing after
1073 the closing single quote. */
1074 static inline char *
1075 string_extract_single_quoted (string, sindex)
1076 char *string;
1077 int *sindex;
1078 {
1079 register int i;
1080 size_t slen;
1081 char *t;
1082 DECLARE_MBSTATE;
1083
1084 /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */
1085 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0;
1086 i = *sindex;
1087 while (string[i] && string[i] != '\'')
1088 ADVANCE_CHAR (string, slen, i);
1089
1090 t = substring (string, *sindex, i);
1091
1092 if (string[i])
1093 i++;
1094 *sindex = i;
1095
1096 return (t);
1097 }
1098
1099 /* Skip over a single-quoted string. We overload the SX_COMPLETE flag to mean
1100 that we are splitting out words for completion and have encountered a $'...'
1101 string, which allows backslash-escaped single quotes. */
1102 static inline int
1103 skip_single_quoted (string, slen, sind, flags)
1104 const char *string;
1105 size_t slen;
1106 int sind;
1107 int flags;
1108 {
1109 register int c;
1110 DECLARE_MBSTATE;
1111
1112 c = sind;
1113 while (string[c] && string[c] != '\'')
1114 {
1115 if ((flags & SX_COMPLETE) && string[c] == '\\' && string[c+1] == '\'' && string[c+2])
1116 ADVANCE_CHAR (string, slen, c);
1117 ADVANCE_CHAR (string, slen, c);
1118 }
1119
1120 if (string[c])
1121 c++;
1122 return c;
1123 }
1124
1125 /* Just like string_extract, but doesn't hack backslashes or any of
1126 that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */
1127 static char *
1128 string_extract_verbatim (string, slen, sindex, charlist, flags)
1129 char *string;
1130 size_t slen;
1131 int *sindex;
1132 char *charlist;
1133 int flags;
1134 {
1135 register int i;
1136 #if defined (HANDLE_MULTIBYTE)
1137 size_t clen;
1138 wchar_t *wcharlist;
1139 #endif
1140 int c;
1141 char *temp;
1142 DECLARE_MBSTATE;
1143
1144 if ((flags & SX_NOCTLESC) && charlist[0] == '\'' && charlist[1] == '\0')
1145 {
1146 temp = string_extract_single_quoted (string, sindex);
1147 --*sindex; /* leave *sindex at separator character */
1148 return temp;
1149 }
1150
1151 i = *sindex;
1152 #if 0
1153 /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need
1154 this only if MB_CUR_MAX > 1. */
1155 slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1;
1156 #endif
1157 #if defined (HANDLE_MULTIBYTE)
1158 clen = strlen (charlist);
1159 wcharlist = 0;
1160 #endif
1161 while (c = string[i])
1162 {
1163 #if defined (HANDLE_MULTIBYTE)
1164 size_t mblength;
1165 #endif
1166 if ((flags & SX_NOCTLESC) == 0 && c == CTLESC)
1167 {
1168 i += 2;
1169 continue;
1170 }
1171 /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL
1172 through, to protect the CTLNULs from later calls to
1173 remove_quoted_nulls. */
1174 else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL)
1175 {
1176 i += 2;
1177 continue;
1178 }
1179
1180 #if defined (HANDLE_MULTIBYTE)
1181 mblength = MBLEN (string + i, slen - i);
1182 if (mblength > 1)
1183 {
1184 wchar_t wc;
1185 mblength = mbtowc (&wc, string + i, slen - i);
1186 if (MB_INVALIDCH (mblength))
1187 {
1188 if (MEMBER (c, charlist))
1189 break;
1190 }
1191 else
1192 {
1193 if (wcharlist == 0)
1194 {
1195 size_t len;
1196 len = mbstowcs (wcharlist, charlist, 0);
1197 if (len == -1)
1198 len = 0;
1199 wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1));
1200 mbstowcs (wcharlist, charlist, len + 1);
1201 }
1202
1203 if (wcschr (wcharlist, wc))
1204 break;
1205 }
1206 }
1207 else
1208 #endif
1209 if (MEMBER (c, charlist))
1210 break;
1211
1212 ADVANCE_CHAR (string, slen, i);
1213 }
1214
1215 #if defined (HANDLE_MULTIBYTE)
1216 FREE (wcharlist);
1217 #endif
1218
1219 temp = substring (string, *sindex, i);
1220 *sindex = i;
1221
1222 return (temp);
1223 }
1224
1225 /* Extract the $( construct in STRING, and return a new string.
1226 Start extracting at (SINDEX) as if we had just seen "$(".
1227 Make (SINDEX) get the position of the matching ")". )
1228 XFLAGS is additional flags to pass to other extraction functions. */
1229 char *
1230 extract_command_subst (string, sindex, xflags)
1231 char *string;
1232 int *sindex;
1233 int xflags;
1234 {
1235 char *ret;
1236
1237 if (string[*sindex] == LPAREN || (xflags & SX_COMPLETE))
1238 return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/
1239 else
1240 {
1241 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1242 ret = xparse_dolparen (string, string+*sindex, sindex, xflags);
1243 return ret;
1244 }
1245 }
1246
1247 /* Extract the $[ construct in STRING, and return a new string. (])
1248 Start extracting at (SINDEX) as if we had just seen "$[".
1249 Make (SINDEX) get the position of the matching "]". */
1250 char *
1251 extract_arithmetic_subst (string, sindex)
1252 char *string;
1253 int *sindex;
1254 {
1255 return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/
1256 }
1257
1258 #if defined (PROCESS_SUBSTITUTION)
1259 /* Extract the <( or >( construct in STRING, and return a new string.
1260 Start extracting at (SINDEX) as if we had just seen "<(".
1261 Make (SINDEX) get the position of the matching ")". */ /*))*/
1262 char *
1263 extract_process_subst (string, starter, sindex, xflags)
1264 char *string;
1265 char *starter;
1266 int *sindex;
1267 int xflags;
1268 {
1269 #if 0
1270 return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND));
1271 #else
1272 xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0);
1273 return (xparse_dolparen (string, string+*sindex, sindex, xflags));
1274 #endif
1275 }
1276 #endif /* PROCESS_SUBSTITUTION */
1277
1278 #if defined (ARRAY_VARS)
1279 /* This can be fooled by unquoted right parens in the passed string. If
1280 each caller verifies that the last character in STRING is a right paren,
1281 we don't even need to call extract_delimited_string. */
1282 char *
1283 extract_array_assignment_list (string, sindex)
1284 char *string;
1285 int *sindex;
1286 {
1287 int slen;
1288 char *ret;
1289
1290 slen = strlen (string); /* ( */
1291 if (string[slen - 1] == ')')
1292 {
1293 ret = substring (string, *sindex, slen - 1);
1294 *sindex = slen - 1;
1295 return ret;
1296 }
1297 return 0;
1298 }
1299 #endif
1300
1301 /* Extract and create a new string from the contents of STRING, a
1302 character string delimited with OPENER and CLOSER. SINDEX is
1303 the address of an int describing the current offset in STRING;
1304 it should point to just after the first OPENER found. On exit,
1305 SINDEX gets the position of the last character of the matching CLOSER.
1306 If OPENER is more than a single character, ALT_OPENER, if non-null,
1307 contains a character string that can also match CLOSER and thus
1308 needs to be skipped. */
1309 static char *
1310 extract_delimited_string (string, sindex, opener, alt_opener, closer, flags)
1311 char *string;
1312 int *sindex;
1313 char *opener, *alt_opener, *closer;
1314 int flags;
1315 {
1316 int i, c, si;
1317 size_t slen;
1318 char *t, *result;
1319 int pass_character, nesting_level, in_comment;
1320 int len_closer, len_opener, len_alt_opener;
1321 DECLARE_MBSTATE;
1322
1323 slen = strlen (string + *sindex) + *sindex;
1324 len_opener = STRLEN (opener);
1325 len_alt_opener = STRLEN (alt_opener);
1326 len_closer = STRLEN (closer);
1327
1328 pass_character = in_comment = 0;
1329
1330 nesting_level = 1;
1331 i = *sindex;
1332
1333 while (nesting_level)
1334 {
1335 c = string[i];
1336
1337 /* If a recursive call or a call to ADVANCE_CHAR leaves the index beyond
1338 the end of the string, catch it and cut the loop. */
1339 if (i > slen)
1340 {
1341 i = slen;
1342 c = string[i = slen];
1343 break;
1344 }
1345
1346 if (c == 0)
1347 break;
1348
1349 if (in_comment)
1350 {
1351 if (c == '\n')
1352 in_comment = 0;
1353 ADVANCE_CHAR (string, slen, i);
1354 continue;
1355 }
1356
1357 if (pass_character) /* previous char was backslash */
1358 {
1359 pass_character = 0;
1360 ADVANCE_CHAR (string, slen, i);
1361 continue;
1362 }
1363
1364 /* Not exactly right yet; should handle shell metacharacters and
1365 multibyte characters, too. See COMMENT_BEGIN define in parse.y */
1366 if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1])))
1367 {
1368 in_comment = 1;
1369 ADVANCE_CHAR (string, slen, i);
1370 continue;
1371 }
1372
1373 if (c == CTLESC || c == '\\')
1374 {
1375 pass_character++;
1376 i++;
1377 continue;
1378 }
1379
1380 /* Process a nested command substitution, but only if we're parsing an
1381 arithmetic substitution. */
1382 if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN)
1383 {
1384 si = i + 2;
1385 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1386 i = si + 1;
1387 continue;
1388 }
1389
1390 /* Process a nested OPENER. */
1391 if (STREQN (string + i, opener, len_opener))
1392 {
1393 si = i + len_opener;
1394 t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC);
1395 i = si + 1;
1396 continue;
1397 }
1398
1399 /* Process a nested ALT_OPENER */
1400 if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener))
1401 {
1402 si = i + len_alt_opener;
1403 t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC);
1404 i = si + 1;
1405 continue;
1406 }
1407
1408 /* If the current substring terminates the delimited string, decrement
1409 the nesting level. */
1410 if (STREQN (string + i, closer, len_closer))
1411 {
1412 i += len_closer - 1; /* move to last byte of the closer */
1413 nesting_level--;
1414 if (nesting_level == 0)
1415 break;
1416 }
1417
1418 /* Pass old-style command substitution through verbatim. */
1419 if (c == '`')
1420 {
1421 si = i + 1;
1422 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1423 i = si + 1;
1424 continue;
1425 }
1426
1427 /* Pass single-quoted and double-quoted strings through verbatim. */
1428 if (c == '\'' || c == '"')
1429 {
1430 si = i + 1;
1431 i = (c == '\'') ? skip_single_quoted (string, slen, si, 0)
1432 : skip_double_quoted (string, slen, si, 0);
1433 continue;
1434 }
1435
1436 /* move past this character, which was not special. */
1437 ADVANCE_CHAR (string, slen, i);
1438 }
1439
1440 if (c == 0 && nesting_level)
1441 {
1442 if (no_longjmp_on_fatal_error == 0)
1443 {
1444 last_command_exit_value = EXECUTION_FAILURE;
1445 report_error (_("bad substitution: no closing `%s' in %s"), closer, string);
1446 exp_jump_to_top_level (DISCARD);
1447 }
1448 else
1449 {
1450 *sindex = i;
1451 return (char *)NULL;
1452 }
1453 }
1454
1455 si = i - *sindex - len_closer + 1;
1456 if (flags & SX_NOALLOC)
1457 result = (char *)NULL;
1458 else
1459 {
1460 result = (char *)xmalloc (1 + si);
1461 strncpy (result, string + *sindex, si);
1462 result[si] = '\0';
1463 }
1464 *sindex = i;
1465
1466 return (result);
1467 }
1468
1469 /* Extract a parameter expansion expression within ${ and } from STRING.
1470 Obey the Posix.2 rules for finding the ending `}': count braces while
1471 skipping over enclosed quoted strings and command substitutions.
1472 SINDEX is the address of an int describing the current offset in STRING;
1473 it should point to just after the first `{' found. On exit, SINDEX
1474 gets the position of the matching `}'. QUOTED is non-zero if this
1475 occurs inside double quotes. */
1476 /* XXX -- this is very similar to extract_delimited_string -- XXX */
1477 static char *
1478 extract_dollar_brace_string (string, sindex, quoted, flags)
1479 char *string;
1480 int *sindex, quoted, flags;
1481 {
1482 register int i, c;
1483 size_t slen;
1484 int pass_character, nesting_level, si, dolbrace_state;
1485 char *result, *t;
1486 DECLARE_MBSTATE;
1487
1488 pass_character = 0;
1489 nesting_level = 1;
1490 slen = strlen (string + *sindex) + *sindex;
1491
1492 /* The handling of dolbrace_state needs to agree with the code in parse.y:
1493 parse_matched_pair(). The different initial value is to handle the
1494 case where this function is called to parse the word in
1495 ${param op word} (SX_WORD). */
1496 dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM;
1497 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP))
1498 dolbrace_state = DOLBRACE_QUOTE;
1499
1500 i = *sindex;
1501 while (c = string[i])
1502 {
1503 if (pass_character)
1504 {
1505 pass_character = 0;
1506 ADVANCE_CHAR (string, slen, i);
1507 continue;
1508 }
1509
1510 /* CTLESCs and backslashes quote the next character. */
1511 if (c == CTLESC || c == '\\')
1512 {
1513 pass_character++;
1514 i++;
1515 continue;
1516 }
1517
1518 if (string[i] == '$' && string[i+1] == LBRACE)
1519 {
1520 nesting_level++;
1521 i += 2;
1522 continue;
1523 }
1524
1525 if (c == RBRACE)
1526 {
1527 nesting_level--;
1528 if (nesting_level == 0)
1529 break;
1530 i++;
1531 continue;
1532 }
1533
1534 /* Pass the contents of old-style command substitutions through
1535 verbatim. */
1536 if (c == '`')
1537 {
1538 si = i + 1;
1539 t = string_extract (string, &si, "`", flags|SX_NOALLOC);
1540
1541 CHECK_STRING_OVERRUN (i, si, slen, c);
1542
1543 i = si + 1;
1544 continue;
1545 }
1546
1547 /* Pass the contents of new-style command substitutions and
1548 arithmetic substitutions through verbatim. */
1549 if (string[i] == '$' && string[i+1] == LPAREN)
1550 {
1551 si = i + 2;
1552 t = extract_command_subst (string, &si, flags|SX_NOALLOC);
1553 i = si + 1;
1554 continue;
1555 }
1556
1557 /* Pass the contents of double-quoted strings through verbatim. */
1558 if (c == '"')
1559 {
1560 si = i + 1;
1561 i = skip_double_quoted (string, slen, si, 0);
1562 /* skip_XXX_quoted leaves index one past close quote */
1563 continue;
1564 }
1565
1566 if (c == '\'')
1567 {
1568 /*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/
1569 if (posixly_correct && shell_compatibility_level > 42 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
1570 ADVANCE_CHAR (string, slen, i);
1571 else
1572 {
1573 si = i + 1;
1574 i = skip_single_quoted (string, slen, si, 0);
1575 }
1576
1577 continue;
1578 }
1579
1580 /* move past this character, which was not special. */
1581 ADVANCE_CHAR (string, slen, i);
1582
1583 /* This logic must agree with parse.y:parse_matched_pair, since they
1584 share the same defines. */
1585 if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1)
1586 dolbrace_state = DOLBRACE_QUOTE;
1587 else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1)
1588 dolbrace_state = DOLBRACE_QUOTE;
1589 else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1)
1590 dolbrace_state = DOLBRACE_QUOTE2; /* XXX */
1591 else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1)
1592 dolbrace_state = DOLBRACE_QUOTE;
1593 else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1)
1594 dolbrace_state = DOLBRACE_QUOTE;
1595 /* This is intended to handle all of the [:]op expansions and the substring/
1596 length/pattern removal/pattern substitution expansions. */
1597 else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0)
1598 dolbrace_state = DOLBRACE_OP;
1599 else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0)
1600 dolbrace_state = DOLBRACE_WORD;
1601 }
1602
1603 if (c == 0 && nesting_level)
1604 {
1605 if (no_longjmp_on_fatal_error == 0)
1606 { /* { */
1607 last_command_exit_value = EXECUTION_FAILURE;
1608 report_error (_("bad substitution: no closing `%s' in %s"), "}", string);
1609 exp_jump_to_top_level (DISCARD);
1610 }
1611 else
1612 {
1613 *sindex = i;
1614 return ((char *)NULL);
1615 }
1616 }
1617
1618 result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i);
1619 *sindex = i;
1620
1621 return (result);
1622 }
1623
1624 /* Remove backslashes which are quoting backquotes from STRING. Modifies
1625 STRING, and returns a pointer to it. */
1626 char *
1627 de_backslash (string)
1628 char *string;
1629 {
1630 register size_t slen;
1631 register int i, j, prev_i;
1632 DECLARE_MBSTATE;
1633
1634 slen = strlen (string);
1635 i = j = 0;
1636
1637 /* Loop copying string[i] to string[j], i >= j. */
1638 while (i < slen)
1639 {
1640 if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' ||
1641 string[i + 1] == '$'))
1642 i++;
1643 prev_i = i;
1644 ADVANCE_CHAR (string, slen, i);
1645 if (j < prev_i)
1646 do string[j++] = string[prev_i++]; while (prev_i < i);
1647 else
1648 j = i;
1649 }
1650 string[j] = '\0';
1651
1652 return (string);
1653 }
1654
1655 #if 0
1656 /*UNUSED*/
1657 /* Replace instances of \! in a string with !. */
1658 void
1659 unquote_bang (string)
1660 char *string;
1661 {
1662 register int i, j;
1663 register char *temp;
1664
1665 temp = (char *)xmalloc (1 + strlen (string));
1666
1667 for (i = 0, j = 0; (temp[j] = string[i]); i++, j++)
1668 {
1669 if (string[i] == '\\' && string[i + 1] == '!')
1670 {
1671 temp[j] = '!';
1672 i++;
1673 }
1674 }
1675 strcpy (string, temp);
1676 free (temp);
1677 }
1678 #endif
1679
1680 #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = oldjmp; return (x); } while (0)
1681
1682 /* This function assumes s[i] == open; returns with s[ret] == close; used to
1683 parse array subscripts. FLAGS & 1 means to not attempt to skip over
1684 matched pairs of quotes or backquotes, or skip word expansions; it is
1685 intended to be used after expansion has been performed and during final
1686 assignment parsing (see arrayfunc.c:assign_compound_array_list()). */
1687 static int
1688 skip_matched_pair (string, start, open, close, flags)
1689 const char *string;
1690 int start, open, close, flags;
1691 {
1692 int i, pass_next, backq, si, c, count, oldjmp;
1693 size_t slen;
1694 char *temp, *ss;
1695 DECLARE_MBSTATE;
1696
1697 slen = strlen (string + start) + start;
1698 oldjmp = no_longjmp_on_fatal_error;
1699 no_longjmp_on_fatal_error = 1;
1700
1701 i = start + 1; /* skip over leading bracket */
1702 count = 1;
1703 pass_next = backq = 0;
1704 ss = (char *)string;
1705 while (c = string[i])
1706 {
1707 if (pass_next)
1708 {
1709 pass_next = 0;
1710 if (c == 0)
1711 CQ_RETURN(i);
1712 ADVANCE_CHAR (string, slen, i);
1713 continue;
1714 }
1715 else if (c == '\\')
1716 {
1717 pass_next = 1;
1718 i++;
1719 continue;
1720 }
1721 else if (backq)
1722 {
1723 if (c == '`')
1724 backq = 0;
1725 ADVANCE_CHAR (string, slen, i);
1726 continue;
1727 }
1728 else if ((flags & 1) == 0 && c == '`')
1729 {
1730 backq = 1;
1731 i++;
1732 continue;
1733 }
1734 else if ((flags & 1) == 0 && c == open)
1735 {
1736 count++;
1737 i++;
1738 continue;
1739 }
1740 else if (c == close)
1741 {
1742 count--;
1743 if (count == 0)
1744 break;
1745 i++;
1746 continue;
1747 }
1748 else if ((flags & 1) == 0 && (c == '\'' || c == '"'))
1749 {
1750 i = (c == '\'') ? skip_single_quoted (ss, slen, ++i, 0)
1751 : skip_double_quoted (ss, slen, ++i, 0);
1752 /* no increment, the skip functions increment past the closing quote. */
1753 }
1754 else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE))
1755 {
1756 si = i + 2;
1757 if (string[si] == '\0')
1758 CQ_RETURN(si);
1759
1760 if (string[i+1] == LPAREN)
1761 temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1762 else
1763 temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC);
1764
1765 CHECK_STRING_OVERRUN (i, si, slen, c);
1766
1767 i = si;
1768 if (string[i] == '\0') /* don't increment i past EOS in loop */
1769 break;
1770 i++;
1771 continue;
1772 }
1773 else
1774 ADVANCE_CHAR (string, slen, i);
1775 }
1776
1777 CQ_RETURN(i);
1778 }
1779
1780 #if defined (ARRAY_VARS)
1781 int
1782 skipsubscript (string, start, flags)
1783 const char *string;
1784 int start, flags;
1785 {
1786 return (skip_matched_pair (string, start, '[', ']', flags));
1787 }
1788 #endif
1789
1790 /* Skip characters in STRING until we find a character in DELIMS, and return
1791 the index of that character. START is the index into string at which we
1792 begin. This is similar in spirit to strpbrk, but it returns an index into
1793 STRING and takes a starting index. This little piece of code knows quite
1794 a lot of shell syntax. It's very similar to skip_double_quoted and other
1795 functions of that ilk. */
1796 int
1797 skip_to_delim (string, start, delims, flags)
1798 char *string;
1799 int start;
1800 char *delims;
1801 int flags;
1802 {
1803 int i, pass_next, backq, dquote, si, c, oldjmp;
1804 int invert, skipquote, skipcmd, noprocsub, completeflag;
1805 int arithexp, skipcol;
1806 size_t slen;
1807 char *temp, open[3];
1808 DECLARE_MBSTATE;
1809
1810 slen = strlen (string + start) + start;
1811 oldjmp = no_longjmp_on_fatal_error;
1812 if (flags & SD_NOJMP)
1813 no_longjmp_on_fatal_error = 1;
1814 invert = (flags & SD_INVERT);
1815 skipcmd = (flags & SD_NOSKIPCMD) == 0;
1816 noprocsub = (flags & SD_NOPROCSUB);
1817 completeflag = (flags & SD_COMPLETE) ? SX_COMPLETE : 0;
1818
1819 arithexp = (flags & SD_ARITHEXP);
1820 skipcol = 0;
1821
1822 i = start;
1823 pass_next = backq = dquote = 0;
1824 while (c = string[i])
1825 {
1826 /* If this is non-zero, we should not let quote characters be delimiters
1827 and the current character is a single or double quote. We should not
1828 test whether or not it's a delimiter until after we skip single- or
1829 double-quoted strings. */
1830 skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"'));
1831 if (pass_next)
1832 {
1833 pass_next = 0;
1834 if (c == 0)
1835 CQ_RETURN(i);
1836 ADVANCE_CHAR (string, slen, i);
1837 continue;
1838 }
1839 else if (c == '\\')
1840 {
1841 pass_next = 1;
1842 i++;
1843 continue;
1844 }
1845 else if (backq)
1846 {
1847 if (c == '`')
1848 backq = 0;
1849 ADVANCE_CHAR (string, slen, i);
1850 continue;
1851 }
1852 else if (c == '`')
1853 {
1854 backq = 1;
1855 i++;
1856 continue;
1857 }
1858 else if (arithexp && skipcol && c == ':')
1859 {
1860 skipcol--;
1861 i++;
1862 continue;
1863 }
1864 else if (arithexp && c == '?')
1865 {
1866 skipcol++;
1867 i++;
1868 continue;
1869 }
1870 else if (skipquote == 0 && invert == 0 && member (c, delims))
1871 break;
1872 /* the usual case is to use skip_xxx_quoted, but we don't skip over double
1873 quoted strings when looking for the history expansion character as a
1874 delimiter. */
1875 /* special case for programmable completion which takes place before
1876 parser converts backslash-escaped single quotes between $'...' to
1877 `regular' single-quoted strings. */
1878 else if (completeflag && i > 0 && string[i-1] == '$' && c == '\'')
1879 i = skip_single_quoted (string, slen, ++i, SX_COMPLETE);
1880 else if (c == '\'')
1881 i = skip_single_quoted (string, slen, ++i, 0);
1882 else if (c == '"')
1883 i = skip_double_quoted (string, slen, ++i, completeflag);
1884 else if (c == LPAREN && arithexp)
1885 {
1886 si = i + 1;
1887 if (string[si] == '\0')
1888 CQ_RETURN(si);
1889
1890 temp = extract_delimited_string (string, &si, "(", "(", ")", SX_NOALLOC); /* ) */
1891 i = si;
1892 if (string[i] == '\0') /* don't increment i past EOS in loop */
1893 break;
1894 i++;
1895 continue;
1896 }
1897 else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE))
1898 {
1899 si = i + 2;
1900 if (string[si] == '\0')
1901 CQ_RETURN(si);
1902
1903 if (string[i+1] == LPAREN)
1904 temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */
1905 else
1906 temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC);
1907 i = si;
1908 if (string[i] == '\0') /* don't increment i past EOS in loop */
1909 break;
1910 i++;
1911 continue;
1912 }
1913 #if defined (PROCESS_SUBSTITUTION)
1914 else if (skipcmd && noprocsub == 0 && (c == '<' || c == '>') && string[i+1] == LPAREN)
1915 {
1916 si = i + 2;
1917 if (string[si] == '\0')
1918 CQ_RETURN(si);
1919 #if 1
1920 temp = extract_delimited_string (string, &si, (c == '<') ? "<(" : ">(", "(", ")", SX_COMMAND|SX_NOALLOC); /* )) */
1921 #else
1922 temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si, 0);
1923 free (temp); /* XXX - not using SX_NOALLOC here yet */
1924 #endif
1925 i = si;
1926 if (string[i] == '\0')
1927 break;
1928 i++;
1929 continue;
1930 }
1931 #endif /* PROCESS_SUBSTITUTION */
1932 #if defined (EXTENDED_GLOB)
1933 else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@"))
1934 {
1935 si = i + 2;
1936 if (string[si] == '\0')
1937 CQ_RETURN(si);
1938
1939 open[0] = c;
1940 open[1] = LPAREN;
1941 open[2] = '\0';
1942 temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */
1943
1944 i = si;
1945 if (string[i] == '\0') /* don't increment i past EOS in loop */
1946 break;
1947 i++;
1948 continue;
1949 }
1950 #endif
1951 else if ((flags & SD_GLOB) && c == LBRACK)
1952 {
1953 si = i + 1;
1954 if (string[si] == '\0')
1955 CQ_RETURN(si);
1956
1957 temp = extract_delimited_string (string, &si, "[", "[", "]", SX_NOALLOC); /* ] */
1958
1959 i = si;
1960 if (string[i] == '\0') /* don't increment i past EOS in loop */
1961 break;
1962 i++;
1963 continue;
1964 }
1965 else if ((skipquote || invert) && (member (c, delims) == 0))
1966 break;
1967 else
1968 ADVANCE_CHAR (string, slen, i);
1969 }
1970
1971 CQ_RETURN(i);
1972 }
1973
1974 #if defined (BANG_HISTORY)
1975 /* Skip to the history expansion character (delims[0]), paying attention to
1976 quoted strings and command and process substitution. This is a stripped-
1977 down version of skip_to_delims. The essential difference is that this
1978 resets the quoting state when starting a command substitution */
1979 int
1980 skip_to_histexp (string, start, delims, flags)
1981 char *string;
1982 int start;
1983 char *delims;
1984 int flags;
1985 {
1986 int i, pass_next, backq, dquote, si, c, oldjmp;
1987 int histexp_comsub, histexp_backq, old_dquote;
1988 size_t slen;
1989 char *temp, open[3];
1990 DECLARE_MBSTATE;
1991
1992 slen = strlen (string + start) + start;
1993 oldjmp = no_longjmp_on_fatal_error;
1994 if (flags & SD_NOJMP)
1995 no_longjmp_on_fatal_error = 1;
1996
1997 histexp_comsub = histexp_backq = old_dquote = 0;
1998
1999 i = start;
2000 pass_next = backq = dquote = 0;
2001 while (c = string[i])
2002 {
2003 if (pass_next)
2004 {
2005 pass_next = 0;
2006 if (c == 0)
2007 CQ_RETURN(i);
2008 ADVANCE_CHAR (string, slen, i);
2009 continue;
2010 }
2011 else if (c == '\\')
2012 {
2013 pass_next = 1;
2014 i++;
2015 continue;
2016 }
2017 else if (backq && c == '`')
2018 {
2019 backq = 0;
2020 histexp_backq--;
2021 dquote = old_dquote;
2022 i++;
2023 continue;
2024 }
2025 else if (c == '`')
2026 {
2027 backq = 1;
2028 histexp_backq++;
2029 old_dquote = dquote; /* simple - one level for now */
2030 dquote = 0;
2031 i++;
2032 continue;
2033 }
2034 /* When in double quotes, act as if the double quote is a member of
2035 history_no_expand_chars, like the history library does */
2036 else if (dquote && c == delims[0] && string[i+1] == '"')
2037 {
2038 i++;
2039 continue;
2040 }
2041 else if (c == delims[0])
2042 break;
2043 /* the usual case is to use skip_xxx_quoted, but we don't skip over double
2044 quoted strings when looking for the history expansion character as a
2045 delimiter. */
2046 else if (dquote && c == '\'')
2047 {
2048 i++;
2049 continue;
2050 }
2051 else if (c == '\'')
2052 i = skip_single_quoted (string, slen, ++i, 0);
2053 /* The posixly_correct test makes posix-mode shells allow double quotes
2054 to quote the history expansion character */
2055 else if (posixly_correct == 0 && c == '"')
2056 {
2057 dquote = 1 - dquote;
2058 i++;
2059 continue;
2060 }
2061 else if (c == '"')
2062 i = skip_double_quoted (string, slen, ++i, 0);
2063 #if defined (PROCESS_SUBSTITUTION)
2064 else if ((c == '$' || c == '<' || c == '>') && string[i+1] == LPAREN && string[i+2] != LPAREN)
2065 #else
2066 else if (c == '$' && string[i+1] == LPAREN && string[i+2] != LPAREN)
2067 #endif
2068 {
2069 if (string[i+2] == '\0')
2070 CQ_RETURN(i+2);
2071 i += 2;
2072 histexp_comsub++;
2073 old_dquote = dquote;
2074 dquote = 0;
2075 }
2076 else if (histexp_comsub && c == RPAREN)
2077 {
2078 histexp_comsub--;
2079 dquote = old_dquote;
2080 i++;
2081 continue;
2082 }
2083 else if (backq) /* placeholder */
2084 {
2085 ADVANCE_CHAR (string, slen, i);
2086 continue;
2087 }
2088 else
2089 ADVANCE_CHAR (string, slen, i);
2090 }
2091
2092 CQ_RETURN(i);
2093 }
2094 #endif /* BANG_HISTORY */
2095
2096 #if defined (READLINE)
2097 /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is
2098 an unclosed quoted string), or if the character at EINDEX is quoted
2099 by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various
2100 single and double-quoted string parsing functions should not return an
2101 error if there are unclosed quotes or braces. The characters that this
2102 recognizes need to be the same as the contents of
2103 rl_completer_quote_characters. */
2104
2105 int
2106 char_is_quoted (string, eindex)
2107 char *string;
2108 int eindex;
2109 {
2110 int i, pass_next, c, oldjmp;
2111 size_t slen;
2112 DECLARE_MBSTATE;
2113
2114 slen = strlen (string);
2115 oldjmp = no_longjmp_on_fatal_error;
2116 no_longjmp_on_fatal_error = 1;
2117 i = pass_next = 0;
2118 while (i <= eindex)
2119 {
2120 c = string[i];
2121
2122 if (pass_next)
2123 {
2124 pass_next = 0;
2125 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
2126 CQ_RETURN(1);
2127 ADVANCE_CHAR (string, slen, i);
2128 continue;
2129 }
2130 else if (c == '\\')
2131 {
2132 pass_next = 1;
2133 i++;
2134 continue;
2135 }
2136 else if (c == '$' && string[i+1] == '\'' && string[i+2])
2137 {
2138 i += 2;
2139 i = skip_single_quoted (string, slen, i, SX_COMPLETE);
2140 if (i > eindex)
2141 CQ_RETURN (i);
2142 }
2143 else if (c == '\'' || c == '"')
2144 {
2145 i = (c == '\'') ? skip_single_quoted (string, slen, ++i, 0)
2146 : skip_double_quoted (string, slen, ++i, SX_COMPLETE);
2147 if (i > eindex)
2148 CQ_RETURN(1);
2149 /* no increment, the skip_xxx functions go one past end */
2150 }
2151 else
2152 ADVANCE_CHAR (string, slen, i);
2153 }
2154
2155 CQ_RETURN(0);
2156 }
2157
2158 int
2159 unclosed_pair (string, eindex, openstr)
2160 char *string;
2161 int eindex;
2162 char *openstr;
2163 {
2164 int i, pass_next, openc, olen;
2165 size_t slen;
2166 DECLARE_MBSTATE;
2167
2168 slen = strlen (string);
2169 olen = strlen (openstr);
2170 i = pass_next = openc = 0;
2171 while (i <= eindex)
2172 {
2173 if (pass_next)
2174 {
2175 pass_next = 0;
2176 if (i >= eindex) /* XXX was if (i >= eindex - 1) */
2177 return 0;
2178 ADVANCE_CHAR (string, slen, i);
2179 continue;
2180 }
2181 else if (string[i] == '\\')
2182 {
2183 pass_next = 1;
2184 i++;
2185 continue;
2186 }
2187 else if (STREQN (string + i, openstr, olen))
2188 {
2189 openc = 1 - openc;
2190 i += olen;
2191 }
2192 /* XXX - may want to handle $'...' specially here */
2193 else if (string[i] == '\'' || string[i] == '"')
2194 {
2195 i = (string[i] == '\'') ? skip_single_quoted (string, slen, i, 0)
2196 : skip_double_quoted (string, slen, i, SX_COMPLETE);
2197 if (i > eindex)
2198 return 0;
2199 }
2200 else
2201 ADVANCE_CHAR (string, slen, i);
2202 }
2203 return (openc);
2204 }
2205
2206 /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the
2207 individual words. If DELIMS is NULL, the current value of $IFS is used
2208 to split the string, and the function follows the shell field splitting
2209 rules. SENTINEL is an index to look for. NWP, if non-NULL,
2210 gets the number of words in the returned list. CWP, if non-NULL, gets
2211 the index of the word containing SENTINEL. Non-whitespace chars in
2212 DELIMS delimit separate fields. */
2213 WORD_LIST *
2214 split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp)
2215 char *string;
2216 int slen;
2217 char *delims;
2218 int sentinel, flags;
2219 int *nwp, *cwp;
2220 {
2221 int ts, te, i, nw, cw, ifs_split, dflags;
2222 char *token, *d, *d2;
2223 WORD_LIST *ret, *tl;
2224
2225 if (string == 0 || *string == '\0')
2226 {
2227 if (nwp)
2228 *nwp = 0;
2229 if (cwp)
2230 *cwp = 0;
2231 return ((WORD_LIST *)NULL);
2232 }
2233
2234 d = (delims == 0) ? ifs_value : delims;
2235 ifs_split = delims == 0;
2236
2237 /* Make d2 the non-whitespace characters in delims */
2238 d2 = 0;
2239 if (delims)
2240 {
2241 size_t slength;
2242 #if defined (HANDLE_MULTIBYTE)
2243 size_t mblength = 1;
2244 #endif
2245 DECLARE_MBSTATE;
2246
2247 slength = strlen (delims);
2248 d2 = (char *)xmalloc (slength + 1);
2249 i = ts = 0;
2250 while (delims[i])
2251 {
2252 #if defined (HANDLE_MULTIBYTE)
2253 mbstate_t state_bak;
2254 state_bak = state;
2255 mblength = MBRLEN (delims + i, slength, &state);
2256 if (MB_INVALIDCH (mblength))
2257 state = state_bak;
2258 else if (mblength > 1)
2259 {
2260 memcpy (d2 + ts, delims + i, mblength);
2261 ts += mblength;
2262 i += mblength;
2263 slength -= mblength;
2264 continue;
2265 }
2266 #endif
2267 if (whitespace (delims[i]) == 0)
2268 d2[ts++] = delims[i];
2269
2270 i++;
2271 slength--;
2272 }
2273 d2[ts] = '\0';
2274 }
2275
2276 ret = (WORD_LIST *)NULL;
2277
2278 /* Remove sequences of whitespace characters at the start of the string, as
2279 long as those characters are delimiters. */
2280 for (i = 0; member (string[i], d) && spctabnl (string[i]); i++)
2281 ;
2282 if (string[i] == '\0')
2283 {
2284 FREE (d2);
2285 return (ret);
2286 }
2287
2288 ts = i;
2289 nw = 0;
2290 cw = -1;
2291 dflags = flags|SD_NOJMP;
2292 while (1)
2293 {
2294 te = skip_to_delim (string, ts, d, dflags);
2295
2296 /* If we have a non-whitespace delimiter character, use it to make a
2297 separate field. This is just about what $IFS splitting does and
2298 is closer to the behavior of the shell parser. */
2299 if (ts == te && d2 && member (string[ts], d2))
2300 {
2301 te = ts + 1;
2302 /* If we're using IFS splitting, the non-whitespace delimiter char
2303 and any additional IFS whitespace delimits a field. */
2304 if (ifs_split)
2305 while (member (string[te], d) && spctabnl (string[te]))
2306 te++;
2307 else
2308 while (member (string[te], d2))
2309 te++;
2310 }
2311
2312 token = substring (string, ts, te);
2313
2314 ret = add_string_to_list (token, ret);
2315 free (token);
2316 nw++;
2317
2318 if (sentinel >= ts && sentinel <= te)
2319 cw = nw;
2320
2321 /* If the cursor is at whitespace just before word start, set the
2322 sentinel word to the current word. */
2323 if (cwp && cw == -1 && sentinel == ts-1)
2324 cw = nw;
2325
2326 /* If the cursor is at whitespace between two words, make a new, empty
2327 word, add it before (well, after, since the list is in reverse order)
2328 the word we just added, and set the current word to that one. */
2329 if (cwp && cw == -1 && sentinel < ts)
2330 {
2331 tl = make_word_list (make_word (""), ret->next);
2332 ret->next = tl;
2333 cw = nw;
2334 nw++;
2335 }
2336
2337 if (string[te] == 0)
2338 break;
2339
2340 i = te;
2341 while (member (string[i], d) && (ifs_split || spctabnl(string[i])))
2342 i++;
2343
2344 if (string[i])
2345 ts = i;
2346 else
2347 break;
2348 }
2349
2350 /* Special case for SENTINEL at the end of STRING. If we haven't found
2351 the word containing SENTINEL yet, and the index we're looking for is at
2352 the end of STRING (or past the end of the previously-found token,
2353 possible if the end of the line is composed solely of IFS whitespace)
2354 add an additional null argument and set the current word pointer to that. */
2355 if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te))
2356 {
2357 if (whitespace (string[sentinel - 1]))
2358 {
2359 token = "";
2360 ret = add_string_to_list (token, ret);
2361 nw++;
2362 }
2363 cw = nw;
2364 }
2365
2366 if (nwp)
2367 *nwp = nw;
2368 if (cwp)
2369 *cwp = cw;
2370
2371 FREE (d2);
2372
2373 return (REVERSE_LIST (ret, WORD_LIST *));
2374 }
2375 #endif /* READLINE */
2376
2377 #if 0
2378 /* UNUSED */
2379 /* Extract the name of the variable to bind to from the assignment string. */
2380 char *
2381 assignment_name (string)
2382 char *string;
2383 {
2384 int offset;
2385 char *temp;
2386
2387 offset = assignment (string, 0);
2388 if (offset == 0)
2389 return (char *)NULL;
2390 temp = substring (string, 0, offset);
2391 return (temp);
2392 }
2393 #endif
2394
2395 /* **************************************************************** */
2396 /* */
2397 /* Functions to convert strings to WORD_LISTs and vice versa */
2398 /* */
2399 /* **************************************************************** */
2400
2401 /* Return a single string of all the words in LIST. SEP is the separator
2402 to put between individual elements of LIST in the output string. */
2403 char *
2404 string_list_internal (list, sep)
2405 WORD_LIST *list;
2406 char *sep;
2407 {
2408 register WORD_LIST *t;
2409 char *result, *r;
2410 size_t word_len, sep_len, result_size;
2411
2412 if (list == 0)
2413 return ((char *)NULL);
2414
2415 /* Short-circuit quickly if we don't need to separate anything. */
2416 if (list->next == 0)
2417 return (savestring (list->word->word));
2418
2419 /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */
2420 sep_len = STRLEN (sep);
2421 result_size = 0;
2422
2423 for (t = list; t; t = t->next)
2424 {
2425 if (t != list)
2426 result_size += sep_len;
2427 result_size += strlen (t->word->word);
2428 }
2429
2430 r = result = (char *)xmalloc (result_size + 1);
2431
2432 for (t = list; t; t = t->next)
2433 {
2434 if (t != list && sep_len)
2435 {
2436 if (sep_len > 1)
2437 {
2438 FASTCOPY (sep, r, sep_len);
2439 r += sep_len;
2440 }
2441 else
2442 *r++ = sep[0];
2443 }
2444
2445 word_len = strlen (t->word->word);
2446 FASTCOPY (t->word->word, r, word_len);
2447 r += word_len;
2448 }
2449
2450 *r = '\0';
2451 return (result);
2452 }
2453
2454 /* Return a single string of all the words present in LIST, separating
2455 each word with a space. */
2456 char *
2457 string_list (list)
2458 WORD_LIST *list;
2459 {
2460 return (string_list_internal (list, " "));
2461 }
2462
2463 /* An external interface that can be used by the rest of the shell to
2464 obtain a string containing the first character in $IFS. Handles all
2465 the multibyte complications. If LENP is non-null, it is set to the
2466 length of the returned string. */
2467 char *
2468 ifs_firstchar (lenp)
2469 int *lenp;
2470 {
2471 char *ret;
2472 int len;
2473
2474 ret = xmalloc (MB_LEN_MAX + 1);
2475 #if defined (HANDLE_MULTIBYTE)
2476 if (ifs_firstc_len == 1)
2477 {
2478 ret[0] = ifs_firstc[0];
2479 ret[1] = '\0';
2480 len = ret[0] ? 1 : 0;
2481 }
2482 else
2483 {
2484 memcpy (ret, ifs_firstc, ifs_firstc_len);
2485 ret[len = ifs_firstc_len] = '\0';
2486 }
2487 #else
2488 ret[0] = ifs_firstc;
2489 ret[1] = '\0';
2490 len = ret[0] ? 0 : 1;
2491 #endif
2492
2493 if (lenp)
2494 *lenp = len;
2495
2496 return ret;
2497 }
2498
2499 /* Return a single string of all the words present in LIST, obeying the
2500 quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the
2501 expansion [of $*] appears within a double quoted string, it expands
2502 to a single field with the value of each parameter separated by the
2503 first character of the IFS variable, or by a <space> if IFS is unset." */
2504 char *
2505 string_list_dollar_star (list)
2506 WORD_LIST *list;
2507 {
2508 char *ret;
2509 #if defined (HANDLE_MULTIBYTE)
2510 # if defined (__GNUC__)
2511 char sep[MB_CUR_MAX + 1];
2512 # else
2513 char *sep = 0;
2514 # endif
2515 #else
2516 char sep[2];
2517 #endif
2518
2519 #if defined (HANDLE_MULTIBYTE)
2520 # if !defined (__GNUC__)
2521 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2522 # endif /* !__GNUC__ */
2523 if (ifs_firstc_len == 1)
2524 {
2525 sep[0] = ifs_firstc[0];
2526 sep[1] = '\0';
2527 }
2528 else
2529 {
2530 memcpy (sep, ifs_firstc, ifs_firstc_len);
2531 sep[ifs_firstc_len] = '\0';
2532 }
2533 #else
2534 sep[0] = ifs_firstc;
2535 sep[1] = '\0';
2536 #endif
2537
2538 ret = string_list_internal (list, sep);
2539 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2540 free (sep);
2541 #endif
2542 return ret;
2543 }
2544
2545 /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
2546 is non-zero, the $@ appears within double quotes, and we should quote
2547 the list before converting it into a string. If IFS is unset, and the
2548 word is not quoted, we just need to quote CTLESC and CTLNUL characters
2549 in the words in the list, because the default value of $IFS is
2550 <space><tab><newline>, IFS characters in the words in the list should
2551 also be split. If IFS is null, and the word is not quoted, we need
2552 to quote the words in the list to preserve the positional parameters
2553 exactly.
2554 Valid values for the FLAGS argument are the PF_ flags in command.h,
2555 the only one we care about is PF_ASSIGNRHS. $@ is supposed to expand
2556 to the positional parameters separated by spaces no matter what IFS is
2557 set to if in a context where word splitting is not performed. The only
2558 one that we didn't handle before is assignment statement arguments to
2559 declaration builtins like `declare'. */
2560 char *
2561 string_list_dollar_at (list, quoted, flags)
2562 WORD_LIST *list;
2563 int quoted;
2564 int flags;
2565 {
2566 char *ifs, *ret;
2567 #if defined (HANDLE_MULTIBYTE)
2568 # if defined (__GNUC__)
2569 char sep[MB_CUR_MAX + 1];
2570 # else
2571 char *sep = 0;
2572 # endif /* !__GNUC__ */
2573 #else
2574 char sep[2];
2575 #endif
2576 WORD_LIST *tlist;
2577
2578 /* XXX this could just be ifs = ifs_value; */
2579 ifs = ifs_var ? value_cell (ifs_var) : (char *)0;
2580
2581 #if defined (HANDLE_MULTIBYTE)
2582 # if !defined (__GNUC__)
2583 sep = (char *)xmalloc (MB_CUR_MAX + 1);
2584 # endif /* !__GNUC__ */
2585 /* XXX - bash-4.4/bash-5.0 testing PF_ASSIGNRHS */
2586 if (flags & PF_ASSIGNRHS)
2587 {
2588 sep[0] = ' ';
2589 sep[1] = '\0';
2590 }
2591 else if (ifs && *ifs)
2592 {
2593 if (ifs_firstc_len == 1)
2594 {
2595 sep[0] = ifs_firstc[0];
2596 sep[1] = '\0';
2597 }
2598 else
2599 {
2600 memcpy (sep, ifs_firstc, ifs_firstc_len);
2601 sep[ifs_firstc_len] = '\0';
2602 }
2603 }
2604 else
2605 {
2606 sep[0] = ' ';
2607 sep[1] = '\0';
2608 }
2609 #else
2610 /* XXX - bash-4.4/bash-5.0 test PF_ASSIGNRHS */
2611 sep[0] = ((flags & PF_ASSIGNRHS) || ifs == 0 || *ifs == 0) ? ' ' : *ifs;
2612 sep[1] = '\0';
2613 #endif
2614
2615 /* XXX -- why call quote_list if ifs == 0? we can get away without doing
2616 it now that quote_escapes quotes spaces */
2617 tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
2618 ? quote_list (list)
2619 : list_quote_escapes (list);
2620
2621 ret = string_list_internal (tlist, sep);
2622 #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__)
2623 free (sep);
2624 #endif
2625 return ret;
2626 }
2627
2628 /* Turn the positional parameters into a string, understanding quoting and
2629 the various subtleties of using the first character of $IFS as the
2630 separator. Calls string_list_dollar_at, string_list_dollar_star, and
2631 string_list as appropriate. */
2632 char *
2633 string_list_pos_params (pchar, list, quoted)
2634 int pchar;
2635 WORD_LIST *list;
2636 int quoted;
2637 {
2638 char *ret;
2639 WORD_LIST *tlist;
2640
2641 if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES))
2642 {
2643 tlist = quote_list (list);
2644 word_list_remove_quoted_nulls (tlist);
2645 ret = string_list_dollar_star (tlist);
2646 }
2647 else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT))
2648 {
2649 tlist = quote_list (list);
2650 word_list_remove_quoted_nulls (tlist);
2651 ret = string_list (tlist);
2652 }
2653 else if (pchar == '*')
2654 {
2655 /* Even when unquoted, string_list_dollar_star does the right thing
2656 making sure that the first character of $IFS is used as the
2657 separator. */
2658 ret = string_list_dollar_star (list);
2659 }
2660 else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
2661 /* We use string_list_dollar_at, but only if the string is quoted, since
2662 that quotes the escapes if it's not, which we don't want. We could
2663 use string_list (the old code did), but that doesn't do the right
2664 thing if the first character of $IFS is not a space. We use
2665 string_list_dollar_star if the string is unquoted so we make sure that
2666 the elements of $@ are separated by the first character of $IFS for
2667 later splitting. */
2668 ret = string_list_dollar_at (list, quoted, 0);
2669 else if (pchar == '@')
2670 ret = string_list_dollar_star (list);
2671 else
2672 ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list);
2673
2674 return ret;
2675 }
2676
2677 /* Return the list of words present in STRING. Separate the string into
2678 words at any of the characters found in SEPARATORS. If QUOTED is
2679 non-zero then word in the list will have its quoted flag set, otherwise
2680 the quoted flag is left as make_word () deemed fit.
2681
2682 This obeys the P1003.2 word splitting semantics. If `separators' is
2683 exactly <space><tab><newline>, then the splitting algorithm is that of
2684 the Bourne shell, which treats any sequence of characters from `separators'
2685 as a delimiter. If IFS is unset, which results in `separators' being set
2686 to "", no splitting occurs. If separators has some other value, the
2687 following rules are applied (`IFS white space' means zero or more
2688 occurrences of <space>, <tab>, or <newline>, as long as those characters
2689 are in `separators'):
2690
2691 1) IFS white space is ignored at the start and the end of the
2692 string.
2693 2) Each occurrence of a character in `separators' that is not
2694 IFS white space, along with any adjacent occurrences of
2695 IFS white space delimits a field.
2696 3) Any nonzero-length sequence of IFS white space delimits a field.
2697 */
2698
2699 /* BEWARE! list_string strips null arguments. Don't call it twice and
2700 expect to have "" preserved! */
2701
2702 /* This performs word splitting and quoted null character removal on
2703 STRING. */
2704 #define issep(c) \
2705 (((separators)[0]) ? ((separators)[1] ? isifs(c) \
2706 : (c) == (separators)[0]) \
2707 : 0)
2708
2709 WORD_LIST *
2710 list_string (string, separators, quoted)
2711 register char *string, *separators;
2712 int quoted;
2713 {
2714 WORD_LIST *result;
2715 WORD_DESC *t;
2716 char *current_word, *s;
2717 int sindex, sh_style_split, whitesep, xflags;
2718 size_t slen;
2719
2720 if (!string || !*string)
2721 return ((WORD_LIST *)NULL);
2722
2723 sh_style_split = separators && separators[0] == ' ' &&
2724 separators[1] == '\t' &&
2725 separators[2] == '\n' &&
2726 separators[3] == '\0';
2727 for (xflags = 0, s = ifs_value; s && *s; s++)
2728 {
2729 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2730 else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2731 }
2732
2733 slen = 0;
2734 /* Remove sequences of whitespace at the beginning of STRING, as
2735 long as those characters appear in IFS. Do not do this if
2736 STRING is quoted or if there are no separator characters. */
2737 if (!quoted || !separators || !*separators)
2738 {
2739 for (s = string; *s && spctabnl (*s) && issep (*s); s++);
2740
2741 if (!*s)
2742 return ((WORD_LIST *)NULL);
2743
2744 string = s;
2745 }
2746
2747 /* OK, now STRING points to a word that does not begin with white space.
2748 The splitting algorithm is:
2749 extract a word, stopping at a separator
2750 skip sequences of spc, tab, or nl as long as they are separators
2751 This obeys the field splitting rules in Posix.2. */
2752 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 1;
2753 for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; )
2754 {
2755 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2756 unless multibyte chars are possible. */
2757 current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags);
2758 if (current_word == 0)
2759 break;
2760
2761 /* If we have a quoted empty string, add a quoted null argument. We
2762 want to preserve the quoted null character iff this is a quoted
2763 empty string; otherwise the quoted null characters are removed
2764 below. */
2765 if (QUOTED_NULL (current_word))
2766 {
2767 t = alloc_word_desc ();
2768 t->word = make_quoted_char ('\0');
2769 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2770 result = make_word_list (t, result);
2771 }
2772 else if (current_word[0] != '\0')
2773 {
2774 /* If we have something, then add it regardless. However,
2775 perform quoted null character removal on the current word. */
2776 remove_quoted_nulls (current_word);
2777 result = add_string_to_list (current_word, result);
2778 result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */
2779 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
2780 result->word->flags |= W_QUOTED;
2781 }
2782
2783 /* If we're not doing sequences of separators in the traditional
2784 Bourne shell style, then add a quoted null argument. */
2785 else if (!sh_style_split && !spctabnl (string[sindex]))
2786 {
2787 t = alloc_word_desc ();
2788 t->word = make_quoted_char ('\0');
2789 t->flags |= W_QUOTED|W_HASQUOTEDNULL;
2790 result = make_word_list (t, result);
2791 }
2792
2793 free (current_word);
2794
2795 /* Note whether or not the separator is IFS whitespace, used later. */
2796 whitesep = string[sindex] && spctabnl (string[sindex]);
2797
2798 /* Move past the current separator character. */
2799 if (string[sindex])
2800 {
2801 DECLARE_MBSTATE;
2802 ADVANCE_CHAR (string, slen, sindex);
2803 }
2804
2805 /* Now skip sequences of space, tab, or newline characters if they are
2806 in the list of separators. */
2807 while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex]))
2808 sindex++;
2809
2810 /* If the first separator was IFS whitespace and the current character
2811 is a non-whitespace IFS character, it should be part of the current
2812 field delimiter, not a separate delimiter that would result in an
2813 empty field. Look at POSIX.2, 3.6.5, (3)(b). */
2814 if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex]))
2815 {
2816 sindex++;
2817 /* An IFS character that is not IFS white space, along with any
2818 adjacent IFS white space, shall delimit a field. (SUSv3) */
2819 while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex]))
2820 sindex++;
2821 }
2822 }
2823 return (REVERSE_LIST (result, WORD_LIST *));
2824 }
2825
2826 /* Parse a single word from STRING, using SEPARATORS to separate fields.
2827 ENDPTR is set to the first character after the word. This is used by
2828 the `read' builtin. This is never called with SEPARATORS != $IFS;
2829 it should be simplified.
2830
2831 XXX - this function is very similar to list_string; they should be
2832 combined - XXX */
2833 char *
2834 get_word_from_string (stringp, separators, endptr)
2835 char **stringp, *separators, **endptr;
2836 {
2837 register char *s;
2838 char *current_word;
2839 int sindex, sh_style_split, whitesep, xflags;
2840 size_t slen;
2841
2842 if (!stringp || !*stringp || !**stringp)
2843 return ((char *)NULL);
2844
2845 sh_style_split = separators && separators[0] == ' ' &&
2846 separators[1] == '\t' &&
2847 separators[2] == '\n' &&
2848 separators[3] == '\0';
2849 for (xflags = 0, s = ifs_value; s && *s; s++)
2850 {
2851 if (*s == CTLESC) xflags |= SX_NOCTLESC;
2852 if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL;
2853 }
2854
2855 s = *stringp;
2856 slen = 0;
2857
2858 /* Remove sequences of whitespace at the beginning of STRING, as
2859 long as those characters appear in IFS. */
2860 if (sh_style_split || !separators || !*separators)
2861 {
2862 for (; *s && spctabnl (*s) && isifs (*s); s++);
2863
2864 /* If the string is nothing but whitespace, update it and return. */
2865 if (!*s)
2866 {
2867 *stringp = s;
2868 if (endptr)
2869 *endptr = s;
2870 return ((char *)NULL);
2871 }
2872 }
2873
2874 /* OK, S points to a word that does not begin with white space.
2875 Now extract a word, stopping at a separator, save a pointer to
2876 the first character after the word, then skip sequences of spc,
2877 tab, or nl as long as they are separators.
2878
2879 This obeys the field splitting rules in Posix.2. */
2880 sindex = 0;
2881 /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim
2882 unless multibyte chars are possible. */
2883 slen = (MB_CUR_MAX > 1) ? STRLEN (s) : 1;
2884 current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags);
2885
2886 /* Set ENDPTR to the first character after the end of the word. */
2887 if (endptr)
2888 *endptr = s + sindex;
2889
2890 /* Note whether or not the separator is IFS whitespace, used later. */
2891 whitesep = s[sindex] && spctabnl (s[sindex]);
2892
2893 /* Move past the current separator character. */
2894 if (s[sindex])
2895 {
2896 DECLARE_MBSTATE;
2897 ADVANCE_CHAR (s, slen, sindex);
2898 }
2899
2900 /* Now skip sequences of space, tab, or newline characters if they are
2901 in the list of separators. */
2902 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2903 sindex++;
2904
2905 /* If the first separator was IFS whitespace and the current character is
2906 a non-whitespace IFS character, it should be part of the current field
2907 delimiter, not a separate delimiter that would result in an empty field.
2908 Look at POSIX.2, 3.6.5, (3)(b). */
2909 if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex]))
2910 {
2911 sindex++;
2912 /* An IFS character that is not IFS white space, along with any adjacent
2913 IFS white space, shall delimit a field. */
2914 while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex]))
2915 sindex++;
2916 }
2917
2918 /* Update STRING to point to the next field. */
2919 *stringp = s + sindex;
2920 return (current_word);
2921 }
2922
2923 /* Remove IFS white space at the end of STRING. Start at the end
2924 of the string and walk backwards until the beginning of the string
2925 or we find a character that's not IFS white space and not CTLESC.
2926 Only let CTLESC escape a white space character if SAW_ESCAPE is
2927 non-zero. */
2928 char *
2929 strip_trailing_ifs_whitespace (string, separators, saw_escape)
2930 char *string, *separators;
2931 int saw_escape;
2932 {
2933 char *s;
2934
2935 s = string + STRLEN (string) - 1;
2936 while (s > string && ((spctabnl (*s) && isifs (*s)) ||
2937 (saw_escape && *s == CTLESC && spctabnl (s[1]))))
2938 s--;
2939 *++s = '\0';
2940 return string;
2941 }
2942
2943 #if 0
2944 /* UNUSED */
2945 /* Split STRING into words at whitespace. Obeys shell-style quoting with
2946 backslashes, single and double quotes. */
2947 WORD_LIST *
2948 list_string_with_quotes (string)
2949 char *string;
2950 {
2951 WORD_LIST *list;
2952 char *token, *s;
2953 size_t s_len;
2954 int c, i, tokstart, len;
2955
2956 for (s = string; s && *s && spctabnl (*s); s++)
2957 ;
2958 if (s == 0 || *s == 0)
2959 return ((WORD_LIST *)NULL);
2960
2961 s_len = strlen (s);
2962 tokstart = i = 0;
2963 list = (WORD_LIST *)NULL;
2964 while (1)
2965 {
2966 c = s[i];
2967 if (c == '\\')
2968 {
2969 i++;
2970 if (s[i])
2971 i++;
2972 }
2973 else if (c == '\'')
2974 i = skip_single_quoted (s, s_len, ++i, 0);
2975 else if (c == '"')
2976 i = skip_double_quoted (s, s_len, ++i, 0);
2977 else if (c == 0 || spctabnl (c))
2978 {
2979 /* We have found the end of a token. Make a word out of it and
2980 add it to the word list. */
2981 token = substring (s, tokstart, i);
2982 list = add_string_to_list (token, list);
2983 free (token);
2984 while (spctabnl (s[i]))
2985 i++;
2986 if (s[i])
2987 tokstart = i;
2988 else
2989 break;
2990 }
2991 else
2992 i++; /* normal character */
2993 }
2994 return (REVERSE_LIST (list, WORD_LIST *));
2995 }
2996 #endif
2997
2998 /********************************************************/
2999 /* */
3000 /* Functions to perform assignment statements */
3001 /* */
3002 /********************************************************/
3003
3004 #if defined (ARRAY_VARS)
3005 static SHELL_VAR *
3006 do_compound_assignment (name, value, flags)
3007 char *name, *value;
3008 int flags;
3009 {
3010 SHELL_VAR *v;
3011 int mklocal, mkassoc, mkglobal;
3012 WORD_LIST *list;
3013
3014 mklocal = flags & ASS_MKLOCAL;
3015 mkassoc = flags & ASS_MKASSOC;
3016 mkglobal = flags & ASS_MKGLOBAL;
3017
3018 if (mklocal && variable_context)
3019 {
3020 v = find_variable (name);
3021 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3022 {
3023 if (v && readonly_p (v))
3024 err_readonly (name);
3025 return (v); /* XXX */
3026 }
3027 list = expand_compound_array_assignment (v, value, flags);
3028 if (mkassoc)
3029 v = make_local_assoc_variable (name);
3030 else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context)
3031 v = make_local_array_variable (name, 0);
3032 if (v)
3033 assign_compound_array_list (v, list, flags);
3034 if (list)
3035 dispose_words (list);
3036 }
3037 /* In a function but forcing assignment in global context */
3038 else if (mkglobal && variable_context)
3039 {
3040 v = find_global_variable (name);
3041 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3042 {
3043 if (v && readonly_p (v))
3044 err_readonly (name);
3045 return (v); /* XXX */
3046 }
3047 list = expand_compound_array_assignment (v, value, flags);
3048 if (v == 0 && mkassoc)
3049 v = make_new_assoc_variable (name);
3050 else if (v && mkassoc && assoc_p (v) == 0)
3051 v = convert_var_to_assoc (v);
3052 else if (v == 0)
3053 v = make_new_array_variable (name);
3054 else if (v && mkassoc == 0 && array_p (v) == 0)
3055 v = convert_var_to_array (v);
3056 if (v)
3057 assign_compound_array_list (v, list, flags);
3058 if (list)
3059 dispose_words (list);
3060 }
3061 else
3062 {
3063 v = assign_array_from_string (name, value, flags);
3064 if (v && ((readonly_p (v) && (flags & ASS_FORCE) == 0) || noassign_p (v)))
3065 {
3066 if (v && readonly_p (v))
3067 err_readonly (name);
3068 return (v); /* XXX */
3069 }
3070 }
3071
3072 return (v);
3073 }
3074 #endif
3075
3076 /* Given STRING, an assignment string, get the value of the right side
3077 of the `=', and bind it to the left side. If EXPAND is true, then
3078 perform parameter expansion, command substitution, and arithmetic
3079 expansion on the right-hand side. Perform tilde expansion in any
3080 case. Do not perform word splitting on the result of expansion. */
3081 static int
3082 do_assignment_internal (word, expand)
3083 const WORD_DESC *word;
3084 int expand;
3085 {
3086 int offset, appendop, assign_list, aflags, retval;
3087 char *name, *value, *temp;
3088 SHELL_VAR *entry;
3089 #if defined (ARRAY_VARS)
3090 char *t;
3091 int ni;
3092 #endif
3093 const char *string;
3094
3095 if (word == 0 || word->word == 0)
3096 return 0;
3097
3098 appendop = assign_list = aflags = 0;
3099 string = word->word;
3100 offset = assignment (string, 0);
3101 name = savestring (string);
3102 value = (char *)NULL;
3103
3104 if (name[offset] == '=')
3105 {
3106 if (name[offset - 1] == '+')
3107 {
3108 appendop = 1;
3109 name[offset - 1] = '\0';
3110 }
3111
3112 name[offset] = 0; /* might need this set later */
3113 temp = name + offset + 1;
3114
3115 #if defined (ARRAY_VARS)
3116 if (expand && (word->flags & W_COMPASSIGN))
3117 {
3118 assign_list = ni = 1;
3119 value = extract_array_assignment_list (temp, &ni);
3120 }
3121 else
3122 #endif
3123 if (expand && temp[0])
3124 value = expand_string_if_necessary (temp, 0, expand_string_assignment);
3125 else
3126 value = savestring (temp);
3127 }
3128
3129 if (value == 0)
3130 {
3131 value = (char *)xmalloc (1);
3132 value[0] = '\0';
3133 }
3134
3135 if (echo_command_at_execute)
3136 {
3137 if (appendop)
3138 name[offset - 1] = '+';
3139 xtrace_print_assignment (name, value, assign_list, 1);
3140 if (appendop)
3141 name[offset - 1] = '\0';
3142 }
3143
3144 #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0)
3145
3146 if (appendop)
3147 aflags |= ASS_APPEND;
3148
3149 #if defined (ARRAY_VARS)
3150 if (t = mbschr (name, '[')) /*]*/
3151 {
3152 if (assign_list)
3153 {
3154 report_error (_("%s: cannot assign list to array member"), name);
3155 ASSIGN_RETURN (0);
3156 }
3157 entry = assign_array_element (name, value, aflags);
3158 if (entry == 0)
3159 ASSIGN_RETURN (0);
3160 }
3161 else if (assign_list)
3162 {
3163 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0)
3164 aflags |= ASS_MKLOCAL;
3165 if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL))
3166 aflags |= ASS_MKGLOBAL;
3167 if (word->flags & W_ASSIGNASSOC)
3168 aflags |= ASS_MKASSOC;
3169 entry = do_compound_assignment (name, value, aflags);
3170 }
3171 else
3172 #endif /* ARRAY_VARS */
3173 entry = bind_variable (name, value, aflags);
3174
3175 stupidly_hack_special_variables (name);
3176
3177 /* Return 1 if the assignment seems to have been performed correctly. */
3178 if (entry == 0 || readonly_p (entry))
3179 retval = 0; /* assignment failure */
3180 else if (noassign_p (entry))
3181 {
3182 last_command_exit_value = EXECUTION_FAILURE;
3183 retval = 1; /* error status, but not assignment failure */
3184 }
3185 else
3186 retval = 1;
3187
3188 if (entry && retval != 0 && noassign_p (entry) == 0)
3189 VUNSETATTR (entry, att_invisible);
3190
3191 ASSIGN_RETURN (retval);
3192 }
3193
3194 /* Perform the assignment statement in STRING, and expand the
3195 right side by doing tilde, command and parameter expansion. */
3196 int
3197 do_assignment (string)
3198 char *string;
3199 {
3200 WORD_DESC td;
3201
3202 td.flags = W_ASSIGNMENT;
3203 td.word = string;
3204
3205 return do_assignment_internal (&td, 1);
3206 }
3207
3208 int
3209 do_word_assignment (word, flags)
3210 WORD_DESC *word;
3211 int flags;
3212 {
3213 return do_assignment_internal (word, 1);
3214 }
3215
3216 /* Given STRING, an assignment string, get the value of the right side
3217 of the `=', and bind it to the left side. Do not perform any word
3218 expansions on the right hand side. */
3219 int
3220 do_assignment_no_expand (string)
3221 char *string;
3222 {
3223 WORD_DESC td;
3224
3225 td.flags = W_ASSIGNMENT;
3226 td.word = string;
3227
3228 return (do_assignment_internal (&td, 0));
3229 }
3230
3231 /***************************************************
3232 * *
3233 * Functions to manage the positional parameters *
3234 * *
3235 ***************************************************/
3236
3237 /* Return the word list that corresponds to `$*'. */
3238 WORD_LIST *
3239 list_rest_of_args ()
3240 {
3241 register WORD_LIST *list, *args;
3242 int i;
3243
3244 /* Break out of the loop as soon as one of the dollar variables is null. */
3245 for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++)
3246 list = make_word_list (make_bare_word (dollar_vars[i]), list);
3247
3248 for (args = rest_of_args; args; args = args->next)
3249 list = make_word_list (make_bare_word (args->word->word), list);
3250
3251 return (REVERSE_LIST (list, WORD_LIST *));
3252 }
3253
3254 int
3255 number_of_args ()
3256 {
3257 register WORD_LIST *list;
3258 int n;
3259
3260 for (n = 0; n < 9 && dollar_vars[n+1]; n++)
3261 ;
3262 for (list = rest_of_args; list; list = list->next)
3263 n++;
3264 return n;
3265 }
3266
3267 /* Return the value of a positional parameter. This handles values > 10. */
3268 char *
3269 get_dollar_var_value (ind)
3270 intmax_t ind;
3271 {
3272 char *temp;
3273 WORD_LIST *p;
3274
3275 if (ind < 10)
3276 temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL;
3277 else /* We want something like ${11} */
3278 {
3279 ind -= 10;
3280 for (p = rest_of_args; p && ind--; p = p->next)
3281 ;
3282 temp = p ? savestring (p->word->word) : (char *)NULL;
3283 }
3284 return (temp);
3285 }
3286
3287 /* Make a single large string out of the dollar digit variables,
3288 and the rest_of_args. If DOLLAR_STAR is 1, then obey the special
3289 case of "$*" with respect to IFS. */
3290 char *
3291 string_rest_of_args (dollar_star)
3292 int dollar_star;
3293 {
3294 register WORD_LIST *list;
3295 char *string;
3296
3297 list = list_rest_of_args ();
3298 string = dollar_star ? string_list_dollar_star (list) : string_list (list);
3299 dispose_words (list);
3300 return (string);
3301 }
3302
3303 /* Return a string containing the positional parameters from START to
3304 END, inclusive. If STRING[0] == '*', we obey the rules for $*,
3305 which only makes a difference if QUOTED is non-zero. If QUOTED includes
3306 Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise
3307 no quoting chars are added. */
3308 static char *
3309 pos_params (string, start, end, quoted)
3310 char *string;
3311 int start, end, quoted;
3312 {
3313 WORD_LIST *save, *params, *h, *t;
3314 char *ret;
3315 int i;
3316
3317 /* see if we can short-circuit. if start == end, we want 0 parameters. */
3318 if (start == end)
3319 return ((char *)NULL);
3320
3321 save = params = list_rest_of_args ();
3322 if (save == 0 && start > 0)
3323 return ((char *)NULL);
3324
3325 if (start == 0) /* handle ${@:0[:x]} specially */
3326 {
3327 t = make_word_list (make_word (dollar_vars[0]), params);
3328 save = params = t;
3329 }
3330
3331 for (i = start ? 1 : 0; params && i < start; i++)
3332 params = params->next;
3333 if (params == 0)
3334 {
3335 dispose_words (save);
3336 return ((char *)NULL);
3337 }
3338 for (h = t = params; params && i < end; i++)
3339 {
3340 t = params;
3341 params = params->next;
3342 }
3343
3344 t->next = (WORD_LIST *)NULL;
3345
3346 ret = string_list_pos_params (string[0], h, quoted);
3347
3348 if (t != params)
3349 t->next = params;
3350
3351 dispose_words (save);
3352 return (ret);
3353 }
3354
3355 /******************************************************************/
3356 /* */
3357 /* Functions to expand strings to strings or WORD_LISTs */
3358 /* */
3359 /******************************************************************/
3360
3361 #if defined (PROCESS_SUBSTITUTION)
3362 #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~')
3363 #else
3364 #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~')
3365 #endif
3366
3367 /* If there are any characters in STRING that require full expansion,
3368 then call FUNC to expand STRING; otherwise just perform quote
3369 removal if necessary. This returns a new string. */
3370 static char *
3371 expand_string_if_necessary (string, quoted, func)
3372 char *string;
3373 int quoted;
3374 EXPFUNC *func;
3375 {
3376 WORD_LIST *list;
3377 size_t slen;
3378 int i, saw_quote;
3379 char *ret;
3380 DECLARE_MBSTATE;
3381
3382 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3383 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3384 i = saw_quote = 0;
3385 while (string[i])
3386 {
3387 if (EXP_CHAR (string[i]))
3388 break;
3389 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3390 saw_quote = 1;
3391 ADVANCE_CHAR (string, slen, i);
3392 }
3393
3394 if (string[i])
3395 {
3396 list = (*func) (string, quoted);
3397 if (list)
3398 {
3399 ret = string_list (list);
3400 dispose_words (list);
3401 }
3402 else
3403 ret = (char *)NULL;
3404 }
3405 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3406 ret = string_quote_removal (string, quoted);
3407 else
3408 ret = savestring (string);
3409
3410 return ret;
3411 }
3412
3413 static inline char *
3414 expand_string_to_string_internal (string, quoted, func)
3415 char *string;
3416 int quoted;
3417 EXPFUNC *func;
3418 {
3419 WORD_LIST *list;
3420 char *ret;
3421
3422 if (string == 0 || *string == '\0')
3423 return ((char *)NULL);
3424
3425 list = (*func) (string, quoted);
3426 if (list)
3427 {
3428 ret = string_list (list);
3429 dispose_words (list);
3430 }
3431 else
3432 ret = (char *)NULL;
3433
3434 return (ret);
3435 }
3436
3437 char *
3438 expand_string_to_string (string, quoted)
3439 char *string;
3440 int quoted;
3441 {
3442 return (expand_string_to_string_internal (string, quoted, expand_string));
3443 }
3444
3445 char *
3446 expand_string_unsplit_to_string (string, quoted)
3447 char *string;
3448 int quoted;
3449 {
3450 return (expand_string_to_string_internal (string, quoted, expand_string_unsplit));
3451 }
3452
3453 char *
3454 expand_assignment_string_to_string (string, quoted)
3455 char *string;
3456 int quoted;
3457 {
3458 return (expand_string_to_string_internal (string, quoted, expand_string_assignment));
3459 }
3460
3461 char *
3462 expand_arith_string (string, quoted)
3463 char *string;
3464 int quoted;
3465 {
3466 WORD_DESC td;
3467 WORD_LIST *list, *tlist;
3468 size_t slen;
3469 int i, saw_quote;
3470 char *ret;
3471 DECLARE_MBSTATE;
3472
3473 /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */
3474 slen = (MB_CUR_MAX > 1) ? strlen (string) : 0;
3475 i = saw_quote = 0;
3476 while (string[i])
3477 {
3478 if (EXP_CHAR (string[i]))
3479 break;
3480 else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"')
3481 saw_quote = 1;
3482 ADVANCE_CHAR (string, slen, i);
3483 }
3484
3485 if (string[i])
3486 {
3487 /* This is expanded version of expand_string_internal as it's called by
3488 expand_string_leave_quoted */
3489 td.flags = W_NOPROCSUB; /* don't want process substitution */
3490 td.word = savestring (string);
3491 list = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3492 /* This takes care of the calls from expand_string_leave_quoted and
3493 expand_string */
3494 if (list)
3495 {
3496 tlist = word_list_split (list);
3497 dispose_words (list);
3498 list = tlist;
3499 if (list)
3500 dequote_list (list);
3501 }
3502 /* This comes from expand_string_if_necessary */
3503 if (list)
3504 {
3505 ret = string_list (list);
3506 dispose_words (list);
3507 }
3508 else
3509 ret = (char *)NULL;
3510 FREE (td.word);
3511 }
3512 else if (saw_quote && (quoted & Q_ARITH))
3513 ret = string_quote_removal (string, quoted);
3514 else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
3515 ret = string_quote_removal (string, quoted);
3516 else
3517 ret = savestring (string);
3518
3519 return ret;
3520 }
3521
3522 #if defined (COND_COMMAND)
3523 /* Just remove backslashes in STRING. Returns a new string. */
3524 char *
3525 remove_backslashes (string)
3526 char *string;
3527 {
3528 char *r, *ret, *s;
3529
3530 r = ret = (char *)xmalloc (strlen (string) + 1);
3531 for (s = string; s && *s; )
3532 {
3533 if (*s == '\\')
3534 s++;
3535 if (*s == 0)
3536 break;
3537 *r++ = *s++;
3538 }
3539 *r = '\0';
3540 return ret;
3541 }
3542
3543 /* This needs better error handling. */
3544 /* Expand W for use as an argument to a unary or binary operator in a
3545 [[...]] expression. If SPECIAL is 1, this is the rhs argument
3546 to the != or == operator, and should be treated as a pattern. In
3547 this case, we quote the string specially for the globbing code. If
3548 SPECIAL is 2, this is an rhs argument for the =~ operator, and should
3549 be quoted appropriately for regcomp/regexec. The caller is responsible
3550 for removing the backslashes if the unquoted word is needed later. */
3551 char *
3552 cond_expand_word (w, special)
3553 WORD_DESC *w;
3554 int special;
3555 {
3556 char *r, *p;
3557 WORD_LIST *l;
3558 int qflags;
3559
3560 if (w->word == 0 || w->word[0] == '\0')
3561 return ((char *)NULL);
3562
3563 expand_no_split_dollar_star = 1;
3564 w->flags |= W_NOSPLIT2;
3565 l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0);
3566 expand_no_split_dollar_star = 0;
3567 if (l)
3568 {
3569 if (special == 0) /* LHS */
3570 {
3571 dequote_list (l);
3572 r = string_list (l);
3573 }
3574 else
3575 {
3576 /* Need to figure out whether or not we should call dequote_escapes
3577 or a new dequote_ctlnul function here, and under what
3578 circumstances. */
3579 qflags = QGLOB_CVTNULL;
3580 if (special == 2)
3581 qflags |= QGLOB_REGEXP;
3582 word_list_remove_quoted_nulls (l);
3583 p = string_list (l);
3584 r = quote_string_for_globbing (p, qflags);
3585 free (p);
3586 }
3587 dispose_words (l);
3588 }
3589 else
3590 r = (char *)NULL;
3591
3592 return r;
3593 }
3594 #endif
3595
3596 /* Call expand_word_internal to expand W and handle error returns.
3597 A convenience function for functions that don't want to handle
3598 any errors or free any memory before aborting. */
3599 static WORD_LIST *
3600 call_expand_word_internal (w, q, i, c, e)
3601 WORD_DESC *w;
3602 int q, i, *c, *e;
3603 {
3604 WORD_LIST *result;
3605
3606 result = expand_word_internal (w, q, i, c, e);
3607 if (result == &expand_word_error || result == &expand_word_fatal)
3608 {
3609 /* By convention, each time this error is returned, w->word has
3610 already been freed (it sometimes may not be in the fatal case,
3611 but that doesn't result in a memory leak because we're going
3612 to exit in most cases). */
3613 w->word = (char *)NULL;
3614 last_command_exit_value = EXECUTION_FAILURE;
3615 exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF);
3616 /* NOTREACHED */
3617 return (NULL);
3618 }
3619 else
3620 return (result);
3621 }
3622
3623 /* Perform parameter expansion, command substitution, and arithmetic
3624 expansion on STRING, as if it were a word. Leave the result quoted.
3625 Since this does not perform word splitting, it leaves quoted nulls
3626 in the result. */
3627 static WORD_LIST *
3628 expand_string_internal (string, quoted)
3629 char *string;
3630 int quoted;
3631 {
3632 WORD_DESC td;
3633 WORD_LIST *tresult;
3634
3635 if (string == 0 || *string == 0)
3636 return ((WORD_LIST *)NULL);
3637
3638 td.flags = 0;
3639 td.word = savestring (string);
3640
3641 tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3642
3643 FREE (td.word);
3644 return (tresult);
3645 }
3646
3647 /* Expand STRING by performing parameter expansion, command substitution,
3648 and arithmetic expansion. Dequote the resulting WORD_LIST before
3649 returning it, but do not perform word splitting. The call to
3650 remove_quoted_nulls () is in here because word splitting normally
3651 takes care of quote removal. */
3652 WORD_LIST *
3653 expand_string_unsplit (string, quoted)
3654 char *string;
3655 int quoted;
3656 {
3657 WORD_LIST *value;
3658
3659 if (string == 0 || *string == '\0')
3660 return ((WORD_LIST *)NULL);
3661
3662 expand_no_split_dollar_star = 1;
3663 value = expand_string_internal (string, quoted);
3664 expand_no_split_dollar_star = 0;
3665
3666 if (value)
3667 {
3668 if (value->word)
3669 {
3670 remove_quoted_nulls (value->word->word);
3671 value->word->flags &= ~W_HASQUOTEDNULL;
3672 }
3673 dequote_list (value);
3674 }
3675 return (value);
3676 }
3677
3678 /* Expand the rhs of an assignment statement */
3679 WORD_LIST *
3680 expand_string_assignment (string, quoted)
3681 char *string;
3682 int quoted;
3683 {
3684 WORD_DESC td;
3685 WORD_LIST *value;
3686
3687 if (string == 0 || *string == '\0')
3688 return ((WORD_LIST *)NULL);
3689
3690 expand_no_split_dollar_star = 1;
3691
3692 td.flags = W_ASSIGNRHS;
3693 td.word = savestring (string);
3694 value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3695 FREE (td.word);
3696
3697 expand_no_split_dollar_star = 0;
3698
3699 if (value)
3700 {
3701 if (value->word)
3702 {
3703 remove_quoted_nulls (value->word->word);
3704 value->word->flags &= ~W_HASQUOTEDNULL;
3705 }
3706 dequote_list (value);
3707 }
3708 return (value);
3709 }
3710
3711
3712 /* Expand one of the PS? prompt strings. This is a sort of combination of
3713 expand_string_unsplit and expand_string_internal, but returns the
3714 passed string when an error occurs. Might want to trap other calls
3715 to jump_to_top_level here so we don't endlessly loop. */
3716 WORD_LIST *
3717 expand_prompt_string (string, quoted, wflags)
3718 char *string;
3719 int quoted;
3720 int wflags;
3721 {
3722 WORD_LIST *value;
3723 WORD_DESC td;
3724
3725 if (string == 0 || *string == 0)
3726 return ((WORD_LIST *)NULL);
3727
3728 td.flags = wflags;
3729 td.word = savestring (string);
3730
3731 no_longjmp_on_fatal_error = 1;
3732 value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL);
3733 no_longjmp_on_fatal_error = 0;
3734
3735 if (value == &expand_word_error || value == &expand_word_fatal)
3736 {
3737 value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL);
3738 return value;
3739 }
3740 FREE (td.word);
3741 if (value)
3742 {
3743 if (value->word)
3744 {
3745 remove_quoted_nulls (value->word->word);
3746 value->word->flags &= ~W_HASQUOTEDNULL;
3747 }
3748 dequote_list (value);
3749 }
3750 return (value);
3751 }
3752
3753 /* Expand STRING just as if you were expanding a word, but do not dequote
3754 the resultant WORD_LIST. This is called only from within this file,
3755 and is used to correctly preserve quoted characters when expanding
3756 things like ${1+"$@"}. This does parameter expansion, command
3757 substitution, arithmetic expansion, and word splitting. */
3758 static WORD_LIST *
3759 expand_string_leave_quoted (string, quoted)
3760 char *string;
3761 int quoted;
3762 {
3763 WORD_LIST *tlist;
3764 WORD_LIST *tresult;
3765
3766 if (string == 0 || *string == '\0')
3767 return ((WORD_LIST *)NULL);
3768
3769 tlist = expand_string_internal (string, quoted);
3770
3771 if (tlist)
3772 {
3773 tresult = word_list_split (tlist);
3774 dispose_words (tlist);
3775 return (tresult);
3776 }
3777 return ((WORD_LIST *)NULL);
3778 }
3779
3780 /* This does not perform word splitting or dequote the WORD_LIST
3781 it returns. */
3782 static WORD_LIST *
3783 expand_string_for_rhs (string, quoted, dollar_at_p, expanded_p)
3784 char *string;
3785 int quoted, *dollar_at_p, *expanded_p;
3786 {
3787 WORD_DESC td;
3788 WORD_LIST *tresult;
3789
3790 if (string == 0 || *string == '\0')
3791 return (WORD_LIST *)NULL;
3792
3793 expand_no_split_dollar_star = 1;
3794 td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */
3795 td.word = string;
3796 tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, expanded_p);
3797 expand_no_split_dollar_star = 0;
3798
3799 return (tresult);
3800 }
3801
3802 /* Expand STRING just as if you were expanding a word. This also returns
3803 a list of words. Note that filename globbing is *NOT* done for word
3804 or string expansion, just when the shell is expanding a command. This
3805 does parameter expansion, command substitution, arithmetic expansion,
3806 and word splitting. Dequote the resultant WORD_LIST before returning. */
3807 WORD_LIST *
3808 expand_string (string, quoted)
3809 char *string;
3810 int quoted;
3811 {
3812 WORD_LIST *result;
3813
3814 if (string == 0 || *string == '\0')
3815 return ((WORD_LIST *)NULL);
3816
3817 result = expand_string_leave_quoted (string, quoted);
3818 return (result ? dequote_list (result) : result);
3819 }
3820
3821 /***************************************************
3822 * *
3823 * Functions to handle quoting chars *
3824 * *
3825 ***************************************************/
3826
3827 /* Conventions:
3828
3829 A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string.
3830 The parser passes CTLNUL as CTLESC CTLNUL. */
3831
3832 /* Quote escape characters in string s, but no other characters. This is
3833 used to protect CTLESC and CTLNUL in variable values from the rest of
3834 the word expansion process after the variable is expanded (word splitting
3835 and filename generation). If IFS is null, we quote spaces as well, just
3836 in case we split on spaces later (in the case of unquoted $@, we will
3837 eventually attempt to split the entire word on spaces). Corresponding
3838 code exists in dequote_escapes. Even if we don't end up splitting on
3839 spaces, quoting spaces is not a problem. This should never be called on
3840 a string that is quoted with single or double quotes or part of a here
3841 document (effectively double-quoted). */
3842 char *
3843 quote_escapes (string)
3844 char *string;
3845 {
3846 register char *s, *t;
3847 size_t slen;
3848 char *result, *send;
3849 int quote_spaces, skip_ctlesc, skip_ctlnul;
3850 DECLARE_MBSTATE;
3851
3852 slen = strlen (string);
3853 send = string + slen;
3854
3855 quote_spaces = (ifs_value && *ifs_value == 0);
3856
3857 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
3858 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
3859
3860 t = result = (char *)xmalloc ((slen * 2) + 1);
3861 s = string;
3862
3863 while (*s)
3864 {
3865 if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' '))
3866 *t++ = CTLESC;
3867 COPY_CHAR_P (t, s, send);
3868 }
3869 *t = '\0';
3870
3871 return (result);
3872 }
3873
3874 static WORD_LIST *
3875 list_quote_escapes (list)
3876 WORD_LIST *list;
3877 {
3878 register WORD_LIST *w;
3879 char *t;
3880
3881 for (w = list; w; w = w->next)
3882 {
3883 t = w->word->word;
3884 w->word->word = quote_escapes (t);
3885 free (t);
3886 }
3887 return list;
3888 }
3889
3890 /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL.
3891
3892 The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL.
3893 This is necessary to make unquoted CTLESC and CTLNUL characters in the
3894 data stream pass through properly.
3895
3896 We need to remove doubled CTLESC characters inside quoted strings before
3897 quoting the entire string, so we do not double the number of CTLESC
3898 characters.
3899
3900 Also used by parts of the pattern substitution code. */
3901 char *
3902 dequote_escapes (string)
3903 char *string;
3904 {
3905 register char *s, *t, *s1;
3906 size_t slen;
3907 char *result, *send;
3908 int quote_spaces;
3909 DECLARE_MBSTATE;
3910
3911 if (string == 0)
3912 return string;
3913
3914 slen = strlen (string);
3915 send = string + slen;
3916
3917 t = result = (char *)xmalloc (slen + 1);
3918
3919 if (strchr (string, CTLESC) == 0)
3920 return (strcpy (result, string));
3921
3922 quote_spaces = (ifs_value && *ifs_value == 0);
3923
3924 s = string;
3925 while (*s)
3926 {
3927 if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' ')))
3928 {
3929 s++;
3930 if (*s == '\0')
3931 break;
3932 }
3933 COPY_CHAR_P (t, s, send);
3934 }
3935 *t = '\0';
3936
3937 return result;
3938 }
3939
3940 static WORD_LIST *
3941 list_dequote_escapes (list)
3942 WORD_LIST *list;
3943 {
3944 register WORD_LIST *w;
3945 char *t;
3946
3947 for (w = list; w; w = w->next)
3948 {
3949 t = w->word->word;
3950 w->word->word = dequote_escapes (t);
3951 free (t);
3952 }
3953 return list;
3954 }
3955
3956 /* Return a new string with the quoted representation of character C.
3957 This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be
3958 set in any resultant WORD_DESC where this value is the word. */
3959 static char *
3960 make_quoted_char (c)
3961 int c;
3962 {
3963 char *temp;
3964
3965 temp = (char *)xmalloc (3);
3966 if (c == 0)
3967 {
3968 temp[0] = CTLNUL;
3969 temp[1] = '\0';
3970 }
3971 else
3972 {
3973 temp[0] = CTLESC;
3974 temp[1] = c;
3975 temp[2] = '\0';
3976 }
3977 return (temp);
3978 }
3979
3980 /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so
3981 the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where
3982 this value is the word. */
3983 char *
3984 quote_string (string)
3985 char *string;
3986 {
3987 register char *t;
3988 size_t slen;
3989 char *result, *send;
3990
3991 if (*string == 0)
3992 {
3993 result = (char *)xmalloc (2);
3994 result[0] = CTLNUL;
3995 result[1] = '\0';
3996 }
3997 else
3998 {
3999 DECLARE_MBSTATE;
4000
4001 slen = strlen (string);
4002 send = string + slen;
4003
4004 result = (char *)xmalloc ((slen * 2) + 1);
4005
4006 for (t = result; string < send; )
4007 {
4008 *t++ = CTLESC;
4009 COPY_CHAR_P (t, string, send);
4010 }
4011 *t = '\0';
4012 }
4013 return (result);
4014 }
4015
4016 /* De-quote quoted characters in STRING. */
4017 char *
4018 dequote_string (string)
4019 char *string;
4020 {
4021 register char *s, *t;
4022 size_t slen;
4023 char *result, *send;
4024 DECLARE_MBSTATE;
4025
4026 #if defined (DEBUG)
4027 if (string[0] == CTLESC && string[1] == 0)
4028 internal_inform ("dequote_string: string with bare CTLESC");
4029 #endif
4030
4031 slen = strlen (string);
4032
4033 t = result = (char *)xmalloc (slen + 1);
4034
4035 if (QUOTED_NULL (string))
4036 {
4037 result[0] = '\0';
4038 return (result);
4039 }
4040
4041 /* A string consisting of only a single CTLESC should pass through unchanged */
4042 if (string[0] == CTLESC && string[1] == 0)
4043 {
4044 result[0] = CTLESC;
4045 result[1] = '\0';
4046 return (result);
4047 }
4048
4049 /* If no character in the string can be quoted, don't bother examining
4050 each character. Just return a copy of the string passed to us. */
4051 if (strchr (string, CTLESC) == NULL)
4052 return (strcpy (result, string));
4053
4054 send = string + slen;
4055 s = string;
4056 while (*s)
4057 {
4058 if (*s == CTLESC)
4059 {
4060 s++;
4061 if (*s == '\0')
4062 break;
4063 }
4064 COPY_CHAR_P (t, s, send);
4065 }
4066
4067 *t = '\0';
4068 return (result);
4069 }
4070
4071 /* Quote the entire WORD_LIST list. */
4072 static WORD_LIST *
4073 quote_list (list)
4074 WORD_LIST *list;
4075 {
4076 register WORD_LIST *w;
4077 char *t;
4078
4079 for (w = list; w; w = w->next)
4080 {
4081 t = w->word->word;
4082 w->word->word = quote_string (t);
4083 if (*t == 0)
4084 w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */
4085 w->word->flags |= W_QUOTED;
4086 free (t);
4087 }
4088 return list;
4089 }
4090
4091 /* De-quote quoted characters in each word in LIST. */
4092 WORD_LIST *
4093 dequote_list (list)
4094 WORD_LIST *list;
4095 {
4096 register char *s;
4097 register WORD_LIST *tlist;
4098
4099 for (tlist = list; tlist; tlist = tlist->next)
4100 {
4101 s = dequote_string (tlist->word->word);
4102 if (QUOTED_NULL (tlist->word->word))
4103 tlist->word->flags &= ~W_HASQUOTEDNULL;
4104 free (tlist->word->word);
4105 tlist->word->word = s;
4106 }
4107 return list;
4108 }
4109
4110 /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed
4111 string. */
4112 char *
4113 remove_quoted_escapes (string)
4114 char *string;
4115 {
4116 char *t;
4117
4118 if (string)
4119 {
4120 t = dequote_escapes (string);
4121 strcpy (string, t);
4122 free (t);
4123 }
4124
4125 return (string);
4126 }
4127
4128 /* Perform quoted null character removal on STRING. We don't allow any
4129 quoted null characters in the middle or at the ends of strings because
4130 of how expand_word_internal works. remove_quoted_nulls () turns
4131 STRING into an empty string iff it only consists of a quoted null,
4132 and removes all unquoted CTLNUL characters. */
4133 char *
4134 remove_quoted_nulls (string)
4135 char *string;
4136 {
4137 register size_t slen;
4138 register int i, j, prev_i;
4139 DECLARE_MBSTATE;
4140
4141 if (strchr (string, CTLNUL) == 0) /* XXX */
4142 return string; /* XXX */
4143
4144 slen = strlen (string);
4145 i = j = 0;
4146
4147 while (i < slen)
4148 {
4149 if (string[i] == CTLESC)
4150 {
4151 /* Old code had j++, but we cannot assume that i == j at this
4152 point -- what if a CTLNUL has already been removed from the
4153 string? We don't want to drop the CTLESC or recopy characters
4154 that we've already copied down. */
4155 i++; string[j++] = CTLESC;
4156 if (i == slen)
4157 break;
4158 }
4159 else if (string[i] == CTLNUL)
4160 {
4161 i++;
4162 continue;
4163 }
4164
4165 prev_i = i;
4166 ADVANCE_CHAR (string, slen, i);
4167 if (j < prev_i)
4168 {
4169 do string[j++] = string[prev_i++]; while (prev_i < i);
4170 }
4171 else
4172 j = i;
4173 }
4174 string[j] = '\0';
4175
4176 return (string);
4177 }
4178
4179 /* Perform quoted null character removal on each element of LIST.
4180 This modifies LIST. */
4181 void
4182 word_list_remove_quoted_nulls (list)
4183 WORD_LIST *list;
4184 {
4185 register WORD_LIST *t;
4186
4187 for (t = list; t; t = t->next)
4188 {
4189 remove_quoted_nulls (t->word->word);
4190 t->word->flags &= ~W_HASQUOTEDNULL;
4191 }
4192 }
4193
4194 /* **************************************************************** */
4195 /* */
4196 /* Functions for Matching and Removing Patterns */
4197 /* */
4198 /* **************************************************************** */
4199
4200 #if defined (HANDLE_MULTIBYTE)
4201 #if 0 /* Currently unused */
4202 static unsigned char *
4203 mb_getcharlens (string, len)
4204 char *string;
4205 int len;
4206 {
4207 int i, offset, last;
4208 unsigned char *ret;
4209 char *p;
4210 DECLARE_MBSTATE;
4211
4212 i = offset = 0;
4213 last = 0;
4214 ret = (unsigned char *)xmalloc (len);
4215 memset (ret, 0, len);
4216 while (string[last])
4217 {
4218 ADVANCE_CHAR (string, len, offset);
4219 ret[last] = offset - last;
4220 last = offset;
4221 }
4222 return ret;
4223 }
4224 #endif
4225 #endif
4226
4227 /* Remove the portion of PARAM matched by PATTERN according to OP, where OP
4228 can have one of 4 values:
4229 RP_LONG_LEFT remove longest matching portion at start of PARAM
4230 RP_SHORT_LEFT remove shortest matching portion at start of PARAM
4231 RP_LONG_RIGHT remove longest matching portion at end of PARAM
4232 RP_SHORT_RIGHT remove shortest matching portion at end of PARAM
4233 */
4234
4235 #define RP_LONG_LEFT 1
4236 #define RP_SHORT_LEFT 2
4237 #define RP_LONG_RIGHT 3
4238 #define RP_SHORT_RIGHT 4
4239
4240 /* Returns its first argument if nothing matched; new memory otherwise */
4241 static char *
4242 remove_upattern (param, pattern, op)
4243 char *param, *pattern;
4244 int op;
4245 {
4246 register size_t len;
4247 register char *end;
4248 register char *p, *ret, c;
4249
4250 len = STRLEN (param);
4251 end = param + len;
4252
4253 switch (op)
4254 {
4255 case RP_LONG_LEFT: /* remove longest match at start */
4256 for (p = end; p >= param; p--)
4257 {
4258 c = *p; *p = '\0';
4259 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4260 {
4261 *p = c;
4262 return (savestring (p));
4263 }
4264 *p = c;
4265
4266 }
4267 break;
4268
4269 case RP_SHORT_LEFT: /* remove shortest match at start */
4270 for (p = param; p <= end; p++)
4271 {
4272 c = *p; *p = '\0';
4273 if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4274 {
4275 *p = c;
4276 return (savestring (p));
4277 }
4278 *p = c;
4279 }
4280 break;
4281
4282 case RP_LONG_RIGHT: /* remove longest match at end */
4283 for (p = param; p <= end; p++)
4284 {
4285 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4286 {
4287 c = *p; *p = '\0';
4288 ret = savestring (param);
4289 *p = c;
4290 return (ret);
4291 }
4292 }
4293 break;
4294
4295 case RP_SHORT_RIGHT: /* remove shortest match at end */
4296 for (p = end; p >= param; p--)
4297 {
4298 if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4299 {
4300 c = *p; *p = '\0';
4301 ret = savestring (param);
4302 *p = c;
4303 return (ret);
4304 }
4305 }
4306 break;
4307 }
4308
4309 return (param); /* no match, return original string */
4310 }
4311
4312 #if defined (HANDLE_MULTIBYTE)
4313 /* Returns its first argument if nothing matched; new memory otherwise */
4314 static wchar_t *
4315 remove_wpattern (wparam, wstrlen, wpattern, op)
4316 wchar_t *wparam;
4317 size_t wstrlen;
4318 wchar_t *wpattern;
4319 int op;
4320 {
4321 wchar_t wc, *ret;
4322 int n;
4323
4324 switch (op)
4325 {
4326 case RP_LONG_LEFT: /* remove longest match at start */
4327 for (n = wstrlen; n >= 0; n--)
4328 {
4329 wc = wparam[n]; wparam[n] = L'\0';
4330 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4331 {
4332 wparam[n] = wc;
4333 return (wcsdup (wparam + n));
4334 }
4335 wparam[n] = wc;
4336 }
4337 break;
4338
4339 case RP_SHORT_LEFT: /* remove shortest match at start */
4340 for (n = 0; n <= wstrlen; n++)
4341 {
4342 wc = wparam[n]; wparam[n] = L'\0';
4343 if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4344 {
4345 wparam[n] = wc;
4346 return (wcsdup (wparam + n));
4347 }
4348 wparam[n] = wc;
4349 }
4350 break;
4351
4352 case RP_LONG_RIGHT: /* remove longest match at end */
4353 for (n = 0; n <= wstrlen; n++)
4354 {
4355 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4356 {
4357 wc = wparam[n]; wparam[n] = L'\0';
4358 ret = wcsdup (wparam);
4359 wparam[n] = wc;
4360 return (ret);
4361 }
4362 }
4363 break;
4364
4365 case RP_SHORT_RIGHT: /* remove shortest match at end */
4366 for (n = wstrlen; n >= 0; n--)
4367 {
4368 if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH)
4369 {
4370 wc = wparam[n]; wparam[n] = L'\0';
4371 ret = wcsdup (wparam);
4372 wparam[n] = wc;
4373 return (ret);
4374 }
4375 }
4376 break;
4377 }
4378
4379 return (wparam); /* no match, return original string */
4380 }
4381 #endif /* HANDLE_MULTIBYTE */
4382
4383 static char *
4384 remove_pattern (param, pattern, op)
4385 char *param, *pattern;
4386 int op;
4387 {
4388 char *xret;
4389
4390 if (param == NULL)
4391 return (param);
4392 if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */
4393 return (savestring (param));
4394
4395 #if defined (HANDLE_MULTIBYTE)
4396 if (MB_CUR_MAX > 1)
4397 {
4398 wchar_t *ret, *oret;
4399 size_t n;
4400 wchar_t *wparam, *wpattern;
4401 mbstate_t ps;
4402
4403 n = xdupmbstowcs (&wpattern, NULL, pattern);
4404 if (n == (size_t)-1)
4405 {
4406 xret = remove_upattern (param, pattern, op);
4407 return ((xret == param) ? savestring (param) : xret);
4408 }
4409 n = xdupmbstowcs (&wparam, NULL, param);
4410
4411 if (n == (size_t)-1)
4412 {
4413 free (wpattern);
4414 xret = remove_upattern (param, pattern, op);
4415 return ((xret == param) ? savestring (param) : xret);
4416 }
4417 oret = ret = remove_wpattern (wparam, n, wpattern, op);
4418 /* Don't bother to convert wparam back to multibyte string if nothing
4419 matched; just return copy of original string */
4420 if (ret == wparam)
4421 {
4422 free (wparam);
4423 free (wpattern);
4424 return (savestring (param));
4425 }
4426
4427 free (wparam);
4428 free (wpattern);
4429
4430 n = strlen (param);
4431 xret = (char *)xmalloc (n + 1);
4432 memset (&ps, '\0', sizeof (mbstate_t));
4433 n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps);
4434 xret[n] = '\0'; /* just to make sure */
4435 free (oret);
4436 return xret;
4437 }
4438 else
4439 #endif
4440 {
4441 xret = remove_upattern (param, pattern, op);
4442 return ((xret == param) ? savestring (param) : xret);
4443 }
4444 }
4445
4446 /* Match PAT anywhere in STRING and return the match boundaries.
4447 This returns 1 in case of a successful match, 0 otherwise. SP
4448 and EP are pointers into the string where the match begins and
4449 ends, respectively. MTYPE controls what kind of match is attempted.
4450 MATCH_BEG and MATCH_END anchor the match at the beginning and end
4451 of the string, respectively. The longest match is returned. */
4452 static int
4453 match_upattern (string, pat, mtype, sp, ep)
4454 char *string, *pat;
4455 int mtype;
4456 char **sp, **ep;
4457 {
4458 int c, mlen;
4459 size_t len;
4460 register char *p, *p1, *npat;
4461 char *end;
4462 int n1;
4463
4464 /* If the pattern doesn't match anywhere in the string, go ahead and
4465 short-circuit right away. A minor optimization, saves a bunch of
4466 unnecessary calls to strmatch (up to N calls for a string of N
4467 characters) if the match is unsuccessful. To preserve the semantics
4468 of the substring matches below, we make sure that the pattern has
4469 `*' as first and last character, making a new pattern if necessary. */
4470 /* XXX - check this later if I ever implement `**' with special meaning,
4471 since this will potentially result in `**' at the beginning or end */
4472 len = STRLEN (pat);
4473 if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*')
4474 {
4475 int unescaped_backslash;
4476 char *pp;
4477
4478 p = npat = (char *)xmalloc (len + 3);
4479 p1 = pat;
4480 if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob))
4481 *p++ = '*';
4482 while (*p1)
4483 *p++ = *p1++;
4484 #if 1
4485 /* Need to also handle a pattern that ends with an unescaped backslash.
4486 For right now, we ignore it because the pattern matching code will
4487 fail the match anyway */
4488 /* If the pattern ends with a `*' we leave it alone if it's preceded by
4489 an even number of backslashes, but if it's escaped by a backslash
4490 we need to add another `*'. */
4491 if (p1[-1] == '*' && (unescaped_backslash = p1[-2] == '\\'))
4492 {
4493 pp = p1 - 3;
4494 while (pp >= pat && *pp-- == '\\')
4495 unescaped_backslash = 1 - unescaped_backslash;
4496 if (unescaped_backslash)
4497 *p++ = '*';
4498 }
4499 else if (p1[-1] != '*')
4500 *p++ = '*';
4501 #else
4502 if (p1[-1] != '*' || p1[-2] == '\\')
4503 *p++ = '*';
4504 #endif
4505 *p = '\0';
4506 }
4507 else
4508 npat = pat;
4509 c = strmatch (npat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
4510 if (npat != pat)
4511 free (npat);
4512 if (c == FNM_NOMATCH)
4513 return (0);
4514
4515 len = STRLEN (string);
4516 end = string + len;
4517
4518 mlen = umatchlen (pat, len);
4519
4520 switch (mtype)
4521 {
4522 case MATCH_ANY:
4523 for (p = string; p <= end; p++)
4524 {
4525 if (match_pattern_char (pat, p, FNMATCH_IGNCASE))
4526 {
4527 p1 = (mlen == -1) ? end : p + mlen;
4528 /* p1 - p = length of portion of string to be considered
4529 p = current position in string
4530 mlen = number of characters consumed by match (-1 for entire string)
4531 end = end of string
4532 we want to break immediately if the potential match len
4533 is greater than the number of characters remaining in the
4534 string
4535 */
4536 if (p1 > end)
4537 break;
4538 for ( ; p1 >= p; p1--)
4539 {
4540 c = *p1; *p1 = '\0';
4541 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4542 {
4543 *p1 = c;
4544 *sp = p;
4545 *ep = p1;
4546 return 1;
4547 }
4548 *p1 = c;
4549 #if 1
4550 /* If MLEN != -1, we have a fixed length pattern. */
4551 if (mlen != -1)
4552 break;
4553 #endif
4554 }
4555 }
4556 }
4557
4558 return (0);
4559
4560 case MATCH_BEG:
4561 if (match_pattern_char (pat, string, FNMATCH_IGNCASE) == 0)
4562 return (0);
4563
4564 for (p = (mlen == -1) ? end : string + mlen; p >= string; p--)
4565 {
4566 c = *p; *p = '\0';
4567 if (strmatch (pat, string, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4568 {
4569 *p = c;
4570 *sp = string;
4571 *ep = p;
4572 return 1;
4573 }
4574 *p = c;
4575 /* If MLEN != -1, we have a fixed length pattern. */
4576 if (mlen != -1)
4577 break;
4578 }
4579
4580 return (0);
4581
4582 case MATCH_END:
4583 for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++)
4584 {
4585 if (strmatch (pat, p, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4586 {
4587 *sp = p;
4588 *ep = end;
4589 return 1;
4590 }
4591 /* If MLEN != -1, we have a fixed length pattern. */
4592 if (mlen != -1)
4593 break;
4594 }
4595
4596 return (0);
4597 }
4598
4599 return (0);
4600 }
4601
4602 #if defined (HANDLE_MULTIBYTE)
4603
4604 #define WFOLD(c) (match_ignore_case && iswupper (c) ? towlower (c) : (c))
4605
4606 /* Match WPAT anywhere in WSTRING and return the match boundaries.
4607 This returns 1 in case of a successful match, 0 otherwise. Wide
4608 character version. */
4609 static int
4610 match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep)
4611 wchar_t *wstring;
4612 char **indices;
4613 size_t wstrlen;
4614 wchar_t *wpat;
4615 int mtype;
4616 char **sp, **ep;
4617 {
4618 wchar_t wc, *wp, *nwpat, *wp1;
4619 size_t len;
4620 int mlen;
4621 int n, n1, n2, simple;
4622
4623 simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'[');
4624 #if defined (EXTENDED_GLOB)
4625 if (extended_glob)
4626 simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/
4627 #endif
4628
4629 /* If the pattern doesn't match anywhere in the string, go ahead and
4630 short-circuit right away. A minor optimization, saves a bunch of
4631 unnecessary calls to strmatch (up to N calls for a string of N
4632 characters) if the match is unsuccessful. To preserve the semantics
4633 of the substring matches below, we make sure that the pattern has
4634 `*' as first and last character, making a new pattern if necessary. */
4635 len = wcslen (wpat);
4636 if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*')
4637 {
4638 int unescaped_backslash;
4639 wchar_t *wpp;
4640
4641 wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t));
4642 wp1 = wpat;
4643 if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob))
4644 *wp++ = L'*';
4645 while (*wp1 != L'\0')
4646 *wp++ = *wp1++;
4647 #if 1
4648 /* See comments above in match_upattern. */
4649 if (wp1[-1] == L'*' && (unescaped_backslash = wp1[-2] == L'\\'))
4650 {
4651 wpp = wp1 - 3;
4652 while (wpp >= wpat && *wpp-- == L'\\')
4653 unescaped_backslash = 1 - unescaped_backslash;
4654 if (unescaped_backslash)
4655 *wp++ = L'*';
4656 }
4657 else if (wp1[-1] != L'*')
4658 *wp++ = L'*';
4659 #else
4660 if (wp1[-1] != L'*' || wp1[-2] == L'\\')
4661 *wp++ = L'*';
4662 #endif
4663 *wp = '\0';
4664 }
4665 else
4666 nwpat = wpat;
4667 len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE);
4668 if (nwpat != wpat)
4669 free (nwpat);
4670 if (len == FNM_NOMATCH)
4671 return (0);
4672
4673 mlen = wmatchlen (wpat, wstrlen);
4674
4675 /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */
4676 switch (mtype)
4677 {
4678 case MATCH_ANY:
4679 for (n = 0; n <= wstrlen; n++)
4680 {
4681 n2 = simple ? (WFOLD(*wpat) == WFOLD(wstring[n])) : match_pattern_wchar (wpat, wstring + n, FNMATCH_IGNCASE);
4682 if (n2)
4683 {
4684 n1 = (mlen == -1) ? wstrlen : n + mlen;
4685 if (n1 > wstrlen)
4686 break;
4687
4688 for ( ; n1 >= n; n1--)
4689 {
4690 wc = wstring[n1]; wstring[n1] = L'\0';
4691 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4692 {
4693 wstring[n1] = wc;
4694 *sp = indices[n];
4695 *ep = indices[n1];
4696 return 1;
4697 }
4698 wstring[n1] = wc;
4699 /* If MLEN != -1, we have a fixed length pattern. */
4700 if (mlen != -1)
4701 break;
4702 }
4703 }
4704 }
4705
4706 return (0);
4707
4708 case MATCH_BEG:
4709 if (match_pattern_wchar (wpat, wstring, FNMATCH_IGNCASE) == 0)
4710 return (0);
4711
4712 for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--)
4713 {
4714 wc = wstring[n]; wstring[n] = L'\0';
4715 if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4716 {
4717 wstring[n] = wc;
4718 *sp = indices[0];
4719 *ep = indices[n];
4720 return 1;
4721 }
4722 wstring[n] = wc;
4723 /* If MLEN != -1, we have a fixed length pattern. */
4724 if (mlen != -1)
4725 break;
4726 }
4727
4728 return (0);
4729
4730 case MATCH_END:
4731 for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++)
4732 {
4733 if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG | FNMATCH_IGNCASE) == 0)
4734 {
4735 *sp = indices[n];
4736 *ep = indices[wstrlen];
4737 return 1;
4738 }
4739 /* If MLEN != -1, we have a fixed length pattern. */
4740 if (mlen != -1)
4741 break;
4742 }
4743
4744 return (0);
4745 }
4746
4747 return (0);
4748 }
4749 #undef WFOLD
4750 #endif /* HANDLE_MULTIBYTE */
4751
4752 static int
4753 match_pattern (string, pat, mtype, sp, ep)
4754 char *string, *pat;
4755 int mtype;
4756 char **sp, **ep;
4757 {
4758 #if defined (HANDLE_MULTIBYTE)
4759 int ret;
4760 size_t n;
4761 wchar_t *wstring, *wpat;
4762 char **indices;
4763 size_t slen, plen, mslen, mplen;
4764 #endif
4765
4766 if (string == 0 || pat == 0 || *pat == 0)
4767 return (0);
4768
4769 #if defined (HANDLE_MULTIBYTE)
4770 if (MB_CUR_MAX > 1)
4771 {
4772 if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0)
4773 return (match_upattern (string, pat, mtype, sp, ep));
4774
4775 n = xdupmbstowcs (&wpat, NULL, pat);
4776 if (n == (size_t)-1)
4777 return (match_upattern (string, pat, mtype, sp, ep));
4778 n = xdupmbstowcs (&wstring, &indices, string);
4779 if (n == (size_t)-1)
4780 {
4781 free (wpat);
4782 return (match_upattern (string, pat, mtype, sp, ep));
4783 }
4784 ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep);
4785
4786 free (wpat);
4787 free (wstring);
4788 free (indices);
4789
4790 return (ret);
4791 }
4792 else
4793 #endif
4794 return (match_upattern (string, pat, mtype, sp, ep));
4795 }
4796
4797 static int
4798 getpatspec (c, value)
4799 int c;
4800 char *value;
4801 {
4802 if (c == '#')
4803 return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT);
4804 else /* c == '%' */
4805 return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT);
4806 }
4807
4808 /* Posix.2 says that the WORD should be run through tilde expansion,
4809 parameter expansion, command substitution and arithmetic expansion.
4810 This leaves the result quoted, so quote_string_for_globbing () has
4811 to be called to fix it up for strmatch (). If QUOTED is non-zero,
4812 it means that the entire expression was enclosed in double quotes.
4813 This means that quoting characters in the pattern do not make any
4814 special pattern characters quoted. For example, the `*' in the
4815 following retains its special meaning: "${foo#'*'}". */
4816 static char *
4817 getpattern (value, quoted, expandpat)
4818 char *value;
4819 int quoted, expandpat;
4820 {
4821 char *pat, *tword;
4822 WORD_LIST *l;
4823 #if 0
4824 int i;
4825 #endif
4826 /* There is a problem here: how to handle single or double quotes in the
4827 pattern string when the whole expression is between double quotes?
4828 POSIX.2 says that enclosing double quotes do not cause the pattern to
4829 be quoted, but does that leave us a problem with @ and array[@] and their
4830 expansions inside a pattern? */
4831 #if 0
4832 if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword)
4833 {
4834 i = 0;
4835 pat = string_extract_double_quoted (tword, &i, SX_STRIPDQ);
4836 free (tword);
4837 tword = pat;
4838 }
4839 #endif
4840
4841 /* expand_string_for_rhs () leaves WORD quoted and does not perform
4842 word splitting. */
4843 l = *value ? expand_string_for_rhs (value,
4844 (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted,
4845 (int *)NULL, (int *)NULL)
4846 : (WORD_LIST *)0;
4847 pat = string_list (l);
4848 dispose_words (l);
4849 if (pat)
4850 {
4851 tword = quote_string_for_globbing (pat, QGLOB_CVTNULL);
4852 free (pat);
4853 pat = tword;
4854 }
4855 return (pat);
4856 }
4857
4858 #if 0
4859 /* Handle removing a pattern from a string as a result of ${name%[%]value}
4860 or ${name#[#]value}. */
4861 static char *
4862 variable_remove_pattern (value, pattern, patspec, quoted)
4863 char *value, *pattern;
4864 int patspec, quoted;
4865 {
4866 char *tword;
4867
4868 tword = remove_pattern (value, pattern, patspec);
4869
4870 return (tword);
4871 }
4872 #endif
4873
4874 static char *
4875 list_remove_pattern (list, pattern, patspec, itype, quoted)
4876 WORD_LIST *list;
4877 char *pattern;
4878 int patspec, itype, quoted;
4879 {
4880 WORD_LIST *new, *l;
4881 WORD_DESC *w;
4882 char *tword;
4883
4884 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
4885 {
4886 tword = remove_pattern (l->word->word, pattern, patspec);
4887 w = alloc_word_desc ();
4888 w->word = tword ? tword : savestring ("");
4889 new = make_word_list (w, new);
4890 }
4891
4892 l = REVERSE_LIST (new, WORD_LIST *);
4893 tword = string_list_pos_params (itype, l, quoted);
4894 dispose_words (l);
4895
4896 return (tword);
4897 }
4898
4899 static char *
4900 parameter_list_remove_pattern (itype, pattern, patspec, quoted)
4901 int itype;
4902 char *pattern;
4903 int patspec, quoted;
4904 {
4905 char *ret;
4906 WORD_LIST *list;
4907
4908 list = list_rest_of_args ();
4909 if (list == 0)
4910 return ((char *)NULL);
4911 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4912 dispose_words (list);
4913 return (ret);
4914 }
4915
4916 #if defined (ARRAY_VARS)
4917 static char *
4918 array_remove_pattern (var, pattern, patspec, varname, quoted)
4919 SHELL_VAR *var;
4920 char *pattern;
4921 int patspec;
4922 char *varname; /* so we can figure out how it's indexed */
4923 int quoted;
4924 {
4925 ARRAY *a;
4926 HASH_TABLE *h;
4927 int itype;
4928 char *ret;
4929 WORD_LIST *list;
4930 SHELL_VAR *v;
4931
4932 /* compute itype from varname here */
4933 v = array_variable_part (varname, &ret, 0);
4934
4935 /* XXX */
4936 if (v && invisible_p (v))
4937 return ((char *)NULL);
4938
4939 itype = ret[0];
4940
4941 a = (v && array_p (v)) ? array_cell (v) : 0;
4942 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
4943
4944 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
4945 if (list == 0)
4946 return ((char *)NULL);
4947 ret = list_remove_pattern (list, pattern, patspec, itype, quoted);
4948 dispose_words (list);
4949
4950 return ret;
4951 }
4952 #endif /* ARRAY_VARS */
4953
4954 static char *
4955 parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags)
4956 char *varname, *value;
4957 int ind;
4958 char *patstr;
4959 int rtype, quoted, flags;
4960 {
4961 int vtype, patspec, starsub;
4962 char *temp1, *val, *pattern;
4963 SHELL_VAR *v;
4964
4965 if (value == 0)
4966 return ((char *)NULL);
4967
4968 this_command_name = varname;
4969
4970 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
4971 if (vtype == -1)
4972 return ((char *)NULL);
4973
4974 starsub = vtype & VT_STARSUB;
4975 vtype &= ~VT_STARSUB;
4976
4977 patspec = getpatspec (rtype, patstr);
4978 if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT)
4979 patstr++;
4980
4981 /* Need to pass getpattern newly-allocated memory in case of expansion --
4982 the expansion code will free the passed string on an error. */
4983 temp1 = savestring (patstr);
4984 pattern = getpattern (temp1, quoted, 1);
4985 free (temp1);
4986
4987 temp1 = (char *)NULL; /* shut up gcc */
4988 switch (vtype)
4989 {
4990 case VT_VARIABLE:
4991 case VT_ARRAYMEMBER:
4992 temp1 = remove_pattern (val, pattern, patspec);
4993 if (vtype == VT_VARIABLE)
4994 FREE (val);
4995 if (temp1)
4996 {
4997 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
4998 ? quote_string (temp1)
4999 : quote_escapes (temp1);
5000 free (temp1);
5001 temp1 = val;
5002 }
5003 break;
5004 #if defined (ARRAY_VARS)
5005 case VT_ARRAYVAR:
5006 temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted);
5007 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5008 {
5009 val = quote_escapes (temp1);
5010 free (temp1);
5011 temp1 = val;
5012 }
5013 break;
5014 #endif
5015 case VT_POSPARMS:
5016 temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted);
5017 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5018 {
5019 val = quote_escapes (temp1);
5020 free (temp1);
5021 temp1 = val;
5022 }
5023 break;
5024 }
5025
5026 FREE (pattern);
5027 return temp1;
5028 }
5029
5030 static char *
5031 string_var_assignment (v, s)
5032 SHELL_VAR *v;
5033 char *s;
5034 {
5035 char flags[MAX_ATTRIBUTES], *ret, *val;
5036 int i;
5037
5038 val = sh_quote_reusable (s, 0);
5039 i = var_attribute_string (v, 0, flags);
5040 ret = (char *)xmalloc (i + strlen (val) + strlen (v->name) + 16 + MAX_ATTRIBUTES);
5041 if (i > 0)
5042 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
5043 else
5044 sprintf (ret, "%s=%s", v->name, val);
5045 free (val);
5046 return ret;
5047 }
5048
5049 #if defined (ARRAY_VARS)
5050 static char *
5051 array_var_assignment (v, itype, quoted)
5052 SHELL_VAR *v;
5053 int itype, quoted;
5054 {
5055 char *ret, *val, flags[MAX_ATTRIBUTES];
5056 int i;
5057
5058 if (v == 0)
5059 return (char *)NULL;
5060 val = array_p (v) ? array_to_assign (array_cell (v), 0)
5061 : assoc_to_assign (assoc_cell (v), 0);
5062 if (val == 0)
5063 {
5064 val = (char *)xmalloc (3);
5065 val[0] = '(';
5066 val[1] = ')';
5067 val[2] = 0;
5068 }
5069 else
5070 {
5071 ret = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) ? quote_string (val) : quote_escapes (val);
5072 free (val);
5073 val = ret;
5074 }
5075 i = var_attribute_string (v, 0, flags);
5076 ret = (char *)xmalloc (i + strlen (val) + strlen (v->name) + 16);
5077 sprintf (ret, "declare -%s %s=%s", flags, v->name, val);
5078 free (val);
5079 return ret;
5080 }
5081 #endif
5082
5083 static char *
5084 pos_params_assignment (list, itype, quoted)
5085 WORD_LIST *list;
5086 int itype;
5087 int quoted;
5088 {
5089 char *temp, *ret;
5090
5091 /* first, we transform the list to quote each word. */
5092 temp = list_transform ('Q', (SHELL_VAR *)0, list, itype, quoted);
5093 ret = (char *)xmalloc (strlen (temp) + 8);
5094 strcpy (ret, "set -- ");
5095 strcpy (ret + 7, temp);
5096 free (temp);
5097 return ret;
5098 }
5099
5100 static char *
5101 string_transform (xc, v, s)
5102 int xc;
5103 SHELL_VAR *v;
5104 char *s;
5105 {
5106 char *ret, flags[MAX_ATTRIBUTES];
5107 int i;
5108
5109 if (((xc == 'A' || xc == 'a') && v == 0) || (xc != 'a' && s == 0))
5110 return (char *)NULL;
5111
5112 switch (xc)
5113 {
5114 /* Transformations that interrogate the variable */
5115 case 'a':
5116 i = var_attribute_string (v, 0, flags);
5117 ret = (i > 0) ? savestring (flags) : (char *)NULL;
5118 break;
5119 case 'A':
5120 ret = string_var_assignment (v, s);
5121 break;
5122 /* Transformations that modify the variable's value */
5123 case 'E':
5124 ret = ansiexpand (s, 0, strlen (s), (int *)0);
5125 break;
5126 case 'P':
5127 ret = decode_prompt_string (s);
5128 break;
5129 case 'Q':
5130 ret = sh_quote_reusable (s, 0);
5131 break;
5132 default:
5133 ret = (char *)NULL;
5134 break;
5135 }
5136 return ret;
5137 }
5138
5139 static char *
5140 list_transform (xc, v, list, itype, quoted)
5141 int xc;
5142 SHELL_VAR *v;
5143 WORD_LIST *list;
5144 int itype, quoted;
5145 {
5146 WORD_LIST *new, *l;
5147 WORD_DESC *w;
5148 char *tword;
5149
5150 for (new = (WORD_LIST *)NULL, l = list; l; l = l->next)
5151 {
5152 tword = string_transform (xc, v, l->word->word);
5153 w = alloc_word_desc ();
5154 w->word = tword ? tword : savestring (""); /* XXX */
5155 new = make_word_list (w, new);
5156 }
5157
5158 l = REVERSE_LIST (new, WORD_LIST *);
5159 tword = string_list_pos_params (itype, l, quoted);
5160 dispose_words (l);
5161
5162 return (tword);
5163 }
5164
5165 static char *
5166 parameter_list_transform (xc, itype, quoted)
5167 int xc;
5168 int itype;
5169 int quoted;
5170 {
5171 char *ret;
5172 WORD_LIST *list;
5173
5174 list = list_rest_of_args ();
5175 if (list == 0)
5176 return ((char *)NULL);
5177 if (xc == 'A')
5178 return (pos_params_assignment (list, itype, quoted));
5179 ret = list_transform (xc, (SHELL_VAR *)0, list, itype, quoted);
5180 dispose_words (list);
5181 return (ret);
5182 }
5183
5184 #if defined (ARRAY_VARS)
5185 static char *
5186 array_transform (xc, var, varname, quoted)
5187 int xc;
5188 SHELL_VAR *var;
5189 char *varname; /* so we can figure out how it's indexed */
5190 int quoted;
5191 {
5192 ARRAY *a;
5193 HASH_TABLE *h;
5194 int itype;
5195 char *ret;
5196 WORD_LIST *list;
5197 SHELL_VAR *v;
5198
5199 /* compute itype from varname here */
5200 v = array_variable_part (varname, &ret, 0);
5201
5202 /* XXX */
5203 if (v && invisible_p (v))
5204 return ((char *)NULL);
5205
5206 itype = ret[0];
5207
5208 if (xc == 'A')
5209 return (array_var_assignment (v, itype, quoted));
5210
5211 a = (v && array_p (v)) ? array_cell (v) : 0;
5212 h = (v && assoc_p (v)) ? assoc_cell (v) : 0;
5213
5214 list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0);
5215 if (list == 0)
5216 return ((char *)NULL);
5217 ret = list_transform (xc, v, list, itype, quoted);
5218 dispose_words (list);
5219
5220 return ret;
5221 }
5222 #endif /* ARRAY_VARS */
5223
5224 static char *
5225 parameter_brace_transform (varname, value, ind, xform, rtype, quoted, flags)
5226 char *varname, *value;
5227 int ind;
5228 char *xform;
5229 int rtype, quoted, flags;
5230 {
5231 int vtype, xc;
5232 char *temp1, *val;
5233 SHELL_VAR *v;
5234
5235 xc = xform[0];
5236 if (value == 0 && xc != 'A' && xc != 'a')
5237 return ((char *)NULL);
5238
5239 this_command_name = varname;
5240
5241 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
5242 if (vtype == -1)
5243 return ((char *)NULL);
5244
5245 /* check for valid values of xc */
5246 switch (xc)
5247 {
5248 case 'a': /* expand to a string with just attributes */
5249 case 'A': /* expand as an assignment statement with attributes */
5250 case 'E': /* expand like $'...' */
5251 case 'P': /* expand like prompt string */
5252 case 'Q': /* quote reusably */
5253 break;
5254 default:
5255 return &expand_param_error;
5256 }
5257
5258 temp1 = (char *)NULL; /* shut up gcc */
5259 switch (vtype & ~VT_STARSUB)
5260 {
5261 case VT_VARIABLE:
5262 case VT_ARRAYMEMBER:
5263 temp1 = string_transform (xc, v, val);
5264 if (vtype == VT_VARIABLE)
5265 FREE (val);
5266 if (temp1)
5267 {
5268 val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
5269 ? quote_string (temp1)
5270 : quote_escapes (temp1);
5271 free (temp1);
5272 temp1 = val;
5273 }
5274 break;
5275 #if defined (ARRAY_VARS)
5276 case VT_ARRAYVAR:
5277 temp1 = array_transform (xc, v, varname, quoted);
5278 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5279 {
5280 val = quote_escapes (temp1);
5281 free (temp1);
5282 temp1 = val;
5283 }
5284 break;
5285 #endif
5286 case VT_POSPARMS:
5287 temp1 = parameter_list_transform (xc, varname[0], quoted);
5288 if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0))
5289 {
5290 val = quote_escapes (temp1);
5291 free (temp1);
5292 temp1 = val;
5293 }
5294 break;
5295 }
5296
5297 return temp1;
5298 }
5299
5300 /*******************************************
5301 * *
5302 * Functions to expand WORD_DESCs *
5303 * *
5304 *******************************************/
5305
5306 /* Expand WORD, performing word splitting on the result. This does
5307 parameter expansion, command substitution, arithmetic expansion,
5308 word splitting, and quote removal. */
5309
5310 WORD_LIST *
5311 expand_word (word, quoted)
5312 WORD_DESC *word;
5313 int quoted;
5314 {
5315 WORD_LIST *result, *tresult;
5316
5317 tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
5318 result = word_list_split (tresult);
5319 dispose_words (tresult);
5320 return (result ? dequote_list (result) : result);
5321 }
5322
5323 /* Expand WORD, but do not perform word splitting on the result. This
5324 does parameter expansion, command substitution, arithmetic expansion,
5325 and quote removal. */
5326 WORD_LIST *
5327 expand_word_unsplit (word, quoted)
5328 WORD_DESC *word;
5329 int quoted;
5330 {
5331 WORD_LIST *result;
5332
5333 expand_no_split_dollar_star = 1;
5334 #if defined (HANDLE_MULTIBYTE)
5335 if (ifs_firstc[0] == 0)
5336 #else
5337 if (ifs_firstc == 0)
5338 #endif
5339 word->flags |= W_NOSPLIT;
5340 word->flags |= W_NOSPLIT2;
5341 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
5342 expand_no_split_dollar_star = 0;
5343
5344 return (result ? dequote_list (result) : result);
5345 }
5346
5347 /* Perform shell expansions on WORD, but do not perform word splitting or
5348 quote removal on the result. Virtually identical to expand_word_unsplit;
5349 could be combined if implementations don't diverge. */
5350 WORD_LIST *
5351 expand_word_leave_quoted (word, quoted)
5352 WORD_DESC *word;
5353 int quoted;
5354 {
5355 WORD_LIST *result;
5356
5357 expand_no_split_dollar_star = 1;
5358 #if defined (HANDLE_MULTIBYTE)
5359 if (ifs_firstc[0] == 0)
5360 #else
5361 if (ifs_firstc == 0)
5362 #endif
5363 word->flags |= W_NOSPLIT;
5364 word->flags |= W_NOSPLIT2;
5365 result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL);
5366 expand_no_split_dollar_star = 0;
5367
5368 return result;
5369 }
5370
5371 #if defined (PROCESS_SUBSTITUTION)
5372
5373 /*****************************************************************/
5374 /* */
5375 /* Hacking Process Substitution */
5376 /* */
5377 /*****************************************************************/
5378
5379 #if !defined (HAVE_DEV_FD)
5380 /* Named pipes must be removed explicitly with `unlink'. This keeps a list
5381 of FIFOs the shell has open. unlink_fifo_list will walk the list and
5382 unlink all of them. add_fifo_list adds the name of an open FIFO to the
5383 list. NFIFO is a count of the number of FIFOs in the list. */
5384 #define FIFO_INCR 20
5385
5386 struct temp_fifo {
5387 char *file;
5388 pid_t proc;
5389 };
5390
5391 static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL;
5392 static int nfifo;
5393 static int fifo_list_size;
5394
5395 void
5396 clear_fifo_list ()
5397 {
5398 }
5399
5400 char *
5401 copy_fifo_list (sizep)
5402 int *sizep;
5403 {
5404 if (sizep)
5405 *sizep = 0;
5406 return (char *)NULL;
5407 }
5408
5409 static void
5410 add_fifo_list (pathname)
5411 char *pathname;
5412 {
5413 if (nfifo >= fifo_list_size - 1)
5414 {
5415 fifo_list_size += FIFO_INCR;
5416 fifo_list = (struct temp_fifo *)xrealloc (fifo_list,
5417 fifo_list_size * sizeof (struct temp_fifo));
5418 }
5419
5420 fifo_list[nfifo].file = savestring (pathname);
5421 nfifo++;
5422 }
5423
5424 void
5425 unlink_fifo (i)
5426 int i;
5427 {
5428 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
5429 {
5430 unlink (fifo_list[i].file);
5431 free (fifo_list[i].file);
5432 fifo_list[i].file = (char *)NULL;
5433 fifo_list[i].proc = -1;
5434 }
5435 }
5436
5437 void
5438 unlink_fifo_list ()
5439 {
5440 int saved, i, j;
5441
5442 if (nfifo == 0)
5443 return;
5444
5445 for (i = saved = 0; i < nfifo; i++)
5446 {
5447 if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1))
5448 {
5449 unlink (fifo_list[i].file);
5450 free (fifo_list[i].file);
5451 fifo_list[i].file = (char *)NULL;
5452 fifo_list[i].proc = -1;
5453 }
5454 else
5455 saved++;
5456 }
5457
5458 /* If we didn't remove some of the FIFOs, compact the list. */
5459 if (saved)
5460 {
5461 for (i = j = 0; i < nfifo; i++)
5462 if (fifo_list[i].file)
5463 {
5464 fifo_list[j].file = fifo_list[i].file;
5465 fifo_list[j].proc = fifo_list[i].proc;
5466 j++;
5467 }
5468 nfifo = j;
5469 }
5470 else
5471 nfifo = 0;
5472 }
5473
5474 /* Take LIST, which is a bitmap denoting active FIFOs in fifo_list
5475 from some point in the past, and close all open FIFOs in fifo_list
5476 that are not marked as active in LIST. If LIST is NULL, close
5477 everything in fifo_list. LSIZE is the number of elements in LIST, in
5478 case it's larger than fifo_list_size (size of fifo_list). */
5479 void
5480 close_new_fifos (list, lsize)
5481 char *list;
5482 int lsize;
5483 {
5484 int i;
5485
5486 if (list == 0)
5487 {
5488 unlink_fifo_list ();
5489 return;
5490 }
5491
5492 for (i = 0; i < lsize; i++)
5493 if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1)
5494 unlink_fifo (i);
5495
5496 for (i = lsize; i < fifo_list_size; i++)
5497 unlink_fifo (i);
5498 }
5499
5500 int
5501 fifos_pending ()
5502 {
5503 return nfifo;
5504 }
5505
5506 int
5507 num_fifos ()
5508 {
5509 return nfifo;
5510 }
5511
5512 static char *
5513 make_named_pipe ()
5514 {
5515 char *tname;
5516
5517 tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR);
5518 if (mkfifo (tname, 0600) < 0)
5519 {
5520 free (tname);
5521 return ((char *)NULL);
5522 }
5523
5524 add_fifo_list (tname);
5525 return (tname);
5526 }
5527
5528 #else /* HAVE_DEV_FD */
5529
5530 /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell
5531 has open to children. NFDS is a count of the number of bits currently
5532 set in DEV_FD_LIST. TOTFDS is a count of the highest possible number
5533 of open files. */
5534 static char *dev_fd_list = (char *)NULL;
5535 static int nfds;
5536 static int totfds; /* The highest possible number of open files. */
5537
5538 void
5539 clear_fifo (i)
5540 int i;
5541 {
5542 if (dev_fd_list[i])
5543 {
5544 dev_fd_list[i] = 0;
5545 nfds--;
5546 }
5547 }
5548
5549 void
5550 clear_fifo_list ()
5551 {
5552 register int i;
5553
5554 if (nfds == 0)
5555 return;
5556
5557 for (i = 0; nfds && i < totfds; i++)
5558 clear_fifo (i);
5559
5560 nfds = 0;
5561 }
5562
5563 char *
5564 copy_fifo_list (sizep)
5565 int *sizep;
5566 {
5567 char *ret;
5568
5569 if (nfds == 0 || totfds == 0)
5570 {
5571 if (sizep)
5572 *sizep = 0;
5573 return (char *)NULL;
5574 }
5575
5576 if (sizep)
5577 *sizep = totfds;
5578 ret = (char *)xmalloc (totfds);
5579 return (memcpy (ret, dev_fd_list, totfds));
5580 }
5581
5582 static void
5583 add_fifo_list (fd)
5584 int fd;
5585 {
5586 if (dev_fd_list == 0 || fd >= totfds)
5587 {
5588 int ofds;
5589
5590 ofds = totfds;
5591 totfds = getdtablesize ();
5592 if (totfds < 0 || totfds > 256)
5593 totfds = 256;
5594 if (fd >= totfds)
5595 totfds = fd + 2;
5596
5597 dev_fd_list = (char *)xrealloc (dev_fd_list, totfds);
5598 memset (dev_fd_list + ofds, '\0', totfds - ofds);
5599 }
5600
5601 dev_fd_list[fd] = 1;
5602 nfds++;
5603 }
5604
5605 int
5606 fifos_pending ()
5607 {
5608 return 0; /* used for cleanup; not needed with /dev/fd */
5609 }
5610
5611 int
5612 num_fifos ()
5613 {
5614 return nfds;
5615 }
5616
5617 void
5618 unlink_fifo (fd)
5619 int fd;
5620 {
5621 if (dev_fd_list[fd])
5622 {
5623 close (fd);
5624 dev_fd_list[fd] = 0;
5625 nfds--;
5626 }
5627 }
5628
5629 void
5630 unlink_fifo_list ()
5631 {
5632 register int i;
5633
5634 if (nfds == 0)
5635 return;
5636
5637 for (i = 0; nfds && i < totfds; i++)
5638 unlink_fifo (i);
5639
5640 nfds = 0;
5641 }
5642
5643 /* Take LIST, which is a snapshot copy of dev_fd_list from some point in
5644 the past, and close all open fds in dev_fd_list that are not marked
5645 as open in LIST. If LIST is NULL, close everything in dev_fd_list.
5646 LSIZE is the number of elements in LIST, in case it's larger than
5647 totfds (size of dev_fd_list). */
5648 void
5649 close_new_fifos (list, lsize)
5650 char *list;
5651 int lsize;
5652 {
5653 int i;
5654
5655 if (list == 0)
5656 {
5657 unlink_fifo_list ();
5658 return;
5659 }
5660
5661 for (i = 0; i < lsize; i++)
5662 if (list[i] == 0 && i < totfds && dev_fd_list[i])
5663 unlink_fifo (i);
5664
5665 for (i = lsize; i < totfds; i++)
5666 unlink_fifo (i);
5667 }
5668
5669 #if defined (NOTDEF)
5670 print_dev_fd_list ()
5671 {
5672 register int i;
5673
5674 fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ());
5675 fflush (stderr);
5676
5677 for (i = 0; i < totfds; i++)
5678 {
5679 if (dev_fd_list[i])
5680 fprintf (stderr, " %d", i);
5681 }
5682 fprintf (stderr, "\n");
5683 }
5684 #endif /* NOTDEF */
5685
5686 static char *
5687 make_dev_fd_filename (fd)
5688 int fd;
5689 {
5690 char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p;
5691
5692 ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8);
5693
5694 strcpy (ret, DEV_FD_PREFIX);
5695 p = inttostr (fd, intbuf, sizeof (intbuf));
5696 strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p);
5697
5698 add_fifo_list (fd);
5699 return (ret);
5700 }
5701
5702 #endif /* HAVE_DEV_FD */
5703
5704 /* Return a filename that will open a connection to the process defined by
5705 executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return
5706 a filename in /dev/fd corresponding to a descriptor that is one of the
5707 ends of the pipe. If not defined, we use named pipes on systems that have
5708 them. Systems without /dev/fd and named pipes are out of luck.
5709
5710 OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or
5711 use the read end of the pipe and dup that file descriptor to fd 0 in
5712 the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for
5713 writing or use the write end of the pipe in the child, and dup that
5714 file descriptor to fd 1 in the child. The parent does the opposite. */
5715
5716 static char *
5717 process_substitute (string, open_for_read_in_child)
5718 char *string;
5719 int open_for_read_in_child;
5720 {
5721 char *pathname;
5722 int fd, result;
5723 pid_t old_pid, pid;
5724 #if defined (HAVE_DEV_FD)
5725 int parent_pipe_fd, child_pipe_fd;
5726 int fildes[2];
5727 #endif /* HAVE_DEV_FD */
5728 #if defined (JOB_CONTROL)
5729 pid_t old_pipeline_pgrp;
5730 #endif
5731
5732 if (!string || !*string || wordexp_only)
5733 return ((char *)NULL);
5734
5735 #if !defined (HAVE_DEV_FD)
5736 pathname = make_named_pipe ();
5737 #else /* HAVE_DEV_FD */
5738 if (pipe (fildes) < 0)
5739 {
5740 sys_error ("%s", _("cannot make pipe for process substitution"));
5741 return ((char *)NULL);
5742 }
5743 /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of
5744 the pipe in the parent, otherwise the read end. */
5745 parent_pipe_fd = fildes[open_for_read_in_child];
5746 child_pipe_fd = fildes[1 - open_for_read_in_child];
5747 /* Move the parent end of the pipe to some high file descriptor, to
5748 avoid clashes with FDs used by the script. */
5749 parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64);
5750
5751 pathname = make_dev_fd_filename (parent_pipe_fd);
5752 #endif /* HAVE_DEV_FD */
5753
5754 if (pathname == 0)
5755 {
5756 sys_error ("%s", _("cannot make pipe for process substitution"));
5757 return ((char *)NULL);
5758 }
5759
5760 old_pid = last_made_pid;
5761
5762 #if defined (JOB_CONTROL)
5763 old_pipeline_pgrp = pipeline_pgrp;
5764 if (pipeline_pgrp == 0 || (subshell_environment & (SUBSHELL_PIPE|SUBSHELL_FORK|SUBSHELL_ASYNC)) == 0)
5765 pipeline_pgrp = shell_pgrp;
5766 save_pipeline (1);
5767 #endif /* JOB_CONTROL */
5768
5769 pid = make_child ((char *)NULL, 1);
5770 if (pid == 0)
5771 {
5772 reset_terminating_signals (); /* XXX */
5773 free_pushed_string_input ();
5774 /* Cancel traps, in trap.c. */
5775 restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */
5776 QUIT; /* catch any interrupts we got post-fork */
5777 setup_async_signals ();
5778 subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB;
5779
5780 /* if we're expanding a redirection, we shouldn't have access to the
5781 temporary environment, but commands in the subshell should have
5782 access to their own temporary environment. */
5783 if (expanding_redir)
5784 flush_temporary_env ();
5785 }
5786
5787 #if defined (JOB_CONTROL)
5788 set_sigchld_handler ();
5789 stop_making_children ();
5790 /* XXX - should we only do this in the parent? (as in command subst) */
5791 pipeline_pgrp = old_pipeline_pgrp;
5792 #else
5793 stop_making_children ();
5794 #endif /* JOB_CONTROL */
5795
5796 if (pid < 0)
5797 {
5798 sys_error ("%s", _("cannot make child for process substitution"));
5799 free (pathname);
5800 #if defined (HAVE_DEV_FD)
5801 close (parent_pipe_fd);
5802 close (child_pipe_fd);
5803 #endif /* HAVE_DEV_FD */
5804 return ((char *)NULL);
5805 }
5806
5807 if (pid > 0)
5808 {
5809 #if defined (JOB_CONTROL)
5810 if (last_procsub_child)
5811 discard_last_procsub_child ();
5812 last_procsub_child = restore_pipeline (0);
5813 #endif
5814
5815 #if !defined (HAVE_DEV_FD)
5816 fifo_list[nfifo-1].proc = pid;
5817 #endif
5818
5819 last_made_pid = old_pid;
5820
5821 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
5822 close_pgrp_pipe ();
5823 #endif /* JOB_CONTROL && PGRP_PIPE */
5824
5825 #if defined (HAVE_DEV_FD)
5826 close (child_pipe_fd);
5827 #endif /* HAVE_DEV_FD */
5828
5829 return (pathname);
5830 }
5831
5832 set_sigint_handler ();
5833
5834 #if defined (JOB_CONTROL)
5835 set_job_control (0);
5836 #endif /* JOB_CONTROL */
5837
5838 #if !defined (HAVE_DEV_FD)
5839 /* Open the named pipe in the child. */
5840 fd = open (pathname, open_for_read_in_child ? O_RDONLY : O_WRONLY);
5841 if (fd < 0)
5842 {
5843 /* Two separate strings for ease of translation. */
5844 if (open_for_read_in_child)
5845 sys_error (_("cannot open named pipe %s for reading"), pathname);
5846 else
5847 sys_error (_("cannot open named pipe %s for writing"), pathname);
5848
5849 exit (127);
5850 }
5851 if (open_for_read_in_child)
5852 {
5853 if (sh_unset_nodelay_mode (fd) < 0)
5854 {
5855 sys_error (_("cannot reset nodelay mode for fd %d"), fd);
5856 exit (127);
5857 }
5858 }
5859 #else /* HAVE_DEV_FD */
5860 fd = child_pipe_fd;
5861 #endif /* HAVE_DEV_FD */
5862
5863 /* Discard buffered stdio output before replacing the underlying file
5864 descriptor. */
5865 if (open_for_read_in_child == 0)
5866 fpurge (stdout);
5867
5868 if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0)
5869 {
5870 sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname,
5871 open_for_read_in_child ? 0 : 1);
5872 exit (127);
5873 }
5874
5875 if (fd != (open_for_read_in_child ? 0 : 1))
5876 close (fd);
5877
5878 /* Need to close any files that this process has open to pipes inherited
5879 from its parent. */
5880 if (current_fds_to_close)
5881 {
5882 close_fd_bitmap (current_fds_to_close);
5883 current_fds_to_close = (struct fd_bitmap *)NULL;
5884 }
5885
5886 #if defined (HAVE_DEV_FD)
5887 /* Make sure we close the parent's end of the pipe and clear the slot
5888 in the fd list so it is not closed later, if reallocated by, for
5889 instance, pipe(2). */
5890 close (parent_pipe_fd);
5891 dev_fd_list[parent_pipe_fd] = 0;
5892 #endif /* HAVE_DEV_FD */
5893
5894 /* subshells shouldn't have this flag, which controls using the temporary
5895 environment for variable lookups. We have already flushed the temporary
5896 environment above in the case we're expanding a redirection, so processes
5897 executed by this command need to be able to set it independently of their
5898 parent. */
5899 expanding_redir = 0;
5900
5901 subshell_level++;
5902 result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST));
5903 subshell_level--;
5904
5905 #if !defined (HAVE_DEV_FD)
5906 /* Make sure we close the named pipe in the child before we exit. */
5907 close (open_for_read_in_child ? 0 : 1);
5908 #endif /* !HAVE_DEV_FD */
5909
5910 last_command_exit_value = result;
5911 result = run_exit_trap ();
5912 exit (result);
5913 /*NOTREACHED*/
5914 }
5915 #endif /* PROCESS_SUBSTITUTION */
5916
5917 /***********************************/
5918 /* */
5919 /* Command Substitution */
5920 /* */
5921 /***********************************/
5922
5923 static char *
5924 read_comsub (fd, quoted, rflag)
5925 int fd, quoted;
5926 int *rflag;
5927 {
5928 char *istring, buf[128], *bufp, *s;
5929 int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul;
5930 ssize_t bufn;
5931 int nullbyte;
5932
5933 istring = (char *)NULL;
5934 istring_index = istring_size = bufn = tflag = 0;
5935
5936 for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++)
5937 skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL;
5938
5939 nullbyte = 0;
5940
5941 /* Read the output of the command through the pipe. This may need to be
5942 changed to understand multibyte characters in the future. */
5943 while (1)
5944 {
5945 if (fd < 0)
5946 break;
5947 if (--bufn <= 0)
5948 {
5949 bufn = zread (fd, buf, sizeof (buf));
5950 if (bufn <= 0)
5951 break;
5952 bufp = buf;
5953 }
5954 c = *bufp++;
5955
5956 if (c == 0)
5957 {
5958 #if 1
5959 if (nullbyte == 0)
5960 {
5961 internal_warning ("%s", _("command substitution: ignored null byte in input"));
5962 nullbyte = 1;
5963 }
5964 #endif
5965 continue;
5966 }
5967
5968 /* Add the character to ISTRING, possibly after resizing it. */
5969 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE);
5970
5971 /* This is essentially quote_string inline */
5972 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */)
5973 istring[istring_index++] = CTLESC;
5974 /* Escape CTLESC and CTLNUL in the output to protect those characters
5975 from the rest of the word expansions (word splitting and globbing.)
5976 This is essentially quote_escapes inline. */
5977 else if (skip_ctlesc == 0 && c == CTLESC)
5978 {
5979 tflag |= W_HASCTLESC;
5980 istring[istring_index++] = CTLESC;
5981 }
5982 else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0)))
5983 istring[istring_index++] = CTLESC;
5984
5985 istring[istring_index++] = c;
5986
5987 #if 0
5988 #if defined (__CYGWIN__)
5989 if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r')
5990 {
5991 istring_index--;
5992 istring[istring_index - 1] = '\n';
5993 }
5994 #endif
5995 #endif
5996 }
5997
5998 if (istring)
5999 istring[istring_index] = '\0';
6000
6001 /* If we read no output, just return now and save ourselves some
6002 trouble. */
6003 if (istring_index == 0)
6004 {
6005 FREE (istring);
6006 if (rflag)
6007 *rflag = tflag;
6008 return (char *)NULL;
6009 }
6010
6011 /* Strip trailing newlines from the output of the command. */
6012 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6013 {
6014 while (istring_index > 0)
6015 {
6016 if (istring[istring_index - 1] == '\n')
6017 {
6018 --istring_index;
6019
6020 /* If the newline was quoted, remove the quoting char. */
6021 if (istring[istring_index - 1] == CTLESC)
6022 --istring_index;
6023 }
6024 else
6025 break;
6026 }
6027 istring[istring_index] = '\0';
6028 }
6029 else
6030 strip_trailing (istring, istring_index - 1, 1);
6031
6032 if (rflag)
6033 *rflag = tflag;
6034 return istring;
6035 }
6036
6037 /* Perform command substitution on STRING. This returns a WORD_DESC * with the
6038 contained string possibly quoted. */
6039 WORD_DESC *
6040 command_substitute (string, quoted)
6041 char *string;
6042 int quoted;
6043 {
6044 pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid;
6045 char *istring, *s;
6046 int result, fildes[2], function_value, pflags, rc, tflag;
6047 WORD_DESC *ret;
6048
6049 istring = (char *)NULL;
6050
6051 /* Don't fork () if there is no need to. In the case of no command to
6052 run, just return NULL. */
6053 #if 1
6054 for (s = string; s && *s && (shellblank (*s) || *s == '\n'); s++)
6055 ;
6056 if (s == 0 || *s == 0)
6057 return ((WORD_DESC *)NULL);
6058 #else
6059 if (!string || !*string || (string[0] == '\n' && !string[1]))
6060 return ((WORD_DESC *)NULL);
6061 #endif
6062
6063 if (wordexp_only && read_but_dont_execute)
6064 {
6065 last_command_exit_value = EX_WEXPCOMSUB;
6066 jump_to_top_level (EXITPROG);
6067 }
6068
6069 /* We're making the assumption here that the command substitution will
6070 eventually run a command from the file system. Since we'll run
6071 maybe_make_export_env in this subshell before executing that command,
6072 the parent shell and any other shells it starts will have to remake
6073 the environment. If we make it before we fork, other shells won't
6074 have to. Don't bother if we have any temporary variable assignments,
6075 though, because the export environment will be remade after this
6076 command completes anyway, but do it if all the words to be expanded
6077 are variable assignments. */
6078 if (subst_assign_varlist == 0 || garglist == 0)
6079 maybe_make_export_env (); /* XXX */
6080
6081 /* Flags to pass to parse_and_execute() */
6082 pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0;
6083
6084 /* Pipe the output of executing STRING into the current shell. */
6085 if (pipe (fildes) < 0)
6086 {
6087 sys_error ("%s", _("cannot make pipe for command substitution"));
6088 goto error_exit;
6089 }
6090
6091 old_pid = last_made_pid;
6092 #if defined (JOB_CONTROL)
6093 old_pipeline_pgrp = pipeline_pgrp;
6094 /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */
6095 if ((subshell_environment & SUBSHELL_PIPE) == 0)
6096 pipeline_pgrp = shell_pgrp;
6097 cleanup_the_pipeline ();
6098 #endif /* JOB_CONTROL */
6099
6100 old_async_pid = last_asynchronous_pid;
6101 pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC);
6102 last_asynchronous_pid = old_async_pid;
6103
6104 if (pid == 0)
6105 {
6106 /* Reset the signal handlers in the child, but don't free the
6107 trap strings. Set a flag noting that we have to free the
6108 trap strings if we run trap to change a signal disposition. */
6109 reset_signal_handlers ();
6110 if (ISINTERRUPT)
6111 {
6112 kill (getpid (), SIGINT);
6113 CLRINTERRUPT; /* if we're ignoring SIGINT somehow */
6114 }
6115 QUIT; /* catch any interrupts we got post-fork */
6116 subshell_environment |= SUBSHELL_RESETTRAP;
6117 }
6118
6119 #if defined (JOB_CONTROL)
6120 /* XXX DO THIS ONLY IN PARENT ? XXX */
6121 set_sigchld_handler ();
6122 stop_making_children ();
6123 if (pid != 0)
6124 pipeline_pgrp = old_pipeline_pgrp;
6125 #else
6126 stop_making_children ();
6127 #endif /* JOB_CONTROL */
6128
6129 if (pid < 0)
6130 {
6131 sys_error (_("cannot make child for command substitution"));
6132 error_exit:
6133
6134 last_made_pid = old_pid;
6135
6136 FREE (istring);
6137 close (fildes[0]);
6138 close (fildes[1]);
6139 return ((WORD_DESC *)NULL);
6140 }
6141
6142 if (pid == 0)
6143 {
6144 /* The currently executing shell is not interactive. */
6145 interactive = 0;
6146
6147 set_sigint_handler (); /* XXX */
6148
6149 free_pushed_string_input ();
6150
6151 /* Discard buffered stdio output before replacing the underlying file
6152 descriptor. */
6153 fpurge (stdout);
6154
6155 if (dup2 (fildes[1], 1) < 0)
6156 {
6157 sys_error ("%s", _("command_substitute: cannot duplicate pipe as fd 1"));
6158 exit (EXECUTION_FAILURE);
6159 }
6160
6161 /* If standard output is closed in the parent shell
6162 (such as after `exec >&-'), file descriptor 1 will be
6163 the lowest available file descriptor, and end up in
6164 fildes[0]. This can happen for stdin and stderr as well,
6165 but stdout is more important -- it will cause no output
6166 to be generated from this command. */
6167 if ((fildes[1] != fileno (stdin)) &&
6168 (fildes[1] != fileno (stdout)) &&
6169 (fildes[1] != fileno (stderr)))
6170 close (fildes[1]);
6171
6172 if ((fildes[0] != fileno (stdin)) &&
6173 (fildes[0] != fileno (stdout)) &&
6174 (fildes[0] != fileno (stderr)))
6175 close (fildes[0]);
6176
6177 #ifdef __CYGWIN__
6178 /* Let stdio know the fd may have changed from text to binary mode, and
6179 make sure to preserve stdout line buffering. */
6180 freopen (NULL, "w", stdout);
6181 sh_setlinebuf (stdout);
6182 #endif /* __CYGWIN__ */
6183
6184 /* This is a subshell environment. */
6185 subshell_environment |= SUBSHELL_COMSUB;
6186
6187 /* Many shells do not appear to inherit the -v option for command
6188 substitutions. */
6189 change_flag ('v', FLAG_OFF);
6190
6191 /* When inherit_errexit option is not enabled, command substitution does
6192 not inherit the -e flag. It is enabled when Posix mode is enabled */
6193 if (inherit_errexit == 0)
6194 {
6195 builtin_ignoring_errexit = 0;
6196 change_flag ('e', FLAG_OFF);
6197 }
6198 set_shellopts ();
6199
6200 /* If we are expanding a redirection, we can dispose of any temporary
6201 environment we received, since redirections are not supposed to have
6202 access to the temporary environment. We will have to see whether this
6203 affects temporary environments supplied to `eval', but the temporary
6204 environment gets copied to builtin_env at some point. */
6205 if (expanding_redir)
6206 {
6207 flush_temporary_env ();
6208 expanding_redir = 0;
6209 }
6210
6211 remove_quoted_escapes (string);
6212
6213 startup_state = 2; /* see if we can avoid a fork */
6214 /* Give command substitution a place to jump back to on failure,
6215 so we don't go back up to main (). */
6216 result = setjmp_nosigs (top_level);
6217
6218 /* If we're running a command substitution inside a shell function,
6219 trap `return' so we don't return from the function in the subshell
6220 and go off to never-never land. */
6221 if (result == 0 && return_catch_flag)
6222 function_value = setjmp_nosigs (return_catch);
6223 else
6224 function_value = 0;
6225
6226 if (result == ERREXIT)
6227 rc = last_command_exit_value;
6228 else if (result == EXITPROG)
6229 rc = last_command_exit_value;
6230 else if (result)
6231 rc = EXECUTION_FAILURE;
6232 else if (function_value)
6233 rc = return_catch_value;
6234 else
6235 {
6236 subshell_level++;
6237 rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST);
6238 subshell_level--;
6239 }
6240
6241 last_command_exit_value = rc;
6242 rc = run_exit_trap ();
6243 #if defined (PROCESS_SUBSTITUTION)
6244 unlink_fifo_list ();
6245 #endif
6246 exit (rc);
6247 }
6248 else
6249 {
6250 #if defined (JOB_CONTROL) && defined (PGRP_PIPE)
6251 close_pgrp_pipe ();
6252 #endif /* JOB_CONTROL && PGRP_PIPE */
6253
6254 close (fildes[1]);
6255
6256 tflag = 0;
6257 istring = read_comsub (fildes[0], quoted, &tflag);
6258
6259 close (fildes[0]);
6260
6261 current_command_subst_pid = pid;
6262 last_command_exit_value = wait_for (pid);
6263 last_command_subst_pid = pid;
6264 last_made_pid = old_pid;
6265
6266 #if defined (JOB_CONTROL)
6267 /* If last_command_exit_value > 128, then the substituted command
6268 was terminated by a signal. If that signal was SIGINT, then send
6269 SIGINT to ourselves. This will break out of loops, for instance. */
6270 if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT)
6271 kill (getpid (), SIGINT);
6272
6273 /* wait_for gives the terminal back to shell_pgrp. If some other
6274 process group should have it, give it away to that group here.
6275 pipeline_pgrp is non-zero only while we are constructing a
6276 pipeline, so what we are concerned about is whether or not that
6277 pipeline was started in the background. A pipeline started in
6278 the background should never get the tty back here. We duplicate
6279 the conditions that wait_for tests to make sure we only give
6280 the terminal back to pipeline_pgrp under the conditions that wait_for
6281 gave it to shell_pgrp. If wait_for doesn't mess with the terminal
6282 pgrp, we should not either. */
6283 if (interactive && pipeline_pgrp != (pid_t)0 && running_in_background == 0 &&
6284 (subshell_environment & (SUBSHELL_ASYNC|SUBSHELL_PIPE)) == 0)
6285 give_terminal_to (pipeline_pgrp, 0);
6286 #endif /* JOB_CONTROL */
6287
6288 ret = alloc_word_desc ();
6289 ret->word = istring;
6290 ret->flags = tflag;
6291
6292 return ret;
6293 }
6294 }
6295
6296 /********************************************************
6297 * *
6298 * Utility functions for parameter expansion *
6299 * *
6300 ********************************************************/
6301
6302 #if defined (ARRAY_VARS)
6303
6304 static arrayind_t
6305 array_length_reference (s)
6306 char *s;
6307 {
6308 int len;
6309 arrayind_t ind;
6310 char *akey;
6311 char *t, c;
6312 ARRAY *array;
6313 HASH_TABLE *h;
6314 SHELL_VAR *var;
6315
6316 var = array_variable_part (s, &t, &len);
6317
6318 /* If unbound variables should generate an error, report one and return
6319 failure. */
6320 if ((var == 0 || invisible_p (var) || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error)
6321 {
6322 c = *--t;
6323 *t = '\0';
6324 last_command_exit_value = EXECUTION_FAILURE;
6325 err_unboundvar (s);
6326 *t = c;
6327 return (-1);
6328 }
6329 else if (var == 0 || invisible_p (var))
6330 return 0;
6331
6332 /* We support a couple of expansions for variables that are not arrays.
6333 We'll return the length of the value for v[0], and 1 for v[@] or
6334 v[*]. Return 0 for everything else. */
6335
6336 array = array_p (var) ? array_cell (var) : (ARRAY *)NULL;
6337 h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL;
6338
6339 if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']')
6340 {
6341 if (assoc_p (var))
6342 return (h ? assoc_num_elements (h) : 0);
6343 else if (array_p (var))
6344 return (array ? array_num_elements (array) : 0);
6345 else
6346 return (var_isset (var) ? 1 : 0);
6347 }
6348
6349 if (assoc_p (var))
6350 {
6351 t[len - 1] = '\0';
6352 akey = expand_assignment_string_to_string (t, 0); /* [ */
6353 t[len - 1] = ']';
6354 if (akey == 0 || *akey == 0)
6355 {
6356 err_badarraysub (t);
6357 FREE (akey);
6358 return (-1);
6359 }
6360 t = assoc_reference (assoc_cell (var), akey);
6361 free (akey);
6362 }
6363 else
6364 {
6365 ind = array_expand_index (var, t, len);
6366 /* negative subscripts to indexed arrays count back from end */
6367 if (var && array_p (var) && ind < 0)
6368 ind = array_max_index (array_cell (var)) + 1 + ind;
6369 if (ind < 0)
6370 {
6371 err_badarraysub (t);
6372 return (-1);
6373 }
6374 if (array_p (var))
6375 t = array_reference (array, ind);
6376 else
6377 t = (ind == 0) ? value_cell (var) : (char *)NULL;
6378 }
6379
6380 len = MB_STRLEN (t);
6381 return (len);
6382 }
6383 #endif /* ARRAY_VARS */
6384
6385 static int
6386 valid_brace_expansion_word (name, var_is_special)
6387 char *name;
6388 int var_is_special;
6389 {
6390 if (DIGIT (*name) && all_digits (name))
6391 return 1;
6392 else if (var_is_special)
6393 return 1;
6394 #if defined (ARRAY_VARS)
6395 else if (valid_array_reference (name, 0))
6396 return 1;
6397 #endif /* ARRAY_VARS */
6398 else if (legal_identifier (name))
6399 return 1;
6400 else
6401 return 0;
6402 }
6403
6404 static int
6405 chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at)
6406 char *name;
6407 int quoted;
6408 int *quoted_dollar_atp, *contains_dollar_at;
6409 {
6410 char *temp1;
6411
6412 if (name == 0)
6413 {
6414 if (quoted_dollar_atp)
6415 *quoted_dollar_atp = 0;
6416 if (contains_dollar_at)
6417 *contains_dollar_at = 0;
6418 return 0;
6419 }
6420
6421 /* check for $@ and $* */
6422 if (name[0] == '@' && name[1] == 0)
6423 {
6424 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6425 *quoted_dollar_atp = 1;
6426 if (contains_dollar_at)
6427 *contains_dollar_at = 1;
6428 return 1;
6429 }
6430 else if (name[0] == '*' && name[1] == '\0' && quoted == 0)
6431 {
6432 if (contains_dollar_at)
6433 *contains_dollar_at = 1;
6434 return 1;
6435 }
6436
6437 /* Now check for ${array[@]} and ${array[*]} */
6438 #if defined (ARRAY_VARS)
6439 else if (valid_array_reference (name, 0))
6440 {
6441 temp1 = mbschr (name, '[');
6442 if (temp1 && temp1[1] == '@' && temp1[2] == ']')
6443 {
6444 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
6445 *quoted_dollar_atp = 1;
6446 if (contains_dollar_at)
6447 *contains_dollar_at = 1;
6448 return 1;
6449 } /* [ */
6450 /* ${array[*]}, when unquoted, should be treated like ${array[@]},
6451 which should result in separate words even when IFS is unset. */
6452 if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0)
6453 {
6454 if (contains_dollar_at)
6455 *contains_dollar_at = 1;
6456 return 1;
6457 }
6458 }
6459 #endif
6460 return 0;
6461 }
6462
6463 /* Parameter expand NAME, and return a new string which is the expansion,
6464 or NULL if there was no expansion.
6465 VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in
6466 the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that
6467 NAME was found inside of a double-quoted expression. */
6468 static WORD_DESC *
6469 parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp)
6470 char *name;
6471 int var_is_special, quoted, pflags;
6472 arrayind_t *indp;
6473 {
6474 WORD_DESC *ret;
6475 char *temp, *tt;
6476 intmax_t arg_index;
6477 SHELL_VAR *var;
6478 int atype, rflags;
6479 arrayind_t ind;
6480
6481 ret = 0;
6482 temp = 0;
6483 rflags = 0;
6484
6485 if (indp)
6486 *indp = INTMAX_MIN;
6487
6488 /* Handle multiple digit arguments, as in ${11}. */
6489 if (legal_number (name, &arg_index))
6490 {
6491 tt = get_dollar_var_value (arg_index);
6492 if (tt)
6493 temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6494 ? quote_string (tt)
6495 : quote_escapes (tt);
6496 else
6497 temp = (char *)NULL;
6498 FREE (tt);
6499 }
6500 else if (var_is_special) /* ${@} */
6501 {
6502 int sindex;
6503 tt = (char *)xmalloc (2 + strlen (name));
6504 tt[sindex = 0] = '$';
6505 strcpy (tt + 1, name);
6506
6507 ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL,
6508 (int *)NULL, (int *)NULL, pflags);
6509 free (tt);
6510 }
6511 #if defined (ARRAY_VARS)
6512 else if (valid_array_reference (name, 0))
6513 {
6514 expand_arrayref:
6515 if (pflags & PF_ASSIGNRHS)
6516 {
6517 var = array_variable_part (name, &tt, (int *)0);
6518 if (ALL_ELEMENT_SUB (tt[0]) && tt[1] == ']')
6519 {
6520 /* Only treat as double quoted if array variable */
6521 if (var && (array_p (var) || assoc_p (var)))
6522 /* XXX - bash-4.4/bash-5.0 pass AV_ASSIGNRHS */
6523 temp = array_value (name, quoted|Q_DOUBLE_QUOTES, AV_ASSIGNRHS, &atype, &ind);
6524 else
6525 temp = array_value (name, quoted, 0, &atype, &ind);
6526 }
6527 else
6528 temp = array_value (name, quoted, 0, &atype, &ind);
6529 }
6530 else
6531 temp = array_value (name, quoted, 0, &atype, &ind);
6532 if (atype == 0 && temp)
6533 {
6534 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6535 ? quote_string (temp)
6536 : quote_escapes (temp);
6537 rflags |= W_ARRAYIND;
6538 if (indp)
6539 *indp = ind;
6540 }
6541 else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6542 rflags |= W_HASQUOTEDNULL;
6543 }
6544 #endif
6545 else if (var = find_variable (name))
6546 {
6547 if (var_isset (var) && invisible_p (var) == 0)
6548 {
6549 #if defined (ARRAY_VARS)
6550 if (assoc_p (var))
6551 temp = assoc_reference (assoc_cell (var), "0");
6552 else if (array_p (var))
6553 temp = array_reference (array_cell (var), 0);
6554 else
6555 temp = value_cell (var);
6556 #else
6557 temp = value_cell (var);
6558 #endif
6559
6560 if (temp)
6561 temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
6562 ? quote_string (temp)
6563 : quote_escapes (temp);
6564 }
6565 else
6566 temp = (char *)NULL;
6567 }
6568 else if (var = find_variable_last_nameref (name, 0))
6569 {
6570 temp = nameref_cell (var);
6571 #if defined (ARRAY_VARS)
6572 /* Handle expanding nameref whose value is x[n] */
6573 if (temp && *temp && valid_array_reference (temp, 0))
6574 {
6575 name = temp;
6576 goto expand_arrayref;
6577 }
6578 else
6579 #endif
6580 /* y=2 ; typeset -n x=y; echo ${x} is not the same as echo ${2} in ksh */
6581 if (temp && *temp && legal_identifier (temp) == 0)
6582 {
6583 last_command_exit_value = EXECUTION_FAILURE;
6584 report_error (_("%s: invalid variable name for name reference"), temp);
6585 temp = &expand_param_error;
6586 }
6587 else
6588 temp = (char *)NULL;
6589 }
6590 else
6591 temp = (char *)NULL;
6592
6593 if (ret == 0)
6594 {
6595 ret = alloc_word_desc ();
6596 ret->word = temp;
6597 ret->flags |= rflags;
6598 }
6599 return ret;
6600 }
6601
6602 static char *
6603 parameter_brace_find_indir (name, var_is_special, quoted, find_nameref)
6604 char *name;
6605 int var_is_special, quoted, find_nameref;
6606 {
6607 char *temp, *t;
6608 WORD_DESC *w;
6609 SHELL_VAR *v;
6610
6611 if (find_nameref && var_is_special == 0 && (v = find_variable_last_nameref (name, 0)) &&
6612 nameref_p (v) && (t = nameref_cell (v)) && *t)
6613 return (savestring (t));
6614
6615 /* If var_is_special == 0, and name is not an array reference, this does
6616 more expansion than necessary. It should really look up the variable's
6617 value and not try to expand it. */
6618 w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0);
6619 t = w->word;
6620 /* Have to dequote here if necessary */
6621 if (t)
6622 {
6623 temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
6624 ? dequote_string (t)
6625 : dequote_escapes (t);
6626 free (t);
6627 t = temp;
6628 }
6629 dispose_word_desc (w);
6630
6631 return t;
6632 }
6633
6634 /* Expand an indirect reference to a variable: ${!NAME} expands to the
6635 value of the variable whose name is the value of NAME. */
6636 static WORD_DESC *
6637 parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at)
6638 char *name;
6639 int var_is_special, quoted;
6640 int *quoted_dollar_atp, *contains_dollar_at;
6641 {
6642 char *temp, *t;
6643 WORD_DESC *w;
6644 SHELL_VAR *v;
6645
6646 /* See if it's a nameref first, behave in ksh93-compatible fashion.
6647 There is at least one incompatibility: given ${!foo[0]} where foo=bar,
6648 bash performs an indirect lookup on foo[0] and expands the result;
6649 ksh93 expands bar[0]. We could do that here -- there are enough usable
6650 primitives to do that -- but do not at this point. */
6651 if (var_is_special == 0 && (v = find_variable_last_nameref (name, 0)))
6652 {
6653 if (nameref_p (v) && (t = nameref_cell (v)) && *t)
6654 {
6655 w = alloc_word_desc ();
6656 w->word = savestring (t);
6657 w->flags = 0;
6658 return w;
6659 }
6660 }
6661
6662 t = parameter_brace_find_indir (name, var_is_special, quoted, 0);
6663
6664 chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at);
6665 if (t == 0)
6666 return (WORD_DESC *)NULL;
6667
6668 if (valid_brace_expansion_word (t, SPECIAL_VAR (t, 0)) == 0)
6669 {
6670 report_error (_("%s: bad substitution"), t);
6671 free (t);
6672 w = alloc_word_desc ();
6673 w->word = &expand_param_error;
6674 w->flags = 0;
6675 return (w);
6676 }
6677
6678 w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0);
6679 free (t);
6680
6681 return w;
6682 }
6683
6684 /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE},
6685 depending on the value of C, the separating character. C can be one of
6686 "-", "+", or "=". QUOTED is true if the entire brace expression occurs
6687 between double quotes. */
6688 static WORD_DESC *
6689 parameter_brace_expand_rhs (name, value, c, quoted, pflags, qdollaratp, hasdollarat)
6690 char *name, *value;
6691 int c, quoted, pflags, *qdollaratp, *hasdollarat;
6692 {
6693 WORD_DESC *w;
6694 WORD_LIST *l;
6695 char *t, *t1, *temp, *vname;
6696 int l_hasdollat, sindex;
6697
6698 /*itrace("parameter_brace_expand_rhs: %s:%s pflags = %d", name, value, pflags);*/
6699 /* If the entire expression is between double quotes, we want to treat
6700 the value as a double-quoted string, with the exception that we strip
6701 embedded unescaped double quotes (for sh backwards compatibility). */
6702 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value)
6703 {
6704 sindex = 0;
6705 temp = string_extract_double_quoted (value, &sindex, SX_STRIPDQ);
6706 }
6707 else
6708 temp = value;
6709
6710 w = alloc_word_desc ();
6711 l_hasdollat = 0;
6712 /* XXX was 0 not quoted */
6713 l = *temp ? expand_string_for_rhs (temp, quoted, &l_hasdollat, (int *)NULL)
6714 : (WORD_LIST *)0;
6715 if (hasdollarat)
6716 *hasdollarat = l_hasdollat || (l && l->next);
6717 if (temp != value)
6718 free (temp);
6719 if (l)
6720 {
6721 /* If l->next is not null, we know that TEMP contained "$@", since that
6722 is the only expansion that creates more than one word. */
6723 if (qdollaratp && ((l_hasdollat && quoted) || l->next))
6724 {
6725 /*itrace("parameter_brace_expand_rhs: %s:%s: l != NULL, set *qdollaratp", name, value);*/
6726 *qdollaratp = 1;
6727 }
6728
6729 /* The expansion of TEMP returned something. We need to treat things
6730 slightly differently if L_HASDOLLAT is non-zero. If we have "$@",
6731 the individual words have already been quoted. We need to turn them
6732 into a string with the words separated by the first character of
6733 $IFS without any additional quoting, so string_list_dollar_at won't
6734 do the right thing. If IFS is null, we want "$@" to split into
6735 separate arguments, not be concatenated, so we use string_list_internal
6736 and mark the word to be split on spaces later. We use
6737 string_list_dollar_star for "$@" otherwise. */
6738 if (l->next && ifs_is_null)
6739 {
6740 temp = string_list_internal (l, " ");
6741 w->flags |= W_SPLITSPACE;
6742 }
6743 else
6744 temp = (l_hasdollat || l->next) ? string_list_dollar_star (l) : string_list (l);
6745
6746 /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is
6747 a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the
6748 flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the
6749 expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
6750 (which is more paranoia than anything else), we need to return the
6751 quoted null string and set the flags to indicate it. */
6752 if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL))
6753 {
6754 w->flags |= W_HASQUOTEDNULL;
6755 /*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null, turning off qdollaratp", name, value);*/
6756 /* If we return a quoted null with L_HASDOLLARAT, we either have a
6757 construct like "${@-$@}" or "${@-${@-$@}}" with no positional
6758 parameters or a quoted expansion of "$@" with $1 == ''. In either
6759 case, we don't want to enable special handling of $@. */
6760 if (qdollaratp && l_hasdollat)
6761 *qdollaratp = 0;
6762 }
6763 dispose_words (l);
6764 }
6765 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && l_hasdollat)
6766 {
6767 /* Posix interp 221 changed the rules on this. The idea is that
6768 something like "$xxx$@" should expand the same as "${foo-$xxx$@}"
6769 when foo and xxx are unset. The problem is that it's not in any
6770 way backwards compatible and few other shells do it. We're eventually
6771 going to try and split the difference (heh) a little bit here. */
6772 /* l_hasdollat == 1 means we saw a quoted dollar at. */
6773
6774 /* The brace expansion occurred between double quotes and there was
6775 a $@ in TEMP. It does not matter if the $@ is quoted, as long as
6776 it does not expand to anything. In this case, we want to return
6777 a quoted empty string. Posix interp 888 */
6778 temp = make_quoted_char ('\0');
6779 w->flags |= W_HASQUOTEDNULL;
6780 /*itrace("parameter_brace_expand_rhs (%s:%s): returning quoted null", name, value);*/
6781 }
6782 else
6783 temp = (char *)NULL;
6784
6785 if (c == '-' || c == '+')
6786 {
6787 w->word = temp;
6788 return w;
6789 }
6790
6791 /* c == '=' */
6792 t = temp ? savestring (temp) : savestring ("");
6793 t1 = dequote_string (t);
6794 free (t);
6795
6796 /* bash-4.4/5.0 */
6797 vname = name;
6798 if (*name == '!' &&
6799 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) || VALID_INDIR_PARAM (name[1])))
6800 {
6801 vname = parameter_brace_find_indir (name + 1, SPECIAL_VAR (name, 1), quoted, 1);
6802 if (vname == 0 || *vname == 0)
6803 {
6804 report_error (_("%s: invalid indirect expansion"), name);
6805 free (vname);
6806 dispose_word (w);
6807 return &expand_wdesc_error;
6808 }
6809 if (legal_identifier (vname) == 0)
6810 {
6811 report_error (_("%s: invalid variable name"), vname);
6812 free (vname);
6813 dispose_word (w);
6814 return &expand_wdesc_error;
6815 }
6816 }
6817
6818 #if defined (ARRAY_VARS)
6819 if (valid_array_reference (vname, 0))
6820 assign_array_element (vname, t1, 0);
6821 else
6822 #endif /* ARRAY_VARS */
6823 bind_variable (vname, t1, 0);
6824
6825 stupidly_hack_special_variables (vname);
6826
6827 if (vname != name)
6828 free (vname);
6829
6830 /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */
6831 free (temp);
6832
6833 w->word = t1;
6834 return w;
6835 }
6836
6837 /* Deal with the right hand side of a ${name:?value} expansion in the case
6838 that NAME is null or not set. If VALUE is non-null it is expanded and
6839 used as the error message to print, otherwise a standard message is
6840 printed. */
6841 static void
6842 parameter_brace_expand_error (name, value)
6843 char *name, *value;
6844 {
6845 WORD_LIST *l;
6846 char *temp;
6847
6848 last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */
6849 if (value && *value)
6850 {
6851 l = expand_string (value, 0);
6852 temp = string_list (l);
6853 report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */
6854 FREE (temp);
6855 dispose_words (l);
6856 }
6857 else
6858 report_error (_("%s: parameter null or not set"), name);
6859
6860 /* Free the data we have allocated during this expansion, since we
6861 are about to longjmp out. */
6862 free (name);
6863 FREE (value);
6864 }
6865
6866 /* Return 1 if NAME is something for which parameter_brace_expand_length is
6867 OK to do. */
6868 static int
6869 valid_length_expression (name)
6870 char *name;
6871 {
6872 return (name[1] == '\0' || /* ${#} */
6873 ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */
6874 (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */
6875 #if defined (ARRAY_VARS)
6876 valid_array_reference (name + 1, 0) || /* ${#a[7]} */
6877 #endif
6878 legal_identifier (name + 1)); /* ${#PS1} */
6879 }
6880
6881 /* Handle the parameter brace expansion that requires us to return the
6882 length of a parameter. */
6883 static intmax_t
6884 parameter_brace_expand_length (name)
6885 char *name;
6886 {
6887 char *t, *newname;
6888 intmax_t number, arg_index;
6889 WORD_LIST *list;
6890 #if defined (ARRAY_VARS)
6891 SHELL_VAR *var;
6892 #endif
6893
6894 if (name[1] == '\0') /* ${#} */
6895 number = number_of_args ();
6896 else if (DOLLAR_AT_STAR (name[1]) && name[2] == '\0') /* ${#@}, ${#*} */
6897 number = number_of_args ();
6898 else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0')
6899 {
6900 /* Take the lengths of some of the shell's special parameters. */
6901 switch (name[1])
6902 {
6903 case '-':
6904 t = which_set_flags ();
6905 break;
6906 case '?':
6907 t = itos (last_command_exit_value);
6908 break;
6909 case '$':
6910 t = itos (dollar_dollar_pid);
6911 break;
6912 case '!':
6913 if (last_asynchronous_pid == NO_PID)
6914 t = (char *)NULL; /* XXX - error if set -u set? */
6915 else
6916 t = itos (last_asynchronous_pid);
6917 break;
6918 case '#':
6919 t = itos (number_of_args ());
6920 break;
6921 }
6922 number = STRLEN (t);
6923 FREE (t);
6924 }
6925 #if defined (ARRAY_VARS)
6926 else if (valid_array_reference (name + 1, 0))
6927 number = array_length_reference (name + 1);
6928 #endif /* ARRAY_VARS */
6929 else
6930 {
6931 number = 0;
6932
6933 if (legal_number (name + 1, &arg_index)) /* ${#1} */
6934 {
6935 t = get_dollar_var_value (arg_index);
6936 if (t == 0 && unbound_vars_is_error)
6937 return INTMAX_MIN;
6938 number = MB_STRLEN (t);
6939 FREE (t);
6940 }
6941 #if defined (ARRAY_VARS)
6942 else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var)))
6943 {
6944 if (assoc_p (var))
6945 t = assoc_reference (assoc_cell (var), "0");
6946 else
6947 t = array_reference (array_cell (var), 0);
6948 if (t == 0 && unbound_vars_is_error)
6949 return INTMAX_MIN;
6950 number = MB_STRLEN (t);
6951 }
6952 #endif
6953 else /* ${#PS1} */
6954 {
6955 newname = savestring (name);
6956 newname[0] = '$';
6957 list = expand_string (newname, Q_DOUBLE_QUOTES);
6958 t = list ? string_list (list) : (char *)NULL;
6959 free (newname);
6960 if (list)
6961 dispose_words (list);
6962
6963 number = t ? MB_STRLEN (t) : 0;
6964 FREE (t);
6965 }
6966 }
6967
6968 return (number);
6969 }
6970
6971 /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression,
6972 so we do some ad-hoc parsing of an arithmetic expression to find
6973 the first DELIM, instead of using strchr(3). Two rules:
6974 1. If the substring contains a `(', read until closing `)'.
6975 2. If the substring contains a `?', read past one `:' for each `?'.
6976 The SD_ARITHEXP flag to skip_to_delim takes care of doing this.
6977 */
6978
6979 static char *
6980 skiparith (substr, delim)
6981 char *substr;
6982 int delim;
6983 {
6984 int i;
6985 char delims[2];
6986
6987 delims[0] = delim;
6988 delims[1] = '\0';
6989
6990 i = skip_to_delim (substr, 0, delims, SD_ARITHEXP);
6991 return (substr + i);
6992 }
6993
6994 /* Verify and limit the start and end of the desired substring. If
6995 VTYPE == 0, a regular shell variable is being used; if it is 1,
6996 then the positional parameters are being used; if it is 2, then
6997 VALUE is really a pointer to an array variable that should be used.
6998 Return value is 1 if both values were OK, 0 if there was a problem
6999 with an invalid expression, or -1 if the values were out of range. */
7000 static int
7001 verify_substring_values (v, value, substr, vtype, e1p, e2p)
7002 SHELL_VAR *v;
7003 char *value, *substr;
7004 int vtype;
7005 intmax_t *e1p, *e2p;
7006 {
7007 char *t, *temp1, *temp2;
7008 arrayind_t len;
7009 int expok;
7010 #if defined (ARRAY_VARS)
7011 ARRAY *a;
7012 HASH_TABLE *h;
7013 #endif
7014
7015 /* duplicate behavior of strchr(3) */
7016 t = skiparith (substr, ':');
7017 if (*t && *t == ':')
7018 *t = '\0';
7019 else
7020 t = (char *)0;
7021
7022 temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES);
7023 *e1p = evalexp (temp1, &expok);
7024 free (temp1);
7025 if (expok == 0)
7026 return (0);
7027
7028 len = -1; /* paranoia */
7029 switch (vtype)
7030 {
7031 case VT_VARIABLE:
7032 case VT_ARRAYMEMBER:
7033 len = MB_STRLEN (value);
7034 break;
7035 case VT_POSPARMS:
7036 len = number_of_args () + 1;
7037 if (*e1p == 0)
7038 len++; /* add one arg if counting from $0 */
7039 break;
7040 #if defined (ARRAY_VARS)
7041 case VT_ARRAYVAR:
7042 /* For arrays, the first value deals with array indices. Negative
7043 offsets count from one past the array's maximum index. Associative
7044 arrays treat the number of elements as the maximum index. */
7045 if (assoc_p (v))
7046 {
7047 h = assoc_cell (v);
7048 len = assoc_num_elements (h) + (*e1p < 0);
7049 }
7050 else
7051 {
7052 a = (ARRAY *)value;
7053 len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */
7054 }
7055 break;
7056 #endif
7057 }
7058
7059 if (len == -1) /* paranoia */
7060 return -1;
7061
7062 if (*e1p < 0) /* negative offsets count from end */
7063 *e1p += len;
7064
7065 if (*e1p > len || *e1p < 0)
7066 return (-1);
7067
7068 #if defined (ARRAY_VARS)
7069 /* For arrays, the second offset deals with the number of elements. */
7070 if (vtype == VT_ARRAYVAR)
7071 len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a);
7072 #endif
7073
7074 if (t)
7075 {
7076 t++;
7077 temp2 = savestring (t);
7078 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES);
7079 free (temp2);
7080 t[-1] = ':';
7081 *e2p = evalexp (temp1, &expok);
7082 free (temp1);
7083 if (expok == 0)
7084 return (0);
7085 #if 1
7086 if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0)
7087 #else
7088 /* bash-4.3: allow positional parameter length < 0 to count backwards
7089 from end of positional parameters */
7090 if (vtype == VT_ARRAYVAR && *e2p < 0)
7091 #endif
7092 {
7093 internal_error (_("%s: substring expression < 0"), t);
7094 return (0);
7095 }
7096 #if defined (ARRAY_VARS)
7097 /* In order to deal with sparse arrays, push the intelligence about how
7098 to deal with the number of elements desired down to the array-
7099 specific functions. */
7100 if (vtype != VT_ARRAYVAR)
7101 #endif
7102 {
7103 if (*e2p < 0)
7104 {
7105 *e2p += len;
7106 if (*e2p < 0 || *e2p < *e1p)
7107 {
7108 internal_error (_("%s: substring expression < 0"), t);
7109 return (0);
7110 }
7111 }
7112 else
7113 *e2p += *e1p; /* want E2 chars starting at E1 */
7114 if (*e2p > len)
7115 *e2p = len;
7116 }
7117 }
7118 else
7119 *e2p = len;
7120
7121 return (1);
7122 }
7123
7124 /* Return the type of variable specified by VARNAME (simple variable,
7125 positional param, or array variable). Also return the value specified
7126 by VARNAME (value of a variable or a reference to an array element).
7127 QUOTED is the standard description of quoting state, using Q_* defines.
7128 FLAGS is currently a set of flags to pass to array_value. If IND is
7129 non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is
7130 passed to array_value so the array index is not computed again.
7131 If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL
7132 characters in the value are quoted with CTLESC and takes appropriate
7133 steps. For convenience, *VALP is set to the dequoted VALUE. */
7134 static int
7135 get_var_and_type (varname, value, ind, quoted, flags, varp, valp)
7136 char *varname, *value;
7137 arrayind_t ind;
7138 int quoted, flags;
7139 SHELL_VAR **varp;
7140 char **valp;
7141 {
7142 int vtype, want_indir;
7143 char *temp, *vname;
7144 WORD_DESC *wd;
7145 SHELL_VAR *v;
7146 arrayind_t lind;
7147
7148 want_indir = *varname == '!' &&
7149 (legal_variable_starter ((unsigned char)varname[1]) || DIGIT (varname[1])
7150 || VALID_INDIR_PARAM (varname[1]));
7151 if (want_indir)
7152 vname = parameter_brace_find_indir (varname+1, SPECIAL_VAR (varname, 1), quoted, 1);
7153 /* XXX - what if vname == 0 || *vname == 0 ? */
7154 else
7155 vname = varname;
7156
7157 if (vname == 0)
7158 {
7159 vtype = VT_VARIABLE;
7160 *varp = (SHELL_VAR *)NULL;
7161 *valp = (char *)NULL;
7162 return (vtype);
7163 }
7164
7165 /* This sets vtype to VT_VARIABLE or VT_POSPARMS */
7166 vtype = STR_DOLLAR_AT_STAR (vname);
7167 if (vtype == VT_POSPARMS && vname[0] == '*')
7168 vtype |= VT_STARSUB;
7169 *varp = (SHELL_VAR *)NULL;
7170
7171 #if defined (ARRAY_VARS)
7172 if (valid_array_reference (vname, 0))
7173 {
7174 v = array_variable_part (vname, &temp, (int *)0);
7175 /* If we want to signal array_value to use an already-computed index,
7176 set LIND to that index */
7177 lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0;
7178 if (v && invisible_p (v))
7179 {
7180 vtype = VT_ARRAYMEMBER;
7181 *varp = (SHELL_VAR *)NULL;
7182 *valp = (char *)NULL;
7183 }
7184 if (v && (array_p (v) || assoc_p (v)))
7185 { /* [ */
7186 if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')
7187 {
7188 /* Callers have to differentiate between indexed and associative */
7189 vtype = VT_ARRAYVAR;
7190 if (temp[0] == '*')
7191 vtype |= VT_STARSUB;
7192 *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v);
7193 }
7194 else
7195 {
7196 vtype = VT_ARRAYMEMBER;
7197 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7198 }
7199 *varp = v;
7200 }
7201 else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']'))
7202 {
7203 vtype = VT_VARIABLE;
7204 *varp = v;
7205 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7206 *valp = dequote_string (value);
7207 else
7208 *valp = dequote_escapes (value);
7209 }
7210 else
7211 {
7212 vtype = VT_ARRAYMEMBER;
7213 *varp = v;
7214 *valp = array_value (vname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind);
7215 }
7216 }
7217 else if ((v = find_variable (vname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v)))
7218 {
7219 vtype = VT_ARRAYMEMBER;
7220 *varp = v;
7221 *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0);
7222 }
7223 else
7224 #endif
7225 {
7226 if (value && vtype == VT_VARIABLE)
7227 {
7228 *varp = find_variable (vname);
7229 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7230 *valp = dequote_string (value);
7231 else
7232 *valp = dequote_escapes (value);
7233 }
7234 else
7235 *valp = value;
7236 }
7237
7238 if (want_indir)
7239 free (vname);
7240
7241 return vtype;
7242 }
7243
7244 /******************************************************/
7245 /* */
7246 /* Functions to extract substrings of variable values */
7247 /* */
7248 /******************************************************/
7249
7250 #if defined (HANDLE_MULTIBYTE)
7251 /* Character-oriented rather than strictly byte-oriented substrings. S and
7252 E, rather being strict indices into STRING, indicate character (possibly
7253 multibyte character) positions that require calculation.
7254 Used by the ${param:offset[:length]} expansion. */
7255 static char *
7256 mb_substring (string, s, e)
7257 char *string;
7258 int s, e;
7259 {
7260 char *tt;
7261 int start, stop, i;
7262 size_t slen;
7263 DECLARE_MBSTATE;
7264
7265 start = 0;
7266 /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */
7267 slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0;
7268
7269 i = s;
7270 while (string[start] && i--)
7271 ADVANCE_CHAR (string, slen, start);
7272 stop = start;
7273 i = e - s;
7274 while (string[stop] && i--)
7275 ADVANCE_CHAR (string, slen, stop);
7276 tt = substring (string, start, stop);
7277 return tt;
7278 }
7279 #endif
7280
7281 /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME
7282 is `@', use the positional parameters; otherwise, use the value of
7283 VARNAME. If VARNAME is an array variable, use the array elements. */
7284
7285 static char *
7286 parameter_brace_substring (varname, value, ind, substr, quoted, flags)
7287 char *varname, *value;
7288 int ind;
7289 char *substr;
7290 int quoted, flags;
7291 {
7292 intmax_t e1, e2;
7293 int vtype, r, starsub;
7294 char *temp, *val, *tt, *oname;
7295 SHELL_VAR *v;
7296
7297 if (value == 0 && ((varname[0] != '@' && varname[0] != '*') || varname[1]))
7298 return ((char *)NULL);
7299
7300 oname = this_command_name;
7301 this_command_name = varname;
7302
7303 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7304 if (vtype == -1)
7305 {
7306 this_command_name = oname;
7307 return ((char *)NULL);
7308 }
7309
7310 starsub = vtype & VT_STARSUB;
7311 vtype &= ~VT_STARSUB;
7312
7313 r = verify_substring_values (v, val, substr, vtype, &e1, &e2);
7314 this_command_name = oname;
7315 if (r <= 0)
7316 {
7317 if (vtype == VT_VARIABLE)
7318 FREE (val);
7319 return ((r == 0) ? &expand_param_error : (char *)NULL);
7320 }
7321
7322 switch (vtype)
7323 {
7324 case VT_VARIABLE:
7325 case VT_ARRAYMEMBER:
7326 #if defined (HANDLE_MULTIBYTE)
7327 if (MB_CUR_MAX > 1)
7328 tt = mb_substring (val, e1, e2);
7329 else
7330 #endif
7331 tt = substring (val, e1, e2);
7332
7333 if (vtype == VT_VARIABLE)
7334 FREE (val);
7335 if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))
7336 temp = quote_string (tt);
7337 else
7338 temp = tt ? quote_escapes (tt) : (char *)NULL;
7339 FREE (tt);
7340 break;
7341 case VT_POSPARMS:
7342 tt = pos_params (varname, e1, e2, quoted);
7343 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
7344 {
7345 temp = tt ? quote_escapes (tt) : (char *)NULL;
7346 FREE (tt);
7347 }
7348 else
7349 temp = tt;
7350 break;
7351 #if defined (ARRAY_VARS)
7352 case VT_ARRAYVAR:
7353 if (assoc_p (v))
7354 /* we convert to list and take first e2 elements starting at e1th
7355 element -- officially undefined for now */
7356 temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted);
7357 else
7358 /* We want E2 to be the number of elements desired (arrays can be sparse,
7359 so verify_substring_values just returns the numbers specified and we
7360 rely on array_subrange to understand how to deal with them). */
7361 temp = array_subrange (array_cell (v), e1, e2, starsub, quoted);
7362 /* array_subrange now calls array_quote_escapes as appropriate, so the
7363 caller no longer needs to. */
7364 break;
7365 #endif
7366 default:
7367 temp = (char *)NULL;
7368 }
7369
7370 return temp;
7371 }
7372
7373 /****************************************************************/
7374 /* */
7375 /* Functions to perform pattern substitution on variable values */
7376 /* */
7377 /****************************************************************/
7378
7379 #if 0 /* Unused */
7380 static int
7381 shouldexp_replacement (s)
7382 char *s;
7383 {
7384 register char *p;
7385
7386 for (p = s; p && *p; p++)
7387 {
7388 if (*p == '\\')
7389 p++;
7390 else if (*p == '&')
7391 return 1;
7392 }
7393 return 0;
7394 }
7395 #endif
7396
7397 char *
7398 pat_subst (string, pat, rep, mflags)
7399 char *string, *pat, *rep;
7400 int mflags;
7401 {
7402 char *ret, *s, *e, *str, *rstr, *mstr;
7403 int rptr, mtype, rxpand, mlen;
7404 size_t rsize, l, replen, rslen;
7405
7406 if (string == 0)
7407 return (savestring (""));
7408
7409 mtype = mflags & MATCH_TYPEMASK;
7410
7411 #if 0 /* bash-4.2 ? */
7412 rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0;
7413 #else
7414 rxpand = 0;
7415 #endif
7416
7417 /* Special cases:
7418 * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING
7419 * with REP and return the result.
7420 * 2. A null pattern with mtype == MATCH_END means to append REP to
7421 * STRING and return the result.
7422 * 3. A null STRING with a matching pattern means to append REP to
7423 * STRING and return the result.
7424 * These don't understand or process `&' in the replacement string.
7425 */
7426 if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END))
7427 {
7428 replen = STRLEN (rep);
7429 l = STRLEN (string);
7430 ret = (char *)xmalloc (replen + l + 2);
7431 if (replen == 0)
7432 strcpy (ret, string);
7433 else if (mtype == MATCH_BEG)
7434 {
7435 strcpy (ret, rep);
7436 strcpy (ret + replen, string);
7437 }
7438 else
7439 {
7440 strcpy (ret, string);
7441 strcpy (ret + l, rep);
7442 }
7443 return (ret);
7444 }
7445 else if (*string == 0 && (match_pattern (string, pat, mtype, &s, &e) != 0))
7446 {
7447 replen = STRLEN (rep);
7448 ret = (char *)xmalloc (replen + 1);
7449 if (replen == 0)
7450 ret[0] = '\0';
7451 else
7452 strcpy (ret, rep);
7453 return (ret);
7454 }
7455
7456 ret = (char *)xmalloc (rsize = 64);
7457 ret[0] = '\0';
7458
7459 for (replen = STRLEN (rep), rptr = 0, str = string; *str;)
7460 {
7461 if (match_pattern (str, pat, mtype, &s, &e) == 0)
7462 break;
7463 l = s - str;
7464
7465 if (rep && rxpand)
7466 {
7467 int x;
7468 mlen = e - s;
7469 mstr = xmalloc (mlen + 1);
7470 for (x = 0; x < mlen; x++)
7471 mstr[x] = s[x];
7472 mstr[mlen] = '\0';
7473 rstr = strcreplace (rep, '&', mstr, 0);
7474 free (mstr);
7475 rslen = strlen (rstr);
7476 }
7477 else
7478 {
7479 rstr = rep;
7480 rslen = replen;
7481 }
7482
7483 RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64);
7484
7485 /* OK, now copy the leading unmatched portion of the string (from
7486 str to s) to ret starting at rptr (the current offset). Then copy
7487 the replacement string at ret + rptr + (s - str). Increment
7488 rptr (if necessary) and str and go on. */
7489 if (l)
7490 {
7491 strncpy (ret + rptr, str, l);
7492 rptr += l;
7493 }
7494 if (replen)
7495 {
7496 strncpy (ret + rptr, rstr, rslen);
7497 rptr += rslen;
7498 }
7499 str = e; /* e == end of match */
7500
7501 if (rstr != rep)
7502 free (rstr);
7503
7504 if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY)
7505 break;
7506
7507 if (s == e)
7508 {
7509 /* On a zero-length match, make sure we copy one character, since
7510 we increment one character to avoid infinite recursion. */
7511 RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64);
7512 ret[rptr++] = *str++;
7513 e++; /* avoid infinite recursion on zero-length match */
7514 }
7515 }
7516
7517 /* Now copy the unmatched portion of the input string */
7518 if (str && *str)
7519 {
7520 RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64);
7521 strcpy (ret + rptr, str);
7522 }
7523 else
7524 ret[rptr] = '\0';
7525
7526 return ret;
7527 }
7528
7529 /* Do pattern match and replacement on the positional parameters. */
7530 static char *
7531 pos_params_pat_subst (string, pat, rep, mflags)
7532 char *string, *pat, *rep;
7533 int mflags;
7534 {
7535 WORD_LIST *save, *params;
7536 WORD_DESC *w;
7537 char *ret;
7538 int pchar, qflags;
7539
7540 save = params = list_rest_of_args ();
7541 if (save == 0)
7542 return ((char *)NULL);
7543
7544 for ( ; params; params = params->next)
7545 {
7546 ret = pat_subst (params->word->word, pat, rep, mflags);
7547 w = alloc_word_desc ();
7548 w->word = ret ? ret : savestring ("");
7549 dispose_word (params->word);
7550 params->word = w;
7551 }
7552
7553 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7554 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7555
7556 ret = string_list_pos_params (pchar, save, qflags);
7557
7558 dispose_words (save);
7559
7560 return (ret);
7561 }
7562
7563 /* Perform pattern substitution on VALUE, which is the expansion of
7564 VARNAME. PATSUB is an expression supplying the pattern to match
7565 and the string to substitute. QUOTED is a flags word containing
7566 the type of quoting currently in effect. */
7567 static char *
7568 parameter_brace_patsub (varname, value, ind, patsub, quoted, pflags, flags)
7569 char *varname, *value;
7570 int ind;
7571 char *patsub;
7572 int quoted, pflags, flags;
7573 {
7574 int vtype, mflags, starsub, delim;
7575 char *val, *temp, *pat, *rep, *p, *lpatsub, *tt;
7576 SHELL_VAR *v;
7577
7578 if (value == 0)
7579 return ((char *)NULL);
7580
7581 this_command_name = varname;
7582
7583 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7584 if (vtype == -1)
7585 return ((char *)NULL);
7586
7587 starsub = vtype & VT_STARSUB;
7588 vtype &= ~VT_STARSUB;
7589
7590 mflags = 0;
7591 /* PATSUB is never NULL when this is called. */
7592 if (*patsub == '/')
7593 {
7594 mflags |= MATCH_GLOBREP;
7595 patsub++;
7596 }
7597
7598 /* Malloc this because expand_string_if_necessary or one of the expansion
7599 functions in its call chain may free it on a substitution error. */
7600 lpatsub = savestring (patsub);
7601
7602 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7603 mflags |= MATCH_QUOTED;
7604
7605 if (starsub)
7606 mflags |= MATCH_STARSUB;
7607
7608 if (pflags & PF_ASSIGNRHS)
7609 mflags |= MATCH_ASSIGNRHS;
7610
7611 /* If the pattern starts with a `/', make sure we skip over it when looking
7612 for the replacement delimiter. */
7613 delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0);
7614 if (lpatsub[delim] == '/')
7615 {
7616 lpatsub[delim] = 0;
7617 rep = lpatsub + delim + 1;
7618 }
7619 else
7620 rep = (char *)NULL;
7621
7622 if (rep && *rep == '\0')
7623 rep = (char *)NULL;
7624
7625 /* Perform the same expansions on the pattern as performed by the
7626 pattern removal expansions. */
7627 pat = getpattern (lpatsub, quoted, 1);
7628
7629 if (rep)
7630 {
7631 /* We want to perform quote removal on the expanded replacement even if
7632 the entire expansion is double-quoted because the parser and string
7633 extraction functions treated quotes in the replacement string as
7634 special. THIS IS NOT BACKWARDS COMPATIBLE WITH BASH-4.2. */
7635 if (shell_compatibility_level > 42)
7636 rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit);
7637 /* This is the bash-4.2 code. */
7638 else if ((mflags & MATCH_QUOTED) == 0)
7639 rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit);
7640 else
7641 rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit);
7642 }
7643
7644 /* ksh93 doesn't allow the match specifier to be a part of the expanded
7645 pattern. This is an extension. Make sure we don't anchor the pattern
7646 at the beginning or end of the string if we're doing global replacement,
7647 though. */
7648 p = pat;
7649 if (mflags & MATCH_GLOBREP)
7650 mflags |= MATCH_ANY;
7651 else if (pat && pat[0] == '#')
7652 {
7653 mflags |= MATCH_BEG;
7654 p++;
7655 }
7656 else if (pat && pat[0] == '%')
7657 {
7658 mflags |= MATCH_END;
7659 p++;
7660 }
7661 else
7662 mflags |= MATCH_ANY;
7663
7664 /* OK, we now want to substitute REP for PAT in VAL. If
7665 flags & MATCH_GLOBREP is non-zero, the substitution is done
7666 everywhere, otherwise only the first occurrence of PAT is
7667 replaced. The pattern matching code doesn't understand
7668 CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable
7669 values passed in (VT_VARIABLE) so the pattern substitution
7670 code works right. We need to requote special chars after
7671 we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the
7672 other cases if QUOTED == 0, since the posparams and arrays
7673 indexed by * or @ do special things when QUOTED != 0. */
7674
7675 switch (vtype)
7676 {
7677 case VT_VARIABLE:
7678 case VT_ARRAYMEMBER:
7679 temp = pat_subst (val, p, rep, mflags);
7680 if (vtype == VT_VARIABLE)
7681 FREE (val);
7682 if (temp)
7683 {
7684 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7685 free (temp);
7686 temp = tt;
7687 }
7688 break;
7689 case VT_POSPARMS:
7690 temp = pos_params_pat_subst (val, p, rep, mflags);
7691 if (temp && (mflags & MATCH_QUOTED) == 0)
7692 {
7693 tt = quote_escapes (temp);
7694 free (temp);
7695 temp = tt;
7696 }
7697 break;
7698 #if defined (ARRAY_VARS)
7699 case VT_ARRAYVAR:
7700 temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags)
7701 : array_patsub (array_cell (v), p, rep, mflags);
7702 /* Don't call quote_escapes anymore; array_patsub calls
7703 array_quote_escapes as appropriate before adding the
7704 space separators; ditto for assoc_patsub. */
7705 break;
7706 #endif
7707 }
7708
7709 FREE (pat);
7710 FREE (rep);
7711 free (lpatsub);
7712
7713 return temp;
7714 }
7715
7716 /****************************************************************/
7717 /* */
7718 /* Functions to perform case modification on variable values */
7719 /* */
7720 /****************************************************************/
7721
7722 /* Do case modification on the positional parameters. */
7723
7724 static char *
7725 pos_params_modcase (string, pat, modop, mflags)
7726 char *string, *pat;
7727 int modop;
7728 int mflags;
7729 {
7730 WORD_LIST *save, *params;
7731 WORD_DESC *w;
7732 char *ret;
7733 int pchar, qflags;
7734
7735 save = params = list_rest_of_args ();
7736 if (save == 0)
7737 return ((char *)NULL);
7738
7739 for ( ; params; params = params->next)
7740 {
7741 ret = sh_modcase (params->word->word, pat, modop);
7742 w = alloc_word_desc ();
7743 w->word = ret ? ret : savestring ("");
7744 dispose_word (params->word);
7745 params->word = w;
7746 }
7747
7748 pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@';
7749 qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0;
7750
7751 ret = string_list_pos_params (pchar, save, qflags);
7752 dispose_words (save);
7753
7754 return (ret);
7755 }
7756
7757 /* Perform case modification on VALUE, which is the expansion of
7758 VARNAME. MODSPEC is an expression supplying the type of modification
7759 to perform. QUOTED is a flags word containing the type of quoting
7760 currently in effect. */
7761 static char *
7762 parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags)
7763 char *varname, *value;
7764 int ind, modspec;
7765 char *patspec;
7766 int quoted, flags;
7767 {
7768 int vtype, starsub, modop, mflags, x;
7769 char *val, *temp, *pat, *p, *lpat, *tt;
7770 SHELL_VAR *v;
7771
7772 if (value == 0)
7773 return ((char *)NULL);
7774
7775 this_command_name = varname;
7776
7777 vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val);
7778 if (vtype == -1)
7779 return ((char *)NULL);
7780
7781 starsub = vtype & VT_STARSUB;
7782 vtype &= ~VT_STARSUB;
7783
7784 modop = 0;
7785 mflags = 0;
7786 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
7787 mflags |= MATCH_QUOTED;
7788 if (starsub)
7789 mflags |= MATCH_STARSUB;
7790
7791 p = patspec;
7792 if (modspec == '^')
7793 {
7794 x = p && p[0] == modspec;
7795 modop = x ? CASE_UPPER : CASE_UPFIRST;
7796 p += x;
7797 }
7798 else if (modspec == ',')
7799 {
7800 x = p && p[0] == modspec;
7801 modop = x ? CASE_LOWER : CASE_LOWFIRST;
7802 p += x;
7803 }
7804 else if (modspec == '~')
7805 {
7806 x = p && p[0] == modspec;
7807 modop = x ? CASE_TOGGLEALL : CASE_TOGGLE;
7808 p += x;
7809 }
7810
7811 lpat = p ? savestring (p) : 0;
7812 /* Perform the same expansions on the pattern as performed by the
7813 pattern removal expansions. FOR LATER */
7814 pat = lpat ? getpattern (lpat, quoted, 1) : 0;
7815
7816 /* OK, now we do the case modification. */
7817 switch (vtype)
7818 {
7819 case VT_VARIABLE:
7820 case VT_ARRAYMEMBER:
7821 temp = sh_modcase (val, pat, modop);
7822 if (vtype == VT_VARIABLE)
7823 FREE (val);
7824 if (temp)
7825 {
7826 tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp);
7827 free (temp);
7828 temp = tt;
7829 }
7830 break;
7831
7832 case VT_POSPARMS:
7833 temp = pos_params_modcase (val, pat, modop, mflags);
7834 if (temp && (mflags & MATCH_QUOTED) == 0)
7835 {
7836 tt = quote_escapes (temp);
7837 free (temp);
7838 temp = tt;
7839 }
7840 break;
7841
7842 #if defined (ARRAY_VARS)
7843 case VT_ARRAYVAR:
7844 temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags)
7845 : array_modcase (array_cell (v), pat, modop, mflags);
7846 /* Don't call quote_escapes; array_modcase calls array_quote_escapes
7847 as appropriate before adding the space separators; ditto for
7848 assoc_modcase. */
7849 break;
7850 #endif
7851 }
7852
7853 FREE (pat);
7854 free (lpat);
7855
7856 return temp;
7857 }
7858
7859 /* Check for unbalanced parens in S, which is the contents of $(( ... )). If
7860 any occur, this must be a nested command substitution, so return 0.
7861 Otherwise, return 1. A valid arithmetic expression must always have a
7862 ( before a matching ), so any cases where there are more right parens
7863 means that this must not be an arithmetic expression, though the parser
7864 will not accept it without a balanced total number of parens. */
7865 static int
7866 chk_arithsub (s, len)
7867 const char *s;
7868 int len;
7869 {
7870 int i, count;
7871 DECLARE_MBSTATE;
7872
7873 i = count = 0;
7874 while (i < len)
7875 {
7876 if (s[i] == LPAREN)
7877 count++;
7878 else if (s[i] == RPAREN)
7879 {
7880 count--;
7881 if (count < 0)
7882 return 0;
7883 }
7884
7885 switch (s[i])
7886 {
7887 default:
7888 ADVANCE_CHAR (s, len, i);
7889 break;
7890
7891 case '\\':
7892 i++;
7893 if (s[i])
7894 ADVANCE_CHAR (s, len, i);
7895 break;
7896
7897 case '\'':
7898 i = skip_single_quoted (s, len, ++i, 0);
7899 break;
7900
7901 case '"':
7902 i = skip_double_quoted ((char *)s, len, ++i, 0);
7903 break;
7904 }
7905 }
7906
7907 return (count == 0);
7908 }
7909
7910 /****************************************************************/
7911 /* */
7912 /* Functions to perform parameter expansion on a string */
7913 /* */
7914 /****************************************************************/
7915
7916 /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */
7917 static WORD_DESC *
7918 parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at)
7919 char *string;
7920 int *indexp, quoted, pflags, *quoted_dollar_atp, *contains_dollar_at;
7921 {
7922 int check_nullness, var_is_set, var_is_null, var_is_special;
7923 int want_substring, want_indir, want_patsub, want_casemod;
7924 char *name, *value, *temp, *temp1;
7925 WORD_DESC *tdesc, *ret;
7926 int t_index, sindex, c, tflag, modspec, all_element_arrayref;
7927 intmax_t number;
7928 arrayind_t ind;
7929
7930 temp = temp1 = value = (char *)NULL;
7931 var_is_set = var_is_null = var_is_special = check_nullness = 0;
7932 want_substring = want_indir = want_patsub = want_casemod = 0;
7933
7934 all_element_arrayref = 0;
7935
7936 sindex = *indexp;
7937 t_index = ++sindex;
7938 /* ${#var} doesn't have any of the other parameter expansions on it. */
7939 if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */
7940 name = string_extract (string, &t_index, "}", SX_VARNAME);
7941 else
7942 #if defined (CASEMOD_EXPANSIONS)
7943 /* To enable case-toggling expansions using the `~' operator character
7944 change the 1 to 0. */
7945 # if defined (CASEMOD_CAPCASE)
7946 name = string_extract (string, &t_index, "#%^,~:-=?+/@}", SX_VARNAME);
7947 # else
7948 name = string_extract (string, &t_index, "#%^,:-=?+/@}", SX_VARNAME);
7949 # endif /* CASEMOD_CAPCASE */
7950 #else
7951 name = string_extract (string, &t_index, "#%:-=?+/@}", SX_VARNAME);
7952 #endif /* CASEMOD_EXPANSIONS */
7953
7954 /* Handle ${@[stuff]} now that @ is a word expansion operator. Not exactly
7955 the cleanest code ever. */
7956 if (*name == 0 && sindex == t_index && string[sindex] == '@')
7957 {
7958 name = (char *)xrealloc (name, 2);
7959 name[0] = '@';
7960 name[1] = '\0';
7961 t_index++;
7962 }
7963 else if (*name == '!' && t_index > sindex && string[t_index] == '@' && string[t_index+1] == '}')
7964 {
7965 name = (char *)xrealloc (name, t_index - sindex + 2);
7966 name[t_index - sindex] = '@';
7967 name[t_index - sindex + 1] = '\0';
7968 t_index++;
7969 }
7970
7971 ret = 0;
7972 tflag = 0;
7973
7974 ind = INTMAX_MIN;
7975
7976 /* If the name really consists of a special variable, then make sure
7977 that we have the entire name. We don't allow indirect references
7978 to special variables except `#', `?', `@' and `*'. This clause is
7979 designed to handle ${#SPECIAL} and ${!SPECIAL}, not anything more
7980 general. */
7981 if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) ||
7982 (sindex == t_index && string[sindex] == '#' && VALID_SPECIAL_LENGTH_PARAM (string[sindex + 1])) ||
7983 (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index])))
7984 {
7985 t_index++;
7986 temp1 = string_extract (string, &t_index, "#%:-=?+/@}", 0);
7987 name = (char *)xrealloc (name, 3 + (strlen (temp1)));
7988 *name = string[sindex];
7989 if (string[sindex] == '!')
7990 {
7991 /* indirect reference of $#, $?, $@, or $* */
7992 name[1] = string[sindex + 1];
7993 strcpy (name + 2, temp1);
7994 }
7995 else
7996 strcpy (name + 1, temp1);
7997 free (temp1);
7998 }
7999 sindex = t_index;
8000
8001 /* Find out what character ended the variable name. Then
8002 do the appropriate thing. */
8003 if (c = string[sindex])
8004 sindex++;
8005
8006 /* If c is followed by one of the valid parameter expansion
8007 characters, move past it as normal. If not, assume that
8008 a substring specification is being given, and do not move
8009 past it. */
8010 if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex]))
8011 {
8012 check_nullness++;
8013 if (c = string[sindex])
8014 sindex++;
8015 }
8016 else if (c == ':' && string[sindex] != RBRACE)
8017 want_substring = 1;
8018 else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */
8019 want_patsub = 1;
8020 #if defined (CASEMOD_EXPANSIONS)
8021 else if (c == '^' || c == ',' || c == '~')
8022 {
8023 modspec = c;
8024 want_casemod = 1;
8025 }
8026 #endif
8027
8028 /* Catch the valid and invalid brace expressions that made it through the
8029 tests above. */
8030 /* ${#-} is a valid expansion and means to take the length of $-.
8031 Similarly for ${#?} and ${##}... */
8032 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8033 VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE)
8034 {
8035 name = (char *)xrealloc (name, 3);
8036 name[1] = c;
8037 name[2] = '\0';
8038 c = string[sindex++];
8039 }
8040
8041 /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */
8042 if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 &&
8043 member (c, "%:=+/") && string[sindex] == RBRACE)
8044 {
8045 temp = (char *)NULL;
8046 goto bad_substitution; /* XXX - substitution error */
8047 }
8048
8049 /* Indirect expansion begins with a `!'. A valid indirect expansion is
8050 either a variable name, one of the positional parameters or a special
8051 variable that expands to one of the positional parameters. */
8052 want_indir = *name == '!' &&
8053 (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1])
8054 || VALID_INDIR_PARAM (name[1]));
8055
8056 /* Determine the value of this variable. */
8057
8058 /* Check for special variables, directly referenced. */
8059 if (SPECIAL_VAR (name, want_indir))
8060 var_is_special++;
8061
8062 /* Check for special expansion things, like the length of a parameter */
8063 if (*name == '#' && name[1])
8064 {
8065 /* If we are not pointing at the character just after the
8066 closing brace, then we haven't gotten all of the name.
8067 Since it begins with a special character, this is a bad
8068 substitution. Also check NAME for validity before trying
8069 to go on. */
8070 if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0))
8071 {
8072 temp = (char *)NULL;
8073 goto bad_substitution; /* substitution error */
8074 }
8075
8076 number = parameter_brace_expand_length (name);
8077 if (number == INTMAX_MIN && unbound_vars_is_error)
8078 {
8079 last_command_exit_value = EXECUTION_FAILURE;
8080 err_unboundvar (name+1);
8081 free (name);
8082 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8083 }
8084 free (name);
8085
8086 *indexp = sindex;
8087 if (number < 0)
8088 return (&expand_wdesc_error);
8089 else
8090 {
8091 ret = alloc_word_desc ();
8092 ret->word = itos (number);
8093 return ret;
8094 }
8095 }
8096
8097 /* ${@} is identical to $@. */
8098 if (name[0] == '@' && name[1] == '\0')
8099 {
8100 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8101 *quoted_dollar_atp = 1;
8102
8103 if (contains_dollar_at)
8104 *contains_dollar_at = 1;
8105
8106 tflag |= W_DOLLARAT;
8107 }
8108
8109 /* Process ${!PREFIX*} expansion. */
8110 if (want_indir && string[sindex - 1] == RBRACE &&
8111 (string[sindex - 2] == '*' || string[sindex - 2] == '@') &&
8112 legal_variable_starter ((unsigned char) name[1]))
8113 {
8114 char **x;
8115 WORD_LIST *xlist;
8116
8117 temp1 = savestring (name + 1);
8118 number = strlen (temp1);
8119 temp1[number - 1] = '\0';
8120 x = all_variables_matching_prefix (temp1);
8121 xlist = strvec_to_word_list (x, 0, 0);
8122 if (string[sindex - 2] == '*')
8123 temp = string_list_dollar_star (xlist);
8124 else
8125 {
8126 temp = string_list_dollar_at (xlist, quoted, 0);
8127 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8128 *quoted_dollar_atp = 1;
8129 if (contains_dollar_at)
8130 *contains_dollar_at = 1;
8131
8132 tflag |= W_DOLLARAT;
8133 }
8134 free (x);
8135 dispose_words (xlist);
8136 free (temp1);
8137 *indexp = sindex;
8138
8139 free (name);
8140
8141 ret = alloc_word_desc ();
8142 ret->word = temp;
8143 ret->flags = tflag; /* XXX */
8144 return ret;
8145 }
8146
8147 #if defined (ARRAY_VARS)
8148 /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */
8149 if (want_indir && string[sindex - 1] == RBRACE &&
8150 string[sindex - 2] == ']' && valid_array_reference (name+1, 0))
8151 {
8152 char *x, *x1;
8153
8154 temp1 = savestring (name + 1);
8155 x = array_variable_name (temp1, &x1, (int *)0); /* [ */
8156 FREE (x);
8157 if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']')
8158 {
8159 temp = array_keys (temp1, quoted); /* handles assoc vars too */
8160 if (x1[0] == '@')
8161 {
8162 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8163 *quoted_dollar_atp = 1;
8164 if (contains_dollar_at)
8165 *contains_dollar_at = 1;
8166
8167 tflag |= W_DOLLARAT;
8168 }
8169
8170 free (name);
8171 free (temp1);
8172 *indexp = sindex;
8173
8174 ret = alloc_word_desc ();
8175 ret->word = temp;
8176 ret->flags = tflag; /* XXX */
8177 return ret;
8178 }
8179
8180 free (temp1);
8181 }
8182 #endif /* ARRAY_VARS */
8183
8184 /* Make sure that NAME is valid before trying to go on. */
8185 if (valid_brace_expansion_word (want_indir ? name + 1 : name,
8186 var_is_special) == 0)
8187 {
8188 temp = (char *)NULL;
8189 goto bad_substitution; /* substitution error */
8190 }
8191
8192 if (want_indir)
8193 {
8194 tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at);
8195 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8196 {
8197 temp = (char *)NULL;
8198 goto bad_substitution;
8199 }
8200 /* Turn off the W_ARRAYIND flag because there is no way for this function
8201 to return the index we're supposed to be using. */
8202 if (tdesc && tdesc->flags)
8203 tdesc->flags &= ~W_ARRAYIND;
8204 }
8205 else
8206 tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&(PF_NOSPLIT2|PF_ASSIGNRHS)), &ind);
8207
8208 if (tdesc)
8209 {
8210 temp = tdesc->word;
8211 tflag = tdesc->flags;
8212 dispose_word_desc (tdesc);
8213 }
8214 else
8215 temp = (char *)0;
8216
8217 if (temp == &expand_param_error || temp == &expand_param_fatal)
8218 {
8219 FREE (name);
8220 FREE (value);
8221 return (temp == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8222 }
8223
8224 #if defined (ARRAY_VARS)
8225 if (valid_array_reference (name, 0))
8226 {
8227 int qflags;
8228 char *t;
8229
8230 qflags = quoted;
8231 /* If in a context where word splitting will not take place, treat as
8232 if double-quoted. Has effects with $* and ${array[*]} */
8233 if (pflags & PF_ASSIGNRHS)
8234 qflags |= Q_DOUBLE_QUOTES;
8235 chk_atstar (name, qflags, quoted_dollar_atp, contains_dollar_at);
8236 /* We duplicate a little code here */
8237 t = mbschr (name, '[');
8238 if (t && ALL_ELEMENT_SUB (t[1]) && t[2] == ']')
8239 all_element_arrayref = 1;
8240 }
8241 #endif
8242
8243 var_is_set = temp != (char *)0;
8244 var_is_null = check_nullness && (var_is_set == 0 || *temp == 0);
8245 /* XXX - this may not need to be restricted to special variables */
8246 if (check_nullness)
8247 var_is_null |= var_is_set && var_is_special && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp);
8248
8249 /* Get the rest of the stuff inside the braces. */
8250 if (c && c != RBRACE)
8251 {
8252 /* Extract the contents of the ${ ... } expansion
8253 according to the Posix.2 rules. */
8254 value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD);
8255 if (string[sindex] == RBRACE)
8256 sindex++;
8257 else
8258 goto bad_substitution; /* substitution error */
8259 }
8260 else
8261 value = (char *)NULL;
8262
8263 *indexp = sindex;
8264
8265 /* All the cases where an expansion can possibly generate an unbound
8266 variable error. */
8267 if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE)
8268 {
8269 if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1]) && all_element_arrayref == 0)
8270 {
8271 last_command_exit_value = EXECUTION_FAILURE;
8272 err_unboundvar (name);
8273 FREE (value);
8274 FREE (temp);
8275 free (name);
8276 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8277 }
8278 }
8279
8280 /* If this is a substring spec, process it and add the result. */
8281 if (want_substring)
8282 {
8283 temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8284 FREE (name);
8285 FREE (value);
8286 FREE (temp);
8287
8288 if (temp1 == &expand_param_error)
8289 return (&expand_wdesc_error);
8290 else if (temp1 == &expand_param_fatal)
8291 return (&expand_wdesc_fatal);
8292
8293 ret = alloc_word_desc ();
8294 ret->word = temp1;
8295 /* We test quoted_dollar_atp because we want variants with double-quoted
8296 "$@" to take a different code path. In fact, we make sure at the end
8297 of expand_word_internal that we're only looking at these flags if
8298 quoted_dollar_at == 0. */
8299 if (temp1 &&
8300 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8301 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8302 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8303 return ret;
8304 }
8305 else if (want_patsub)
8306 {
8307 temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, pflags, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8308 FREE (name);
8309 FREE (value);
8310 FREE (temp);
8311
8312 if (temp1 == &expand_param_error)
8313 return (&expand_wdesc_error);
8314 else if (temp1 == &expand_param_fatal)
8315 return (&expand_wdesc_fatal);
8316
8317 ret = alloc_word_desc ();
8318 ret->word = temp1;
8319 if (temp1 &&
8320 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8321 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8322 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8323 return ret;
8324 }
8325 #if defined (CASEMOD_EXPANSIONS)
8326 else if (want_casemod)
8327 {
8328 temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8329 FREE (name);
8330 FREE (value);
8331 FREE (temp);
8332
8333 if (temp1 == &expand_param_error)
8334 return (&expand_wdesc_error);
8335 else if (temp1 == &expand_param_fatal)
8336 return (&expand_wdesc_fatal);
8337
8338 ret = alloc_word_desc ();
8339 ret->word = temp1;
8340 if (temp1 &&
8341 (quoted_dollar_atp == 0 || *quoted_dollar_atp == 0) &&
8342 QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8343 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8344 return ret;
8345 }
8346 #endif
8347
8348 /* Do the right thing based on which character ended the variable name. */
8349 switch (c)
8350 {
8351 default:
8352 case '\0':
8353 bad_substitution:
8354 last_command_exit_value = EXECUTION_FAILURE;
8355 report_error (_("%s: bad substitution"), string ? string : "??");
8356 FREE (value);
8357 FREE (temp);
8358 free (name);
8359 if (shell_compatibility_level <= 43)
8360 return &expand_wdesc_error;
8361 else
8362 return ((posixly_correct && interactive_shell == 0) ? &expand_wdesc_fatal : &expand_wdesc_error);
8363
8364 case RBRACE:
8365 break;
8366
8367 case '@':
8368 temp1 = parameter_brace_transform (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8369 free (temp);
8370 free (value);
8371 free (name);
8372 if (temp1 == &expand_param_error || temp1 == &expand_param_fatal)
8373 {
8374 last_command_exit_value = EXECUTION_FAILURE;
8375 report_error (_("%s: bad substitution"), string ? string : "??");
8376 return (temp1 == &expand_param_error ? &expand_wdesc_error : &expand_wdesc_fatal);
8377 }
8378
8379 ret = alloc_word_desc ();
8380 ret->word = temp1;
8381 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8382 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8383 return ret;
8384
8385 case '#': /* ${param#[#]pattern} */
8386 case '%': /* ${param%[%]pattern} */
8387 if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0')
8388 {
8389 FREE (value);
8390 break;
8391 }
8392 temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0);
8393 free (temp);
8394 free (value);
8395 free (name);
8396
8397 ret = alloc_word_desc ();
8398 ret->word = temp1;
8399 if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8400 ret->flags |= W_QUOTED|W_HASQUOTEDNULL;
8401 return ret;
8402
8403 case '-':
8404 case '=':
8405 case '?':
8406 case '+':
8407 if (var_is_set && var_is_null == 0)
8408 {
8409 /* If the operator is `+', we don't want the value of the named
8410 variable for anything, just the value of the right hand side. */
8411 if (c == '+')
8412 {
8413 /* XXX -- if we're double-quoted and the named variable is "$@",
8414 we want to turn off any special handling of "$@" --
8415 we're not using it, so whatever is on the rhs applies. */
8416 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8417 *quoted_dollar_atp = 0;
8418 if (contains_dollar_at)
8419 *contains_dollar_at = 0;
8420
8421 FREE (temp);
8422 if (value)
8423 {
8424 /* From Posix discussion on austin-group list. Issue 221
8425 requires that backslashes escaping `}' inside
8426 double-quoted ${...} be removed. */
8427 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8428 quoted |= Q_DOLBRACE;
8429 ret = parameter_brace_expand_rhs (name, value, c,
8430 quoted,
8431 pflags,
8432 quoted_dollar_atp,
8433 contains_dollar_at);
8434 /* XXX - fix up later, esp. noting presence of
8435 W_HASQUOTEDNULL in ret->flags */
8436 free (value);
8437 }
8438 else
8439 temp = (char *)NULL;
8440 }
8441 else
8442 {
8443 FREE (value);
8444 }
8445 /* Otherwise do nothing; just use the value in TEMP. */
8446 }
8447 else /* VAR not set or VAR is NULL. */
8448 {
8449 FREE (temp);
8450 temp = (char *)NULL;
8451 if (c == '=' && var_is_special)
8452 {
8453 last_command_exit_value = EXECUTION_FAILURE;
8454 report_error (_("$%s: cannot assign in this way"), name);
8455 free (name);
8456 free (value);
8457 return &expand_wdesc_error;
8458 }
8459 else if (c == '?')
8460 {
8461 parameter_brace_expand_error (name, value);
8462 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8463 }
8464 else if (c != '+')
8465 {
8466 /* XXX -- if we're double-quoted and the named variable is "$@",
8467 we want to turn off any special handling of "$@" --
8468 we're not using it, so whatever is on the rhs applies. */
8469 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp)
8470 *quoted_dollar_atp = 0;
8471 if (contains_dollar_at)
8472 *contains_dollar_at = 0;
8473
8474 /* From Posix discussion on austin-group list. Issue 221 requires
8475 that backslashes escaping `}' inside double-quoted ${...} be
8476 removed. */
8477 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
8478 quoted |= Q_DOLBRACE;
8479 ret = parameter_brace_expand_rhs (name, value, c, quoted, pflags,
8480 quoted_dollar_atp,
8481 contains_dollar_at);
8482 /* XXX - fix up later, esp. noting presence of
8483 W_HASQUOTEDNULL in tdesc->flags */
8484 }
8485 free (value);
8486 }
8487
8488 break;
8489 }
8490 free (name);
8491
8492 if (ret == 0)
8493 {
8494 ret = alloc_word_desc ();
8495 ret->flags = tflag;
8496 ret->word = temp;
8497 }
8498 return (ret);
8499 }
8500
8501 /* Expand a single ${xxx} expansion. The braces are optional. When
8502 the braces are used, parameter_brace_expand() does the work,
8503 possibly calling param_expand recursively. */
8504 static WORD_DESC *
8505 param_expand (string, sindex, quoted, expanded_something,
8506 contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p,
8507 pflags)
8508 char *string;
8509 int *sindex, quoted, *expanded_something, *contains_dollar_at;
8510 int *quoted_dollar_at_p, *had_quoted_null_p, pflags;
8511 {
8512 char *temp, *temp1, uerror[3], *savecmd;
8513 int zindex, t_index, expok;
8514 unsigned char c;
8515 intmax_t number;
8516 SHELL_VAR *var;
8517 WORD_LIST *list;
8518 WORD_DESC *tdesc, *ret;
8519 int tflag;
8520
8521 /*itrace("param_expand: `%s' pflags = %d", string+*sindex, pflags);*/
8522 zindex = *sindex;
8523 c = string[++zindex];
8524
8525 temp = (char *)NULL;
8526 ret = tdesc = (WORD_DESC *)NULL;
8527 tflag = 0;
8528
8529 /* Do simple cases first. Switch on what follows '$'. */
8530 switch (c)
8531 {
8532 /* $0 .. $9? */
8533 case '0':
8534 case '1':
8535 case '2':
8536 case '3':
8537 case '4':
8538 case '5':
8539 case '6':
8540 case '7':
8541 case '8':
8542 case '9':
8543 temp1 = dollar_vars[TODIGIT (c)];
8544 if (unbound_vars_is_error && temp1 == (char *)NULL)
8545 {
8546 uerror[0] = '$';
8547 uerror[1] = c;
8548 uerror[2] = '\0';
8549 last_command_exit_value = EXECUTION_FAILURE;
8550 err_unboundvar (uerror);
8551 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8552 }
8553 if (temp1)
8554 temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8555 ? quote_string (temp1)
8556 : quote_escapes (temp1);
8557 else
8558 temp = (char *)NULL;
8559
8560 break;
8561
8562 /* $$ -- pid of the invoking shell. */
8563 case '$':
8564 temp = itos (dollar_dollar_pid);
8565 break;
8566
8567 /* $# -- number of positional parameters. */
8568 case '#':
8569 temp = itos (number_of_args ());
8570 break;
8571
8572 /* $? -- return value of the last synchronous command. */
8573 case '?':
8574 temp = itos (last_command_exit_value);
8575 break;
8576
8577 /* $- -- flags supplied to the shell on invocation or by `set'. */
8578 case '-':
8579 temp = which_set_flags ();
8580 break;
8581
8582 /* $! -- Pid of the last asynchronous command. */
8583 case '!':
8584 /* If no asynchronous pids have been created, expand to nothing.
8585 If `set -u' has been executed, and no async processes have
8586 been created, this is an expansion error. */
8587 if (last_asynchronous_pid == NO_PID)
8588 {
8589 if (expanded_something)
8590 *expanded_something = 0;
8591 temp = (char *)NULL;
8592 if (unbound_vars_is_error)
8593 {
8594 uerror[0] = '$';
8595 uerror[1] = c;
8596 uerror[2] = '\0';
8597 last_command_exit_value = EXECUTION_FAILURE;
8598 err_unboundvar (uerror);
8599 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8600 }
8601 }
8602 else
8603 temp = itos (last_asynchronous_pid);
8604 break;
8605
8606 /* The only difference between this and $@ is when the arg is quoted. */
8607 case '*': /* `$*' */
8608 list = list_rest_of_args ();
8609
8610 #if 0
8611 /* According to austin-group posix proposal by Geoff Clare in
8612 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8613
8614 "The shell shall write a message to standard error and
8615 immediately exit when it tries to expand an unset parameter
8616 other than the '@' and '*' special parameters."
8617 */
8618
8619 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
8620 {
8621 uerror[0] = '$';
8622 uerror[1] = '*';
8623 uerror[2] = '\0';
8624 last_command_exit_value = EXECUTION_FAILURE;
8625 err_unboundvar (uerror);
8626 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8627 }
8628 #endif
8629
8630 /* If there are no command-line arguments, this should just
8631 disappear if there are other characters in the expansion,
8632 even if it's quoted. */
8633 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0)
8634 temp = (char *)NULL;
8635 else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE))
8636 {
8637 /* If we have "$*" we want to make a string of the positional
8638 parameters, separated by the first character of $IFS, and
8639 quote the whole string, including the separators. If IFS
8640 is unset, the parameters are separated by ' '; if $IFS is
8641 null, the parameters are concatenated. */
8642 temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list);
8643 if (temp)
8644 {
8645 temp1 = (quoted & Q_DOUBLE_QUOTES) ? quote_string (temp) : temp;
8646 if (*temp == 0)
8647 tflag |= W_HASQUOTEDNULL;
8648 if (temp != temp1)
8649 free (temp);
8650 temp = temp1;
8651 }
8652 }
8653 else
8654 {
8655 /* We check whether or not we're eventually going to split $* here,
8656 for example when IFS is empty and we are processing the rhs of
8657 an assignment statement. In that case, we don't separate the
8658 arguments at all. Otherwise, if the $* is not quoted it is
8659 identical to $@ */
8660 # if defined (HANDLE_MULTIBYTE)
8661 if (expand_no_split_dollar_star && ifs_firstc[0] == 0)
8662 # else
8663 if (expand_no_split_dollar_star && ifs_firstc == 0)
8664 # endif
8665 temp = string_list_dollar_star (list);
8666 else
8667 {
8668 temp = string_list_dollar_at (list, quoted, 0);
8669 if (quoted == 0 && (ifs_is_set == 0 || ifs_is_null))
8670 tflag |= W_SPLITSPACE;
8671 /* If we're not quoted but we still don't want word splitting, make
8672 we quote the IFS characters to protect them from splitting (e.g.,
8673 when $@ is in the string as well). */
8674 else if (temp && quoted == 0 && ifs_is_set && (pflags & PF_ASSIGNRHS))
8675 {
8676 temp1 = quote_string (temp);
8677 free (temp);
8678 temp = temp1;
8679 }
8680 }
8681
8682 if (expand_no_split_dollar_star == 0 && contains_dollar_at)
8683 *contains_dollar_at = 1;
8684 }
8685
8686 dispose_words (list);
8687 break;
8688
8689 /* When we have "$@" what we want is "$1" "$2" "$3" ... This
8690 means that we have to turn quoting off after we split into
8691 the individually quoted arguments so that the final split
8692 on the first character of $IFS is still done. */
8693 case '@': /* `$@' */
8694 list = list_rest_of_args ();
8695
8696 #if 0
8697 /* According to austin-group posix proposal by Geoff Clare in
8698 <20090505091501.GA10097@squonk.masqnet> of 5 May 2009:
8699
8700 "The shell shall write a message to standard error and
8701 immediately exit when it tries to expand an unset parameter
8702 other than the '@' and '*' special parameters."
8703 */
8704
8705 if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0)
8706 {
8707 uerror[0] = '$';
8708 uerror[1] = '@';
8709 uerror[2] = '\0';
8710 last_command_exit_value = EXECUTION_FAILURE;
8711 err_unboundvar (uerror);
8712 return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal);
8713 }
8714 #endif
8715
8716 /* We want to flag the fact that we saw this. We can't turn
8717 off quoting entirely, because other characters in the
8718 string might need it (consider "\"$@\""), but we need some
8719 way to signal that the final split on the first character
8720 of $IFS should be done, even though QUOTED is 1. */
8721 /* XXX - should this test include Q_PATQUOTE? */
8722 if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8723 *quoted_dollar_at_p = 1;
8724 if (contains_dollar_at)
8725 *contains_dollar_at = 1;
8726
8727 /* We want to separate the positional parameters with the first
8728 character of $IFS in case $IFS is something other than a space.
8729 We also want to make sure that splitting is done no matter what --
8730 according to POSIX.2, this expands to a list of the positional
8731 parameters no matter what IFS is set to. */
8732 /* XXX - what to do when in a context where word splitting is not
8733 performed? Even when IFS is not the default, posix seems to imply
8734 that we behave like unquoted $* ? Maybe we should use PF_NOSPLIT2
8735 here. */
8736 /* XXX - bash-4.4/bash-5.0 passing PFLAGS */
8737 temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted, pflags);
8738
8739 tflag |= W_DOLLARAT;
8740 dispose_words (list);
8741 break;
8742
8743 case LBRACE:
8744 tdesc = parameter_brace_expand (string, &zindex, quoted, pflags,
8745 quoted_dollar_at_p,
8746 contains_dollar_at);
8747
8748 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8749 return (tdesc);
8750 temp = tdesc ? tdesc->word : (char *)0;
8751
8752 /* XXX */
8753 /* Quoted nulls should be removed if there is anything else
8754 in the string. */
8755 /* Note that we saw the quoted null so we can add one back at
8756 the end of this function if there are no other characters
8757 in the string, discard TEMP, and go on. The exception to
8758 this is when we have "${@}" and $1 is '', since $@ needs
8759 special handling. */
8760 if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp))
8761 {
8762 if (had_quoted_null_p)
8763 *had_quoted_null_p = 1;
8764 if (*quoted_dollar_at_p == 0)
8765 {
8766 free (temp);
8767 tdesc->word = temp = (char *)NULL;
8768 }
8769
8770 }
8771
8772 ret = tdesc;
8773 goto return0;
8774
8775 /* Do command or arithmetic substitution. */
8776 case LPAREN:
8777 /* We have to extract the contents of this paren substitution. */
8778 t_index = zindex + 1;
8779 /* XXX - might want to check for string[t_index+2] == LPAREN and parse
8780 as arithmetic substitution immediately. */
8781 temp = extract_command_subst (string, &t_index, (pflags&PF_COMPLETE) ? SX_COMPLETE : 0);
8782 zindex = t_index;
8783
8784 /* For Posix.2-style `$(( ))' arithmetic substitution,
8785 extract the expression and pass it to the evaluator. */
8786 if (temp && *temp == LPAREN)
8787 {
8788 char *temp2;
8789 temp1 = temp + 1;
8790 temp2 = savestring (temp1);
8791 t_index = strlen (temp2) - 1;
8792
8793 if (temp2[t_index] != RPAREN)
8794 {
8795 free (temp2);
8796 goto comsub;
8797 }
8798
8799 /* Cut off ending `)' */
8800 temp2[t_index] = '\0';
8801
8802 if (chk_arithsub (temp2, t_index) == 0)
8803 {
8804 free (temp2);
8805 #if 0
8806 internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution"));
8807 #endif
8808 goto comsub;
8809 }
8810
8811 /* Expand variables found inside the expression. */
8812 temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES|Q_ARITH);
8813 free (temp2);
8814
8815 arithsub:
8816 /* No error messages. */
8817 savecmd = this_command_name;
8818 this_command_name = (char *)NULL;
8819 number = evalexp (temp1, &expok);
8820 this_command_name = savecmd;
8821 free (temp);
8822 free (temp1);
8823 if (expok == 0)
8824 {
8825 if (interactive_shell == 0 && posixly_correct)
8826 {
8827 last_command_exit_value = EXECUTION_FAILURE;
8828 return (&expand_wdesc_fatal);
8829 }
8830 else
8831 return (&expand_wdesc_error);
8832 }
8833 temp = itos (number);
8834 break;
8835 }
8836
8837 comsub:
8838 if (pflags & PF_NOCOMSUB)
8839 /* we need zindex+1 because string[zindex] == RPAREN */
8840 temp1 = substring (string, *sindex, zindex+1);
8841 else
8842 {
8843 tdesc = command_substitute (temp, quoted);
8844 temp1 = tdesc ? tdesc->word : (char *)NULL;
8845 if (tdesc)
8846 dispose_word_desc (tdesc);
8847 }
8848 FREE (temp);
8849 temp = temp1;
8850 break;
8851
8852 /* Do POSIX.2d9-style arithmetic substitution. This will probably go
8853 away in a future bash release. */
8854 case '[':
8855 /* Extract the contents of this arithmetic substitution. */
8856 t_index = zindex + 1;
8857 temp = extract_arithmetic_subst (string, &t_index);
8858 zindex = t_index;
8859 if (temp == 0)
8860 {
8861 temp = savestring (string);
8862 if (expanded_something)
8863 *expanded_something = 0;
8864 goto return0;
8865 }
8866
8867 /* Do initial variable expansion. */
8868 temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES|Q_ARITH);
8869
8870 goto arithsub;
8871
8872 default:
8873 /* Find the variable in VARIABLE_LIST. */
8874 temp = (char *)NULL;
8875
8876 for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++)
8877 ;
8878 temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL;
8879
8880 /* If this isn't a variable name, then just output the `$'. */
8881 if (temp1 == 0 || *temp1 == '\0')
8882 {
8883 FREE (temp1);
8884 temp = (char *)xmalloc (2);
8885 temp[0] = '$';
8886 temp[1] = '\0';
8887 if (expanded_something)
8888 *expanded_something = 0;
8889 goto return0;
8890 }
8891
8892 /* If the variable exists, return its value cell. */
8893 var = find_variable (temp1);
8894
8895 if (var && invisible_p (var) == 0 && var_isset (var))
8896 {
8897 #if defined (ARRAY_VARS)
8898 if (assoc_p (var) || array_p (var))
8899 {
8900 temp = array_p (var) ? array_reference (array_cell (var), 0)
8901 : assoc_reference (assoc_cell (var), "0");
8902 if (temp)
8903 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8904 ? quote_string (temp)
8905 : quote_escapes (temp);
8906 else if (unbound_vars_is_error)
8907 goto unbound_variable;
8908 }
8909 else
8910 #endif
8911 {
8912 temp = value_cell (var);
8913
8914 temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)))
8915 ? quote_string (temp)
8916 : quote_escapes (temp);
8917 }
8918
8919 free (temp1);
8920
8921 goto return0;
8922 }
8923 else if (var && (invisible_p (var) || var_isset (var) == 0))
8924 temp = (char *)NULL;
8925 else if ((var = find_variable_last_nameref (temp1, 0)) && var_isset (var) && invisible_p (var) == 0)
8926 {
8927 temp = nameref_cell (var);
8928 #if defined (ARRAY_VARS)
8929 if (temp && *temp && valid_array_reference (temp, 0))
8930 {
8931 tdesc = parameter_brace_expand_word (temp, SPECIAL_VAR (temp, 0), quoted, pflags, (arrayind_t *)NULL);
8932 if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal)
8933 return (tdesc);
8934 ret = tdesc;
8935 goto return0;
8936 }
8937 else
8938 #endif
8939 /* y=2 ; typeset -n x=y; echo $x is not the same as echo $2 in ksh */
8940 if (temp && *temp && legal_identifier (temp) == 0)
8941 {
8942 last_command_exit_value = EXECUTION_FAILURE;
8943 report_error (_("%s: invalid variable name for name reference"), temp);
8944 return (&expand_wdesc_error); /* XXX */
8945 }
8946 else
8947 temp = (char *)NULL;
8948 }
8949
8950 temp = (char *)NULL;
8951
8952 unbound_variable:
8953 if (unbound_vars_is_error)
8954 {
8955 last_command_exit_value = EXECUTION_FAILURE;
8956 err_unboundvar (temp1);
8957 }
8958 else
8959 {
8960 free (temp1);
8961 goto return0;
8962 }
8963
8964 free (temp1);
8965 last_command_exit_value = EXECUTION_FAILURE;
8966 return ((unbound_vars_is_error && interactive_shell == 0)
8967 ? &expand_wdesc_fatal
8968 : &expand_wdesc_error);
8969 }
8970
8971 if (string[zindex])
8972 zindex++;
8973
8974 return0:
8975 *sindex = zindex;
8976
8977 if (ret == 0)
8978 {
8979 ret = alloc_word_desc ();
8980 ret->flags = tflag; /* XXX */
8981 ret->word = temp;
8982 }
8983 return ret;
8984 }
8985
8986 void
8987 invalidate_cached_quoted_dollar_at ()
8988 {
8989 dispose_words (cached_quoted_dollar_at);
8990 cached_quoted_dollar_at = 0;
8991 }
8992
8993 /* Make a word list which is the result of parameter and variable
8994 expansion, command substitution, arithmetic substitution, and
8995 quote removal of WORD. Return a pointer to a WORD_LIST which is
8996 the result of the expansion. If WORD contains a null word, the
8997 word list returned is also null.
8998
8999 QUOTED contains flag values defined in shell.h.
9000
9001 ISEXP is used to tell expand_word_internal that the word should be
9002 treated as the result of an expansion. This has implications for
9003 how IFS characters in the word are treated.
9004
9005 CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null
9006 they point to an integer value which receives information about expansion.
9007 CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero.
9008 EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions,
9009 else zero.
9010
9011 This only does word splitting in the case of $@ expansion. In that
9012 case, we split on ' '. */
9013
9014 /* Values for the local variable quoted_state. */
9015 #define UNQUOTED 0
9016 #define PARTIALLY_QUOTED 1
9017 #define WHOLLY_QUOTED 2
9018
9019 static WORD_LIST *
9020 expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something)
9021 WORD_DESC *word;
9022 int quoted, isexp;
9023 int *contains_dollar_at;
9024 int *expanded_something;
9025 {
9026 WORD_LIST *list;
9027 WORD_DESC *tword;
9028
9029 /* The intermediate string that we build while expanding. */
9030 char *istring;
9031
9032 /* The current size of the above object. */
9033 size_t istring_size;
9034
9035 /* Index into ISTRING. */
9036 int istring_index;
9037
9038 /* Temporary string storage. */
9039 char *temp, *temp1;
9040
9041 /* The text of WORD. */
9042 register char *string;
9043
9044 /* The size of STRING. */
9045 size_t string_size;
9046
9047 /* The index into STRING. */
9048 int sindex;
9049
9050 /* This gets 1 if we see a $@ while quoted. */
9051 int quoted_dollar_at;
9052
9053 /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on
9054 whether WORD contains no quoting characters, a partially quoted
9055 string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */
9056 int quoted_state;
9057
9058 /* State flags */
9059 int had_quoted_null;
9060 int has_dollar_at, temp_has_dollar_at;
9061 int split_on_spaces;
9062 int tflag;
9063 int pflags; /* flags passed to param_expand */
9064 int mb_cur_max;
9065
9066 int assignoff; /* If assignment, offset of `=' */
9067
9068 register unsigned char c; /* Current character. */
9069 int t_index; /* For calls to string_extract_xxx. */
9070
9071 char twochars[2];
9072
9073 DECLARE_MBSTATE;
9074
9075 /* OK, let's see if we can optimize a common idiom: "$@" */
9076 if (STREQ (word->word, "\"$@\"") &&
9077 (word->flags == (W_HASDOLLAR|W_QUOTED)) &&
9078 dollar_vars[1]) /* XXX - check IFS here as well? */
9079 {
9080 if (contains_dollar_at)
9081 *contains_dollar_at = 1;
9082 if (expanded_something)
9083 *expanded_something = 1;
9084 if (cached_quoted_dollar_at)
9085 return (copy_word_list (cached_quoted_dollar_at));
9086 list = list_rest_of_args ();
9087 list = quote_list (list);
9088 cached_quoted_dollar_at = copy_word_list (list);
9089 return (list);
9090 }
9091
9092 istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE);
9093 istring[istring_index = 0] = '\0';
9094 quoted_dollar_at = had_quoted_null = has_dollar_at = 0;
9095 split_on_spaces = 0;
9096 quoted_state = UNQUOTED;
9097
9098 string = word->word;
9099 if (string == 0)
9100 goto finished_with_string;
9101 mb_cur_max = MB_CUR_MAX;
9102
9103 /* Don't need the string length for the SADD... and COPY_ macros unless
9104 multibyte characters are possible. */
9105 string_size = (mb_cur_max > 1) ? strlen (string) : 1;
9106
9107 if (contains_dollar_at)
9108 *contains_dollar_at = 0;
9109
9110 assignoff = -1;
9111
9112 /* Begin the expansion. */
9113
9114 for (sindex = 0; ;)
9115 {
9116 c = string[sindex];
9117
9118 /* Case on top-level character. */
9119 switch (c)
9120 {
9121 case '\0':
9122 goto finished_with_string;
9123
9124 case CTLESC:
9125 sindex++;
9126 #if HANDLE_MULTIBYTE
9127 if (mb_cur_max > 1 && string[sindex])
9128 {
9129 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
9130 }
9131 else
9132 #endif
9133 {
9134 temp = (char *)xmalloc (3);
9135 temp[0] = CTLESC;
9136 temp[1] = c = string[sindex];
9137 temp[2] = '\0';
9138 }
9139
9140 dollar_add_string:
9141 if (string[sindex])
9142 sindex++;
9143
9144 add_string:
9145 if (temp)
9146 {
9147 istring = sub_append_string (temp, istring, &istring_index, &istring_size);
9148 temp = (char *)0;
9149 }
9150
9151 break;
9152
9153 #if defined (PROCESS_SUBSTITUTION)
9154 /* Process substitution. */
9155 case '<':
9156 case '>':
9157 {
9158 /* bash-4.4/bash-5.0
9159 XXX - technically this should only be expanded at the start
9160 of a word */
9161 if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct)
9162 {
9163 sindex--; /* add_character: label increments sindex */
9164 goto add_character;
9165 }
9166 else
9167 t_index = sindex + 1; /* skip past both '<' and LPAREN */
9168
9169 temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index, 0); /*))*/
9170 sindex = t_index;
9171
9172 /* If the process substitution specification is `<()', we want to
9173 open the pipe for writing in the child and produce output; if
9174 it is `>()', we want to open the pipe for reading in the child
9175 and consume input. */
9176 temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0;
9177
9178 FREE (temp1);
9179
9180 goto dollar_add_string;
9181 }
9182 #endif /* PROCESS_SUBSTITUTION */
9183
9184 case '=':
9185 /* Posix.2 section 3.6.1 says that tildes following `=' in words
9186 which are not assignment statements are not expanded. If the
9187 shell isn't in posix mode, though, we perform tilde expansion
9188 on `likely candidate' unquoted assignment statements (flags
9189 include W_ASSIGNMENT but not W_QUOTED). A likely candidate
9190 contains an unquoted :~ or =~. Something to think about: we
9191 now have a flag that says to perform tilde expansion on arguments
9192 to `assignment builtins' like declare and export that look like
9193 assignment statements. We now do tilde expansion on such words
9194 even in POSIX mode. */
9195 if (word->flags & (W_ASSIGNRHS|W_NOTILDE))
9196 {
9197 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
9198 goto add_ifs_character;
9199 else
9200 goto add_character;
9201 }
9202 /* If we're not in posix mode or forcing assignment-statement tilde
9203 expansion, note where the `=' appears in the word and prepare to
9204 do tilde expansion following the first `='. */
9205 if ((word->flags & W_ASSIGNMENT) &&
9206 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9207 assignoff == -1 && sindex > 0)
9208 assignoff = sindex;
9209 if (sindex == assignoff && string[sindex+1] == '~') /* XXX */
9210 word->flags |= W_ITILDE;
9211 #if 0
9212 else if ((word->flags & W_ASSIGNMENT) &&
9213 (posixly_correct == 0 || (word->flags & W_TILDEEXP)) &&
9214 string[sindex+1] == '~')
9215 word->flags |= W_ITILDE;
9216 #endif
9217
9218 /* XXX - bash-4.4/bash-5.0 */
9219 if (word->flags & W_ASSIGNARG)
9220 word->flags |= W_ASSIGNRHS; /* affects $@ */
9221
9222 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
9223 goto add_ifs_character;
9224 else
9225 goto add_character;
9226
9227 case ':':
9228 if (word->flags & W_NOTILDE)
9229 {
9230 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
9231 goto add_ifs_character;
9232 else
9233 goto add_character;
9234 }
9235
9236 if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) &&
9237 string[sindex+1] == '~')
9238 word->flags |= W_ITILDE;
9239
9240 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c))
9241 goto add_ifs_character;
9242 else
9243 goto add_character;
9244
9245 case '~':
9246 /* If the word isn't supposed to be tilde expanded, or we're not
9247 at the start of a word or after an unquoted : or = in an
9248 assignment statement, we don't do tilde expansion. If we don't want
9249 tilde expansion when expanding words to be passed to the arithmetic
9250 evaluator, remove the check for Q_ARITH. */
9251 if ((word->flags & (W_NOTILDE|W_DQUOTE)) ||
9252 (sindex > 0 && ((word->flags & W_ITILDE) == 0)) ||
9253 ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0)))
9254 {
9255 word->flags &= ~W_ITILDE;
9256 if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0)
9257 goto add_ifs_character;
9258 else
9259 goto add_character;
9260 }
9261
9262 if (word->flags & W_ASSIGNRHS)
9263 tflag = 2;
9264 else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP))
9265 tflag = 1;
9266 else
9267 tflag = 0;
9268
9269 temp = bash_tilde_find_word (string + sindex, tflag, &t_index);
9270
9271 word->flags &= ~W_ITILDE;
9272
9273 if (temp && *temp && t_index > 0)
9274 {
9275 temp1 = bash_tilde_expand (temp, tflag);
9276 if (temp1 && *temp1 == '~' && STREQ (temp, temp1))
9277 {
9278 FREE (temp);
9279 FREE (temp1);
9280 goto add_character; /* tilde expansion failed */
9281 }
9282 free (temp);
9283 temp = temp1;
9284 sindex += t_index;
9285 goto add_quoted_string; /* XXX was add_string */
9286 }
9287 else
9288 {
9289 FREE (temp);
9290 goto add_character;
9291 }
9292
9293 case '$':
9294 if (expanded_something)
9295 *expanded_something = 1;
9296
9297 temp_has_dollar_at = 0;
9298 pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0;
9299 if (word->flags & W_NOSPLIT2)
9300 pflags |= PF_NOSPLIT2;
9301 if (word->flags & W_ASSIGNRHS)
9302 pflags |= PF_ASSIGNRHS;
9303 if (word->flags & W_COMPLETE)
9304 pflags |= PF_COMPLETE;
9305 tword = param_expand (string, &sindex, quoted, expanded_something,
9306 &temp_has_dollar_at, &quoted_dollar_at,
9307 &had_quoted_null, pflags);
9308 has_dollar_at += temp_has_dollar_at;
9309 split_on_spaces += (tword->flags & W_SPLITSPACE);
9310
9311 if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal)
9312 {
9313 free (string);
9314 free (istring);
9315 return ((tword == &expand_wdesc_error) ? &expand_word_error
9316 : &expand_word_fatal);
9317 }
9318 if (contains_dollar_at && has_dollar_at)
9319 *contains_dollar_at = 1;
9320
9321 if (tword && (tword->flags & W_HASQUOTEDNULL))
9322 had_quoted_null = 1;
9323
9324 temp = tword ? tword->word : (char *)NULL;
9325 dispose_word_desc (tword);
9326
9327 /* Kill quoted nulls; we will add them back at the end of
9328 expand_word_internal if nothing else in the string */
9329 if (had_quoted_null && temp && QUOTED_NULL (temp))
9330 {
9331 FREE (temp);
9332 temp = (char *)NULL;
9333 }
9334
9335 goto add_string;
9336 break;
9337
9338 case '`': /* Backquoted command substitution. */
9339 {
9340 t_index = sindex++;
9341
9342 temp = string_extract (string, &sindex, "`", SX_REQMATCH);
9343 /* The test of sindex against t_index is to allow bare instances of
9344 ` to pass through, for backwards compatibility. */
9345 if (temp == &extract_string_error || temp == &extract_string_fatal)
9346 {
9347 if (sindex - 1 == t_index)
9348 {
9349 sindex = t_index;
9350 goto add_character;
9351 }
9352 last_command_exit_value = EXECUTION_FAILURE;
9353 report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index);
9354 free (string);
9355 free (istring);
9356 return ((temp == &extract_string_error) ? &expand_word_error
9357 : &expand_word_fatal);
9358 }
9359
9360 if (expanded_something)
9361 *expanded_something = 1;
9362
9363 if (word->flags & W_NOCOMSUB)
9364 /* sindex + 1 because string[sindex] == '`' */
9365 temp1 = substring (string, t_index, sindex + 1);
9366 else
9367 {
9368 de_backslash (temp);
9369 tword = command_substitute (temp, quoted);
9370 temp1 = tword ? tword->word : (char *)NULL;
9371 if (tword)
9372 dispose_word_desc (tword);
9373 }
9374 FREE (temp);
9375 temp = temp1;
9376 goto dollar_add_string;
9377 }
9378
9379 case '\\':
9380 if (string[sindex + 1] == '\n')
9381 {
9382 sindex += 2;
9383 continue;
9384 }
9385
9386 c = string[++sindex];
9387
9388 if (quoted & Q_HERE_DOCUMENT)
9389 tflag = CBSHDOC;
9390 else if (quoted & Q_DOUBLE_QUOTES)
9391 tflag = CBSDQUOTE;
9392 else
9393 tflag = 0;
9394
9395 /* From Posix discussion on austin-group list: Backslash escaping
9396 a } in ${...} is removed. Issue 0000221 */
9397 if ((quoted & Q_DOLBRACE) && c == RBRACE)
9398 {
9399 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9400 }
9401 /* This is the fix for " $@\ " */
9402 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0) && isexp == 0 && isifs (c))
9403 {
9404 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9405 DEFAULT_ARRAY_SIZE);
9406 istring[istring_index++] = CTLESC;
9407 istring[istring_index++] = '\\';
9408 istring[istring_index] = '\0';
9409
9410 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9411 }
9412 else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0))
9413 {
9414 SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size);
9415 }
9416 else if (c == 0)
9417 {
9418 c = CTLNUL;
9419 sindex--; /* add_character: label increments sindex */
9420 goto add_character;
9421 }
9422 else
9423 {
9424 SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size);
9425 }
9426
9427 sindex++;
9428 add_twochars:
9429 /* BEFORE jumping here, we need to increment sindex if appropriate */
9430 RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size,
9431 DEFAULT_ARRAY_SIZE);
9432 istring[istring_index++] = twochars[0];
9433 istring[istring_index++] = twochars[1];
9434 istring[istring_index] = '\0';
9435
9436 break;
9437
9438 case '"':
9439 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) && ((quoted & Q_ARITH) == 0))
9440 goto add_character;
9441
9442 t_index = ++sindex;
9443 temp = string_extract_double_quoted (string, &sindex, (word->flags & W_COMPLETE) ? SX_COMPLETE : 0);
9444
9445 /* If the quotes surrounded the entire string, then the
9446 whole word was quoted. */
9447 quoted_state = (t_index == 1 && string[sindex] == '\0')
9448 ? WHOLLY_QUOTED
9449 : PARTIALLY_QUOTED;
9450
9451 if (temp && *temp)
9452 {
9453 tword = alloc_word_desc ();
9454 tword->word = temp;
9455
9456 /* XXX - bash-4.4/bash-5.0 */
9457 if (word->flags & W_ASSIGNARG)
9458 tword->flags |= word->flags & (W_ASSIGNARG|W_ASSIGNRHS); /* affects $@ */
9459 if (word->flags & W_COMPLETE)
9460 tword->flags |= W_COMPLETE; /* for command substitutions */
9461
9462 temp = (char *)NULL;
9463
9464 temp_has_dollar_at = 0; /* XXX */
9465 /* Need to get W_HASQUOTEDNULL flag through this function. */
9466 list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &temp_has_dollar_at, (int *)NULL);
9467 has_dollar_at += temp_has_dollar_at;
9468
9469 if (list == &expand_word_error || list == &expand_word_fatal)
9470 {
9471 free (istring);
9472 free (string);
9473 /* expand_word_internal has already freed temp_word->word
9474 for us because of the way it prints error messages. */
9475 tword->word = (char *)NULL;
9476 dispose_word (tword);
9477 return list;
9478 }
9479
9480 dispose_word (tword);
9481
9482 /* "$@" (a double-quoted dollar-at) expands into nothing,
9483 not even a NULL word, when there are no positional
9484 parameters. */
9485 if (list == 0 && temp_has_dollar_at) /* XXX - was has_dollar_at */
9486 {
9487 quoted_dollar_at++;
9488 break;
9489 }
9490
9491 /* If we get "$@", we know we have expanded something, so we
9492 need to remember it for the final split on $IFS. This is
9493 a special case; it's the only case where a quoted string
9494 can expand into more than one word. It's going to come back
9495 from the above call to expand_word_internal as a list with
9496 a single word, in which all characters are quoted and
9497 separated by blanks. What we want to do is to turn it back
9498 into a list for the next piece of code. */
9499 if (list)
9500 dequote_list (list);
9501
9502 if (list && list->word && (list->word->flags & W_HASQUOTEDNULL))
9503 had_quoted_null = 1; /* XXX */
9504
9505 if (temp_has_dollar_at) /* XXX - was has_dollar_at */
9506 {
9507 quoted_dollar_at++;
9508 if (contains_dollar_at)
9509 *contains_dollar_at = 1;
9510 if (expanded_something)
9511 *expanded_something = 1;
9512 }
9513 }
9514 else
9515 {
9516 /* What we have is "". This is a minor optimization. */
9517 FREE (temp);
9518 list = (WORD_LIST *)NULL;
9519 }
9520
9521 /* The code above *might* return a list (consider the case of "$@",
9522 where it returns "$1", "$2", etc.). We can't throw away the
9523 rest of the list, and we have to make sure each word gets added
9524 as quoted. We test on tresult->next: if it is non-NULL, we
9525 quote the whole list, save it to a string with string_list, and
9526 add that string. We don't need to quote the results of this
9527 (and it would be wrong, since that would quote the separators
9528 as well), so we go directly to add_string. */
9529 if (list)
9530 {
9531 if (list->next)
9532 {
9533 /* Testing quoted_dollar_at makes sure that "$@" is
9534 split correctly when $IFS does not contain a space. */
9535 temp = quoted_dollar_at
9536 ? string_list_dollar_at (list, Q_DOUBLE_QUOTES, 0)
9537 : string_list (quote_list (list));
9538 dispose_words (list);
9539 goto add_string;
9540 }
9541 else
9542 {
9543 temp = savestring (list->word->word);
9544 tflag = list->word->flags;
9545 dispose_words (list);
9546
9547 /* If the string is not a quoted null string, we want
9548 to remove any embedded unquoted CTLNUL characters.
9549 We do not want to turn quoted null strings back into
9550 the empty string, though. We do this because we
9551 want to remove any quoted nulls from expansions that
9552 contain other characters. For example, if we have
9553 x"$*"y or "x$*y" and there are no positional parameters,
9554 the $* should expand into nothing. */
9555 /* We use the W_HASQUOTEDNULL flag to differentiate the
9556 cases: a quoted null character as above and when
9557 CTLNUL is contained in the (non-null) expansion
9558 of some variable. We use the had_quoted_null flag to
9559 pass the value through this function to its caller. */
9560 if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0)
9561 remove_quoted_nulls (temp); /* XXX */
9562 }
9563 }
9564 else
9565 temp = (char *)NULL;
9566
9567 /* We do not want to add quoted nulls to strings that are only
9568 partially quoted; we can throw them away. The exception to
9569 this is when we are going to be performing word splitting,
9570 since we have to preserve a null argument if the next character
9571 will cause word splitting. */
9572 if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
9573 continue;
9574
9575 add_quoted_string:
9576
9577 if (temp)
9578 {
9579 temp1 = temp;
9580 temp = quote_string (temp);
9581 free (temp1);
9582 goto add_string;
9583 }
9584 else
9585 {
9586 /* Add NULL arg. */
9587 c = CTLNUL;
9588 sindex--; /* add_character: label increments sindex */
9589 goto add_character;
9590 }
9591
9592 /* break; */
9593
9594 case '\'':
9595 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)))
9596 goto add_character;
9597
9598 t_index = ++sindex;
9599 temp = string_extract_single_quoted (string, &sindex);
9600
9601 /* If the entire STRING was surrounded by single quotes,
9602 then the string is wholly quoted. */
9603 quoted_state = (t_index == 1 && string[sindex] == '\0')
9604 ? WHOLLY_QUOTED
9605 : PARTIALLY_QUOTED;
9606
9607 /* If all we had was '', it is a null expansion. */
9608 if (*temp == '\0')
9609 {
9610 free (temp);
9611 temp = (char *)NULL;
9612 }
9613 else
9614 remove_quoted_escapes (temp); /* ??? */
9615
9616 /* We do not want to add quoted nulls to strings that are only
9617 partially quoted; such nulls are discarded. See above for the
9618 exception, which is when the string is going to be split. */
9619 if (temp == 0 && (quoted_state == PARTIALLY_QUOTED) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)))
9620 continue;
9621
9622 /* If we have a quoted null expansion, add a quoted NULL to istring. */
9623 if (temp == 0)
9624 {
9625 c = CTLNUL;
9626 sindex--; /* add_character: label increments sindex */
9627 goto add_character;
9628 }
9629 else
9630 goto add_quoted_string;
9631
9632 /* break; */
9633
9634 default:
9635 /* This is the fix for " $@ " */
9636 add_ifs_character:
9637 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c) && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0))
9638 {
9639 if (string[sindex]) /* from old goto dollar_add_string */
9640 sindex++;
9641 if (c == 0)
9642 {
9643 c = CTLNUL;
9644 goto add_character;
9645 }
9646 else
9647 {
9648 #if HANDLE_MULTIBYTE
9649 if (mb_cur_max > 1)
9650 sindex--;
9651
9652 if (mb_cur_max > 1)
9653 {
9654 SADD_MBQCHAR_BODY(temp, string, sindex, string_size);
9655 }
9656 else
9657 #endif
9658 {
9659 twochars[0] = CTLESC;
9660 twochars[1] = c;
9661 goto add_twochars;
9662 }
9663 }
9664 }
9665
9666 SADD_MBCHAR (temp, string, sindex, string_size);
9667
9668 add_character:
9669 RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size,
9670 DEFAULT_ARRAY_SIZE);
9671 istring[istring_index++] = c;
9672 istring[istring_index] = '\0';
9673
9674 /* Next character. */
9675 sindex++;
9676 }
9677 }
9678
9679 finished_with_string:
9680 /* OK, we're ready to return. If we have a quoted string, and
9681 quoted_dollar_at is not set, we do no splitting at all; otherwise
9682 we split on ' '. The routines that call this will handle what to
9683 do if nothing has been expanded. */
9684
9685 /* Partially and wholly quoted strings which expand to the empty
9686 string are retained as an empty arguments. Unquoted strings
9687 which expand to the empty string are discarded. The single
9688 exception is the case of expanding "$@" when there are no
9689 positional parameters. In that case, we discard the expansion. */
9690
9691 /* Because of how the code that handles "" and '' in partially
9692 quoted strings works, we need to make ISTRING into a QUOTED_NULL
9693 if we saw quoting characters, but the expansion was empty.
9694 "" and '' are tossed away before we get to this point when
9695 processing partially quoted strings. This makes "" and $xxx""
9696 equivalent when xxx is unset. We also look to see whether we
9697 saw a quoted null from a ${} expansion and add one back if we
9698 need to. */
9699
9700 /* If we expand to nothing and there were no single or double quotes
9701 in the word, we throw it away. Otherwise, we return a NULL word.
9702 The single exception is for $@ surrounded by double quotes when
9703 there are no positional parameters. In that case, we also throw
9704 the word away. */
9705
9706 if (*istring == '\0')
9707 {
9708 if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED))
9709 {
9710 istring[0] = CTLNUL;
9711 istring[1] = '\0';
9712 tword = make_bare_word (istring);
9713 tword->flags |= W_HASQUOTEDNULL; /* XXX */
9714 list = make_word_list (tword, (WORD_LIST *)NULL);
9715 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9716 tword->flags |= W_QUOTED;
9717 }
9718 /* According to sh, ksh, and Posix.2, if a word expands into nothing
9719 and a double-quoted "$@" appears anywhere in it, then the entire
9720 word is removed. */
9721 /* XXX - exception appears to be that quoted null strings result in
9722 null arguments */
9723 else if (quoted_state == UNQUOTED || quoted_dollar_at)
9724 list = (WORD_LIST *)NULL;
9725 #if 0
9726 else
9727 {
9728 tword = make_bare_word (istring);
9729 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9730 tword->flags |= W_QUOTED;
9731 list = make_word_list (tword, (WORD_LIST *)NULL);
9732 }
9733 #else
9734 else
9735 list = (WORD_LIST *)NULL;
9736 #endif
9737 }
9738 else if (word->flags & W_NOSPLIT)
9739 {
9740 tword = make_bare_word (istring);
9741 if (word->flags & W_ASSIGNMENT)
9742 tword->flags |= W_ASSIGNMENT; /* XXX */
9743 if (word->flags & W_COMPASSIGN)
9744 tword->flags |= W_COMPASSIGN; /* XXX */
9745 if (word->flags & W_NOGLOB)
9746 tword->flags |= W_NOGLOB; /* XXX */
9747 if (word->flags & W_NOBRACE)
9748 tword->flags |= W_NOBRACE; /* XXX */
9749 if (word->flags & W_NOEXPAND)
9750 tword->flags |= W_NOEXPAND; /* XXX */
9751 if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))
9752 tword->flags |= W_QUOTED;
9753 if (had_quoted_null && QUOTED_NULL (istring))
9754 tword->flags |= W_HASQUOTEDNULL;
9755 list = make_word_list (tword, (WORD_LIST *)NULL);
9756 }
9757 else
9758 {
9759 char *ifs_chars;
9760 char *tstring;
9761
9762 ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL;
9763
9764 /* If we have $@, we need to split the results no matter what. If
9765 IFS is unset or NULL, string_list_dollar_at has separated the
9766 positional parameters with a space, so we split on space (we have
9767 set ifs_chars to " \t\n" above if ifs is unset). If IFS is set,
9768 string_list_dollar_at has separated the positional parameters
9769 with the first character of $IFS, so we split on $IFS. If
9770 SPLIT_ON_SPACES is set, we expanded $* (unquoted) with IFS either
9771 unset or null, and we want to make sure that we split on spaces
9772 regardless of what else has happened to IFS since the expansion,
9773 or we expanded "$@" with IFS null and we need to split the positional
9774 parameters into separate words. */
9775 if (split_on_spaces)
9776 list = list_string (istring, " ", 1); /* XXX quoted == 1? */
9777 /* If we have $@ (has_dollar_at != 0) and we are in a context where we
9778 don't want to split the result (W_NOSPLIT2), and we are not quoted,
9779 we have already separated the arguments with the first character of
9780 $IFS. In this case, we want to return a list with a single word
9781 with the separator possibly replaced with a space (it's what other
9782 shells seem to do).
9783 quoted_dollar_at is internal to this function and is set if we are
9784 passed an argument that is unquoted (quoted == 0) but we encounter a
9785 double-quoted $@ while expanding it. */
9786 else if (has_dollar_at && quoted_dollar_at == 0 && ifs_chars && quoted == 0 && (word->flags & W_NOSPLIT2))
9787 {
9788 /* Only split and rejoin if we have to */
9789 if (*ifs_chars && *ifs_chars != ' ')
9790 {
9791 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
9792 tstring = string_list (list);
9793 }
9794 else
9795 tstring = istring;
9796 tword = make_bare_word (tstring);
9797 if (tstring != istring)
9798 free (tstring);
9799 goto set_word_flags;
9800 }
9801 /* This is the attempt to make $* in an assignment context (a=$*) and
9802 array variables subscripted with * in an assignment context (a=${foo[*]})
9803 behave similarly. It has side effects that, though they increase
9804 compatibility with other shells, are not backwards compatible. */
9805 #if 0
9806 else if (has_dollar_at && quoted == 0 && ifs_chars && (word->flags & W_ASSIGNRHS))
9807 {
9808 tword = make_bare_word (istring);
9809 goto set_word_flags;
9810 }
9811 #endif
9812 else if (has_dollar_at && ifs_chars)
9813 list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1);
9814 else
9815 {
9816 tword = make_bare_word (istring);
9817 set_word_flags:
9818 if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED))
9819 tword->flags |= W_QUOTED;
9820 if (word->flags & W_ASSIGNMENT)
9821 tword->flags |= W_ASSIGNMENT;
9822 if (word->flags & W_COMPASSIGN)
9823 tword->flags |= W_COMPASSIGN;
9824 if (word->flags & W_NOGLOB)
9825 tword->flags |= W_NOGLOB;
9826 if (word->flags & W_NOBRACE)
9827 tword->flags |= W_NOBRACE;
9828 if (word->flags & W_NOEXPAND)
9829 tword->flags |= W_NOEXPAND;
9830 if (had_quoted_null && QUOTED_NULL (istring))
9831 tword->flags |= W_HASQUOTEDNULL; /* XXX */
9832 list = make_word_list (tword, (WORD_LIST *)NULL);
9833 }
9834 }
9835
9836 free (istring);
9837 return (list);
9838 }
9839
9840 /* **************************************************************** */
9841 /* */
9842 /* Functions for Quote Removal */
9843 /* */
9844 /* **************************************************************** */
9845
9846 /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the
9847 backslash quoting rules for within double quotes or a here document. */
9848 char *
9849 string_quote_removal (string, quoted)
9850 char *string;
9851 int quoted;
9852 {
9853 size_t slen;
9854 char *r, *result_string, *temp, *send;
9855 int sindex, tindex, dquote;
9856 unsigned char c;
9857 DECLARE_MBSTATE;
9858
9859 /* The result can be no longer than the original string. */
9860 slen = strlen (string);
9861 send = string + slen;
9862
9863 r = result_string = (char *)xmalloc (slen + 1);
9864
9865 for (dquote = sindex = 0; c = string[sindex];)
9866 {
9867 switch (c)
9868 {
9869 case '\\':
9870 c = string[++sindex];
9871 if (c == 0)
9872 {
9873 *r++ = '\\';
9874 break;
9875 }
9876 if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0)
9877 *r++ = '\\';
9878 /* FALLTHROUGH */
9879
9880 default:
9881 SCOPY_CHAR_M (r, string, send, sindex);
9882 break;
9883
9884 case '\'':
9885 if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote)
9886 {
9887 *r++ = c;
9888 sindex++;
9889 break;
9890 }
9891 tindex = sindex + 1;
9892 temp = string_extract_single_quoted (string, &tindex);
9893 if (temp)
9894 {
9895 strcpy (r, temp);
9896 r += strlen (r);
9897 free (temp);
9898 }
9899 sindex = tindex;
9900 break;
9901
9902 case '"':
9903 dquote = 1 - dquote;
9904 sindex++;
9905 break;
9906 }
9907 }
9908 *r = '\0';
9909 return (result_string);
9910 }
9911
9912 #if 0
9913 /* UNUSED */
9914 /* Perform quote removal on word WORD. This allocates and returns a new
9915 WORD_DESC *. */
9916 WORD_DESC *
9917 word_quote_removal (word, quoted)
9918 WORD_DESC *word;
9919 int quoted;
9920 {
9921 WORD_DESC *w;
9922 char *t;
9923
9924 t = string_quote_removal (word->word, quoted);
9925 w = alloc_word_desc ();
9926 w->word = t ? t : savestring ("");
9927 return (w);
9928 }
9929
9930 /* Perform quote removal on all words in LIST. If QUOTED is non-zero,
9931 the members of the list are treated as if they are surrounded by
9932 double quotes. Return a new list, or NULL if LIST is NULL. */
9933 WORD_LIST *
9934 word_list_quote_removal (list, quoted)
9935 WORD_LIST *list;
9936 int quoted;
9937 {
9938 WORD_LIST *result, *t, *tresult, *e;
9939
9940 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
9941 {
9942 tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL);
9943 #if 0
9944 result = (WORD_LIST *) list_append (result, tresult);
9945 #else
9946 if (result == 0)
9947 result = e = tresult;
9948 else
9949 {
9950 e->next = tresult;
9951 while (e->next)
9952 e = e->next;
9953 }
9954 #endif
9955 }
9956 return (result);
9957 }
9958 #endif
9959
9960 /*******************************************
9961 * *
9962 * Functions to perform word splitting *
9963 * *
9964 *******************************************/
9965
9966 void
9967 setifs (v)
9968 SHELL_VAR *v;
9969 {
9970 char *t;
9971 unsigned char uc;
9972
9973 ifs_var = v;
9974 ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n";
9975
9976 ifs_is_set = ifs_var != 0;
9977 ifs_is_null = ifs_is_set && (*ifs_value == 0);
9978
9979 /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet
9980 handle multibyte chars in IFS */
9981 memset (ifs_cmap, '\0', sizeof (ifs_cmap));
9982 for (t = ifs_value ; t && *t; t++)
9983 {
9984 uc = *t;
9985 ifs_cmap[uc] = 1;
9986 }
9987
9988 #if defined (HANDLE_MULTIBYTE)
9989 if (ifs_value == 0)
9990 {
9991 ifs_firstc[0] = '\0';
9992 ifs_firstc_len = 1;
9993 }
9994 else
9995 {
9996 size_t ifs_len;
9997 ifs_len = strnlen (ifs_value, MB_CUR_MAX);
9998 ifs_firstc_len = MBLEN (ifs_value, ifs_len);
9999 if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len))
10000 {
10001 ifs_firstc[0] = ifs_value[0];
10002 ifs_firstc[1] = '\0';
10003 ifs_firstc_len = 1;
10004 }
10005 else
10006 memcpy (ifs_firstc, ifs_value, ifs_firstc_len);
10007 }
10008 #else
10009 ifs_firstc = ifs_value ? *ifs_value : 0;
10010 #endif
10011 }
10012
10013 char *
10014 getifs ()
10015 {
10016 return ifs_value;
10017 }
10018
10019 /* This splits a single word into a WORD LIST on $IFS, but only if the word
10020 is not quoted. list_string () performs quote removal for us, even if we
10021 don't do any splitting. */
10022 WORD_LIST *
10023 word_split (w, ifs_chars)
10024 WORD_DESC *w;
10025 char *ifs_chars;
10026 {
10027 WORD_LIST *result;
10028
10029 if (w)
10030 {
10031 char *xifs;
10032
10033 xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars;
10034 result = list_string (w->word, xifs, w->flags & W_QUOTED);
10035 }
10036 else
10037 result = (WORD_LIST *)NULL;
10038
10039 return (result);
10040 }
10041
10042 /* Perform word splitting on LIST and return the RESULT. It is possible
10043 to return (WORD_LIST *)NULL. */
10044 static WORD_LIST *
10045 word_list_split (list)
10046 WORD_LIST *list;
10047 {
10048 WORD_LIST *result, *t, *tresult, *e;
10049
10050 for (t = list, result = (WORD_LIST *)NULL; t; t = t->next)
10051 {
10052 tresult = word_split (t->word, ifs_value);
10053 if (result == 0)
10054 result = e = tresult;
10055 else
10056 {
10057 e->next = tresult;
10058 while (e->next)
10059 e = e->next;
10060 }
10061 }
10062 return (result);
10063 }
10064
10065 /**************************************************
10066 * *
10067 * Functions to expand an entire WORD_LIST *
10068 * *
10069 **************************************************/
10070
10071 /* Do any word-expansion-specific cleanup and jump to top_level */
10072 static void
10073 exp_jump_to_top_level (v)
10074 int v;
10075 {
10076 set_pipestatus_from_exit (last_command_exit_value);
10077
10078 /* Cleanup code goes here. */
10079 expand_no_split_dollar_star = 0; /* XXX */
10080 expanding_redir = 0;
10081 assigning_in_environment = 0;
10082
10083 if (parse_and_execute_level == 0)
10084 top_level_cleanup (); /* from sig.c */
10085
10086 jump_to_top_level (v);
10087 }
10088
10089 /* Put NLIST (which is a WORD_LIST * of only one element) at the front of
10090 ELIST, and set ELIST to the new list. */
10091 #define PREPEND_LIST(nlist, elist) \
10092 do { nlist->next = elist; elist = nlist; } while (0)
10093
10094 /* Separate out any initial variable assignments from TLIST. If set -k has
10095 been executed, remove all assignment statements from TLIST. Initial
10096 variable assignments and other environment assignments are placed
10097 on SUBST_ASSIGN_VARLIST. */
10098 static WORD_LIST *
10099 separate_out_assignments (tlist)
10100 WORD_LIST *tlist;
10101 {
10102 register WORD_LIST *vp, *lp;
10103
10104 if (tlist == 0)
10105 return ((WORD_LIST *)NULL);
10106
10107 if (subst_assign_varlist)
10108 dispose_words (subst_assign_varlist); /* Clean up after previous error */
10109
10110 subst_assign_varlist = (WORD_LIST *)NULL;
10111 vp = lp = tlist;
10112
10113 /* Separate out variable assignments at the start of the command.
10114 Loop invariant: vp->next == lp
10115 Loop postcondition:
10116 lp = list of words left after assignment statements skipped
10117 tlist = original list of words
10118 */
10119 while (lp && (lp->word->flags & W_ASSIGNMENT))
10120 {
10121 vp = lp;
10122 lp = lp->next;
10123 }
10124
10125 /* If lp != tlist, we have some initial assignment statements.
10126 We make SUBST_ASSIGN_VARLIST point to the list of assignment
10127 words and TLIST point to the remaining words. */
10128 if (lp != tlist)
10129 {
10130 subst_assign_varlist = tlist;
10131 /* ASSERT(vp->next == lp); */
10132 vp->next = (WORD_LIST *)NULL; /* terminate variable list */
10133 tlist = lp; /* remainder of word list */
10134 }
10135
10136 /* vp == end of variable list */
10137 /* tlist == remainder of original word list without variable assignments */
10138 if (!tlist)
10139 /* All the words in tlist were assignment statements */
10140 return ((WORD_LIST *)NULL);
10141
10142 /* ASSERT(tlist != NULL); */
10143 /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */
10144
10145 /* If the -k option is in effect, we need to go through the remaining
10146 words, separate out the assignment words, and place them on
10147 SUBST_ASSIGN_VARLIST. */
10148 if (place_keywords_in_env)
10149 {
10150 WORD_LIST *tp; /* tp == running pointer into tlist */
10151
10152 tp = tlist;
10153 lp = tlist->next;
10154
10155 /* Loop Invariant: tp->next == lp */
10156 /* Loop postcondition: tlist == word list without assignment statements */
10157 while (lp)
10158 {
10159 if (lp->word->flags & W_ASSIGNMENT)
10160 {
10161 /* Found an assignment statement, add this word to end of
10162 subst_assign_varlist (vp). */
10163 if (!subst_assign_varlist)
10164 subst_assign_varlist = vp = lp;
10165 else
10166 {
10167 vp->next = lp;
10168 vp = lp;
10169 }
10170
10171 /* Remove the word pointed to by LP from TLIST. */
10172 tp->next = lp->next;
10173 /* ASSERT(vp == lp); */
10174 lp->next = (WORD_LIST *)NULL;
10175 lp = tp->next;
10176 }
10177 else
10178 {
10179 tp = lp;
10180 lp = lp->next;
10181 }
10182 }
10183 }
10184 return (tlist);
10185 }
10186
10187 #define WEXP_VARASSIGN 0x001
10188 #define WEXP_BRACEEXP 0x002
10189 #define WEXP_TILDEEXP 0x004
10190 #define WEXP_PARAMEXP 0x008
10191 #define WEXP_PATHEXP 0x010
10192
10193 /* All of the expansions, including variable assignments at the start of
10194 the list. */
10195 #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10196
10197 /* All of the expansions except variable assignments at the start of
10198 the list. */
10199 #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP)
10200
10201 /* All of the `shell expansions': brace expansion, tilde expansion, parameter
10202 expansion, command substitution, arithmetic expansion, word splitting, and
10203 quote removal. */
10204 #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP)
10205
10206 /* Take the list of words in LIST and do the various substitutions. Return
10207 a new list of words which is the expanded list, and without things like
10208 variable assignments. */
10209
10210 WORD_LIST *
10211 expand_words (list)
10212 WORD_LIST *list;
10213 {
10214 return (expand_word_list_internal (list, WEXP_ALL));
10215 }
10216
10217 /* Same as expand_words (), but doesn't hack variable or environment
10218 variables. */
10219 WORD_LIST *
10220 expand_words_no_vars (list)
10221 WORD_LIST *list;
10222 {
10223 return (expand_word_list_internal (list, WEXP_NOVARS));
10224 }
10225
10226 WORD_LIST *
10227 expand_words_shellexp (list)
10228 WORD_LIST *list;
10229 {
10230 return (expand_word_list_internal (list, WEXP_SHELLEXP));
10231 }
10232
10233 static WORD_LIST *
10234 glob_expand_word_list (tlist, eflags)
10235 WORD_LIST *tlist;
10236 int eflags;
10237 {
10238 char **glob_array, *temp_string;
10239 register int glob_index;
10240 WORD_LIST *glob_list, *output_list, *disposables, *next;
10241 WORD_DESC *tword;
10242
10243 output_list = disposables = (WORD_LIST *)NULL;
10244 glob_array = (char **)NULL;
10245 while (tlist)
10246 {
10247 /* For each word, either globbing is attempted or the word is
10248 added to orig_list. If globbing succeeds, the results are
10249 added to orig_list and the word (tlist) is added to the list
10250 of disposable words. If globbing fails and failed glob
10251 expansions are left unchanged (the shell default), the
10252 original word is added to orig_list. If globbing fails and
10253 failed glob expansions are removed, the original word is
10254 added to the list of disposable words. orig_list ends up
10255 in reverse order and requires a call to REVERSE_LIST to
10256 be set right. After all words are examined, the disposable
10257 words are freed. */
10258 next = tlist->next;
10259
10260 /* If the word isn't an assignment and contains an unquoted
10261 pattern matching character, then glob it. */
10262 if ((tlist->word->flags & W_NOGLOB) == 0 &&
10263 unquoted_glob_pattern_p (tlist->word->word))
10264 {
10265 glob_array = shell_glob_filename (tlist->word->word);
10266
10267 /* Handle error cases.
10268 I don't think we should report errors like "No such file
10269 or directory". However, I would like to report errors
10270 like "Read failed". */
10271
10272 if (glob_array == 0 || GLOB_FAILED (glob_array))
10273 {
10274 glob_array = (char **)xmalloc (sizeof (char *));
10275 glob_array[0] = (char *)NULL;
10276 }
10277
10278 /* Dequote the current word in case we have to use it. */
10279 if (glob_array[0] == NULL)
10280 {
10281 temp_string = dequote_string (tlist->word->word);
10282 free (tlist->word->word);
10283 tlist->word->word = temp_string;
10284 }
10285
10286 /* Make the array into a word list. */
10287 glob_list = (WORD_LIST *)NULL;
10288 for (glob_index = 0; glob_array[glob_index]; glob_index++)
10289 {
10290 tword = make_bare_word (glob_array[glob_index]);
10291 glob_list = make_word_list (tword, glob_list);
10292 }
10293
10294 if (glob_list)
10295 {
10296 output_list = (WORD_LIST *)list_append (glob_list, output_list);
10297 PREPEND_LIST (tlist, disposables);
10298 }
10299 else if (fail_glob_expansion != 0)
10300 {
10301 last_command_exit_value = EXECUTION_FAILURE;
10302 report_error (_("no match: %s"), tlist->word->word);
10303 exp_jump_to_top_level (DISCARD);
10304 }
10305 else if (allow_null_glob_expansion == 0)
10306 {
10307 /* Failed glob expressions are left unchanged. */
10308 PREPEND_LIST (tlist, output_list);
10309 }
10310 else
10311 {
10312 /* Failed glob expressions are removed. */
10313 PREPEND_LIST (tlist, disposables);
10314 }
10315 }
10316 else
10317 {
10318 /* Dequote the string. */
10319 temp_string = dequote_string (tlist->word->word);
10320 free (tlist->word->word);
10321 tlist->word->word = temp_string;
10322 PREPEND_LIST (tlist, output_list);
10323 }
10324
10325 strvec_dispose (glob_array);
10326 glob_array = (char **)NULL;
10327
10328 tlist = next;
10329 }
10330
10331 if (disposables)
10332 dispose_words (disposables);
10333
10334 if (output_list)
10335 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10336
10337 return (output_list);
10338 }
10339
10340 #if defined (BRACE_EXPANSION)
10341 static WORD_LIST *
10342 brace_expand_word_list (tlist, eflags)
10343 WORD_LIST *tlist;
10344 int eflags;
10345 {
10346 register char **expansions;
10347 char *temp_string;
10348 WORD_LIST *disposables, *output_list, *next;
10349 WORD_DESC *w;
10350 int eindex;
10351
10352 for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next)
10353 {
10354 next = tlist->next;
10355
10356 if (tlist->word->flags & W_NOBRACE)
10357 {
10358 /*itrace("brace_expand_word_list: %s: W_NOBRACE", tlist->word->word);*/
10359 PREPEND_LIST (tlist, output_list);
10360 continue;
10361 }
10362
10363 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10364 {
10365 /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/
10366 PREPEND_LIST (tlist, output_list);
10367 continue;
10368 }
10369
10370 /* Only do brace expansion if the word has a brace character. If
10371 not, just add the word list element to BRACES and continue. In
10372 the common case, at least when running shell scripts, this will
10373 degenerate to a bunch of calls to `mbschr', and then what is
10374 basically a reversal of TLIST into BRACES, which is corrected
10375 by a call to REVERSE_LIST () on BRACES when the end of TLIST
10376 is reached. */
10377 if (mbschr (tlist->word->word, LBRACE))
10378 {
10379 expansions = brace_expand (tlist->word->word);
10380
10381 for (eindex = 0; temp_string = expansions[eindex]; eindex++)
10382 {
10383 w = alloc_word_desc ();
10384 w->word = temp_string;
10385
10386 /* If brace expansion didn't change the word, preserve
10387 the flags. We may want to preserve the flags
10388 unconditionally someday -- XXX */
10389 if (STREQ (temp_string, tlist->word->word))
10390 w->flags = tlist->word->flags;
10391 else
10392 w = make_word_flags (w, temp_string);
10393
10394 output_list = make_word_list (w, output_list);
10395 }
10396 free (expansions);
10397
10398 /* Add TLIST to the list of words to be freed after brace
10399 expansion has been performed. */
10400 PREPEND_LIST (tlist, disposables);
10401 }
10402 else
10403 PREPEND_LIST (tlist, output_list);
10404 }
10405
10406 if (disposables)
10407 dispose_words (disposables);
10408
10409 if (output_list)
10410 output_list = REVERSE_LIST (output_list, WORD_LIST *);
10411
10412 return (output_list);
10413 }
10414 #endif
10415
10416 #if defined (ARRAY_VARS)
10417 /* Take WORD, a compound associative array assignment, and internally run
10418 'declare -A w', where W is the variable name portion of WORD. */
10419 static int
10420 make_internal_declare (word, option, cmd)
10421 char *word;
10422 char *option;
10423 char *cmd;
10424 {
10425 int t, r;
10426 WORD_LIST *wl;
10427 WORD_DESC *w;
10428
10429 w = make_word (word);
10430
10431 t = assignment (w->word, 0);
10432 if (w->word[t] == '=')
10433 {
10434 w->word[t] = '\0';
10435 if (w->word[t - 1] == '+') /* cut off any append op */
10436 w->word[t - 1] = '\0';
10437 }
10438
10439 wl = make_word_list (w, (WORD_LIST *)NULL);
10440 wl = make_word_list (make_word (option), wl);
10441
10442 r = declare_builtin (wl);
10443
10444 dispose_words (wl);
10445 return r;
10446 }
10447 #endif
10448
10449 static WORD_LIST *
10450 shell_expand_word_list (tlist, eflags)
10451 WORD_LIST *tlist;
10452 int eflags;
10453 {
10454 WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list, *wcmd;
10455 int expanded_something, has_dollar_at;
10456 char *temp_string;
10457
10458 /* We do tilde expansion all the time. This is what 1003.2 says. */
10459 new_list = (WORD_LIST *)NULL;
10460 for (wcmd = tlist; wcmd; wcmd = wcmd->next)
10461 if (wcmd->word->flags & W_ASSNBLTIN)
10462 break;
10463
10464 for (orig_list = tlist; tlist; tlist = next)
10465 {
10466 temp_string = tlist->word->word;
10467
10468 next = tlist->next;
10469
10470 #if defined (ARRAY_VARS)
10471 /* If this is a compound array assignment to a builtin that accepts
10472 such assignments (e.g., `declare'), take the assignment and perform
10473 it separately, handling the semantics of declarations inside shell
10474 functions. This avoids the double-evaluation of such arguments,
10475 because `declare' does some evaluation of compound assignments on
10476 its own. */
10477 if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG))
10478 {
10479 int t;
10480 char opts[16], opti;
10481
10482 opti = 0;
10483 if (tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL|W_ASSIGNARRAY))
10484 opts[opti++] = '-';
10485
10486 if ((tlist->word->flags & (W_ASSIGNASSOC|W_ASSNGLOBAL)) == (W_ASSIGNASSOC|W_ASSNGLOBAL))
10487 {
10488 opts[opti++] = 'g';
10489 opts[opti++] = 'A';
10490 }
10491 else if (tlist->word->flags & W_ASSIGNASSOC)
10492 opts[opti++] = 'A';
10493 else if ((tlist->word->flags & (W_ASSIGNARRAY|W_ASSNGLOBAL)) == (W_ASSIGNARRAY|W_ASSNGLOBAL))
10494 {
10495 opts[opti++] = 'g';
10496 opts[opti++] = 'a';
10497 }
10498 else if (tlist->word->flags & W_ASSIGNARRAY)
10499 opts[opti++] = 'a';
10500 else if (tlist->word->flags & W_ASSNGLOBAL)
10501 opts[opti++] = 'g';
10502
10503 /* If we have special handling note the integer attribute and others
10504 that transform the value upon assignment. What we do is take all
10505 of the option arguments and scan through them looking for options
10506 that cause such transformations, and add them to the `opts' array. */
10507 /* if (opti > 0) */
10508 {
10509 char omap[128];
10510 int oind;
10511 WORD_LIST *l;
10512
10513 memset (omap, '\0', sizeof (omap));
10514 for (l = orig_list->next; l != tlist; l = l->next)
10515 {
10516 if (l->word->word[0] != '-')
10517 break; /* non-option argument */
10518 if (l->word->word[0] == '-' && l->word->word[1] == '-' && l->word->word[2] == 0)
10519 break; /* -- signals end of options */
10520 for (oind = 1; l->word->word[oind]; oind++)
10521 switch (l->word->word[oind])
10522 {
10523 case 'i':
10524 case 'l':
10525 case 'u':
10526 case 'c':
10527 omap[l->word->word[oind]] = 1;
10528 if (opti == 0)
10529 opts[opti++] = '-';
10530 break;
10531 default:
10532 break;
10533 }
10534 }
10535
10536 for (oind = 0; oind < sizeof (omap); oind++)
10537 if (omap[oind])
10538 opts[opti++] = oind;
10539 }
10540
10541 opts[opti] = '\0';
10542 if (opti > 0)
10543 {
10544 t = make_internal_declare (tlist->word->word, opts, wcmd ? wcmd->word->word : (char *)0);
10545 if (t != EXECUTION_SUCCESS)
10546 {
10547 last_command_exit_value = t;
10548 exp_jump_to_top_level (DISCARD);
10549 }
10550 }
10551
10552 t = do_word_assignment (tlist->word, 0);
10553 if (t == 0)
10554 {
10555 last_command_exit_value = EXECUTION_FAILURE;
10556 exp_jump_to_top_level (DISCARD);
10557 }
10558
10559 /* Now transform the word as ksh93 appears to do and go on */
10560 t = assignment (tlist->word->word, 0);
10561 tlist->word->word[t] = '\0';
10562 if (tlist->word->word[t - 1] == '+')
10563 tlist->word->word[t - 1] = '\0'; /* cut off append op */
10564 tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC|W_ASSIGNARRAY);
10565 }
10566 #endif
10567
10568 expanded_something = 0;
10569 expanded = expand_word_internal
10570 (tlist->word, 0, 0, &has_dollar_at, &expanded_something);
10571
10572 if (expanded == &expand_word_error || expanded == &expand_word_fatal)
10573 {
10574 /* By convention, each time this error is returned,
10575 tlist->word->word has already been freed. */
10576 tlist->word->word = (char *)NULL;
10577
10578 /* Dispose our copy of the original list. */
10579 dispose_words (orig_list);
10580 /* Dispose the new list we're building. */
10581 dispose_words (new_list);
10582
10583 last_command_exit_value = EXECUTION_FAILURE;
10584 if (expanded == &expand_word_error)
10585 exp_jump_to_top_level (DISCARD);
10586 else
10587 exp_jump_to_top_level (FORCE_EOF);
10588 }
10589
10590 /* Don't split words marked W_NOSPLIT. */
10591 if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0)
10592 {
10593 temp_list = word_list_split (expanded);
10594 dispose_words (expanded);
10595 }
10596 else
10597 {
10598 /* If no parameter expansion, command substitution, process
10599 substitution, or arithmetic substitution took place, then
10600 do not do word splitting. We still have to remove quoted
10601 null characters from the result. */
10602 word_list_remove_quoted_nulls (expanded);
10603 temp_list = expanded;
10604 }
10605
10606 expanded = REVERSE_LIST (temp_list, WORD_LIST *);
10607 new_list = (WORD_LIST *)list_append (expanded, new_list);
10608 }
10609
10610 if (orig_list)
10611 dispose_words (orig_list);
10612
10613 if (new_list)
10614 new_list = REVERSE_LIST (new_list, WORD_LIST *);
10615
10616 return (new_list);
10617 }
10618
10619 /* The workhorse for expand_words () and expand_words_no_vars ().
10620 First arg is LIST, a WORD_LIST of words.
10621 Second arg EFLAGS is a flags word controlling which expansions are
10622 performed.
10623
10624 This does all of the substitutions: brace expansion, tilde expansion,
10625 parameter expansion, command substitution, arithmetic expansion,
10626 process substitution, word splitting, and pathname expansion, according
10627 to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits
10628 set, or for which no expansion is done, do not undergo word splitting.
10629 Words with the W_NOGLOB bit set do not undergo pathname expansion; words
10630 with W_NOBRACE set do not undergo brace expansion (see
10631 brace_expand_word_list above). */
10632 static WORD_LIST *
10633 expand_word_list_internal (list, eflags)
10634 WORD_LIST *list;
10635 int eflags;
10636 {
10637 WORD_LIST *new_list, *temp_list;
10638 int tint;
10639 char *savecmd;
10640
10641 tempenv_assign_error = 0;
10642 if (list == 0)
10643 return ((WORD_LIST *)NULL);
10644
10645 garglist = new_list = copy_word_list (list);
10646 if (eflags & WEXP_VARASSIGN)
10647 {
10648 garglist = new_list = separate_out_assignments (new_list);
10649 if (new_list == 0)
10650 {
10651 if (subst_assign_varlist)
10652 {
10653 /* All the words were variable assignments, so they are placed
10654 into the shell's environment. */
10655 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
10656 {
10657 savecmd = this_command_name;
10658 this_command_name = (char *)NULL; /* no arithmetic errors */
10659 tint = do_word_assignment (temp_list->word, 0);
10660 this_command_name = savecmd;
10661 /* Variable assignment errors in non-interactive shells
10662 running in Posix.2 mode cause the shell to exit. */
10663 if (tint == 0)
10664 {
10665 last_command_exit_value = EXECUTION_FAILURE;
10666 if (interactive_shell == 0 && posixly_correct)
10667 exp_jump_to_top_level (FORCE_EOF);
10668 else
10669 exp_jump_to_top_level (DISCARD);
10670 }
10671 }
10672 dispose_words (subst_assign_varlist);
10673 subst_assign_varlist = (WORD_LIST *)NULL;
10674 }
10675 return ((WORD_LIST *)NULL);
10676 }
10677 }
10678
10679 /* Begin expanding the words that remain. The expansions take place on
10680 things that aren't really variable assignments. */
10681
10682 #if defined (BRACE_EXPANSION)
10683 /* Do brace expansion on this word if there are any brace characters
10684 in the string. */
10685 if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list)
10686 new_list = brace_expand_word_list (new_list, eflags);
10687 #endif /* BRACE_EXPANSION */
10688
10689 /* Perform the `normal' shell expansions: tilde expansion, parameter and
10690 variable substitution, command substitution, arithmetic expansion,
10691 and word splitting. */
10692 new_list = shell_expand_word_list (new_list, eflags);
10693
10694 /* Okay, we're almost done. Now let's just do some filename
10695 globbing. */
10696 if (new_list)
10697 {
10698 if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0)
10699 /* Glob expand the word list unless globbing has been disabled. */
10700 new_list = glob_expand_word_list (new_list, eflags);
10701 else
10702 /* Dequote the words, because we're not performing globbing. */
10703 new_list = dequote_list (new_list);
10704 }
10705
10706 if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist)
10707 {
10708 sh_wassign_func_t *assign_func;
10709 int is_special_builtin, is_builtin_or_func;
10710
10711 /* If the remainder of the words expand to nothing, Posix.2 requires
10712 that the variable and environment assignments affect the shell's
10713 environment. */
10714 assign_func = new_list ? assign_in_env : do_word_assignment;
10715 tempenv_assign_error = 0;
10716
10717 is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word)));
10718 /* Posix says that special builtins exit if a variable assignment error
10719 occurs in an assignment preceding it. */
10720 is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word));
10721
10722 for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next)
10723 {
10724 savecmd = this_command_name;
10725 this_command_name = (char *)NULL;
10726 assigning_in_environment = (assign_func == assign_in_env);
10727 tint = (*assign_func) (temp_list->word, is_builtin_or_func);
10728 assigning_in_environment = 0;
10729 this_command_name = savecmd;
10730 /* Variable assignment errors in non-interactive shells running
10731 in Posix.2 mode cause the shell to exit. */
10732 if (tint == 0)
10733 {
10734 if (assign_func == do_word_assignment)
10735 {
10736 last_command_exit_value = EXECUTION_FAILURE;
10737 if (interactive_shell == 0 && posixly_correct && is_special_builtin)
10738 exp_jump_to_top_level (FORCE_EOF);
10739 else
10740 exp_jump_to_top_level (DISCARD);
10741 }
10742 else
10743 tempenv_assign_error++;
10744 }
10745 }
10746
10747 dispose_words (subst_assign_varlist);
10748 subst_assign_varlist = (WORD_LIST *)NULL;
10749 }
10750
10751 return (new_list);
10752 }