]>
Commit | Line | Data |
---|---|---|
95732b49 JA |
1 | /* subst.c -- The part of the shell that does parameter, command, arithmetic, |
2 | and globbing substitutions. */ | |
726f6388 | 3 | |
bb70624e JA |
4 | /* ``Have a little faith, there's magic in the night. You ain't a |
5 | beauty, but, hey, you're alright.'' */ | |
6 | ||
f1be666c | 7 | /* Copyright (C) 1987-2007 Free Software Foundation, Inc. |
726f6388 JA |
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 it under | |
12 | the terms of the GNU General Public License as published by the Free | |
13 | Software Foundation; either version 2, or (at your option) any later | |
14 | version. | |
15 | ||
16 | Bash is distributed in the hope that it will be useful, but WITHOUT ANY | |
17 | WARRANTY; without even the implied warranty of MERCHANTABILITY or | |
18 | FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License | |
19 | for more details. | |
20 | ||
21 | You should have received a copy of the GNU General Public License along | |
22 | with Bash; see the file COPYING. If not, write to the Free Software | |
bb70624e | 23 | Foundation, 59 Temple Place, Suite 330, Boston, MA 02111 USA. */ |
726f6388 | 24 | |
ccc6cda3 JA |
25 | #include "config.h" |
26 | ||
726f6388 JA |
27 | #include "bashtypes.h" |
28 | #include <stdio.h> | |
f73dda09 | 29 | #include "chartypes.h" |
726f6388 JA |
30 | #include <pwd.h> |
31 | #include <signal.h> | |
32 | #include <errno.h> | |
ccc6cda3 JA |
33 | |
34 | #if defined (HAVE_UNISTD_H) | |
35 | # include <unistd.h> | |
36 | #endif | |
726f6388 JA |
37 | |
38 | #include "bashansi.h" | |
39 | #include "posixstat.h" | |
b80f6443 | 40 | #include "bashintl.h" |
726f6388 JA |
41 | |
42 | #include "shell.h" | |
43 | #include "flags.h" | |
44 | #include "jobs.h" | |
45 | #include "execute_cmd.h" | |
46 | #include "filecntl.h" | |
ccc6cda3 JA |
47 | #include "trap.h" |
48 | #include "pathexp.h" | |
49 | #include "mailcheck.h" | |
50 | ||
7117c2d2 JA |
51 | #include "shmbutil.h" |
52 | ||
ccc6cda3 JA |
53 | #include "builtins/getopt.h" |
54 | #include "builtins/common.h" | |
726f6388 | 55 | |
cce855bc | 56 | #include <tilde/tilde.h> |
f73dda09 | 57 | #include <glob/strmatch.h> |
ccc6cda3 JA |
58 | |
59 | #if !defined (errno) | |
60 | extern int errno; | |
61 | #endif /* !errno */ | |
726f6388 JA |
62 | |
63 | /* The size that strings change by. */ | |
d166f048 | 64 | #define DEFAULT_INITIAL_ARRAY_SIZE 112 |
ccc6cda3 JA |
65 | #define DEFAULT_ARRAY_SIZE 128 |
66 | ||
67 | /* Variable types. */ | |
68 | #define VT_VARIABLE 0 | |
69 | #define VT_POSPARMS 1 | |
70 | #define VT_ARRAYVAR 2 | |
d166f048 | 71 | #define VT_ARRAYMEMBER 3 |
726f6388 | 72 | |
b80f6443 JA |
73 | #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */ |
74 | ||
ccc6cda3 JA |
75 | /* Flags for quoted_strchr */ |
76 | #define ST_BACKSL 0x01 | |
77 | #define ST_CTLESC 0x02 | |
7117c2d2 JA |
78 | #define ST_SQUOTE 0x04 /* unused yet */ |
79 | #define ST_DQUOTE 0x08 /* unused yet */ | |
80 | ||
81 | /* Flags for the string extraction functions. */ | |
82 | #define EX_NOALLOC 0x01 /* just skip; don't return substring */ | |
83 | #define EX_VARNAME 0x02 /* variable name; for string_extract () */ | |
95732b49 | 84 | #define EX_REQMATCH 0x04 /* closing/matching delimiter required */ |
0628567a | 85 | #define EX_COMMAND 0x08 /* extracting a shell script/command */ |
ccc6cda3 | 86 | |
b80f6443 JA |
87 | /* Flags for the `pflags' argument to param_expand() */ |
88 | #define PF_NOCOMSUB 0x01 /* Do not perform command substitution */ | |
89 | ||
cce855bc JA |
90 | /* These defs make it easier to use the editor. */ |
91 | #define LBRACE '{' | |
92 | #define RBRACE '}' | |
93 | #define LPAREN '(' | |
94 | #define RPAREN ')' | |
726f6388 | 95 | |
28ef6c31 JA |
96 | /* Evaluates to 1 if C is one of the shell's special parameters whose length |
97 | can be taken, but is also one of the special expansion characters. */ | |
98 | #define VALID_SPECIAL_LENGTH_PARAM(c) \ | |
99 | ((c) == '-' || (c) == '?' || (c) == '#') | |
100 | ||
101 | /* Evaluates to 1 if C is one of the shell's special parameters for which an | |
102 | indirect variable reference may be made. */ | |
103 | #define VALID_INDIR_PARAM(c) \ | |
104 | ((c) == '#' || (c) == '?' || (c) == '@' || (c) == '*') | |
105 | ||
106 | /* Evaluates to 1 if C is one of the OP characters that follows the parameter | |
107 | in ${parameter[:]OPword}. */ | |
7117c2d2 | 108 | #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP) |
28ef6c31 | 109 | |
bb70624e JA |
110 | /* Evaluates to 1 if this is one of the shell's special variables. */ |
111 | #define SPECIAL_VAR(name, wi) \ | |
f73dda09 JA |
112 | ((DIGIT (*name) && all_digits (name)) || \ |
113 | (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \ | |
28ef6c31 | 114 | (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1]))) |
bb70624e | 115 | |
f73dda09 JA |
116 | /* An expansion function that takes a string and a quoted flag and returns |
117 | a WORD_LIST *. Used as the type of the third argument to | |
118 | expand_string_if_necessary(). */ | |
119 | typedef WORD_LIST *EXPFUNC __P((char *, int)); | |
120 | ||
726f6388 JA |
121 | /* Process ID of the last command executed within command substitution. */ |
122 | pid_t last_command_subst_pid = NO_PID; | |
b72432fd | 123 | pid_t current_command_subst_pid = NO_PID; |
726f6388 | 124 | |
7117c2d2 JA |
125 | /* Variables used to keep track of the characters in IFS. */ |
126 | SHELL_VAR *ifs_var; | |
127 | char *ifs_value; | |
128 | unsigned char ifs_cmap[UCHAR_MAX + 1]; | |
95732b49 JA |
129 | |
130 | #if defined (HANDLE_MULTIBYTE) | |
131 | unsigned char ifs_firstc[MB_LEN_MAX]; | |
132 | size_t ifs_firstc_len; | |
133 | #else | |
7117c2d2 | 134 | unsigned char ifs_firstc; |
95732b49 | 135 | #endif |
7117c2d2 | 136 | |
726f6388 | 137 | /* Extern functions and variables from different files. */ |
b80f6443 | 138 | extern int last_command_exit_value, last_command_exit_signal; |
7117c2d2 | 139 | extern int subshell_environment; |
f1be666c | 140 | extern int subshell_level, parse_and_execute_level; |
7117c2d2 | 141 | extern int eof_encountered; |
bb70624e | 142 | extern int return_catch_flag, return_catch_value; |
f73dda09 | 143 | extern pid_t dollar_dollar_pid; |
726f6388 | 144 | extern int posixly_correct; |
726f6388 | 145 | extern char *this_command_name; |
ccc6cda3 | 146 | extern struct fd_bitmap *current_fds_to_close; |
cce855bc | 147 | extern int wordexp_only; |
b80f6443 JA |
148 | extern int expanding_redir; |
149 | extern int tempenv_assign_error; | |
726f6388 | 150 | |
0628567a JA |
151 | #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE) |
152 | extern wchar_t *wcsdup __P((const wchar_t *)); | |
153 | #endif | |
154 | ||
ccc6cda3 JA |
155 | /* Non-zero means to allow unmatched globbed filenames to expand to |
156 | a null file. */ | |
157 | int allow_null_glob_expansion; | |
158 | ||
b80f6443 JA |
159 | /* Non-zero means to throw an error when globbing fails to match anything. */ |
160 | int fail_glob_expansion; | |
161 | ||
f73dda09 | 162 | #if 0 |
ccc6cda3 JA |
163 | /* Variables to keep track of which words in an expanded word list (the |
164 | output of expand_word_list_internal) are the result of globbing | |
f73dda09 JA |
165 | expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c. |
166 | (CURRENTLY UNUSED). */ | |
ccc6cda3 JA |
167 | char *glob_argv_flags; |
168 | static int glob_argv_flags_size; | |
f73dda09 | 169 | #endif |
726f6388 JA |
170 | |
171 | static WORD_LIST expand_word_error, expand_word_fatal; | |
95732b49 | 172 | static WORD_DESC expand_wdesc_error, expand_wdesc_fatal; |
726f6388 | 173 | static char expand_param_error, expand_param_fatal; |
95732b49 | 174 | static char extract_string_error, extract_string_fatal; |
726f6388 | 175 | |
28ef6c31 JA |
176 | /* Tell the expansion functions to not longjmp back to top_level on fatal |
177 | errors. Enabled when doing completion and prompt string expansion. */ | |
178 | static int no_longjmp_on_fatal_error = 0; | |
179 | ||
180 | /* Set by expand_word_unsplit; used to inhibit splitting and re-joining | |
181 | $* on $IFS, primarily when doing assignment statements. */ | |
182 | static int expand_no_split_dollar_star = 0; | |
bb70624e JA |
183 | |
184 | /* Used to hold a list of variable assignments preceding a command. Global | |
185 | so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a | |
186 | SIGCHLD trap. */ | |
187 | WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL; | |
28ef6c31 | 188 | |
bb70624e JA |
189 | /* A WORD_LIST of words to be expanded by expand_word_list_internal, |
190 | without any leading variable assignments. */ | |
191 | static WORD_LIST *garglist = (WORD_LIST *)NULL; | |
b72432fd | 192 | |
f73dda09 | 193 | static char *quoted_substring __P((char *, int, int)); |
7117c2d2 JA |
194 | static int quoted_strlen __P((char *)); |
195 | static char *quoted_strchr __P((char *, int, int)); | |
f73dda09 JA |
196 | |
197 | static char *expand_string_if_necessary __P((char *, int, EXPFUNC *)); | |
198 | static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *)); | |
199 | static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *)); | |
200 | static WORD_LIST *expand_string_internal __P((char *, int)); | |
201 | static WORD_LIST *expand_string_leave_quoted __P((char *, int)); | |
202 | static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *)); | |
203 | ||
f73dda09 | 204 | static WORD_LIST *list_quote_escapes __P((WORD_LIST *)); |
7117c2d2 | 205 | static char *dequote_escapes __P((char *)); |
f73dda09 JA |
206 | static char *make_quoted_char __P((int)); |
207 | static WORD_LIST *quote_list __P((WORD_LIST *)); | |
7117c2d2 JA |
208 | static char *remove_quoted_escapes __P((char *)); |
209 | static char *remove_quoted_nulls __P((char *)); | |
f73dda09 JA |
210 | |
211 | static int unquoted_substring __P((char *, char *)); | |
212 | static int unquoted_member __P((int, char *)); | |
213 | ||
95732b49 JA |
214 | #if defined (ARRAY_VARS) |
215 | static SHELL_VAR *do_compound_assignment __P((char *, char *, int)); | |
216 | #endif | |
217 | static int do_assignment_internal __P((const WORD_DESC *, int)); | |
f73dda09 | 218 | |
95732b49 | 219 | static char *string_extract_verbatim __P((char *, size_t, int *, char *)); |
f73dda09 JA |
220 | static char *string_extract __P((char *, int *, char *, int)); |
221 | static char *string_extract_double_quoted __P((char *, int *, int)); | |
7117c2d2 | 222 | static inline char *string_extract_single_quoted __P((char *, int *)); |
0628567a | 223 | static inline int skip_single_quoted __P((const char *, size_t, int)); |
7117c2d2 JA |
224 | static int skip_double_quoted __P((char *, size_t, int)); |
225 | static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int)); | |
226 | static char *extract_dollar_brace_string __P((char *, int *, int, int)); | |
f73dda09 | 227 | |
f73dda09 JA |
228 | static char *pos_params __P((char *, int, int, int)); |
229 | ||
b80f6443 JA |
230 | static unsigned char *mb_getcharlens __P((char *, int)); |
231 | ||
232 | static char *remove_upattern __P((char *, char *, int)); | |
0628567a | 233 | #if defined (HANDLE_MULTIBYTE) |
b80f6443 JA |
234 | static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int)); |
235 | #endif | |
f73dda09 | 236 | static char *remove_pattern __P((char *, char *, int)); |
b80f6443 | 237 | |
f73dda09 | 238 | static int match_pattern_char __P((char *, char *)); |
b80f6443 JA |
239 | static int match_upattern __P((char *, char *, int, char **, char **)); |
240 | #if defined (HANDLE_MULTIBYTE) | |
241 | static int match_pattern_wchar __P((wchar_t *, wchar_t *)); | |
242 | static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **)); | |
243 | #endif | |
f73dda09 JA |
244 | static int match_pattern __P((char *, char *, int, char **, char **)); |
245 | static int getpatspec __P((int, char *)); | |
246 | static char *getpattern __P((char *, int, int)); | |
7117c2d2 | 247 | static char *variable_remove_pattern __P((char *, char *, int, int)); |
f73dda09 | 248 | static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int)); |
7117c2d2 | 249 | static char *parameter_list_remove_pattern __P((int, char *, int, int)); |
f73dda09 | 250 | #ifdef ARRAY_VARS |
7117c2d2 | 251 | static char *array_remove_pattern __P((ARRAY *, char *, int, char *, int)); |
f73dda09 | 252 | #endif |
7117c2d2 | 253 | static char *parameter_brace_remove_pattern __P((char *, char *, char *, int, int)); |
f73dda09 JA |
254 | |
255 | static char *process_substitute __P((char *, int)); | |
256 | ||
257 | static char *read_comsub __P((int, int)); | |
258 | ||
259 | #ifdef ARRAY_VARS | |
260 | static arrayind_t array_length_reference __P((char *)); | |
261 | #endif | |
262 | ||
263 | static int valid_brace_expansion_word __P((char *, int)); | |
b80f6443 | 264 | static int chk_atstar __P((char *, int, int *, int *)); |
0628567a | 265 | static int chk_arithsub __P((const char *, int)); |
b80f6443 | 266 | |
95732b49 JA |
267 | static WORD_DESC *parameter_brace_expand_word __P((char *, int, int)); |
268 | static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *)); | |
269 | static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *)); | |
f73dda09 JA |
270 | static void parameter_brace_expand_error __P((char *, char *)); |
271 | ||
272 | static int valid_length_expression __P((char *)); | |
7117c2d2 | 273 | static intmax_t parameter_brace_expand_length __P((char *)); |
f73dda09 JA |
274 | |
275 | static char *skiparith __P((char *, int)); | |
7117c2d2 | 276 | static int verify_substring_values __P((char *, char *, int, intmax_t *, intmax_t *)); |
b80f6443 JA |
277 | static int get_var_and_type __P((char *, char *, int, SHELL_VAR **, char **)); |
278 | static char *mb_substring __P((char *, int, int)); | |
f73dda09 JA |
279 | static char *parameter_brace_substring __P((char *, char *, char *, int)); |
280 | ||
281 | static char *pos_params_pat_subst __P((char *, char *, char *, int)); | |
282 | ||
283 | static char *parameter_brace_patsub __P((char *, char *, char *, int)); | |
284 | ||
95732b49 JA |
285 | static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int *, int *)); |
286 | static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int)); | |
f73dda09 JA |
287 | |
288 | static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *)); | |
289 | ||
f73dda09 JA |
290 | static WORD_LIST *word_list_split __P((WORD_LIST *)); |
291 | ||
b80f6443 JA |
292 | static void exp_jump_to_top_level __P((int)); |
293 | ||
f73dda09 JA |
294 | static WORD_LIST *separate_out_assignments __P((WORD_LIST *)); |
295 | static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int)); | |
296 | #ifdef BRACE_EXPANSION | |
297 | static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int)); | |
298 | #endif | |
299 | static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int)); | |
300 | static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int)); | |
726f6388 JA |
301 | |
302 | /* **************************************************************** */ | |
303 | /* */ | |
304 | /* Utility Functions */ | |
305 | /* */ | |
306 | /* **************************************************************** */ | |
307 | ||
7117c2d2 | 308 | #ifdef INCLUDE_UNUSED |
ccc6cda3 JA |
309 | static char * |
310 | quoted_substring (string, start, end) | |
311 | char *string; | |
312 | int start, end; | |
313 | { | |
314 | register int len, l; | |
315 | register char *result, *s, *r; | |
316 | ||
317 | len = end - start; | |
318 | ||
319 | /* Move to string[start], skipping quoted characters. */ | |
320 | for (s = string, l = 0; *s && l < start; ) | |
321 | { | |
322 | if (*s == CTLESC) | |
323 | { | |
28ef6c31 JA |
324 | s++; |
325 | continue; | |
ccc6cda3 JA |
326 | } |
327 | l++; | |
328 | if (*s == 0) | |
28ef6c31 | 329 | break; |
ccc6cda3 JA |
330 | } |
331 | ||
f73dda09 | 332 | r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */ |
ccc6cda3 JA |
333 | |
334 | /* Copy LEN characters, including quote characters. */ | |
335 | s = string + l; | |
336 | for (l = 0; l < len; s++) | |
337 | { | |
338 | if (*s == CTLESC) | |
28ef6c31 | 339 | *r++ = *s++; |
ccc6cda3 JA |
340 | *r++ = *s; |
341 | l++; | |
342 | if (*s == 0) | |
28ef6c31 | 343 | break; |
ccc6cda3 JA |
344 | } |
345 | *r = '\0'; | |
346 | return result; | |
347 | } | |
7117c2d2 JA |
348 | #endif |
349 | ||
350 | #ifdef INCLUDE_UNUSED | |
351 | /* Return the length of S, skipping over quoted characters */ | |
352 | static int | |
353 | quoted_strlen (s) | |
354 | char *s; | |
355 | { | |
356 | register char *p; | |
357 | int i; | |
358 | ||
359 | i = 0; | |
360 | for (p = s; *p; p++) | |
361 | { | |
362 | if (*p == CTLESC) | |
363 | { | |
364 | p++; | |
365 | if (*p == 0) | |
366 | return (i + 1); | |
367 | } | |
368 | i++; | |
369 | } | |
370 | ||
371 | return i; | |
372 | } | |
373 | #endif | |
ccc6cda3 JA |
374 | |
375 | /* Find the first occurrence of character C in string S, obeying shell | |
376 | quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped | |
377 | characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters | |
378 | escaped with CTLESC are skipped. */ | |
7117c2d2 | 379 | static char * |
ccc6cda3 JA |
380 | quoted_strchr (s, c, flags) |
381 | char *s; | |
382 | int c, flags; | |
383 | { | |
384 | register char *p; | |
385 | ||
386 | for (p = s; *p; p++) | |
387 | { | |
388 | if (((flags & ST_BACKSL) && *p == '\\') | |
389 | || ((flags & ST_CTLESC) && *p == CTLESC)) | |
390 | { | |
391 | p++; | |
392 | if (*p == '\0') | |
393 | return ((char *)NULL); | |
394 | continue; | |
395 | } | |
396 | else if (*p == c) | |
397 | return p; | |
398 | } | |
399 | return ((char *)NULL); | |
400 | } | |
401 | ||
cce855bc | 402 | /* Return 1 if CHARACTER appears in an unquoted portion of |
7117c2d2 | 403 | STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */ |
cce855bc JA |
404 | static int |
405 | unquoted_member (character, string) | |
406 | int character; | |
726f6388 JA |
407 | char *string; |
408 | { | |
7117c2d2 | 409 | size_t slen; |
cce855bc | 410 | int sindex, c; |
7117c2d2 | 411 | DECLARE_MBSTATE; |
726f6388 | 412 | |
7117c2d2 JA |
413 | slen = strlen (string); |
414 | sindex = 0; | |
415 | while (c = string[sindex]) | |
726f6388 | 416 | { |
cce855bc JA |
417 | if (c == character) |
418 | return (1); | |
419 | ||
420 | switch (c) | |
ccc6cda3 | 421 | { |
cce855bc | 422 | default: |
7117c2d2 | 423 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
424 | break; |
425 | ||
426 | case '\\': | |
427 | sindex++; | |
428 | if (string[sindex]) | |
7117c2d2 | 429 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
430 | break; |
431 | ||
432 | case '\'': | |
7117c2d2 | 433 | sindex = skip_single_quoted (string, slen, ++sindex); |
cce855bc JA |
434 | break; |
435 | ||
436 | case '"': | |
7117c2d2 | 437 | sindex = skip_double_quoted (string, slen, ++sindex); |
cce855bc | 438 | break; |
ccc6cda3 | 439 | } |
726f6388 | 440 | } |
cce855bc | 441 | return (0); |
726f6388 JA |
442 | } |
443 | ||
cce855bc JA |
444 | /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */ |
445 | static int | |
446 | unquoted_substring (substr, string) | |
447 | char *substr, *string; | |
726f6388 | 448 | { |
7117c2d2 | 449 | size_t slen; |
cce855bc | 450 | int sindex, c, sublen; |
7117c2d2 | 451 | DECLARE_MBSTATE; |
726f6388 | 452 | |
cce855bc JA |
453 | if (substr == 0 || *substr == '\0') |
454 | return (0); | |
455 | ||
7117c2d2 | 456 | slen = strlen (string); |
cce855bc JA |
457 | sublen = strlen (substr); |
458 | for (sindex = 0; c = string[sindex]; ) | |
726f6388 | 459 | { |
cce855bc JA |
460 | if (STREQN (string + sindex, substr, sublen)) |
461 | return (1); | |
462 | ||
463 | switch (c) | |
464 | { | |
465 | case '\\': | |
466 | sindex++; | |
467 | ||
468 | if (string[sindex]) | |
7117c2d2 | 469 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
470 | break; |
471 | ||
472 | case '\'': | |
7117c2d2 | 473 | sindex = skip_single_quoted (string, slen, ++sindex); |
cce855bc JA |
474 | break; |
475 | ||
476 | case '"': | |
7117c2d2 | 477 | sindex = skip_double_quoted (string, slen, ++sindex); |
cce855bc JA |
478 | break; |
479 | ||
480 | default: | |
7117c2d2 | 481 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
482 | break; |
483 | } | |
726f6388 | 484 | } |
cce855bc | 485 | return (0); |
ccc6cda3 | 486 | } |
726f6388 | 487 | |
cce855bc JA |
488 | /* Most of the substitutions must be done in parallel. In order |
489 | to avoid using tons of unclear goto's, I have some functions | |
490 | for manipulating malloc'ed strings. They all take INDX, a | |
491 | pointer to an integer which is the offset into the string | |
492 | where manipulation is taking place. They also take SIZE, a | |
493 | pointer to an integer which is the current length of the | |
494 | character array for this string. */ | |
726f6388 | 495 | |
cce855bc JA |
496 | /* Append SOURCE to TARGET at INDEX. SIZE is the current amount |
497 | of space allocated to TARGET. SOURCE can be NULL, in which | |
498 | case nothing happens. Gets rid of SOURCE by freeing it. | |
499 | Returns TARGET in case the location has changed. */ | |
7117c2d2 | 500 | INLINE char * |
cce855bc JA |
501 | sub_append_string (source, target, indx, size) |
502 | char *source, *target; | |
503 | int *indx, *size; | |
504 | { | |
505 | if (source) | |
726f6388 | 506 | { |
cce855bc JA |
507 | int srclen, n; |
508 | ||
509 | srclen = STRLEN (source); | |
510 | if (srclen >= (int)(*size - *indx)) | |
726f6388 | 511 | { |
cce855bc JA |
512 | n = srclen + *indx; |
513 | n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE); | |
f73dda09 | 514 | target = (char *)xrealloc (target, (*size = n)); |
726f6388 | 515 | } |
cce855bc JA |
516 | |
517 | FASTCOPY (source, target + *indx, srclen); | |
518 | *indx += srclen; | |
519 | target[*indx] = '\0'; | |
520 | ||
521 | free (source); | |
726f6388 | 522 | } |
cce855bc JA |
523 | return (target); |
524 | } | |
525 | ||
526 | #if 0 | |
527 | /* UNUSED */ | |
528 | /* Append the textual representation of NUMBER to TARGET. | |
529 | INDX and SIZE are as in SUB_APPEND_STRING. */ | |
530 | char * | |
531 | sub_append_number (number, target, indx, size) | |
7117c2d2 | 532 | intmax_t number; |
f73dda09 | 533 | int *indx, *size; |
cce855bc JA |
534 | char *target; |
535 | { | |
536 | char *temp; | |
537 | ||
538 | temp = itos (number); | |
539 | return (sub_append_string (temp, target, indx, size)); | |
726f6388 | 540 | } |
d166f048 | 541 | #endif |
726f6388 JA |
542 | |
543 | /* Extract a substring from STRING, starting at SINDEX and ending with | |
544 | one of the characters in CHARLIST. Don't make the ending character | |
545 | part of the string. Leave SINDEX pointing at the ending character. | |
7117c2d2 JA |
546 | Understand about backslashes in the string. If (flags & EX_VARNAME) |
547 | is non-zero, and array variables have been compiled into the shell, | |
548 | everything between a `[' and a corresponding `]' is skipped over. | |
549 | If (flags & EX_NOALLOC) is non-zero, don't return the substring, just | |
95732b49 JA |
550 | update SINDEX. If (flags & EX_REQMATCH) is non-zero, the string must |
551 | contain a closing character from CHARLIST. */ | |
726f6388 | 552 | static char * |
7117c2d2 | 553 | string_extract (string, sindex, charlist, flags) |
f73dda09 JA |
554 | char *string; |
555 | int *sindex; | |
556 | char *charlist; | |
7117c2d2 | 557 | int flags; |
726f6388 | 558 | { |
ccc6cda3 | 559 | register int c, i; |
95732b49 | 560 | int found; |
7117c2d2 | 561 | size_t slen; |
726f6388 | 562 | char *temp; |
7117c2d2 | 563 | DECLARE_MBSTATE; |
726f6388 | 564 | |
95732b49 | 565 | slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0; |
7117c2d2 | 566 | i = *sindex; |
95732b49 | 567 | found = 0; |
7117c2d2 | 568 | while (c = string[i]) |
726f6388 JA |
569 | { |
570 | if (c == '\\') | |
7117c2d2 JA |
571 | { |
572 | if (string[i + 1]) | |
573 | i++; | |
574 | else | |
575 | break; | |
576 | } | |
ccc6cda3 | 577 | #if defined (ARRAY_VARS) |
7117c2d2 | 578 | else if ((flags & EX_VARNAME) && c == '[') |
ccc6cda3 JA |
579 | { |
580 | int ni; | |
581 | /* If this is an array subscript, skip over it and continue. */ | |
582 | ni = skipsubscript (string, i); | |
583 | if (string[ni] == ']') | |
584 | i = ni; | |
585 | } | |
586 | #endif | |
587 | else if (MEMBER (c, charlist)) | |
95732b49 JA |
588 | { |
589 | found = 1; | |
726f6388 | 590 | break; |
95732b49 | 591 | } |
7117c2d2 JA |
592 | |
593 | ADVANCE_CHAR (string, slen, i); | |
726f6388 | 594 | } |
bb70624e | 595 | |
95732b49 JA |
596 | /* If we had to have a matching delimiter and didn't find one, return an |
597 | error and let the caller deal with it. */ | |
598 | if ((flags & EX_REQMATCH) && found == 0) | |
599 | { | |
600 | *sindex = i; | |
601 | return (&extract_string_error); | |
602 | } | |
603 | ||
7117c2d2 | 604 | temp = (flags & EX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i); |
726f6388 | 605 | *sindex = i; |
95732b49 | 606 | |
726f6388 JA |
607 | return (temp); |
608 | } | |
609 | ||
ccc6cda3 JA |
610 | /* Extract the contents of STRING as if it is enclosed in double quotes. |
611 | SINDEX, when passed in, is the offset of the character immediately | |
612 | following the opening double quote; on exit, SINDEX is left pointing after | |
613 | the closing double quote. If STRIPDQ is non-zero, unquoted double | |
614 | quotes are stripped and the string is terminated by a null byte. | |
615 | Backslashes between the embedded double quotes are processed. If STRIPDQ | |
616 | is zero, an unquoted `"' terminates the string. */ | |
7117c2d2 | 617 | static char * |
ccc6cda3 | 618 | string_extract_double_quoted (string, sindex, stripdq) |
726f6388 | 619 | char *string; |
ccc6cda3 | 620 | int *sindex, stripdq; |
726f6388 | 621 | { |
7117c2d2 JA |
622 | size_t slen; |
623 | char *send; | |
f73dda09 JA |
624 | int j, i, t; |
625 | unsigned char c; | |
ccc6cda3 JA |
626 | char *temp, *ret; /* The new string we return. */ |
627 | int pass_next, backquote, si; /* State variables for the machine. */ | |
628 | int dquote; | |
7117c2d2 JA |
629 | DECLARE_MBSTATE; |
630 | ||
631 | slen = strlen (string + *sindex) + *sindex; | |
632 | send = string + slen; | |
726f6388 | 633 | |
ccc6cda3 | 634 | pass_next = backquote = dquote = 0; |
7117c2d2 | 635 | temp = (char *)xmalloc (1 + slen - *sindex); |
726f6388 | 636 | |
7117c2d2 JA |
637 | j = 0; |
638 | i = *sindex; | |
639 | while (c = string[i]) | |
726f6388 | 640 | { |
ccc6cda3 JA |
641 | /* Process a character that was quoted by a backslash. */ |
642 | if (pass_next) | |
726f6388 | 643 | { |
ccc6cda3 | 644 | /* Posix.2 sez: |
726f6388 | 645 | |
ccc6cda3 JA |
646 | ``The backslash shall retain its special meaning as an escape |
647 | character only when followed by one of the characters: | |
7117c2d2 | 648 | $ ` " \ <newline>''. |
726f6388 | 649 | |
ccc6cda3 JA |
650 | If STRIPDQ is zero, we handle the double quotes here and let |
651 | expand_word_internal handle the rest. If STRIPDQ is non-zero, | |
652 | we have already been through one round of backslash stripping, | |
653 | and want to strip these backslashes only if DQUOTE is non-zero, | |
654 | indicating that we are inside an embedded double-quoted string. */ | |
655 | ||
656 | /* If we are in an embedded quoted string, then don't strip | |
657 | backslashes before characters for which the backslash | |
658 | retains its special meaning, but remove backslashes in | |
659 | front of other characters. If we are not in an | |
660 | embedded quoted string, don't strip backslashes at all. | |
661 | This mess is necessary because the string was already | |
662 | surrounded by double quotes (and sh has some really weird | |
663 | quoting rules). | |
664 | The returned string will be run through expansion as if | |
665 | it were double-quoted. */ | |
666 | if ((stripdq == 0 && c != '"') || | |
28ef6c31 | 667 | (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0))) |
ccc6cda3 | 668 | temp[j++] = '\\'; |
ccc6cda3 | 669 | pass_next = 0; |
7117c2d2 JA |
670 | |
671 | add_one_character: | |
672 | COPY_CHAR_I (temp, j, string, send, i); | |
ccc6cda3 JA |
673 | continue; |
674 | } | |
726f6388 | 675 | |
ccc6cda3 JA |
676 | /* A backslash protects the next character. The code just above |
677 | handles preserving the backslash in front of any character but | |
678 | a double quote. */ | |
679 | if (c == '\\') | |
726f6388 | 680 | { |
ccc6cda3 | 681 | pass_next++; |
7117c2d2 | 682 | i++; |
726f6388 JA |
683 | continue; |
684 | } | |
685 | ||
ccc6cda3 JA |
686 | /* Inside backquotes, ``the portion of the quoted string from the |
687 | initial backquote and the characters up to the next backquote | |
688 | that is not preceded by a backslash, having escape characters | |
689 | removed, defines that command''. */ | |
690 | if (backquote) | |
726f6388 | 691 | { |
ccc6cda3 JA |
692 | if (c == '`') |
693 | backquote = 0; | |
694 | temp[j++] = c; | |
7117c2d2 | 695 | i++; |
726f6388 JA |
696 | continue; |
697 | } | |
698 | ||
ccc6cda3 | 699 | if (c == '`') |
726f6388 | 700 | { |
ccc6cda3 JA |
701 | temp[j++] = c; |
702 | backquote++; | |
7117c2d2 | 703 | i++; |
ccc6cda3 | 704 | continue; |
726f6388 JA |
705 | } |
706 | ||
ccc6cda3 JA |
707 | /* Pass everything between `$(' and the matching `)' or a quoted |
708 | ${ ... } pair through according to the Posix.2 specification. */ | |
cce855bc | 709 | if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE))) |
726f6388 | 710 | { |
b80f6443 JA |
711 | int free_ret = 1; |
712 | ||
ccc6cda3 | 713 | si = i + 2; |
cce855bc | 714 | if (string[i + 1] == LPAREN) |
0628567a | 715 | ret = extract_delimited_string (string, &si, "$(", "(", ")", EX_COMMAND); /*)*/ |
ccc6cda3 | 716 | else |
7117c2d2 | 717 | ret = extract_dollar_brace_string (string, &si, 1, 0); |
726f6388 | 718 | |
ccc6cda3 JA |
719 | temp[j++] = '$'; |
720 | temp[j++] = string[i + 1]; | |
726f6388 | 721 | |
b80f6443 JA |
722 | /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error |
723 | is set. */ | |
724 | if (ret == 0 && no_longjmp_on_fatal_error) | |
725 | { | |
726 | free_ret = 0; | |
727 | ret = string + i + 2; | |
728 | } | |
729 | ||
ccc6cda3 JA |
730 | for (t = 0; ret[t]; t++, j++) |
731 | temp[j] = ret[t]; | |
b80f6443 | 732 | temp[j] = string[si]; |
726f6388 | 733 | |
b80f6443 JA |
734 | if (string[si]) |
735 | { | |
736 | j++; | |
737 | i = si + 1; | |
738 | } | |
739 | else | |
740 | i = si; | |
741 | ||
742 | if (free_ret) | |
743 | free (ret); | |
ccc6cda3 | 744 | continue; |
726f6388 JA |
745 | } |
746 | ||
ccc6cda3 | 747 | /* Add any character but a double quote to the quoted string we're |
28ef6c31 | 748 | accumulating. */ |
ccc6cda3 | 749 | if (c != '"') |
7117c2d2 | 750 | goto add_one_character; |
ccc6cda3 JA |
751 | |
752 | /* c == '"' */ | |
753 | if (stripdq) | |
726f6388 | 754 | { |
ccc6cda3 | 755 | dquote ^= 1; |
7117c2d2 | 756 | i++; |
ccc6cda3 | 757 | continue; |
726f6388 | 758 | } |
ccc6cda3 JA |
759 | |
760 | break; | |
726f6388 | 761 | } |
ccc6cda3 | 762 | temp[j] = '\0'; |
726f6388 | 763 | |
ccc6cda3 JA |
764 | /* Point to after the closing quote. */ |
765 | if (c) | |
766 | i++; | |
726f6388 JA |
767 | *sindex = i; |
768 | ||
ccc6cda3 JA |
769 | return (temp); |
770 | } | |
771 | ||
772 | /* This should really be another option to string_extract_double_quoted. */ | |
f73dda09 | 773 | static int |
7117c2d2 | 774 | skip_double_quoted (string, slen, sind) |
ccc6cda3 | 775 | char *string; |
7117c2d2 | 776 | size_t slen; |
ccc6cda3 JA |
777 | int sind; |
778 | { | |
f73dda09 | 779 | int c, i; |
ccc6cda3 JA |
780 | char *ret; |
781 | int pass_next, backquote, si; | |
7117c2d2 | 782 | DECLARE_MBSTATE; |
ccc6cda3 JA |
783 | |
784 | pass_next = backquote = 0; | |
7117c2d2 JA |
785 | i = sind; |
786 | while (c = string[i]) | |
726f6388 | 787 | { |
ccc6cda3 JA |
788 | if (pass_next) |
789 | { | |
790 | pass_next = 0; | |
7117c2d2 | 791 | ADVANCE_CHAR (string, slen, i); |
ccc6cda3 JA |
792 | continue; |
793 | } | |
794 | else if (c == '\\') | |
795 | { | |
796 | pass_next++; | |
7117c2d2 | 797 | i++; |
ccc6cda3 JA |
798 | continue; |
799 | } | |
800 | else if (backquote) | |
801 | { | |
802 | if (c == '`') | |
803 | backquote = 0; | |
7117c2d2 | 804 | ADVANCE_CHAR (string, slen, i); |
ccc6cda3 JA |
805 | continue; |
806 | } | |
807 | else if (c == '`') | |
808 | { | |
809 | backquote++; | |
7117c2d2 | 810 | i++; |
ccc6cda3 JA |
811 | continue; |
812 | } | |
cce855bc | 813 | else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE))) |
ccc6cda3 JA |
814 | { |
815 | si = i + 2; | |
cce855bc | 816 | if (string[i + 1] == LPAREN) |
0628567a | 817 | ret = extract_delimited_string (string, &si, "$(", "(", ")", EX_NOALLOC|EX_COMMAND); /* ) */ |
ccc6cda3 | 818 | else |
7117c2d2 | 819 | ret = extract_dollar_brace_string (string, &si, 0, EX_NOALLOC); |
ccc6cda3 | 820 | |
7117c2d2 | 821 | i = si + 1; |
ccc6cda3 JA |
822 | continue; |
823 | } | |
824 | else if (c != '"') | |
7117c2d2 JA |
825 | { |
826 | ADVANCE_CHAR (string, slen, i); | |
827 | continue; | |
828 | } | |
ccc6cda3 JA |
829 | else |
830 | break; | |
726f6388 | 831 | } |
ccc6cda3 JA |
832 | |
833 | if (c) | |
834 | i++; | |
835 | ||
836 | return (i); | |
726f6388 JA |
837 | } |
838 | ||
ccc6cda3 JA |
839 | /* Extract the contents of STRING as if it is enclosed in single quotes. |
840 | SINDEX, when passed in, is the offset of the character immediately | |
841 | following the opening single quote; on exit, SINDEX is left pointing after | |
842 | the closing single quote. */ | |
843 | static inline char * | |
844 | string_extract_single_quoted (string, sindex) | |
845 | char *string; | |
846 | int *sindex; | |
847 | { | |
f73dda09 | 848 | register int i; |
7117c2d2 | 849 | size_t slen; |
ccc6cda3 | 850 | char *t; |
7117c2d2 | 851 | DECLARE_MBSTATE; |
ccc6cda3 | 852 | |
95732b49 JA |
853 | /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */ |
854 | slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0; | |
7117c2d2 JA |
855 | i = *sindex; |
856 | while (string[i] && string[i] != '\'') | |
857 | ADVANCE_CHAR (string, slen, i); | |
ccc6cda3 | 858 | |
bb70624e | 859 | t = substring (string, *sindex, i); |
ccc6cda3 JA |
860 | |
861 | if (string[i]) | |
862 | i++; | |
863 | *sindex = i; | |
864 | ||
865 | return (t); | |
866 | } | |
867 | ||
868 | static inline int | |
7117c2d2 | 869 | skip_single_quoted (string, slen, sind) |
0628567a | 870 | const char *string; |
7117c2d2 | 871 | size_t slen; |
ccc6cda3 JA |
872 | int sind; |
873 | { | |
28ef6c31 | 874 | register int c; |
7117c2d2 JA |
875 | DECLARE_MBSTATE; |
876 | ||
877 | c = sind; | |
878 | while (string[c] && string[c] != '\'') | |
879 | ADVANCE_CHAR (string, slen, c); | |
ccc6cda3 | 880 | |
28ef6c31 JA |
881 | if (string[c]) |
882 | c++; | |
883 | return c; | |
ccc6cda3 JA |
884 | } |
885 | ||
886 | /* Just like string_extract, but doesn't hack backslashes or any of | |
bb70624e | 887 | that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */ |
726f6388 | 888 | static char * |
95732b49 | 889 | string_extract_verbatim (string, slen, sindex, charlist) |
f73dda09 | 890 | char *string; |
95732b49 | 891 | size_t slen; |
ccc6cda3 | 892 | int *sindex; |
f73dda09 | 893 | char *charlist; |
ccc6cda3 JA |
894 | { |
895 | register int i = *sindex; | |
95732b49 JA |
896 | #if defined (HANDLE_MULTIBYTE) |
897 | size_t clen; | |
898 | wchar_t *wcharlist; | |
899 | #endif | |
ccc6cda3 JA |
900 | int c; |
901 | char *temp; | |
95732b49 | 902 | DECLARE_MBSTATE; |
ccc6cda3 JA |
903 | |
904 | if (charlist[0] == '\'' && charlist[1] == '\0') | |
905 | { | |
906 | temp = string_extract_single_quoted (string, sindex); | |
907 | --*sindex; /* leave *sindex at separator character */ | |
908 | return temp; | |
909 | } | |
910 | ||
95732b49 JA |
911 | i = *sindex; |
912 | #if 0 | |
913 | /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need | |
914 | this only if MB_CUR_MAX > 1. */ | |
915 | slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1; | |
916 | #endif | |
917 | #if defined (HANDLE_MULTIBYTE) | |
918 | clen = strlen (charlist); | |
919 | wcharlist = 0; | |
920 | #endif | |
921 | while (c = string[i]) | |
ccc6cda3 | 922 | { |
95732b49 JA |
923 | #if defined (HANDLE_MULTIBYTE) |
924 | size_t mblength; | |
925 | #endif | |
ccc6cda3 JA |
926 | if (c == CTLESC) |
927 | { | |
95732b49 | 928 | i += 2; |
ccc6cda3 JA |
929 | continue; |
930 | } | |
931 | ||
95732b49 JA |
932 | #if defined (HANDLE_MULTIBYTE) |
933 | mblength = MBLEN (string + i, slen - i); | |
934 | if (mblength > 1) | |
935 | { | |
936 | wchar_t wc; | |
937 | mblength = mbtowc (&wc, string + i, slen - i); | |
938 | if (MB_INVALIDCH (mblength)) | |
939 | { | |
940 | if (MEMBER (c, charlist)) | |
941 | break; | |
942 | } | |
943 | else | |
944 | { | |
945 | if (wcharlist == 0) | |
946 | { | |
947 | size_t len; | |
948 | len = mbstowcs (wcharlist, charlist, 0); | |
949 | if (len == -1) | |
950 | len = 0; | |
0628567a JA |
951 | wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1)); |
952 | mbstowcs (wcharlist, charlist, len + 1); | |
95732b49 JA |
953 | } |
954 | ||
955 | if (wcschr (wcharlist, wc)) | |
956 | break; | |
957 | } | |
958 | } | |
959 | else | |
960 | #endif | |
ccc6cda3 JA |
961 | if (MEMBER (c, charlist)) |
962 | break; | |
95732b49 JA |
963 | |
964 | ADVANCE_CHAR (string, slen, i); | |
ccc6cda3 JA |
965 | } |
966 | ||
95732b49 JA |
967 | #if defined (HANDLE_MULTIBYTE) |
968 | FREE (wcharlist); | |
969 | #endif | |
970 | ||
bb70624e | 971 | temp = substring (string, *sindex, i); |
ccc6cda3 JA |
972 | *sindex = i; |
973 | ||
974 | return (temp); | |
975 | } | |
976 | ||
977 | /* Extract the $( construct in STRING, and return a new string. | |
978 | Start extracting at (SINDEX) as if we had just seen "$(". | |
95732b49 | 979 | Make (SINDEX) get the position of the matching ")". ) */ |
ccc6cda3 JA |
980 | char * |
981 | extract_command_subst (string, sindex) | |
726f6388 JA |
982 | char *string; |
983 | int *sindex; | |
984 | { | |
0628567a | 985 | return (extract_delimited_string (string, sindex, "$(", "(", ")", EX_COMMAND)); /*)*/ |
ccc6cda3 JA |
986 | } |
987 | ||
28ef6c31 | 988 | /* Extract the $[ construct in STRING, and return a new string. (]) |
ccc6cda3 JA |
989 | Start extracting at (SINDEX) as if we had just seen "$[". |
990 | Make (SINDEX) get the position of the matching "]". */ | |
991 | char * | |
992 | extract_arithmetic_subst (string, sindex) | |
993 | char *string; | |
994 | int *sindex; | |
995 | { | |
7117c2d2 | 996 | return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/ |
ccc6cda3 JA |
997 | } |
998 | ||
999 | #if defined (PROCESS_SUBSTITUTION) | |
1000 | /* Extract the <( or >( construct in STRING, and return a new string. | |
1001 | Start extracting at (SINDEX) as if we had just seen "<(". | |
cce855bc | 1002 | Make (SINDEX) get the position of the matching ")". */ /*))*/ |
ccc6cda3 JA |
1003 | char * |
1004 | extract_process_subst (string, starter, sindex) | |
1005 | char *string; | |
1006 | char *starter; | |
1007 | int *sindex; | |
1008 | { | |
7117c2d2 | 1009 | return (extract_delimited_string (string, sindex, starter, "(", ")", 0)); |
ccc6cda3 JA |
1010 | } |
1011 | #endif /* PROCESS_SUBSTITUTION */ | |
1012 | ||
1013 | #if defined (ARRAY_VARS) | |
95732b49 JA |
1014 | /* This can be fooled by unquoted right parens in the passed string. If |
1015 | each caller verifies that the last character in STRING is a right paren, | |
1016 | we don't even need to call extract_delimited_string. */ | |
ccc6cda3 JA |
1017 | char * |
1018 | extract_array_assignment_list (string, sindex) | |
1019 | char *string; | |
1020 | int *sindex; | |
1021 | { | |
95732b49 JA |
1022 | int slen; |
1023 | char *ret; | |
1024 | ||
1025 | slen = strlen (string); /* ( */ | |
1026 | if (string[slen - 1] == ')') | |
1027 | { | |
1028 | ret = substring (string, *sindex, slen - 1); | |
1029 | *sindex = slen - 1; | |
1030 | return ret; | |
1031 | } | |
1032 | return 0; | |
ccc6cda3 JA |
1033 | } |
1034 | #endif | |
1035 | ||
1036 | /* Extract and create a new string from the contents of STRING, a | |
1037 | character string delimited with OPENER and CLOSER. SINDEX is | |
1038 | the address of an int describing the current offset in STRING; | |
1039 | it should point to just after the first OPENER found. On exit, | |
1040 | SINDEX gets the position of the last character of the matching CLOSER. | |
1041 | If OPENER is more than a single character, ALT_OPENER, if non-null, | |
1042 | contains a character string that can also match CLOSER and thus | |
1043 | needs to be skipped. */ | |
1044 | static char * | |
7117c2d2 | 1045 | extract_delimited_string (string, sindex, opener, alt_opener, closer, flags) |
ccc6cda3 JA |
1046 | char *string; |
1047 | int *sindex; | |
1048 | char *opener, *alt_opener, *closer; | |
7117c2d2 | 1049 | int flags; |
ccc6cda3 JA |
1050 | { |
1051 | int i, c, si; | |
7117c2d2 | 1052 | size_t slen; |
ccc6cda3 | 1053 | char *t, *result; |
0628567a | 1054 | int pass_character, nesting_level, in_comment; |
ccc6cda3 | 1055 | int len_closer, len_opener, len_alt_opener; |
7117c2d2 | 1056 | DECLARE_MBSTATE; |
ccc6cda3 | 1057 | |
7117c2d2 | 1058 | slen = strlen (string + *sindex) + *sindex; |
ccc6cda3 JA |
1059 | len_opener = STRLEN (opener); |
1060 | len_alt_opener = STRLEN (alt_opener); | |
1061 | len_closer = STRLEN (closer); | |
726f6388 | 1062 | |
0628567a | 1063 | pass_character = in_comment = 0; |
726f6388 JA |
1064 | |
1065 | nesting_level = 1; | |
ccc6cda3 | 1066 | i = *sindex; |
726f6388 | 1067 | |
ccc6cda3 | 1068 | while (nesting_level) |
726f6388 | 1069 | { |
ccc6cda3 JA |
1070 | c = string[i]; |
1071 | ||
1072 | if (c == 0) | |
28ef6c31 | 1073 | break; |
ccc6cda3 | 1074 | |
0628567a JA |
1075 | if (in_comment) |
1076 | { | |
1077 | if (c == '\n') | |
1078 | in_comment = 0; | |
1079 | ADVANCE_CHAR (string, slen, i); | |
1080 | continue; | |
1081 | } | |
1082 | ||
ccc6cda3 | 1083 | if (pass_character) /* previous char was backslash */ |
726f6388 JA |
1084 | { |
1085 | pass_character = 0; | |
7117c2d2 | 1086 | ADVANCE_CHAR (string, slen, i); |
726f6388 JA |
1087 | continue; |
1088 | } | |
1089 | ||
0628567a JA |
1090 | /* Not exactly right yet; should handle shell metacharacters and |
1091 | multibyte characters, too. */ | |
1092 | if ((flags & EX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || whitespace (string[i - 1]))) | |
1093 | { | |
1094 | in_comment = 1; | |
1095 | ADVANCE_CHAR (string, slen, i); | |
1096 | continue; | |
1097 | } | |
1098 | ||
7117c2d2 | 1099 | if (c == CTLESC || c == '\\') |
726f6388 | 1100 | { |
ccc6cda3 JA |
1101 | pass_character++; |
1102 | i++; | |
1103 | continue; | |
726f6388 JA |
1104 | } |
1105 | ||
ccc6cda3 JA |
1106 | /* Process a nested OPENER. */ |
1107 | if (STREQN (string + i, opener, len_opener)) | |
726f6388 | 1108 | { |
ccc6cda3 | 1109 | si = i + len_opener; |
7117c2d2 | 1110 | t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|EX_NOALLOC); |
ccc6cda3 | 1111 | i = si + 1; |
ccc6cda3 | 1112 | continue; |
726f6388 JA |
1113 | } |
1114 | ||
ccc6cda3 JA |
1115 | /* Process a nested ALT_OPENER */ |
1116 | if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener)) | |
726f6388 | 1117 | { |
ccc6cda3 | 1118 | si = i + len_alt_opener; |
7117c2d2 | 1119 | t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|EX_NOALLOC); |
ccc6cda3 | 1120 | i = si + 1; |
726f6388 JA |
1121 | continue; |
1122 | } | |
ccc6cda3 JA |
1123 | |
1124 | /* If the current substring terminates the delimited string, decrement | |
1125 | the nesting level. */ | |
1126 | if (STREQN (string + i, closer, len_closer)) | |
726f6388 | 1127 | { |
7117c2d2 | 1128 | i += len_closer - 1; /* move to last byte of the closer */ |
ccc6cda3 JA |
1129 | nesting_level--; |
1130 | if (nesting_level == 0) | |
1131 | break; | |
726f6388 | 1132 | } |
ccc6cda3 JA |
1133 | |
1134 | /* Pass old-style command substitution through verbatim. */ | |
1135 | if (c == '`') | |
28ef6c31 JA |
1136 | { |
1137 | si = i + 1; | |
7117c2d2 | 1138 | t = string_extract (string, &si, "`", flags|EX_NOALLOC); |
28ef6c31 | 1139 | i = si + 1; |
28ef6c31 JA |
1140 | continue; |
1141 | } | |
ccc6cda3 | 1142 | |
7117c2d2 JA |
1143 | /* Pass single-quoted and double-quoted strings through verbatim. */ |
1144 | if (c == '\'' || c == '"') | |
28ef6c31 JA |
1145 | { |
1146 | si = i + 1; | |
7117c2d2 JA |
1147 | i = (c == '\'') ? skip_single_quoted (string, slen, si) |
1148 | : skip_double_quoted (string, slen, si); | |
28ef6c31 JA |
1149 | continue; |
1150 | } | |
ccc6cda3 | 1151 | |
7117c2d2 JA |
1152 | /* move past this character, which was not special. */ |
1153 | ADVANCE_CHAR (string, slen, i); | |
726f6388 JA |
1154 | } |
1155 | ||
b80f6443 | 1156 | if (c == 0 && nesting_level) |
726f6388 | 1157 | { |
b80f6443 JA |
1158 | if (no_longjmp_on_fatal_error == 0) |
1159 | { | |
1160 | report_error (_("bad substitution: no closing `%s' in %s"), closer, string); | |
1161 | last_command_exit_value = EXECUTION_FAILURE; | |
1162 | exp_jump_to_top_level (DISCARD); | |
1163 | } | |
1164 | else | |
1165 | { | |
1166 | *sindex = i; | |
1167 | return (char *)NULL; | |
1168 | } | |
726f6388 | 1169 | } |
ccc6cda3 | 1170 | |
cce855bc | 1171 | si = i - *sindex - len_closer + 1; |
7117c2d2 JA |
1172 | if (flags & EX_NOALLOC) |
1173 | result = (char *)NULL; | |
1174 | else | |
1175 | { | |
1176 | result = (char *)xmalloc (1 + si); | |
1177 | strncpy (result, string + *sindex, si); | |
1178 | result[si] = '\0'; | |
1179 | } | |
cce855bc JA |
1180 | *sindex = i; |
1181 | ||
726f6388 JA |
1182 | return (result); |
1183 | } | |
1184 | ||
ccc6cda3 JA |
1185 | /* Extract a parameter expansion expression within ${ and } from STRING. |
1186 | Obey the Posix.2 rules for finding the ending `}': count braces while | |
1187 | skipping over enclosed quoted strings and command substitutions. | |
1188 | SINDEX is the address of an int describing the current offset in STRING; | |
1189 | it should point to just after the first `{' found. On exit, SINDEX | |
1190 | gets the position of the matching `}'. QUOTED is non-zero if this | |
1191 | occurs inside double quotes. */ | |
1192 | /* XXX -- this is very similar to extract_delimited_string -- XXX */ | |
726f6388 | 1193 | static char * |
7117c2d2 | 1194 | extract_dollar_brace_string (string, sindex, quoted, flags) |
726f6388 | 1195 | char *string; |
7117c2d2 | 1196 | int *sindex, quoted, flags; |
726f6388 | 1197 | { |
f73dda09 | 1198 | register int i, c; |
7117c2d2 | 1199 | size_t slen; |
ccc6cda3 JA |
1200 | int pass_character, nesting_level, si; |
1201 | char *result, *t; | |
7117c2d2 | 1202 | DECLARE_MBSTATE; |
726f6388 | 1203 | |
ccc6cda3 | 1204 | pass_character = 0; |
ccc6cda3 | 1205 | nesting_level = 1; |
7117c2d2 | 1206 | slen = strlen (string + *sindex) + *sindex; |
ccc6cda3 | 1207 | |
7117c2d2 JA |
1208 | i = *sindex; |
1209 | while (c = string[i]) | |
726f6388 | 1210 | { |
ccc6cda3 | 1211 | if (pass_character) |
726f6388 | 1212 | { |
ccc6cda3 | 1213 | pass_character = 0; |
7117c2d2 | 1214 | ADVANCE_CHAR (string, slen, i); |
ccc6cda3 JA |
1215 | continue; |
1216 | } | |
726f6388 | 1217 | |
cce855bc JA |
1218 | /* CTLESCs and backslashes quote the next character. */ |
1219 | if (c == CTLESC || c == '\\') | |
726f6388 | 1220 | { |
ccc6cda3 | 1221 | pass_character++; |
7117c2d2 | 1222 | i++; |
726f6388 JA |
1223 | continue; |
1224 | } | |
1225 | ||
cce855bc | 1226 | if (string[i] == '$' && string[i+1] == LBRACE) |
726f6388 | 1227 | { |
ccc6cda3 | 1228 | nesting_level++; |
7117c2d2 | 1229 | i += 2; |
726f6388 JA |
1230 | continue; |
1231 | } | |
1232 | ||
cce855bc | 1233 | if (c == RBRACE) |
726f6388 | 1234 | { |
ccc6cda3 JA |
1235 | nesting_level--; |
1236 | if (nesting_level == 0) | |
1237 | break; | |
7117c2d2 | 1238 | i++; |
726f6388 JA |
1239 | continue; |
1240 | } | |
1241 | ||
ccc6cda3 JA |
1242 | /* Pass the contents of old-style command substitutions through |
1243 | verbatim. */ | |
1244 | if (c == '`') | |
726f6388 | 1245 | { |
ccc6cda3 | 1246 | si = i + 1; |
7117c2d2 JA |
1247 | t = string_extract (string, &si, "`", flags|EX_NOALLOC); |
1248 | i = si + 1; | |
ccc6cda3 JA |
1249 | continue; |
1250 | } | |
726f6388 | 1251 | |
cce855bc JA |
1252 | /* Pass the contents of new-style command substitutions and |
1253 | arithmetic substitutions through verbatim. */ | |
1254 | if (string[i] == '$' && string[i+1] == LPAREN) | |
ccc6cda3 | 1255 | { |
726f6388 | 1256 | si = i + 2; |
0628567a | 1257 | t = extract_delimited_string (string, &si, "$(", "(", ")", flags|EX_NOALLOC|EX_COMMAND); /*)*/ |
7117c2d2 | 1258 | i = si + 1; |
726f6388 JA |
1259 | continue; |
1260 | } | |
1261 | ||
cce855bc JA |
1262 | /* Pass the contents of single-quoted and double-quoted strings |
1263 | through verbatim. */ | |
1264 | if (c == '\'' || c == '"') | |
ccc6cda3 JA |
1265 | { |
1266 | si = i + 1; | |
7117c2d2 JA |
1267 | i = (c == '\'') ? skip_single_quoted (string, slen, si) |
1268 | : skip_double_quoted (string, slen, si); | |
cce855bc | 1269 | /* skip_XXX_quoted leaves index one past close quote */ |
ccc6cda3 JA |
1270 | continue; |
1271 | } | |
7117c2d2 JA |
1272 | |
1273 | /* move past this character, which was not special. */ | |
1274 | ADVANCE_CHAR (string, slen, i); | |
cce855bc | 1275 | } |
726f6388 | 1276 | |
b80f6443 | 1277 | if (c == 0 && nesting_level) |
cce855bc | 1278 | { |
b80f6443 JA |
1279 | if (no_longjmp_on_fatal_error == 0) |
1280 | { /* { */ | |
f1be666c | 1281 | report_error (_("bad substitution: no closing `%s' in %s"), "}", string); |
b80f6443 JA |
1282 | last_command_exit_value = EXECUTION_FAILURE; |
1283 | exp_jump_to_top_level (DISCARD); | |
1284 | } | |
1285 | else | |
1286 | { | |
1287 | *sindex = i; | |
1288 | return ((char *)NULL); | |
1289 | } | |
726f6388 | 1290 | } |
726f6388 | 1291 | |
7117c2d2 | 1292 | result = (flags & EX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i); |
726f6388 JA |
1293 | *sindex = i; |
1294 | ||
ccc6cda3 | 1295 | return (result); |
726f6388 JA |
1296 | } |
1297 | ||
ccc6cda3 JA |
1298 | /* Remove backslashes which are quoting backquotes from STRING. Modifies |
1299 | STRING, and returns a pointer to it. */ | |
1300 | char * | |
1301 | de_backslash (string) | |
726f6388 | 1302 | char *string; |
ccc6cda3 | 1303 | { |
7117c2d2 JA |
1304 | register size_t slen; |
1305 | register int i, j, prev_i; | |
1306 | DECLARE_MBSTATE; | |
726f6388 | 1307 | |
7117c2d2 JA |
1308 | slen = strlen (string); |
1309 | i = j = 0; | |
1310 | ||
1311 | /* Loop copying string[i] to string[j], i >= j. */ | |
1312 | while (i < slen) | |
1313 | { | |
1314 | if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' || | |
ccc6cda3 | 1315 | string[i + 1] == '$')) |
7117c2d2 JA |
1316 | i++; |
1317 | prev_i = i; | |
1318 | ADVANCE_CHAR (string, slen, i); | |
1319 | if (j < prev_i) | |
b80f6443 | 1320 | do string[j++] = string[prev_i++]; while (prev_i < i); |
7117c2d2 | 1321 | else |
b80f6443 | 1322 | j = i; |
7117c2d2 JA |
1323 | } |
1324 | string[j] = '\0'; | |
1325 | ||
ccc6cda3 JA |
1326 | return (string); |
1327 | } | |
726f6388 | 1328 | |
ccc6cda3 | 1329 | #if 0 |
cce855bc | 1330 | /*UNUSED*/ |
ccc6cda3 JA |
1331 | /* Replace instances of \! in a string with !. */ |
1332 | void | |
1333 | unquote_bang (string) | |
1334 | char *string; | |
1335 | { | |
1336 | register int i, j; | |
1337 | register char *temp; | |
726f6388 | 1338 | |
f73dda09 | 1339 | temp = (char *)xmalloc (1 + strlen (string)); |
726f6388 | 1340 | |
ccc6cda3 JA |
1341 | for (i = 0, j = 0; (temp[j] = string[i]); i++, j++) |
1342 | { | |
1343 | if (string[i] == '\\' && string[i + 1] == '!') | |
1344 | { | |
1345 | temp[j] = '!'; | |
1346 | i++; | |
1347 | } | |
1348 | } | |
1349 | strcpy (string, temp); | |
1350 | free (temp); | |
726f6388 | 1351 | } |
ccc6cda3 | 1352 | #endif |
726f6388 | 1353 | |
ccc6cda3 | 1354 | #if defined (READLINE) |
726f6388 JA |
1355 | /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is |
1356 | an unclosed quoted string), or if the character at EINDEX is quoted | |
28ef6c31 | 1357 | by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various |
b72432fd | 1358 | single and double-quoted string parsing functions should not return an |
7117c2d2 JA |
1359 | error if there are unclosed quotes or braces. The characters that this |
1360 | recognizes need to be the same as the contents of | |
1361 | rl_completer_quote_characters. */ | |
b72432fd | 1362 | |
28ef6c31 | 1363 | #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0) |
b72432fd | 1364 | |
726f6388 JA |
1365 | int |
1366 | char_is_quoted (string, eindex) | |
1367 | char *string; | |
1368 | int eindex; | |
1369 | { | |
7117c2d2 JA |
1370 | int i, pass_next, c; |
1371 | size_t slen; | |
1372 | DECLARE_MBSTATE; | |
726f6388 | 1373 | |
7117c2d2 | 1374 | slen = strlen (string); |
28ef6c31 | 1375 | no_longjmp_on_fatal_error = 1; |
7117c2d2 JA |
1376 | i = pass_next = 0; |
1377 | while (i <= eindex) | |
726f6388 | 1378 | { |
7117c2d2 JA |
1379 | c = string[i]; |
1380 | ||
726f6388 JA |
1381 | if (pass_next) |
1382 | { | |
1383 | pass_next = 0; | |
1384 | if (i >= eindex) /* XXX was if (i >= eindex - 1) */ | |
b72432fd | 1385 | CQ_RETURN(1); |
7117c2d2 | 1386 | ADVANCE_CHAR (string, slen, i); |
726f6388 JA |
1387 | continue; |
1388 | } | |
7117c2d2 | 1389 | else if (c == '\\') |
ccc6cda3 JA |
1390 | { |
1391 | pass_next = 1; | |
7117c2d2 | 1392 | i++; |
ccc6cda3 JA |
1393 | continue; |
1394 | } | |
7117c2d2 JA |
1395 | else if (c == '\'' || c == '"') |
1396 | { | |
1397 | i = (c == '\'') ? skip_single_quoted (string, slen, ++i) | |
1398 | : skip_double_quoted (string, slen, ++i); | |
1399 | if (i > eindex) | |
1400 | CQ_RETURN(1); | |
1401 | /* no increment, the skip_xxx functions go one past end */ | |
1402 | } | |
1403 | else | |
1404 | ADVANCE_CHAR (string, slen, i); | |
726f6388 | 1405 | } |
7117c2d2 | 1406 | |
b72432fd | 1407 | CQ_RETURN(0); |
726f6388 JA |
1408 | } |
1409 | ||
726f6388 JA |
1410 | int |
1411 | unclosed_pair (string, eindex, openstr) | |
1412 | char *string; | |
1413 | int eindex; | |
1414 | char *openstr; | |
1415 | { | |
ccc6cda3 | 1416 | int i, pass_next, openc, olen; |
7117c2d2 JA |
1417 | size_t slen; |
1418 | DECLARE_MBSTATE; | |
726f6388 | 1419 | |
7117c2d2 | 1420 | slen = strlen (string); |
726f6388 | 1421 | olen = strlen (openstr); |
7117c2d2 JA |
1422 | i = pass_next = openc = 0; |
1423 | while (i <= eindex) | |
726f6388 JA |
1424 | { |
1425 | if (pass_next) | |
1426 | { | |
1427 | pass_next = 0; | |
1428 | if (i >= eindex) /* XXX was if (i >= eindex - 1) */ | |
1429 | return 0; | |
7117c2d2 JA |
1430 | ADVANCE_CHAR (string, slen, i); |
1431 | continue; | |
1432 | } | |
1433 | else if (string[i] == '\\') | |
1434 | { | |
1435 | pass_next = 1; | |
1436 | i++; | |
726f6388 JA |
1437 | continue; |
1438 | } | |
1439 | else if (STREQN (string + i, openstr, olen)) | |
1440 | { | |
1441 | openc = 1 - openc; | |
7117c2d2 | 1442 | i += olen; |
726f6388 | 1443 | } |
ccc6cda3 | 1444 | else if (string[i] == '\'' || string[i] == '"') |
726f6388 | 1445 | { |
7117c2d2 JA |
1446 | i = (string[i] == '\'') ? skip_single_quoted (string, slen, i) |
1447 | : skip_double_quoted (string, slen, i); | |
726f6388 JA |
1448 | if (i > eindex) |
1449 | return 0; | |
1450 | } | |
7117c2d2 JA |
1451 | else |
1452 | ADVANCE_CHAR (string, slen, i); | |
726f6388 JA |
1453 | } |
1454 | return (openc); | |
1455 | } | |
bb70624e JA |
1456 | |
1457 | /* Skip characters in STRING until we find a character in DELIMS, and return | |
1458 | the index of that character. START is the index into string at which we | |
1459 | begin. This is similar in spirit to strpbrk, but it returns an index into | |
1460 | STRING and takes a starting index. This little piece of code knows quite | |
1461 | a lot of shell syntax. It's very similar to skip_double_quoted and other | |
1462 | functions of that ilk. */ | |
1463 | int | |
1464 | skip_to_delim (string, start, delims) | |
1465 | char *string; | |
1466 | int start; | |
1467 | char *delims; | |
1468 | { | |
7117c2d2 JA |
1469 | int i, pass_next, backq, si, c; |
1470 | size_t slen; | |
bb70624e | 1471 | char *temp; |
7117c2d2 | 1472 | DECLARE_MBSTATE; |
bb70624e | 1473 | |
7117c2d2 | 1474 | slen = strlen (string + start) + start; |
28ef6c31 | 1475 | no_longjmp_on_fatal_error = 1; |
7117c2d2 JA |
1476 | i = start; |
1477 | pass_next = backq = 0; | |
1478 | while (c = string[i]) | |
bb70624e JA |
1479 | { |
1480 | if (pass_next) | |
1481 | { | |
1482 | pass_next = 0; | |
7117c2d2 | 1483 | if (c == 0) |
bb70624e | 1484 | CQ_RETURN(i); |
7117c2d2 | 1485 | ADVANCE_CHAR (string, slen, i); |
bb70624e JA |
1486 | continue; |
1487 | } | |
7117c2d2 | 1488 | else if (c == '\\') |
bb70624e JA |
1489 | { |
1490 | pass_next = 1; | |
7117c2d2 | 1491 | i++; |
bb70624e JA |
1492 | continue; |
1493 | } | |
1494 | else if (backq) | |
1495 | { | |
7117c2d2 | 1496 | if (c == '`') |
bb70624e | 1497 | backq = 0; |
7117c2d2 | 1498 | ADVANCE_CHAR (string, slen, i); |
bb70624e JA |
1499 | continue; |
1500 | } | |
7117c2d2 | 1501 | else if (c == '`') |
bb70624e JA |
1502 | { |
1503 | backq = 1; | |
7117c2d2 | 1504 | i++; |
bb70624e JA |
1505 | continue; |
1506 | } | |
7117c2d2 | 1507 | else if (c == '\'' || c == '"') |
bb70624e | 1508 | { |
7117c2d2 JA |
1509 | i = (c == '\'') ? skip_single_quoted (string, slen, ++i) |
1510 | : skip_double_quoted (string, slen, ++i); | |
1511 | /* no increment, the skip functions increment past the closing quote. */ | |
bb70624e | 1512 | } |
7117c2d2 | 1513 | else if (c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE)) |
bb70624e JA |
1514 | { |
1515 | si = i + 2; | |
1516 | if (string[si] == '\0') | |
f73dda09 JA |
1517 | CQ_RETURN(si); |
1518 | ||
bb70624e | 1519 | if (string[i+1] == LPAREN) |
0628567a | 1520 | temp = extract_delimited_string (string, &si, "$(", "(", ")", EX_NOALLOC|EX_COMMAND); /* ) */ |
bb70624e | 1521 | else |
7117c2d2 | 1522 | temp = extract_dollar_brace_string (string, &si, 0, EX_NOALLOC); |
bb70624e | 1523 | i = si; |
f73dda09 JA |
1524 | if (string[i] == '\0') /* don't increment i past EOS in loop */ |
1525 | break; | |
7117c2d2 | 1526 | i++; |
bb70624e JA |
1527 | continue; |
1528 | } | |
7117c2d2 | 1529 | else if (member (c, delims)) |
28ef6c31 | 1530 | break; |
7117c2d2 JA |
1531 | else |
1532 | ADVANCE_CHAR (string, slen, i); | |
bb70624e | 1533 | } |
7117c2d2 | 1534 | |
bb70624e JA |
1535 | CQ_RETURN(i); |
1536 | } | |
1537 | ||
1538 | /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the | |
1539 | individual words. If DELIMS is NULL, the current value of $IFS is used | |
b80f6443 JA |
1540 | to split the string, and the function follows the shell field splitting |
1541 | rules. SENTINEL is an index to look for. NWP, if non-NULL, | |
bb70624e JA |
1542 | gets the number of words in the returned list. CWP, if non-NULL, gets |
1543 | the index of the word containing SENTINEL. Non-whitespace chars in | |
1544 | DELIMS delimit separate fields. */ | |
1545 | WORD_LIST * | |
1546 | split_at_delims (string, slen, delims, sentinel, nwp, cwp) | |
1547 | char *string; | |
1548 | int slen; | |
1549 | char *delims; | |
1550 | int sentinel; | |
1551 | int *nwp, *cwp; | |
1552 | { | |
b80f6443 | 1553 | int ts, te, i, nw, cw, ifs_split; |
f73dda09 | 1554 | char *token, *d, *d2; |
bb70624e JA |
1555 | WORD_LIST *ret, *tl; |
1556 | ||
1557 | if (string == 0 || *string == '\0') | |
1558 | { | |
1559 | if (nwp) | |
1560 | *nwp = 0; | |
1561 | if (cwp) | |
1562 | *cwp = 0; | |
1563 | return ((WORD_LIST *)NULL); | |
1564 | } | |
1565 | ||
7117c2d2 | 1566 | d = (delims == 0) ? ifs_value : delims; |
b80f6443 | 1567 | ifs_split = delims == 0; |
bb70624e JA |
1568 | |
1569 | /* Make d2 the non-whitespace characters in delims */ | |
1570 | d2 = 0; | |
1571 | if (delims) | |
1572 | { | |
95732b49 JA |
1573 | size_t slength; |
1574 | #if defined (HANDLE_MULTIBYTE) | |
1575 | size_t mblength = 1; | |
1576 | #endif | |
1577 | DECLARE_MBSTATE; | |
1578 | ||
1579 | slength = strlen (delims); | |
1580 | d2 = (char *)xmalloc (slength + 1); | |
1581 | i = ts = 0; | |
1582 | while (delims[i]) | |
bb70624e | 1583 | { |
95732b49 | 1584 | #if defined (HANDLE_MULTIBYTE) |
0628567a JA |
1585 | mbstate_t state_bak; |
1586 | state_bak = state; | |
95732b49 JA |
1587 | mblength = MBRLEN (delims + i, slength, &state); |
1588 | if (MB_INVALIDCH (mblength)) | |
1589 | state = state_bak; | |
1590 | else if (mblength > 1) | |
1591 | { | |
1592 | memcpy (d2 + ts, delims + i, mblength); | |
1593 | ts += mblength; | |
1594 | i += mblength; | |
1595 | slength -= mblength; | |
1596 | continue; | |
1597 | } | |
1598 | #endif | |
1599 | if (whitespace (delims[i]) == 0) | |
bb70624e | 1600 | d2[ts++] = delims[i]; |
95732b49 JA |
1601 | |
1602 | i++; | |
1603 | slength--; | |
bb70624e JA |
1604 | } |
1605 | d2[ts] = '\0'; | |
1606 | } | |
1607 | ||
1608 | ret = (WORD_LIST *)NULL; | |
1609 | ||
b80f6443 JA |
1610 | /* Remove sequences of whitspace characters at the start of the string, as |
1611 | long as those characters are delimiters. */ | |
1612 | for (i = 0; member (string[i], d) && spctabnl (string[i]); i++) | |
bb70624e JA |
1613 | ; |
1614 | if (string[i] == '\0') | |
1615 | return (ret); | |
1616 | ||
1617 | ts = i; | |
1618 | nw = 0; | |
1619 | cw = -1; | |
1620 | while (1) | |
1621 | { | |
1622 | te = skip_to_delim (string, ts, d); | |
1623 | ||
1624 | /* If we have a non-whitespace delimiter character, use it to make a | |
1625 | separate field. This is just about what $IFS splitting does and | |
1626 | is closer to the behavior of the shell parser. */ | |
28ef6c31 | 1627 | if (ts == te && d2 && member (string[ts], d2)) |
bb70624e JA |
1628 | { |
1629 | te = ts + 1; | |
b80f6443 JA |
1630 | /* If we're using IFS splitting, the non-whitespace delimiter char |
1631 | and any additional IFS whitespace delimits a field. */ | |
1632 | if (ifs_split) | |
1633 | while (member (string[te], d) && spctabnl (string[te])) | |
1634 | te++; | |
1635 | else | |
1636 | while (member (string[te], d2)) | |
1637 | te++; | |
bb70624e JA |
1638 | } |
1639 | ||
1640 | token = substring (string, ts, te); | |
1641 | ||
1642 | ret = add_string_to_list (token, ret); | |
1643 | free (token); | |
1644 | nw++; | |
1645 | ||
1646 | if (sentinel >= ts && sentinel <= te) | |
1647 | cw = nw; | |
1648 | ||
1649 | /* If the cursor is at whitespace just before word start, set the | |
28ef6c31 | 1650 | sentinel word to the current word. */ |
bb70624e JA |
1651 | if (cwp && cw == -1 && sentinel == ts-1) |
1652 | cw = nw; | |
1653 | ||
1654 | /* If the cursor is at whitespace between two words, make a new, empty | |
28ef6c31 JA |
1655 | word, add it before (well, after, since the list is in reverse order) |
1656 | the word we just added, and set the current word to that one. */ | |
bb70624e | 1657 | if (cwp && cw == -1 && sentinel < ts) |
28ef6c31 | 1658 | { |
7117c2d2 | 1659 | tl = make_word_list (make_word (""), ret->next); |
28ef6c31 JA |
1660 | ret->next = tl; |
1661 | cw = nw; | |
1662 | nw++; | |
1663 | } | |
bb70624e JA |
1664 | |
1665 | if (string[te] == 0) | |
1666 | break; | |
1667 | ||
b80f6443 JA |
1668 | i = te; |
1669 | while (member (string[i], d) && (ifs_split || spctabnl(string[i]))) | |
bb70624e JA |
1670 | i++; |
1671 | ||
1672 | if (string[i]) | |
1673 | ts = i; | |
1674 | else | |
1675 | break; | |
1676 | } | |
1677 | ||
1678 | /* Special case for SENTINEL at the end of STRING. If we haven't found | |
1679 | the word containing SENTINEL yet, and the index we're looking for is at | |
1680 | the end of STRING, add an additional null argument and set the current | |
1681 | word pointer to that. */ | |
1682 | if (cwp && cw == -1 && sentinel >= slen) | |
1683 | { | |
1684 | if (whitespace (string[sentinel - 1])) | |
28ef6c31 JA |
1685 | { |
1686 | token = ""; | |
1687 | ret = add_string_to_list (token, ret); | |
1688 | nw++; | |
1689 | } | |
bb70624e JA |
1690 | cw = nw; |
1691 | } | |
1692 | ||
1693 | if (nwp) | |
1694 | *nwp = nw; | |
1695 | if (cwp) | |
1696 | *cwp = cw; | |
1697 | ||
1698 | return (REVERSE_LIST (ret, WORD_LIST *)); | |
1699 | } | |
726f6388 JA |
1700 | #endif /* READLINE */ |
1701 | ||
ccc6cda3 JA |
1702 | #if 0 |
1703 | /* UNUSED */ | |
726f6388 JA |
1704 | /* Extract the name of the variable to bind to from the assignment string. */ |
1705 | char * | |
1706 | assignment_name (string) | |
1707 | char *string; | |
1708 | { | |
ccc6cda3 | 1709 | int offset; |
726f6388 JA |
1710 | char *temp; |
1711 | ||
b80f6443 | 1712 | offset = assignment (string, 0); |
ccc6cda3 | 1713 | if (offset == 0) |
726f6388 | 1714 | return (char *)NULL; |
bb70624e | 1715 | temp = substring (string, 0, offset); |
726f6388 JA |
1716 | return (temp); |
1717 | } | |
ccc6cda3 | 1718 | #endif |
726f6388 | 1719 | |
cce855bc JA |
1720 | /* **************************************************************** */ |
1721 | /* */ | |
1722 | /* Functions to convert strings to WORD_LISTs and vice versa */ | |
1723 | /* */ | |
1724 | /* **************************************************************** */ | |
1725 | ||
726f6388 JA |
1726 | /* Return a single string of all the words in LIST. SEP is the separator |
1727 | to put between individual elements of LIST in the output string. */ | |
7117c2d2 | 1728 | char * |
726f6388 JA |
1729 | string_list_internal (list, sep) |
1730 | WORD_LIST *list; | |
1731 | char *sep; | |
1732 | { | |
1733 | register WORD_LIST *t; | |
1734 | char *result, *r; | |
1735 | int word_len, sep_len, result_size; | |
1736 | ||
ccc6cda3 | 1737 | if (list == 0) |
726f6388 JA |
1738 | return ((char *)NULL); |
1739 | ||
b80f6443 JA |
1740 | /* Short-circuit quickly if we don't need to separate anything. */ |
1741 | if (list->next == 0) | |
1742 | return (savestring (list->word->word)); | |
1743 | ||
726f6388 JA |
1744 | /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */ |
1745 | sep_len = STRLEN (sep); | |
1746 | result_size = 0; | |
1747 | ||
1748 | for (t = list; t; t = t->next) | |
1749 | { | |
1750 | if (t != list) | |
1751 | result_size += sep_len; | |
1752 | result_size += strlen (t->word->word); | |
1753 | } | |
1754 | ||
f73dda09 | 1755 | r = result = (char *)xmalloc (result_size + 1); |
726f6388 JA |
1756 | |
1757 | for (t = list; t; t = t->next) | |
1758 | { | |
1759 | if (t != list && sep_len) | |
1760 | { | |
ccc6cda3 JA |
1761 | if (sep_len > 1) |
1762 | { | |
1763 | FASTCOPY (sep, r, sep_len); | |
1764 | r += sep_len; | |
1765 | } | |
1766 | else | |
1767 | *r++ = sep[0]; | |
726f6388 JA |
1768 | } |
1769 | ||
1770 | word_len = strlen (t->word->word); | |
1771 | FASTCOPY (t->word->word, r, word_len); | |
1772 | r += word_len; | |
1773 | } | |
1774 | ||
ccc6cda3 | 1775 | *r = '\0'; |
726f6388 JA |
1776 | return (result); |
1777 | } | |
1778 | ||
1779 | /* Return a single string of all the words present in LIST, separating | |
1780 | each word with a space. */ | |
1781 | char * | |
1782 | string_list (list) | |
1783 | WORD_LIST *list; | |
1784 | { | |
1785 | return (string_list_internal (list, " ")); | |
1786 | } | |
1787 | ||
1788 | /* Return a single string of all the words present in LIST, obeying the | |
1789 | quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the | |
1790 | expansion [of $*] appears within a double quoted string, it expands | |
1791 | to a single field with the value of each parameter separated by the | |
1792 | first character of the IFS variable, or by a <space> if IFS is unset." */ | |
f73dda09 | 1793 | char * |
726f6388 JA |
1794 | string_list_dollar_star (list) |
1795 | WORD_LIST *list; | |
1796 | { | |
0628567a | 1797 | char *ret; |
95732b49 | 1798 | #if defined (HANDLE_MULTIBYTE) |
0628567a | 1799 | # if defined (__GNUC__) |
95732b49 | 1800 | char sep[MB_CUR_MAX + 1]; |
0628567a JA |
1801 | # else |
1802 | char *sep = 0; | |
1803 | # endif | |
95732b49 | 1804 | #else |
7117c2d2 | 1805 | char sep[2]; |
95732b49 | 1806 | #endif |
726f6388 | 1807 | |
95732b49 | 1808 | #if defined (HANDLE_MULTIBYTE) |
0628567a JA |
1809 | # if !defined (__GNUC__) |
1810 | sep = (char *)xmalloc (MB_CUR_MAX + 1); | |
1811 | # endif /* !__GNUC__ */ | |
95732b49 JA |
1812 | if (ifs_firstc_len == 1) |
1813 | { | |
1814 | sep[0] = ifs_firstc[0]; | |
1815 | sep[1] = '\0'; | |
1816 | } | |
1817 | else | |
1818 | { | |
1819 | memcpy (sep, ifs_firstc, ifs_firstc_len); | |
1820 | sep[ifs_firstc_len] = '\0'; | |
1821 | } | |
1822 | #else | |
7117c2d2 | 1823 | sep[0] = ifs_firstc; |
726f6388 | 1824 | sep[1] = '\0'; |
95732b49 | 1825 | #endif |
726f6388 | 1826 | |
0628567a JA |
1827 | ret = string_list_internal (list, sep); |
1828 | #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__) | |
1829 | free (sep); | |
1830 | #endif | |
1831 | return ret; | |
726f6388 JA |
1832 | } |
1833 | ||
cce855bc JA |
1834 | /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
1835 | is non-zero, the $@ appears within double quotes, and we should quote | |
1836 | the list before converting it into a string. If IFS is unset, and the | |
1837 | word is not quoted, we just need to quote CTLESC and CTLNUL characters | |
1838 | in the words in the list, because the default value of $IFS is | |
1839 | <space><tab><newline>, IFS characters in the words in the list should | |
1840 | also be split. If IFS is null, and the word is not quoted, we need | |
1841 | to quote the words in the list to preserve the positional parameters | |
1842 | exactly. */ | |
f73dda09 | 1843 | char * |
cce855bc JA |
1844 | string_list_dollar_at (list, quoted) |
1845 | WORD_LIST *list; | |
1846 | int quoted; | |
1847 | { | |
95732b49 JA |
1848 | char *ifs, *ret; |
1849 | #if defined (HANDLE_MULTIBYTE) | |
0628567a | 1850 | # if defined (__GNUC__) |
95732b49 | 1851 | char sep[MB_CUR_MAX + 1]; |
0628567a JA |
1852 | # else |
1853 | char *sep = 0; | |
1854 | # endif /* !__GNUC__ */ | |
95732b49 JA |
1855 | #else |
1856 | char sep[2]; | |
1857 | #endif | |
cce855bc JA |
1858 | WORD_LIST *tlist; |
1859 | ||
7117c2d2 JA |
1860 | /* XXX this could just be ifs = ifs_value; */ |
1861 | ifs = ifs_var ? value_cell (ifs_var) : (char *)0; | |
cce855bc | 1862 | |
95732b49 | 1863 | #if defined (HANDLE_MULTIBYTE) |
0628567a JA |
1864 | # if !defined (__GNUC__) |
1865 | sep = (char *)xmalloc (MB_CUR_MAX + 1); | |
1866 | # endif /* !__GNUC__ */ | |
95732b49 JA |
1867 | if (ifs && *ifs) |
1868 | { | |
1869 | if (ifs_firstc_len == 1) | |
1870 | { | |
1871 | sep[0] = ifs_firstc[0]; | |
1872 | sep[1] = '\0'; | |
1873 | } | |
1874 | else | |
1875 | { | |
1876 | memcpy (sep, ifs_firstc, ifs_firstc_len); | |
1877 | sep[ifs_firstc_len] = '\0'; | |
1878 | } | |
1879 | } | |
1880 | else | |
1881 | { | |
1882 | sep[0] = ' '; | |
1883 | sep[1] = '\0'; | |
1884 | } | |
1885 | #else | |
cce855bc JA |
1886 | sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs; |
1887 | sep[1] = '\0'; | |
95732b49 | 1888 | #endif |
cce855bc | 1889 | |
f1be666c JA |
1890 | /* XXX -- why call quote_list if ifs == 0? we can get away without doing |
1891 | it now that quote_escapes quotes spaces */ | |
1892 | #if 0 | |
cce855bc | 1893 | tlist = ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (ifs && *ifs == 0)) |
f1be666c JA |
1894 | #else |
1895 | tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
1896 | #endif | |
cce855bc JA |
1897 | ? quote_list (list) |
1898 | : list_quote_escapes (list); | |
0628567a JA |
1899 | |
1900 | ret = string_list_internal (tlist, sep); | |
1901 | #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__) | |
1902 | free (sep); | |
1903 | #endif | |
1904 | return ret; | |
cce855bc JA |
1905 | } |
1906 | ||
726f6388 JA |
1907 | /* Return the list of words present in STRING. Separate the string into |
1908 | words at any of the characters found in SEPARATORS. If QUOTED is | |
1909 | non-zero then word in the list will have its quoted flag set, otherwise | |
1910 | the quoted flag is left as make_word () deemed fit. | |
1911 | ||
1912 | This obeys the P1003.2 word splitting semantics. If `separators' is | |
1913 | exactly <space><tab><newline>, then the splitting algorithm is that of | |
1914 | the Bourne shell, which treats any sequence of characters from `separators' | |
1915 | as a delimiter. If IFS is unset, which results in `separators' being set | |
1916 | to "", no splitting occurs. If separators has some other value, the | |
1917 | following rules are applied (`IFS white space' means zero or more | |
1918 | occurrences of <space>, <tab>, or <newline>, as long as those characters | |
1919 | are in `separators'): | |
1920 | ||
1921 | 1) IFS white space is ignored at the start and the end of the | |
1922 | string. | |
1923 | 2) Each occurrence of a character in `separators' that is not | |
1924 | IFS white space, along with any adjacent occurrences of | |
1925 | IFS white space delimits a field. | |
1926 | 3) Any nonzero-length sequence of IFS white space delimits a field. | |
1927 | */ | |
1928 | ||
1929 | /* BEWARE! list_string strips null arguments. Don't call it twice and | |
1930 | expect to have "" preserved! */ | |
1931 | ||
726f6388 JA |
1932 | /* This performs word splitting and quoted null character removal on |
1933 | STRING. */ | |
b80f6443 JA |
1934 | #define issep(c) \ |
1935 | (((separators)[0]) ? ((separators)[1] ? isifs(c) \ | |
1936 | : (c) == (separators)[0]) \ | |
1937 | : 0) | |
726f6388 JA |
1938 | |
1939 | WORD_LIST * | |
1940 | list_string (string, separators, quoted) | |
1941 | register char *string, *separators; | |
1942 | int quoted; | |
1943 | { | |
ccc6cda3 JA |
1944 | WORD_LIST *result; |
1945 | WORD_DESC *t; | |
1946 | char *current_word, *s; | |
28ef6c31 | 1947 | int sindex, sh_style_split, whitesep; |
95732b49 | 1948 | size_t slen; |
726f6388 JA |
1949 | |
1950 | if (!string || !*string) | |
1951 | return ((WORD_LIST *)NULL); | |
1952 | ||
7117c2d2 JA |
1953 | sh_style_split = separators && separators[0] == ' ' && |
1954 | separators[1] == '\t' && | |
1955 | separators[2] == '\n' && | |
1956 | separators[3] == '\0'; | |
726f6388 | 1957 | |
95732b49 | 1958 | slen = 0; |
726f6388 JA |
1959 | /* Remove sequences of whitespace at the beginning of STRING, as |
1960 | long as those characters appear in IFS. Do not do this if | |
1961 | STRING is quoted or if there are no separator characters. */ | |
1962 | if (!quoted || !separators || !*separators) | |
1963 | { | |
1964 | for (s = string; *s && spctabnl (*s) && issep (*s); s++); | |
1965 | ||
1966 | if (!*s) | |
1967 | return ((WORD_LIST *)NULL); | |
1968 | ||
1969 | string = s; | |
1970 | } | |
1971 | ||
1972 | /* OK, now STRING points to a word that does not begin with white space. | |
1973 | The splitting algorithm is: | |
7117c2d2 JA |
1974 | extract a word, stopping at a separator |
1975 | skip sequences of spc, tab, or nl as long as they are separators | |
726f6388 | 1976 | This obeys the field splitting rules in Posix.2. */ |
95732b49 | 1977 | slen = (MB_CUR_MAX > 1) ? strlen (string) : 1; |
ccc6cda3 | 1978 | for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; ) |
726f6388 | 1979 | { |
95732b49 JA |
1980 | /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim |
1981 | unless multibyte chars are possible. */ | |
1982 | current_word = string_extract_verbatim (string, slen, &sindex, separators); | |
ccc6cda3 | 1983 | if (current_word == 0) |
726f6388 JA |
1984 | break; |
1985 | ||
1986 | /* If we have a quoted empty string, add a quoted null argument. We | |
1987 | want to preserve the quoted null character iff this is a quoted | |
1988 | empty string; otherwise the quoted null characters are removed | |
1989 | below. */ | |
1990 | if (QUOTED_NULL (current_word)) | |
1991 | { | |
95732b49 | 1992 | t = alloc_word_desc (); |
726f6388 | 1993 | t->word = make_quoted_char ('\0'); |
95732b49 | 1994 | t->flags |= W_QUOTED|W_HASQUOTEDNULL; |
726f6388 JA |
1995 | result = make_word_list (t, result); |
1996 | } | |
ccc6cda3 | 1997 | else if (current_word[0] != '\0') |
726f6388 JA |
1998 | { |
1999 | /* If we have something, then add it regardless. However, | |
2000 | perform quoted null character removal on the current word. */ | |
2001 | remove_quoted_nulls (current_word); | |
cce855bc | 2002 | result = add_string_to_list (current_word, result); |
95732b49 | 2003 | result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */ |
ccc6cda3 JA |
2004 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) |
2005 | result->word->flags |= W_QUOTED; | |
726f6388 JA |
2006 | } |
2007 | ||
2008 | /* If we're not doing sequences of separators in the traditional | |
2009 | Bourne shell style, then add a quoted null argument. */ | |
726f6388 JA |
2010 | else if (!sh_style_split && !spctabnl (string[sindex])) |
2011 | { | |
95732b49 | 2012 | t = alloc_word_desc (); |
ccc6cda3 | 2013 | t->word = make_quoted_char ('\0'); |
95732b49 | 2014 | t->flags |= W_QUOTED|W_HASQUOTEDNULL; |
ccc6cda3 | 2015 | result = make_word_list (t, result); |
726f6388 JA |
2016 | } |
2017 | ||
2018 | free (current_word); | |
2019 | ||
28ef6c31 JA |
2020 | /* Note whether or not the separator is IFS whitespace, used later. */ |
2021 | whitesep = string[sindex] && spctabnl (string[sindex]); | |
2022 | ||
726f6388 JA |
2023 | /* Move past the current separator character. */ |
2024 | if (string[sindex]) | |
95732b49 JA |
2025 | { |
2026 | DECLARE_MBSTATE; | |
2027 | ADVANCE_CHAR (string, slen, sindex); | |
2028 | } | |
726f6388 JA |
2029 | |
2030 | /* Now skip sequences of space, tab, or newline characters if they are | |
2031 | in the list of separators. */ | |
2032 | while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex])) | |
2033 | sindex++; | |
28ef6c31 | 2034 | |
7117c2d2 JA |
2035 | /* If the first separator was IFS whitespace and the current character |
2036 | is a non-whitespace IFS character, it should be part of the current | |
2037 | field delimiter, not a separate delimiter that would result in an | |
2038 | empty field. Look at POSIX.2, 3.6.5, (3)(b). */ | |
28ef6c31 | 2039 | if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex])) |
95732b49 JA |
2040 | { |
2041 | sindex++; | |
2042 | /* An IFS character that is not IFS white space, along with any | |
2043 | adjacent IFS white space, shall delimit a field. (SUSv3) */ | |
0628567a | 2044 | while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex])) |
95732b49 JA |
2045 | sindex++; |
2046 | } | |
726f6388 JA |
2047 | } |
2048 | return (REVERSE_LIST (result, WORD_LIST *)); | |
2049 | } | |
2050 | ||
2051 | /* Parse a single word from STRING, using SEPARATORS to separate fields. | |
2052 | ENDPTR is set to the first character after the word. This is used by | |
7117c2d2 JA |
2053 | the `read' builtin. This is never called with SEPARATORS != $IFS; |
2054 | it should be simplified. | |
2055 | ||
726f6388 JA |
2056 | XXX - this function is very similar to list_string; they should be |
2057 | combined - XXX */ | |
2058 | char * | |
2059 | get_word_from_string (stringp, separators, endptr) | |
2060 | char **stringp, *separators, **endptr; | |
2061 | { | |
2062 | register char *s; | |
2063 | char *current_word; | |
28ef6c31 | 2064 | int sindex, sh_style_split, whitesep; |
95732b49 | 2065 | size_t slen; |
726f6388 JA |
2066 | |
2067 | if (!stringp || !*stringp || !**stringp) | |
2068 | return ((char *)NULL); | |
ccc6cda3 | 2069 | |
726f6388 JA |
2070 | s = *stringp; |
2071 | ||
7117c2d2 JA |
2072 | sh_style_split = separators && separators[0] == ' ' && |
2073 | separators[1] == '\t' && | |
2074 | separators[2] == '\n' && | |
2075 | separators[3] == '\0'; | |
726f6388 | 2076 | |
95732b49 JA |
2077 | slen = 0; |
2078 | ||
726f6388 JA |
2079 | /* Remove sequences of whitespace at the beginning of STRING, as |
2080 | long as those characters appear in IFS. */ | |
2081 | if (sh_style_split || !separators || !*separators) | |
2082 | { | |
7117c2d2 | 2083 | for (; *s && spctabnl (*s) && isifs (*s); s++); |
726f6388 JA |
2084 | |
2085 | /* If the string is nothing but whitespace, update it and return. */ | |
2086 | if (!*s) | |
2087 | { | |
2088 | *stringp = s; | |
2089 | if (endptr) | |
2090 | *endptr = s; | |
2091 | return ((char *)NULL); | |
2092 | } | |
2093 | } | |
2094 | ||
2095 | /* OK, S points to a word that does not begin with white space. | |
2096 | Now extract a word, stopping at a separator, save a pointer to | |
2097 | the first character after the word, then skip sequences of spc, | |
2098 | tab, or nl as long as they are separators. | |
ccc6cda3 | 2099 | |
726f6388 JA |
2100 | This obeys the field splitting rules in Posix.2. */ |
2101 | sindex = 0; | |
95732b49 JA |
2102 | /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim |
2103 | unless multibyte chars are possible. */ | |
2104 | slen = (MB_CUR_MAX > 1) ? strlen (s) : 1; | |
2105 | current_word = string_extract_verbatim (s, slen, &sindex, separators); | |
726f6388 JA |
2106 | |
2107 | /* Set ENDPTR to the first character after the end of the word. */ | |
2108 | if (endptr) | |
2109 | *endptr = s + sindex; | |
2110 | ||
28ef6c31 JA |
2111 | /* Note whether or not the separator is IFS whitespace, used later. */ |
2112 | whitesep = s[sindex] && spctabnl (s[sindex]); | |
2113 | ||
726f6388 JA |
2114 | /* Move past the current separator character. */ |
2115 | if (s[sindex]) | |
95732b49 JA |
2116 | { |
2117 | DECLARE_MBSTATE; | |
2118 | ADVANCE_CHAR (s, slen, sindex); | |
2119 | } | |
726f6388 JA |
2120 | |
2121 | /* Now skip sequences of space, tab, or newline characters if they are | |
2122 | in the list of separators. */ | |
7117c2d2 | 2123 | while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex])) |
726f6388 JA |
2124 | sindex++; |
2125 | ||
28ef6c31 JA |
2126 | /* If the first separator was IFS whitespace and the current character is |
2127 | a non-whitespace IFS character, it should be part of the current field | |
2128 | delimiter, not a separate delimiter that would result in an empty field. | |
2129 | Look at POSIX.2, 3.6.5, (3)(b). */ | |
7117c2d2 | 2130 | if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex])) |
95732b49 JA |
2131 | { |
2132 | sindex++; | |
2133 | /* An IFS character that is not IFS white space, along with any adjacent | |
2134 | IFS white space, shall delimit a field. */ | |
2135 | while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex])) | |
2136 | sindex++; | |
2137 | } | |
28ef6c31 | 2138 | |
726f6388 JA |
2139 | /* Update STRING to point to the next field. */ |
2140 | *stringp = s + sindex; | |
2141 | return (current_word); | |
2142 | } | |
2143 | ||
2144 | /* Remove IFS white space at the end of STRING. Start at the end | |
2145 | of the string and walk backwards until the beginning of the string | |
2146 | or we find a character that's not IFS white space and not CTLESC. | |
2147 | Only let CTLESC escape a white space character if SAW_ESCAPE is | |
2148 | non-zero. */ | |
2149 | char * | |
2150 | strip_trailing_ifs_whitespace (string, separators, saw_escape) | |
2151 | char *string, *separators; | |
2152 | int saw_escape; | |
2153 | { | |
2154 | char *s; | |
ccc6cda3 | 2155 | |
726f6388 | 2156 | s = string + STRLEN (string) - 1; |
7117c2d2 | 2157 | while (s > string && ((spctabnl (*s) && isifs (*s)) || |
726f6388 JA |
2158 | (saw_escape && *s == CTLESC && spctabnl (s[1])))) |
2159 | s--; | |
2160 | *++s = '\0'; | |
2161 | return string; | |
2162 | } | |
2163 | ||
bb70624e JA |
2164 | #if 0 |
2165 | /* UNUSED */ | |
2166 | /* Split STRING into words at whitespace. Obeys shell-style quoting with | |
2167 | backslashes, single and double quotes. */ | |
ccc6cda3 JA |
2168 | WORD_LIST * |
2169 | list_string_with_quotes (string) | |
2170 | char *string; | |
2171 | { | |
2172 | WORD_LIST *list; | |
2173 | char *token, *s; | |
7117c2d2 | 2174 | size_t s_len; |
ccc6cda3 JA |
2175 | int c, i, tokstart, len; |
2176 | ||
2177 | for (s = string; s && *s && spctabnl (*s); s++) | |
2178 | ; | |
2179 | if (s == 0 || *s == 0) | |
2180 | return ((WORD_LIST *)NULL); | |
2181 | ||
7117c2d2 | 2182 | s_len = strlen (s); |
ccc6cda3 JA |
2183 | tokstart = i = 0; |
2184 | list = (WORD_LIST *)NULL; | |
2185 | while (1) | |
2186 | { | |
2187 | c = s[i]; | |
2188 | if (c == '\\') | |
2189 | { | |
2190 | i++; | |
2191 | if (s[i]) | |
2192 | i++; | |
2193 | } | |
2194 | else if (c == '\'') | |
7117c2d2 | 2195 | i = skip_single_quoted (s, s_len, ++i); |
ccc6cda3 | 2196 | else if (c == '"') |
7117c2d2 | 2197 | i = skip_double_quoted (s, s_len, ++i); |
ccc6cda3 JA |
2198 | else if (c == 0 || spctabnl (c)) |
2199 | { | |
2200 | /* We have found the end of a token. Make a word out of it and | |
2201 | add it to the word list. */ | |
bb70624e | 2202 | token = substring (s, tokstart, i); |
cce855bc | 2203 | list = add_string_to_list (token, list); |
ccc6cda3 JA |
2204 | free (token); |
2205 | while (spctabnl (s[i])) | |
2206 | i++; | |
2207 | if (s[i]) | |
2208 | tokstart = i; | |
2209 | else | |
2210 | break; | |
2211 | } | |
2212 | else | |
2213 | i++; /* normal character */ | |
2214 | } | |
2215 | return (REVERSE_LIST (list, WORD_LIST *)); | |
2216 | } | |
bb70624e | 2217 | #endif |
d166f048 | 2218 | |
cce855bc JA |
2219 | /********************************************************/ |
2220 | /* */ | |
2221 | /* Functions to perform assignment statements */ | |
2222 | /* */ | |
2223 | /********************************************************/ | |
d166f048 | 2224 | |
95732b49 JA |
2225 | #if defined (ARRAY_VARS) |
2226 | static SHELL_VAR * | |
2227 | do_compound_assignment (name, value, flags) | |
2228 | char *name, *value; | |
2229 | int flags; | |
2230 | { | |
2231 | SHELL_VAR *v; | |
0628567a JA |
2232 | int mklocal; |
2233 | WORD_LIST *list; | |
95732b49 JA |
2234 | |
2235 | mklocal = flags & ASS_MKLOCAL; | |
2236 | ||
2237 | if (mklocal && variable_context) | |
2238 | { | |
0628567a | 2239 | list = expand_compound_array_assignment (value, flags); |
95732b49 | 2240 | v = find_variable (name); |
0628567a | 2241 | if (v == 0 || array_p (v) == 0 || v->context != variable_context) |
95732b49 | 2242 | v = make_local_array_variable (name); |
0628567a | 2243 | assign_compound_array_list (v, list, flags); |
95732b49 JA |
2244 | } |
2245 | else | |
2246 | v = assign_array_from_string (name, value, flags); | |
2247 | ||
2248 | return (v); | |
2249 | } | |
2250 | #endif | |
2251 | ||
726f6388 JA |
2252 | /* Given STRING, an assignment string, get the value of the right side |
2253 | of the `=', and bind it to the left side. If EXPAND is true, then | |
2254 | perform parameter expansion, command substitution, and arithmetic | |
2255 | expansion on the right-hand side. Perform tilde expansion in any | |
2256 | case. Do not perform word splitting on the result of expansion. */ | |
2257 | static int | |
95732b49 JA |
2258 | do_assignment_internal (word, expand) |
2259 | const WORD_DESC *word; | |
726f6388 JA |
2260 | int expand; |
2261 | { | |
95732b49 | 2262 | int offset, tlen, appendop, assign_list, aflags; |
0628567a | 2263 | char *name, *value; |
ccc6cda3 JA |
2264 | SHELL_VAR *entry; |
2265 | #if defined (ARRAY_VARS) | |
2266 | char *t; | |
b80f6443 | 2267 | int ni; |
ccc6cda3 | 2268 | #endif |
95732b49 | 2269 | const char *string; |
ccc6cda3 | 2270 | |
95732b49 JA |
2271 | if (word == 0 || word->word == 0) |
2272 | return 0; | |
2273 | ||
2274 | appendop = assign_list = aflags = 0; | |
2275 | string = word->word; | |
b80f6443 | 2276 | offset = assignment (string, 0); |
ccc6cda3 JA |
2277 | name = savestring (string); |
2278 | value = (char *)NULL; | |
726f6388 JA |
2279 | |
2280 | if (name[offset] == '=') | |
2281 | { | |
2282 | char *temp; | |
2283 | ||
95732b49 JA |
2284 | if (name[offset - 1] == '+') |
2285 | { | |
2286 | appendop = 1; | |
2287 | name[offset - 1] = '\0'; | |
2288 | } | |
2289 | ||
2290 | name[offset] = 0; /* might need this set later */ | |
726f6388 | 2291 | temp = name + offset + 1; |
95732b49 | 2292 | tlen = STRLEN (temp); |
726f6388 | 2293 | |
ccc6cda3 | 2294 | #if defined (ARRAY_VARS) |
95732b49 | 2295 | if (expand && (word->flags & W_COMPASSIGN)) |
726f6388 | 2296 | { |
ccc6cda3 | 2297 | assign_list = ni = 1; |
95732b49 | 2298 | value = extract_array_assignment_list (temp, &ni); |
ccc6cda3 JA |
2299 | } |
2300 | else | |
2301 | #endif | |
2302 | ||
ccc6cda3 | 2303 | if (expand && temp[0]) |
95732b49 | 2304 | value = expand_string_if_necessary (temp, 0, expand_string_assignment); |
726f6388 JA |
2305 | else |
2306 | value = savestring (temp); | |
2307 | } | |
2308 | ||
2309 | if (value == 0) | |
d166f048 | 2310 | { |
f73dda09 | 2311 | value = (char *)xmalloc (1); |
d166f048 JA |
2312 | value[0] = '\0'; |
2313 | } | |
726f6388 | 2314 | |
726f6388 | 2315 | if (echo_command_at_execute) |
95732b49 JA |
2316 | { |
2317 | if (appendop) | |
2318 | name[offset - 1] = '+'; | |
2319 | xtrace_print_assignment (name, value, assign_list, 1); | |
2320 | if (appendop) | |
2321 | name[offset - 1] = '\0'; | |
2322 | } | |
726f6388 | 2323 | |
d166f048 | 2324 | #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0) |
ccc6cda3 | 2325 | |
95732b49 JA |
2326 | if (appendop) |
2327 | aflags |= ASS_APPEND; | |
2328 | ||
ccc6cda3 | 2329 | #if defined (ARRAY_VARS) |
7117c2d2 | 2330 | if (t = xstrchr (name, '[')) /*]*/ |
ccc6cda3 JA |
2331 | { |
2332 | if (assign_list) | |
2333 | { | |
b80f6443 | 2334 | report_error (_("%s: cannot assign list to array member"), name); |
ccc6cda3 JA |
2335 | ASSIGN_RETURN (0); |
2336 | } | |
95732b49 | 2337 | entry = assign_array_element (name, value, aflags); |
ccc6cda3 | 2338 | if (entry == 0) |
28ef6c31 | 2339 | ASSIGN_RETURN (0); |
ccc6cda3 JA |
2340 | } |
2341 | else if (assign_list) | |
95732b49 JA |
2342 | { |
2343 | if (word->flags & W_ASSIGNARG) | |
2344 | aflags |= ASS_MKLOCAL; | |
2345 | entry = do_compound_assignment (name, value, aflags); | |
2346 | } | |
ccc6cda3 JA |
2347 | else |
2348 | #endif /* ARRAY_VARS */ | |
95732b49 | 2349 | entry = bind_variable (name, value, aflags); |
ccc6cda3 | 2350 | |
726f6388 JA |
2351 | stupidly_hack_special_variables (name); |
2352 | ||
2353 | if (entry) | |
bb70624e | 2354 | VUNSETATTR (entry, att_invisible); |
726f6388 | 2355 | |
726f6388 | 2356 | /* Return 1 if the assignment seems to have been performed correctly. */ |
28ef6c31 | 2357 | ASSIGN_RETURN (entry ? ((readonly_p (entry) == 0) && noassign_p (entry) == 0) : 0); |
726f6388 JA |
2358 | } |
2359 | ||
2360 | /* Perform the assignment statement in STRING, and expand the | |
95732b49 | 2361 | right side by doing tilde, command and parameter expansion. */ |
ccc6cda3 | 2362 | int |
726f6388 | 2363 | do_assignment (string) |
95732b49 | 2364 | char *string; |
726f6388 | 2365 | { |
95732b49 JA |
2366 | WORD_DESC td; |
2367 | ||
2368 | td.flags = W_ASSIGNMENT; | |
2369 | td.word = string; | |
2370 | ||
2371 | return do_assignment_internal (&td, 1); | |
2372 | } | |
2373 | ||
2374 | int | |
2375 | do_word_assignment (word) | |
2376 | WORD_DESC *word; | |
2377 | { | |
2378 | return do_assignment_internal (word, 1); | |
726f6388 JA |
2379 | } |
2380 | ||
2381 | /* Given STRING, an assignment string, get the value of the right side | |
95732b49 JA |
2382 | of the `=', and bind it to the left side. Do not perform any word |
2383 | expansions on the right hand side. */ | |
ccc6cda3 | 2384 | int |
726f6388 | 2385 | do_assignment_no_expand (string) |
95732b49 | 2386 | char *string; |
726f6388 | 2387 | { |
95732b49 JA |
2388 | WORD_DESC td; |
2389 | ||
2390 | td.flags = W_ASSIGNMENT; | |
2391 | td.word = string; | |
2392 | ||
2393 | return (do_assignment_internal (&td, 0)); | |
726f6388 JA |
2394 | } |
2395 | ||
cce855bc JA |
2396 | /*************************************************** |
2397 | * * | |
2398 | * Functions to manage the positional parameters * | |
2399 | * * | |
2400 | ***************************************************/ | |
726f6388 JA |
2401 | |
2402 | /* Return the word list that corresponds to `$*'. */ | |
2403 | WORD_LIST * | |
2404 | list_rest_of_args () | |
2405 | { | |
ccc6cda3 | 2406 | register WORD_LIST *list, *args; |
726f6388 JA |
2407 | int i; |
2408 | ||
2409 | /* Break out of the loop as soon as one of the dollar variables is null. */ | |
ccc6cda3 JA |
2410 | for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++) |
2411 | list = make_word_list (make_bare_word (dollar_vars[i]), list); | |
2412 | ||
2413 | for (args = rest_of_args; args; args = args->next) | |
2414 | list = make_word_list (make_bare_word (args->word->word), list); | |
726f6388 | 2415 | |
726f6388 JA |
2416 | return (REVERSE_LIST (list, WORD_LIST *)); |
2417 | } | |
2418 | ||
ccc6cda3 JA |
2419 | int |
2420 | number_of_args () | |
2421 | { | |
2422 | register WORD_LIST *list; | |
2423 | int n; | |
2424 | ||
2425 | for (n = 0; n < 9 && dollar_vars[n+1]; n++) | |
2426 | ; | |
2427 | for (list = rest_of_args; list; list = list->next) | |
2428 | n++; | |
2429 | return n; | |
2430 | } | |
2431 | ||
cce855bc JA |
2432 | /* Return the value of a positional parameter. This handles values > 10. */ |
2433 | char * | |
2434 | get_dollar_var_value (ind) | |
7117c2d2 | 2435 | intmax_t ind; |
cce855bc JA |
2436 | { |
2437 | char *temp; | |
2438 | WORD_LIST *p; | |
2439 | ||
2440 | if (ind < 10) | |
2441 | temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL; | |
2442 | else /* We want something like ${11} */ | |
2443 | { | |
2444 | ind -= 10; | |
2445 | for (p = rest_of_args; p && ind--; p = p->next) | |
28ef6c31 | 2446 | ; |
cce855bc JA |
2447 | temp = p ? savestring (p->word->word) : (char *)NULL; |
2448 | } | |
2449 | return (temp); | |
2450 | } | |
2451 | ||
726f6388 JA |
2452 | /* Make a single large string out of the dollar digit variables, |
2453 | and the rest_of_args. If DOLLAR_STAR is 1, then obey the special | |
2454 | case of "$*" with respect to IFS. */ | |
2455 | char * | |
2456 | string_rest_of_args (dollar_star) | |
2457 | int dollar_star; | |
2458 | { | |
ccc6cda3 | 2459 | register WORD_LIST *list; |
726f6388 JA |
2460 | char *string; |
2461 | ||
ccc6cda3 | 2462 | list = list_rest_of_args (); |
726f6388 JA |
2463 | string = dollar_star ? string_list_dollar_star (list) : string_list (list); |
2464 | dispose_words (list); | |
2465 | return (string); | |
2466 | } | |
2467 | ||
cce855bc JA |
2468 | /* Return a string containing the positional parameters from START to |
2469 | END, inclusive. If STRING[0] == '*', we obey the rules for $*, | |
7117c2d2 JA |
2470 | which only makes a difference if QUOTED is non-zero. If QUOTED includes |
2471 | Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise | |
2472 | no quoting chars are added. */ | |
cce855bc JA |
2473 | static char * |
2474 | pos_params (string, start, end, quoted) | |
2475 | char *string; | |
2476 | int start, end, quoted; | |
726f6388 | 2477 | { |
cce855bc JA |
2478 | WORD_LIST *save, *params, *h, *t; |
2479 | char *ret; | |
2480 | int i; | |
726f6388 | 2481 | |
bb70624e JA |
2482 | /* see if we can short-circuit. if start == end, we want 0 parameters. */ |
2483 | if (start == end) | |
2484 | return ((char *)NULL); | |
2485 | ||
cce855bc JA |
2486 | save = params = list_rest_of_args (); |
2487 | if (save == 0) | |
2488 | return ((char *)NULL); | |
2489 | ||
2490 | for (i = 1; params && i < start; i++) | |
2491 | params = params->next; | |
2492 | if (params == 0) | |
2493 | return ((char *)NULL); | |
2494 | for (h = t = params; params && i < end; i++) | |
d166f048 | 2495 | { |
cce855bc JA |
2496 | t = params; |
2497 | params = params->next; | |
d166f048 | 2498 | } |
726f6388 | 2499 | |
cce855bc JA |
2500 | t->next = (WORD_LIST *)NULL; |
2501 | if (string[0] == '*') | |
b80f6443 JA |
2502 | { |
2503 | if (quoted & Q_DOUBLE_QUOTES) | |
2504 | ret = string_list_dollar_star (quote_list (h)); | |
2505 | else if (quoted & Q_HERE_DOCUMENT) | |
2506 | ret = string_list (quote_list (h)); | |
2507 | else | |
2508 | ret = string_list (h); | |
2509 | } | |
cce855bc JA |
2510 | else |
2511 | ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (h) : h); | |
bb70624e JA |
2512 | if (t != params) |
2513 | t->next = params; | |
726f6388 | 2514 | |
cce855bc JA |
2515 | dispose_words (save); |
2516 | return (ret); | |
2517 | } | |
2518 | ||
2519 | /******************************************************************/ | |
2520 | /* */ | |
2521 | /* Functions to expand strings to strings or WORD_LISTs */ | |
2522 | /* */ | |
2523 | /******************************************************************/ | |
2524 | ||
2525 | #if defined (PROCESS_SUBSTITUTION) | |
95732b49 | 2526 | #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~') |
cce855bc | 2527 | #else |
95732b49 | 2528 | #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~') |
cce855bc JA |
2529 | #endif |
2530 | ||
2531 | /* If there are any characters in STRING that require full expansion, | |
2532 | then call FUNC to expand STRING; otherwise just perform quote | |
2533 | removal if necessary. This returns a new string. */ | |
2534 | static char * | |
f73dda09 | 2535 | expand_string_if_necessary (string, quoted, func) |
cce855bc JA |
2536 | char *string; |
2537 | int quoted; | |
f73dda09 | 2538 | EXPFUNC *func; |
cce855bc JA |
2539 | { |
2540 | WORD_LIST *list; | |
7117c2d2 | 2541 | size_t slen; |
cce855bc JA |
2542 | int i, saw_quote; |
2543 | char *ret; | |
7117c2d2 | 2544 | DECLARE_MBSTATE; |
cce855bc | 2545 | |
95732b49 JA |
2546 | /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */ |
2547 | slen = (MB_CUR_MAX > 1) ? strlen (string) : 0; | |
7117c2d2 JA |
2548 | i = saw_quote = 0; |
2549 | while (string[i]) | |
cce855bc JA |
2550 | { |
2551 | if (EXP_CHAR (string[i])) | |
2552 | break; | |
2553 | else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"') | |
2554 | saw_quote = 1; | |
7117c2d2 | 2555 | ADVANCE_CHAR (string, slen, i); |
cce855bc JA |
2556 | } |
2557 | ||
2558 | if (string[i]) | |
2559 | { | |
2560 | list = (*func) (string, quoted); | |
2561 | if (list) | |
2562 | { | |
2563 | ret = string_list (list); | |
2564 | dispose_words (list); | |
2565 | } | |
2566 | else | |
2567 | ret = (char *)NULL; | |
2568 | } | |
2569 | else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) | |
2570 | ret = string_quote_removal (string, quoted); | |
2571 | else | |
2572 | ret = savestring (string); | |
7117c2d2 | 2573 | |
cce855bc JA |
2574 | return ret; |
2575 | } | |
2576 | ||
2577 | static inline char * | |
f73dda09 | 2578 | expand_string_to_string_internal (string, quoted, func) |
cce855bc JA |
2579 | char *string; |
2580 | int quoted; | |
f73dda09 | 2581 | EXPFUNC *func; |
cce855bc JA |
2582 | { |
2583 | WORD_LIST *list; | |
2584 | char *ret; | |
2585 | ||
2586 | if (string == 0 || *string == '\0') | |
2587 | return ((char *)NULL); | |
2588 | ||
2589 | list = (*func) (string, quoted); | |
2590 | if (list) | |
2591 | { | |
2592 | ret = string_list (list); | |
2593 | dispose_words (list); | |
2594 | } | |
2595 | else | |
2596 | ret = (char *)NULL; | |
2597 | ||
2598 | return (ret); | |
2599 | } | |
2600 | ||
f73dda09 JA |
2601 | char * |
2602 | expand_string_to_string (string, quoted) | |
2603 | char *string; | |
2604 | int quoted; | |
2605 | { | |
2606 | return (expand_string_to_string_internal (string, quoted, expand_string)); | |
2607 | } | |
2608 | ||
2609 | char * | |
2610 | expand_string_unsplit_to_string (string, quoted) | |
2611 | char *string; | |
2612 | int quoted; | |
2613 | { | |
2614 | return (expand_string_to_string_internal (string, quoted, expand_string_unsplit)); | |
2615 | } | |
2616 | ||
95732b49 JA |
2617 | char * |
2618 | expand_assignment_string_to_string (string, quoted) | |
2619 | char *string; | |
2620 | int quoted; | |
2621 | { | |
2622 | return (expand_string_to_string_internal (string, quoted, expand_string_assignment)); | |
2623 | } | |
2624 | ||
0628567a JA |
2625 | char * |
2626 | expand_arith_string (string, quoted) | |
2627 | char *string; | |
2628 | { | |
2629 | return (expand_string_if_necessary (string, quoted, expand_string)); | |
2630 | } | |
2631 | ||
cce855bc JA |
2632 | #if defined (COND_COMMAND) |
2633 | /* Just remove backslashes in STRING. Returns a new string. */ | |
2634 | char * | |
2635 | remove_backslashes (string) | |
2636 | char *string; | |
2637 | { | |
2638 | char *r, *ret, *s; | |
2639 | ||
f73dda09 | 2640 | r = ret = (char *)xmalloc (strlen (string) + 1); |
cce855bc JA |
2641 | for (s = string; s && *s; ) |
2642 | { | |
2643 | if (*s == '\\') | |
28ef6c31 | 2644 | s++; |
cce855bc | 2645 | if (*s == 0) |
28ef6c31 | 2646 | break; |
cce855bc JA |
2647 | *r++ = *s++; |
2648 | } | |
2649 | *r = '\0'; | |
2650 | return ret; | |
2651 | } | |
2652 | ||
2653 | /* This needs better error handling. */ | |
2654 | /* Expand W for use as an argument to a unary or binary operator in a | |
f1be666c | 2655 | [[...]] expression. If SPECIAL is 1, this is the rhs argument |
cce855bc | 2656 | to the != or == operator, and should be treated as a pattern. In |
f1be666c JA |
2657 | this case, we quote the string specially for the globbing code. If |
2658 | SPECIAL is 2, this is an rhs argument for the =~ operator, and should | |
2659 | be quoted appropriately for regcomp/regexec. The caller is responsible | |
2660 | for removing the backslashes if the unquoted word is needed later. */ | |
cce855bc JA |
2661 | char * |
2662 | cond_expand_word (w, special) | |
2663 | WORD_DESC *w; | |
2664 | int special; | |
2665 | { | |
2666 | char *r, *p; | |
2667 | WORD_LIST *l; | |
f1be666c | 2668 | int qflags; |
cce855bc JA |
2669 | |
2670 | if (w->word == 0 || w->word[0] == '\0') | |
2671 | return ((char *)NULL); | |
2672 | ||
b72432fd | 2673 | l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0); |
cce855bc JA |
2674 | if (l) |
2675 | { | |
2676 | if (special == 0) | |
2677 | { | |
2678 | dequote_list (l); | |
2679 | r = string_list (l); | |
2680 | } | |
2681 | else | |
28ef6c31 | 2682 | { |
f1be666c JA |
2683 | qflags = QGLOB_CVTNULL; |
2684 | if (special == 2) | |
2685 | qflags |= QGLOB_REGEXP; | |
28ef6c31 | 2686 | p = string_list (l); |
f1be666c | 2687 | r = quote_string_for_globbing (p, qflags); |
28ef6c31 JA |
2688 | free (p); |
2689 | } | |
cce855bc JA |
2690 | dispose_words (l); |
2691 | } | |
2692 | else | |
2693 | r = (char *)NULL; | |
2694 | ||
2695 | return r; | |
2696 | } | |
2697 | #endif | |
2698 | ||
2699 | /* Call expand_word_internal to expand W and handle error returns. | |
2700 | A convenience function for functions that don't want to handle | |
2701 | any errors or free any memory before aborting. */ | |
2702 | static WORD_LIST * | |
b72432fd | 2703 | call_expand_word_internal (w, q, i, c, e) |
cce855bc | 2704 | WORD_DESC *w; |
b72432fd | 2705 | int q, i, *c, *e; |
cce855bc JA |
2706 | { |
2707 | WORD_LIST *result; | |
2708 | ||
b72432fd | 2709 | result = expand_word_internal (w, q, i, c, e); |
bb70624e | 2710 | if (result == &expand_word_error || result == &expand_word_fatal) |
cce855bc JA |
2711 | { |
2712 | /* By convention, each time this error is returned, w->word has | |
bb70624e JA |
2713 | already been freed (it sometimes may not be in the fatal case, |
2714 | but that doesn't result in a memory leak because we're going | |
2715 | to exit in most cases). */ | |
cce855bc | 2716 | w->word = (char *)NULL; |
28ef6c31 | 2717 | last_command_exit_value = EXECUTION_FAILURE; |
b80f6443 | 2718 | exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF); |
cce855bc JA |
2719 | /* NOTREACHED */ |
2720 | } | |
cce855bc JA |
2721 | else |
2722 | return (result); | |
2723 | } | |
2724 | ||
2725 | /* Perform parameter expansion, command substitution, and arithmetic | |
2726 | expansion on STRING, as if it were a word. Leave the result quoted. */ | |
2727 | static WORD_LIST * | |
2728 | expand_string_internal (string, quoted) | |
2729 | char *string; | |
2730 | int quoted; | |
2731 | { | |
2732 | WORD_DESC td; | |
2733 | WORD_LIST *tresult; | |
2734 | ||
2735 | if (string == 0 || *string == 0) | |
2736 | return ((WORD_LIST *)NULL); | |
2737 | ||
28ef6c31 JA |
2738 | td.flags = 0; |
2739 | td.word = savestring (string); | |
2740 | ||
b72432fd | 2741 | tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); |
28ef6c31 JA |
2742 | |
2743 | FREE (td.word); | |
cce855bc | 2744 | return (tresult); |
726f6388 JA |
2745 | } |
2746 | ||
2747 | /* Expand STRING by performing parameter expansion, command substitution, | |
2748 | and arithmetic expansion. Dequote the resulting WORD_LIST before | |
2749 | returning it, but do not perform word splitting. The call to | |
2750 | remove_quoted_nulls () is in here because word splitting normally | |
2751 | takes care of quote removal. */ | |
2752 | WORD_LIST * | |
2753 | expand_string_unsplit (string, quoted) | |
2754 | char *string; | |
2755 | int quoted; | |
2756 | { | |
2757 | WORD_LIST *value; | |
2758 | ||
28ef6c31 | 2759 | if (string == 0 || *string == '\0') |
726f6388 JA |
2760 | return ((WORD_LIST *)NULL); |
2761 | ||
28ef6c31 | 2762 | expand_no_split_dollar_star = 1; |
726f6388 | 2763 | value = expand_string_internal (string, quoted); |
28ef6c31 JA |
2764 | expand_no_split_dollar_star = 0; |
2765 | ||
726f6388 JA |
2766 | if (value) |
2767 | { | |
2768 | if (value->word) | |
95732b49 JA |
2769 | { |
2770 | remove_quoted_nulls (value->word->word); | |
2771 | value->word->flags &= ~W_HASQUOTEDNULL; | |
2772 | } | |
2773 | dequote_list (value); | |
2774 | } | |
2775 | return (value); | |
2776 | } | |
2777 | ||
2778 | /* Expand the rhs of an assignment statement */ | |
2779 | WORD_LIST * | |
2780 | expand_string_assignment (string, quoted) | |
2781 | char *string; | |
2782 | int quoted; | |
2783 | { | |
2784 | WORD_DESC td; | |
2785 | WORD_LIST *value; | |
2786 | ||
2787 | if (string == 0 || *string == '\0') | |
2788 | return ((WORD_LIST *)NULL); | |
2789 | ||
2790 | expand_no_split_dollar_star = 1; | |
2791 | ||
2792 | td.flags = W_ASSIGNRHS; | |
2793 | td.word = savestring (string); | |
2794 | value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); | |
2795 | FREE (td.word); | |
2796 | ||
2797 | expand_no_split_dollar_star = 0; | |
2798 | ||
2799 | if (value) | |
2800 | { | |
2801 | if (value->word) | |
2802 | { | |
2803 | remove_quoted_nulls (value->word->word); | |
2804 | value->word->flags &= ~W_HASQUOTEDNULL; | |
2805 | } | |
726f6388 JA |
2806 | dequote_list (value); |
2807 | } | |
2808 | return (value); | |
2809 | } | |
2810 | ||
bb70624e JA |
2811 | |
2812 | /* Expand one of the PS? prompt strings. This is a sort of combination of | |
2813 | expand_string_unsplit and expand_string_internal, but returns the | |
2814 | passed string when an error occurs. Might want to trap other calls | |
2815 | to jump_to_top_level here so we don't endlessly loop. */ | |
2816 | WORD_LIST * | |
f1be666c | 2817 | expand_prompt_string (string, quoted, wflags) |
bb70624e JA |
2818 | char *string; |
2819 | int quoted; | |
f1be666c | 2820 | int wflags; |
bb70624e JA |
2821 | { |
2822 | WORD_LIST *value; | |
2823 | WORD_DESC td; | |
2824 | ||
2825 | if (string == 0 || *string == 0) | |
2826 | return ((WORD_LIST *)NULL); | |
2827 | ||
f1be666c | 2828 | td.flags = wflags; |
bb70624e | 2829 | td.word = savestring (string); |
28ef6c31 JA |
2830 | |
2831 | no_longjmp_on_fatal_error = 1; | |
bb70624e | 2832 | value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); |
28ef6c31 JA |
2833 | no_longjmp_on_fatal_error = 0; |
2834 | ||
bb70624e JA |
2835 | if (value == &expand_word_error || value == &expand_word_fatal) |
2836 | { | |
2837 | value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL); | |
2838 | return value; | |
2839 | } | |
2840 | FREE (td.word); | |
2841 | if (value) | |
2842 | { | |
2843 | if (value->word) | |
95732b49 JA |
2844 | { |
2845 | remove_quoted_nulls (value->word->word); | |
2846 | value->word->flags &= ~W_HASQUOTEDNULL; | |
2847 | } | |
bb70624e JA |
2848 | dequote_list (value); |
2849 | } | |
2850 | return (value); | |
2851 | } | |
2852 | ||
726f6388 JA |
2853 | /* Expand STRING just as if you were expanding a word, but do not dequote |
2854 | the resultant WORD_LIST. This is called only from within this file, | |
2855 | and is used to correctly preserve quoted characters when expanding | |
2856 | things like ${1+"$@"}. This does parameter expansion, command | |
b72432fd | 2857 | substitution, arithmetic expansion, and word splitting. */ |
726f6388 JA |
2858 | static WORD_LIST * |
2859 | expand_string_leave_quoted (string, quoted) | |
2860 | char *string; | |
2861 | int quoted; | |
2862 | { | |
2863 | WORD_LIST *tlist; | |
2864 | WORD_LIST *tresult; | |
2865 | ||
ccc6cda3 | 2866 | if (string == 0 || *string == '\0') |
726f6388 JA |
2867 | return ((WORD_LIST *)NULL); |
2868 | ||
2869 | tlist = expand_string_internal (string, quoted); | |
2870 | ||
2871 | if (tlist) | |
2872 | { | |
2873 | tresult = word_list_split (tlist); | |
2874 | dispose_words (tlist); | |
2875 | return (tresult); | |
2876 | } | |
2877 | return ((WORD_LIST *)NULL); | |
2878 | } | |
2879 | ||
ccc6cda3 JA |
2880 | /* This does not perform word splitting or dequote the WORD_LIST |
2881 | it returns. */ | |
2882 | static WORD_LIST * | |
2883 | expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at) | |
2884 | char *string; | |
2885 | int quoted, *dollar_at_p, *has_dollar_at; | |
2886 | { | |
2887 | WORD_DESC td; | |
2888 | WORD_LIST *tresult; | |
2889 | ||
2890 | if (string == 0 || *string == '\0') | |
2891 | return (WORD_LIST *)NULL; | |
2892 | ||
28ef6c31 | 2893 | td.flags = 0; |
ccc6cda3 | 2894 | td.word = string; |
b72432fd | 2895 | tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at); |
ccc6cda3 JA |
2896 | return (tresult); |
2897 | } | |
2898 | ||
726f6388 JA |
2899 | /* Expand STRING just as if you were expanding a word. This also returns |
2900 | a list of words. Note that filename globbing is *NOT* done for word | |
2901 | or string expansion, just when the shell is expanding a command. This | |
2902 | does parameter expansion, command substitution, arithmetic expansion, | |
2903 | and word splitting. Dequote the resultant WORD_LIST before returning. */ | |
2904 | WORD_LIST * | |
2905 | expand_string (string, quoted) | |
2906 | char *string; | |
2907 | int quoted; | |
2908 | { | |
2909 | WORD_LIST *result; | |
2910 | ||
28ef6c31 | 2911 | if (string == 0 || *string == '\0') |
726f6388 JA |
2912 | return ((WORD_LIST *)NULL); |
2913 | ||
2914 | result = expand_string_leave_quoted (string, quoted); | |
ccc6cda3 | 2915 | return (result ? dequote_list (result) : result); |
726f6388 JA |
2916 | } |
2917 | ||
2918 | /*************************************************** | |
2919 | * * | |
2920 | * Functions to handle quoting chars * | |
2921 | * * | |
2922 | ***************************************************/ | |
2923 | ||
cce855bc JA |
2924 | /* Conventions: |
2925 | ||
2926 | A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string. | |
2927 | The parser passes CTLNUL as CTLESC CTLNUL. */ | |
2928 | ||
cce855bc JA |
2929 | /* Quote escape characters in string s, but no other characters. This is |
2930 | used to protect CTLESC and CTLNUL in variable values from the rest of | |
f1be666c JA |
2931 | the word expansion process after the variable is expanded. If IFS is |
2932 | null, we quote spaces as well, just in case we split on spaces later | |
2933 | (in the case of unquoted $@, we will eventually attempt to split the | |
2934 | entire word on spaces). Corresponding code exists in dequote_escapes. | |
2935 | Even if we don't end up splitting on spaces, quoting spaces is not a | |
2936 | problem. */ | |
f73dda09 | 2937 | char * |
cce855bc JA |
2938 | quote_escapes (string) |
2939 | char *string; | |
2940 | { | |
2941 | register char *s, *t; | |
7117c2d2 JA |
2942 | size_t slen; |
2943 | char *result, *send; | |
f1be666c | 2944 | int quote_spaces; |
7117c2d2 | 2945 | DECLARE_MBSTATE; |
cce855bc | 2946 | |
7117c2d2 JA |
2947 | slen = strlen (string); |
2948 | send = string + slen; | |
2949 | ||
f1be666c | 2950 | quote_spaces = (ifs_value && *ifs_value == 0); |
7117c2d2 JA |
2951 | t = result = (char *)xmalloc ((slen * 2) + 1); |
2952 | s = string; | |
2953 | ||
2954 | while (*s) | |
cce855bc | 2955 | { |
f1be666c | 2956 | if (*s == CTLESC || *s == CTLNUL || (quote_spaces && *s == ' ')) |
cce855bc | 2957 | *t++ = CTLESC; |
7117c2d2 | 2958 | COPY_CHAR_P (t, s, send); |
cce855bc JA |
2959 | } |
2960 | *t = '\0'; | |
2961 | return (result); | |
2962 | } | |
2963 | ||
2964 | static WORD_LIST * | |
2965 | list_quote_escapes (list) | |
2966 | WORD_LIST *list; | |
2967 | { | |
2968 | register WORD_LIST *w; | |
2969 | char *t; | |
2970 | ||
2971 | for (w = list; w; w = w->next) | |
2972 | { | |
2973 | t = w->word->word; | |
2974 | w->word->word = quote_escapes (t); | |
2975 | free (t); | |
2976 | } | |
2977 | return list; | |
2978 | } | |
2979 | ||
7117c2d2 JA |
2980 | /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL. |
2981 | ||
2982 | The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL. | |
2983 | This is necessary to make unquoted CTLESC and CTLNUL characters in the | |
2984 | data stream pass through properly. | |
2985 | ||
2986 | We need to remove doubled CTLESC characters inside quoted strings before | |
2987 | quoting the entire string, so we do not double the number of CTLESC | |
2988 | characters. | |
2989 | ||
2990 | Also used by parts of the pattern substitution code. */ | |
cce855bc JA |
2991 | static char * |
2992 | dequote_escapes (string) | |
2993 | char *string; | |
2994 | { | |
2995 | register char *s, *t; | |
7117c2d2 JA |
2996 | size_t slen; |
2997 | char *result, *send; | |
f1be666c | 2998 | int quote_spaces; |
7117c2d2 | 2999 | DECLARE_MBSTATE; |
cce855bc | 3000 | |
7117c2d2 JA |
3001 | if (string == 0) |
3002 | return string; | |
3003 | ||
3004 | slen = strlen (string); | |
3005 | send = string + slen; | |
3006 | ||
3007 | t = result = (char *)xmalloc (slen + 1); | |
3008 | s = string; | |
3009 | ||
3010 | if (strchr (string, CTLESC) == 0) | |
3011 | return (strcpy (result, s)); | |
3012 | ||
f1be666c | 3013 | quote_spaces = (ifs_value && *ifs_value == 0); |
7117c2d2 | 3014 | while (*s) |
cce855bc | 3015 | { |
f1be666c | 3016 | if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' '))) |
cce855bc JA |
3017 | { |
3018 | s++; | |
3019 | if (*s == '\0') | |
3020 | break; | |
3021 | } | |
7117c2d2 | 3022 | COPY_CHAR_P (t, s, send); |
cce855bc JA |
3023 | } |
3024 | *t = '\0'; | |
3025 | return result; | |
3026 | } | |
726f6388 | 3027 | |
0628567a JA |
3028 | /* Return a new string with the quoted representation of character C. |
3029 | This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be | |
3030 | set in any resultant WORD_DESC where this value is the word. */ | |
726f6388 JA |
3031 | static char * |
3032 | make_quoted_char (c) | |
3033 | int c; | |
3034 | { | |
3035 | char *temp; | |
3036 | ||
f73dda09 | 3037 | temp = (char *)xmalloc (3); |
726f6388 JA |
3038 | if (c == 0) |
3039 | { | |
3040 | temp[0] = CTLNUL; | |
3041 | temp[1] = '\0'; | |
3042 | } | |
3043 | else | |
3044 | { | |
3045 | temp[0] = CTLESC; | |
3046 | temp[1] = c; | |
3047 | temp[2] = '\0'; | |
3048 | } | |
3049 | return (temp); | |
3050 | } | |
3051 | ||
0628567a JA |
3052 | /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so |
3053 | the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where | |
3054 | this value is the word. */ | |
ccc6cda3 | 3055 | char * |
726f6388 JA |
3056 | quote_string (string) |
3057 | char *string; | |
3058 | { | |
ccc6cda3 | 3059 | register char *t; |
7117c2d2 JA |
3060 | size_t slen; |
3061 | char *result, *send; | |
726f6388 | 3062 | |
ccc6cda3 | 3063 | if (*string == 0) |
726f6388 | 3064 | { |
f73dda09 | 3065 | result = (char *)xmalloc (2); |
726f6388 JA |
3066 | result[0] = CTLNUL; |
3067 | result[1] = '\0'; | |
3068 | } | |
3069 | else | |
3070 | { | |
7117c2d2 | 3071 | DECLARE_MBSTATE; |
726f6388 | 3072 | |
7117c2d2 JA |
3073 | slen = strlen (string); |
3074 | send = string + slen; | |
3075 | ||
3076 | result = (char *)xmalloc ((slen * 2) + 1); | |
3077 | ||
3078 | for (t = result; string < send; ) | |
726f6388 JA |
3079 | { |
3080 | *t++ = CTLESC; | |
7117c2d2 | 3081 | COPY_CHAR_P (t, string, send); |
726f6388 JA |
3082 | } |
3083 | *t = '\0'; | |
3084 | } | |
3085 | return (result); | |
3086 | } | |
3087 | ||
0628567a | 3088 | /* De-quote quoted characters in STRING. */ |
726f6388 JA |
3089 | char * |
3090 | dequote_string (string) | |
3091 | char *string; | |
3092 | { | |
7117c2d2 JA |
3093 | register char *s, *t; |
3094 | size_t slen; | |
3095 | char *result, *send; | |
3096 | DECLARE_MBSTATE; | |
726f6388 | 3097 | |
7117c2d2 JA |
3098 | slen = strlen (string); |
3099 | ||
3100 | t = result = (char *)xmalloc (slen + 1); | |
726f6388 JA |
3101 | |
3102 | if (QUOTED_NULL (string)) | |
3103 | { | |
3104 | result[0] = '\0'; | |
3105 | return (result); | |
3106 | } | |
3107 | ||
3108 | /* If no character in the string can be quoted, don't bother examining | |
3109 | each character. Just return a copy of the string passed to us. */ | |
7117c2d2 JA |
3110 | if (strchr (string, CTLESC) == NULL) |
3111 | return (strcpy (result, string)); | |
726f6388 | 3112 | |
7117c2d2 JA |
3113 | send = string + slen; |
3114 | s = string; | |
3115 | while (*s) | |
726f6388 | 3116 | { |
7117c2d2 | 3117 | if (*s == CTLESC) |
726f6388 | 3118 | { |
7117c2d2 JA |
3119 | s++; |
3120 | if (*s == '\0') | |
726f6388 JA |
3121 | break; |
3122 | } | |
7117c2d2 | 3123 | COPY_CHAR_P (t, s, send); |
726f6388 JA |
3124 | } |
3125 | ||
3126 | *t = '\0'; | |
3127 | return (result); | |
3128 | } | |
3129 | ||
3130 | /* Quote the entire WORD_LIST list. */ | |
ccc6cda3 | 3131 | static WORD_LIST * |
726f6388 JA |
3132 | quote_list (list) |
3133 | WORD_LIST *list; | |
3134 | { | |
3135 | register WORD_LIST *w; | |
ccc6cda3 | 3136 | char *t; |
726f6388 JA |
3137 | |
3138 | for (w = list; w; w = w->next) | |
3139 | { | |
ccc6cda3 | 3140 | t = w->word->word; |
726f6388 JA |
3141 | w->word->word = quote_string (t); |
3142 | free (t); | |
ccc6cda3 | 3143 | w->word->flags |= W_QUOTED; |
0628567a | 3144 | /* XXX - turn on W_HAVEQUOTEDNULL here? */ |
726f6388 | 3145 | } |
ccc6cda3 | 3146 | return list; |
726f6388 JA |
3147 | } |
3148 | ||
0628567a JA |
3149 | /* De-quote quoted characters in each word in LIST. */ |
3150 | WORD_LIST * | |
7117c2d2 JA |
3151 | dequote_list (list) |
3152 | WORD_LIST *list; | |
3153 | { | |
3154 | register char *s; | |
3155 | register WORD_LIST *tlist; | |
3156 | ||
3157 | for (tlist = list; tlist; tlist = tlist->next) | |
3158 | { | |
3159 | s = dequote_string (tlist->word->word); | |
3160 | free (tlist->word->word); | |
3161 | tlist->word->word = s; | |
0628567a | 3162 | /* XXX - turn off W_HAVEQUOTEDNULL here? */ |
7117c2d2 JA |
3163 | } |
3164 | return list; | |
3165 | } | |
3166 | ||
3167 | /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed | |
3168 | string. */ | |
3169 | static char * | |
3170 | remove_quoted_escapes (string) | |
3171 | char *string; | |
3172 | { | |
3173 | char *t; | |
3174 | ||
3175 | if (string) | |
3176 | { | |
3177 | t = dequote_escapes (string); | |
3178 | strcpy (string, t); | |
3179 | free (t); | |
3180 | } | |
3181 | ||
3182 | return (string); | |
3183 | } | |
3184 | ||
cce855bc JA |
3185 | /* Perform quoted null character removal on STRING. We don't allow any |
3186 | quoted null characters in the middle or at the ends of strings because | |
3187 | of how expand_word_internal works. remove_quoted_nulls () turns | |
3188 | STRING into an empty string iff it only consists of a quoted null, | |
3189 | and removes all unquoted CTLNUL characters. */ | |
7117c2d2 | 3190 | static char * |
cce855bc JA |
3191 | remove_quoted_nulls (string) |
3192 | char *string; | |
3193 | { | |
7117c2d2 JA |
3194 | register size_t slen; |
3195 | register int i, j, prev_i; | |
3196 | DECLARE_MBSTATE; | |
3197 | ||
3198 | if (strchr (string, CTLNUL) == 0) /* XXX */ | |
3199 | return string; /* XXX */ | |
3200 | ||
3201 | slen = strlen (string); | |
3202 | i = j = 0; | |
3203 | ||
3204 | while (i < slen) | |
3205 | { | |
3206 | if (string[i] == CTLESC) | |
b80f6443 JA |
3207 | { |
3208 | /* Old code had j++, but we cannot assume that i == j at this | |
3209 | point -- what if a CTLNUL has already been removed from the | |
3210 | string? We don't want to drop the CTLESC or recopy characters | |
3211 | that we've already copied down. */ | |
3212 | i++; string[j++] = CTLESC; | |
3213 | if (i == slen) | |
3214 | break; | |
3215 | } | |
7117c2d2 | 3216 | else if (string[i] == CTLNUL) |
b80f6443 | 3217 | i++; |
7117c2d2 JA |
3218 | |
3219 | prev_i = i; | |
3220 | ADVANCE_CHAR (string, slen, i); | |
3221 | if (j < prev_i) | |
cce855bc | 3222 | { |
7117c2d2 | 3223 | do string[j++] = string[prev_i++]; while (prev_i < i); |
cce855bc | 3224 | } |
7117c2d2 JA |
3225 | else |
3226 | j = i; | |
cce855bc | 3227 | } |
7117c2d2 JA |
3228 | string[j] = '\0'; |
3229 | ||
3230 | return (string); | |
cce855bc JA |
3231 | } |
3232 | ||
3233 | /* Perform quoted null character removal on each element of LIST. | |
3234 | This modifies LIST. */ | |
3235 | void | |
3236 | word_list_remove_quoted_nulls (list) | |
3237 | WORD_LIST *list; | |
3238 | { | |
3239 | register WORD_LIST *t; | |
3240 | ||
3241 | for (t = list; t; t = t->next) | |
95732b49 JA |
3242 | { |
3243 | remove_quoted_nulls (t->word->word); | |
3244 | t->word->flags &= ~W_HASQUOTEDNULL; | |
3245 | } | |
cce855bc JA |
3246 | } |
3247 | ||
3248 | /* **************************************************************** */ | |
3249 | /* */ | |
3250 | /* Functions for Matching and Removing Patterns */ | |
3251 | /* */ | |
3252 | /* **************************************************************** */ | |
3253 | ||
b80f6443 JA |
3254 | #if defined (HANDLE_MULTIBYTE) |
3255 | #if 0 /* Currently unused */ | |
3256 | static unsigned char * | |
3257 | mb_getcharlens (string, len) | |
3258 | char *string; | |
3259 | int len; | |
3260 | { | |
3261 | int i, offset, last; | |
3262 | unsigned char *ret; | |
3263 | char *p; | |
3264 | DECLARE_MBSTATE; | |
3265 | ||
3266 | i = offset = 0; | |
3267 | last = 0; | |
3268 | ret = (unsigned char *)xmalloc (len); | |
3269 | memset (ret, 0, len); | |
3270 | while (string[last]) | |
3271 | { | |
3272 | ADVANCE_CHAR (string, len, offset); | |
3273 | ret[last] = offset - last; | |
3274 | last = offset; | |
3275 | } | |
3276 | return ret; | |
3277 | } | |
3278 | #endif | |
3279 | #endif | |
3280 | ||
cce855bc JA |
3281 | /* Remove the portion of PARAM matched by PATTERN according to OP, where OP |
3282 | can have one of 4 values: | |
3283 | RP_LONG_LEFT remove longest matching portion at start of PARAM | |
726f6388 JA |
3284 | RP_SHORT_LEFT remove shortest matching portion at start of PARAM |
3285 | RP_LONG_RIGHT remove longest matching portion at end of PARAM | |
3286 | RP_SHORT_RIGHT remove shortest matching portion at end of PARAM | |
3287 | */ | |
3288 | ||
3289 | #define RP_LONG_LEFT 1 | |
3290 | #define RP_SHORT_LEFT 2 | |
3291 | #define RP_LONG_RIGHT 3 | |
3292 | #define RP_SHORT_RIGHT 4 | |
3293 | ||
3294 | static char * | |
b80f6443 | 3295 | remove_upattern (param, pattern, op) |
726f6388 JA |
3296 | char *param, *pattern; |
3297 | int op; | |
3298 | { | |
ccc6cda3 JA |
3299 | register int len; |
3300 | register char *end; | |
726f6388 JA |
3301 | register char *p, *ret, c; |
3302 | ||
ccc6cda3 JA |
3303 | len = STRLEN (param); |
3304 | end = param + len; | |
726f6388 JA |
3305 | |
3306 | switch (op) | |
3307 | { | |
3308 | case RP_LONG_LEFT: /* remove longest match at start */ | |
3309 | for (p = end; p >= param; p--) | |
3310 | { | |
3311 | c = *p; *p = '\0'; | |
f73dda09 | 3312 | if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
726f6388 JA |
3313 | { |
3314 | *p = c; | |
3315 | return (savestring (p)); | |
3316 | } | |
3317 | *p = c; | |
b80f6443 | 3318 | |
726f6388 JA |
3319 | } |
3320 | break; | |
3321 | ||
3322 | case RP_SHORT_LEFT: /* remove shortest match at start */ | |
3323 | for (p = param; p <= end; p++) | |
3324 | { | |
3325 | c = *p; *p = '\0'; | |
f73dda09 | 3326 | if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
726f6388 JA |
3327 | { |
3328 | *p = c; | |
3329 | return (savestring (p)); | |
3330 | } | |
3331 | *p = c; | |
3332 | } | |
3333 | break; | |
3334 | ||
ccc6cda3 JA |
3335 | case RP_LONG_RIGHT: /* remove longest match at end */ |
3336 | for (p = param; p <= end; p++) | |
3337 | { | |
f73dda09 | 3338 | if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
ccc6cda3 JA |
3339 | { |
3340 | c = *p; *p = '\0'; | |
3341 | ret = savestring (param); | |
3342 | *p = c; | |
3343 | return (ret); | |
3344 | } | |
3345 | } | |
3346 | break; | |
3347 | ||
3348 | case RP_SHORT_RIGHT: /* remove shortest match at end */ | |
3349 | for (p = end; p >= param; p--) | |
3350 | { | |
f73dda09 | 3351 | if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
ccc6cda3 JA |
3352 | { |
3353 | c = *p; *p = '\0'; | |
3354 | ret = savestring (param); | |
3355 | *p = c; | |
3356 | return (ret); | |
3357 | } | |
3358 | } | |
3359 | break; | |
3360 | } | |
b80f6443 | 3361 | |
ccc6cda3 JA |
3362 | return (savestring (param)); /* no match, return original string */ |
3363 | } | |
3364 | ||
b80f6443 | 3365 | #if defined (HANDLE_MULTIBYTE) |
b80f6443 JA |
3366 | static wchar_t * |
3367 | remove_wpattern (wparam, wstrlen, wpattern, op) | |
3368 | wchar_t *wparam; | |
3369 | size_t wstrlen; | |
3370 | wchar_t *wpattern; | |
3371 | int op; | |
3372 | { | |
0628567a JA |
3373 | wchar_t wc, *ret; |
3374 | int n; | |
b80f6443 JA |
3375 | |
3376 | switch (op) | |
3377 | { | |
3378 | case RP_LONG_LEFT: /* remove longest match at start */ | |
3379 | for (n = wstrlen; n >= 0; n--) | |
3380 | { | |
3381 | wc = wparam[n]; wparam[n] = L'\0'; | |
3382 | if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3383 | { | |
3384 | wparam[n] = wc; | |
3385 | return (wcsdup (wparam + n)); | |
3386 | } | |
3387 | wparam[n] = wc; | |
3388 | } | |
3389 | break; | |
3390 | ||
3391 | case RP_SHORT_LEFT: /* remove shortest match at start */ | |
3392 | for (n = 0; n <= wstrlen; n++) | |
3393 | { | |
3394 | wc = wparam[n]; wparam[n] = L'\0'; | |
3395 | if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3396 | { | |
3397 | wparam[n] = wc; | |
3398 | return (wcsdup (wparam + n)); | |
3399 | } | |
3400 | wparam[n] = wc; | |
3401 | } | |
3402 | break; | |
3403 | ||
3404 | case RP_LONG_RIGHT: /* remove longest match at end */ | |
3405 | for (n = 0; n <= wstrlen; n++) | |
3406 | { | |
3407 | if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3408 | { | |
3409 | wc = wparam[n]; wparam[n] = L'\0'; | |
3410 | ret = wcsdup (wparam); | |
3411 | wparam[n] = wc; | |
3412 | return (ret); | |
3413 | } | |
3414 | } | |
3415 | break; | |
3416 | ||
3417 | case RP_SHORT_RIGHT: /* remove shortest match at end */ | |
3418 | for (n = wstrlen; n >= 0; n--) | |
3419 | { | |
3420 | if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3421 | { | |
3422 | wc = wparam[n]; wparam[n] = L'\0'; | |
3423 | ret = wcsdup (wparam); | |
3424 | wparam[n] = wc; | |
3425 | return (ret); | |
3426 | } | |
3427 | } | |
3428 | break; | |
3429 | } | |
3430 | ||
3431 | return (wcsdup (wparam)); /* no match, return original string */ | |
3432 | } | |
3433 | #endif /* HANDLE_MULTIBYTE */ | |
3434 | ||
3435 | static char * | |
3436 | remove_pattern (param, pattern, op) | |
3437 | char *param, *pattern; | |
3438 | int op; | |
3439 | { | |
3440 | if (param == NULL) | |
3441 | return (param); | |
3442 | if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */ | |
3443 | return (savestring (param)); | |
3444 | ||
3445 | #if defined (HANDLE_MULTIBYTE) | |
3446 | if (MB_CUR_MAX > 1) | |
3447 | { | |
3448 | wchar_t *ret, *oret; | |
3449 | size_t n; | |
3450 | wchar_t *wparam, *wpattern; | |
3451 | mbstate_t ps; | |
3452 | char *xret; | |
3453 | ||
3454 | n = xdupmbstowcs (&wpattern, NULL, pattern); | |
3455 | if (n == (size_t)-1) | |
3456 | return (remove_upattern (param, pattern, op)); | |
3457 | n = xdupmbstowcs (&wparam, NULL, param); | |
3458 | if (n == (size_t)-1) | |
3459 | { | |
3460 | free (wpattern); | |
3461 | return (remove_upattern (param, pattern, op)); | |
3462 | } | |
3463 | oret = ret = remove_wpattern (wparam, n, wpattern, op); | |
3464 | ||
3465 | free (wparam); | |
3466 | free (wpattern); | |
3467 | ||
3468 | n = strlen (param); | |
0628567a | 3469 | xret = (char *)xmalloc (n + 1); |
b80f6443 JA |
3470 | memset (&ps, '\0', sizeof (mbstate_t)); |
3471 | n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps); | |
3472 | xret[n] = '\0'; /* just to make sure */ | |
3473 | free (oret); | |
3474 | return xret; | |
3475 | } | |
3476 | else | |
3477 | #endif | |
3478 | return (remove_upattern (param, pattern, op)); | |
3479 | } | |
3480 | ||
ccc6cda3 | 3481 | /* Return 1 of the first character of STRING could match the first |
f73dda09 | 3482 | character of pattern PAT. Used to avoid n2 calls to strmatch(). */ |
ccc6cda3 JA |
3483 | static int |
3484 | match_pattern_char (pat, string) | |
3485 | char *pat, *string; | |
3486 | { | |
d166f048 | 3487 | char c; |
ccc6cda3 JA |
3488 | |
3489 | if (*string == 0) | |
3490 | return (0); | |
3491 | ||
3492 | switch (c = *pat++) | |
3493 | { | |
3494 | default: | |
3495 | return (*string == c); | |
3496 | case '\\': | |
3497 | return (*string == *pat); | |
3498 | case '?': | |
28ef6c31 | 3499 | return (*pat == LPAREN ? 1 : (*string != '\0')); |
ccc6cda3 JA |
3500 | case '*': |
3501 | return (1); | |
cce855bc JA |
3502 | case '+': |
3503 | case '!': | |
3504 | case '@': | |
28ef6c31 | 3505 | return (*pat == LPAREN ? 1 : (*string == c)); |
ccc6cda3 | 3506 | case '[': |
d166f048 | 3507 | return (*string != '\0'); |
ccc6cda3 JA |
3508 | } |
3509 | } | |
3510 | ||
3511 | /* Match PAT anywhere in STRING and return the match boundaries. | |
3512 | This returns 1 in case of a successful match, 0 otherwise. SP | |
3513 | and EP are pointers into the string where the match begins and | |
3514 | ends, respectively. MTYPE controls what kind of match is attempted. | |
3515 | MATCH_BEG and MATCH_END anchor the match at the beginning and end | |
3516 | of the string, respectively. The longest match is returned. */ | |
3517 | static int | |
b80f6443 | 3518 | match_upattern (string, pat, mtype, sp, ep) |
ccc6cda3 JA |
3519 | char *string, *pat; |
3520 | int mtype; | |
3521 | char **sp, **ep; | |
3522 | { | |
b80f6443 | 3523 | int c, len; |
95732b49 | 3524 | register char *p, *p1, *npat; |
ccc6cda3 JA |
3525 | char *end; |
3526 | ||
95732b49 JA |
3527 | /* If the pattern doesn't match anywhere in the string, go ahead and |
3528 | short-circuit right away. A minor optimization, saves a bunch of | |
3529 | unnecessary calls to strmatch (up to N calls for a string of N | |
3530 | characters) if the match is unsuccessful. To preserve the semantics | |
3531 | of the substring matches below, we make sure that the pattern has | |
3532 | `*' as first and last character, making a new pattern if necessary. */ | |
3533 | /* XXX - check this later if I ever implement `**' with special meaning, | |
3534 | since this will potentially result in `**' at the beginning or end */ | |
3535 | len = STRLEN (pat); | |
3536 | if (pat[0] != '*' || pat[len - 1] != '*') | |
3537 | { | |
0628567a | 3538 | p = npat = (char *)xmalloc (len + 3); |
95732b49 JA |
3539 | p1 = pat; |
3540 | if (*p1 != '*') | |
3541 | *p++ = '*'; | |
3542 | while (*p1) | |
3543 | *p++ = *p1++; | |
3544 | if (p1[-1] != '*' || p[-2] == '\\') | |
3545 | *p++ = '*'; | |
3546 | *p = '\0'; | |
3547 | } | |
3548 | else | |
3549 | npat = pat; | |
3550 | c = strmatch (npat, string, FNMATCH_EXTFLAG); | |
3551 | if (npat != pat) | |
3552 | free (npat); | |
3553 | if (c == FNM_NOMATCH) | |
3554 | return (0); | |
3555 | ||
b80f6443 JA |
3556 | len = STRLEN (string); |
3557 | end = string + len; | |
ccc6cda3 JA |
3558 | |
3559 | switch (mtype) | |
3560 | { | |
3561 | case MATCH_ANY: | |
3562 | for (p = string; p <= end; p++) | |
3563 | { | |
3564 | if (match_pattern_char (pat, p)) | |
3565 | { | |
3566 | for (p1 = end; p1 >= p; p1--) | |
3567 | { | |
3568 | c = *p1; *p1 = '\0'; | |
f73dda09 | 3569 | if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0) |
ccc6cda3 JA |
3570 | { |
3571 | *p1 = c; | |
3572 | *sp = p; | |
3573 | *ep = p1; | |
3574 | return 1; | |
3575 | } | |
3576 | *p1 = c; | |
3577 | } | |
3578 | } | |
3579 | } | |
b80f6443 | 3580 | |
ccc6cda3 JA |
3581 | return (0); |
3582 | ||
3583 | case MATCH_BEG: | |
3584 | if (match_pattern_char (pat, string) == 0) | |
28ef6c31 | 3585 | return (0); |
b80f6443 | 3586 | |
ccc6cda3 JA |
3587 | for (p = end; p >= string; p--) |
3588 | { | |
3589 | c = *p; *p = '\0'; | |
f73dda09 | 3590 | if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0) |
ccc6cda3 JA |
3591 | { |
3592 | *p = c; | |
3593 | *sp = string; | |
3594 | *ep = p; | |
3595 | return 1; | |
3596 | } | |
3597 | *p = c; | |
3598 | } | |
b80f6443 | 3599 | |
ccc6cda3 | 3600 | return (0); |
726f6388 | 3601 | |
ccc6cda3 JA |
3602 | case MATCH_END: |
3603 | for (p = string; p <= end; p++) | |
b80f6443 JA |
3604 | { |
3605 | if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0) | |
3606 | { | |
3607 | *sp = p; | |
3608 | *ep = end; | |
3609 | return 1; | |
3610 | } | |
3611 | ||
3612 | } | |
3613 | ||
3614 | return (0); | |
3615 | } | |
3616 | ||
3617 | return (0); | |
3618 | } | |
3619 | ||
3620 | #if defined (HANDLE_MULTIBYTE) | |
3621 | /* Return 1 of the first character of WSTRING could match the first | |
3622 | character of pattern WPAT. Wide character version. */ | |
3623 | static int | |
3624 | match_pattern_wchar (wpat, wstring) | |
3625 | wchar_t *wpat, *wstring; | |
3626 | { | |
3627 | wchar_t wc; | |
3628 | ||
3629 | if (*wstring == 0) | |
3630 | return (0); | |
3631 | ||
3632 | switch (wc = *wpat++) | |
3633 | { | |
3634 | default: | |
3635 | return (*wstring == wc); | |
3636 | case L'\\': | |
3637 | return (*wstring == *wpat); | |
3638 | case L'?': | |
3639 | return (*wpat == LPAREN ? 1 : (*wstring != L'\0')); | |
3640 | case L'*': | |
3641 | return (1); | |
3642 | case L'+': | |
3643 | case L'!': | |
3644 | case L'@': | |
3645 | return (*wpat == LPAREN ? 1 : (*wstring == wc)); | |
3646 | case L'[': | |
3647 | return (*wstring != L'\0'); | |
3648 | } | |
3649 | } | |
3650 | ||
3651 | /* Match WPAT anywhere in WSTRING and return the match boundaries. | |
3652 | This returns 1 in case of a successful match, 0 otherwise. Wide | |
3653 | character version. */ | |
3654 | static int | |
3655 | match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep) | |
3656 | wchar_t *wstring; | |
3657 | char **indices; | |
3658 | size_t wstrlen; | |
3659 | wchar_t *wpat; | |
3660 | int mtype; | |
3661 | char **sp, **ep; | |
3662 | { | |
95732b49 | 3663 | wchar_t wc, *wp, *nwpat, *wp1; |
b80f6443 JA |
3664 | int len; |
3665 | #if 0 | |
3666 | size_t n, n1; /* Apple's gcc seems to miscompile this badly */ | |
3667 | #else | |
3668 | int n, n1; | |
3669 | #endif | |
3670 | ||
95732b49 JA |
3671 | /* If the pattern doesn't match anywhere in the string, go ahead and |
3672 | short-circuit right away. A minor optimization, saves a bunch of | |
3673 | unnecessary calls to strmatch (up to N calls for a string of N | |
3674 | characters) if the match is unsuccessful. To preserve the semantics | |
3675 | of the substring matches below, we make sure that the pattern has | |
3676 | `*' as first and last character, making a new pattern if necessary. */ | |
3677 | /* XXX - check this later if I ever implement `**' with special meaning, | |
3678 | since this will potentially result in `**' at the beginning or end */ | |
3679 | len = wcslen (wpat); | |
3680 | if (wpat[0] != L'*' || wpat[len - 1] != L'*') | |
3681 | { | |
0628567a | 3682 | wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t)); |
95732b49 JA |
3683 | wp1 = wpat; |
3684 | if (*wp1 != L'*') | |
3685 | *wp++ = L'*'; | |
3686 | while (*wp1 != L'\0') | |
3687 | *wp++ = *wp1++; | |
3688 | if (wp1[-1] != L'*' || wp1[-2] == L'\\') | |
3689 | *wp++ = L'*'; | |
3690 | *wp = '\0'; | |
3691 | } | |
3692 | else | |
3693 | nwpat = wpat; | |
3694 | len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG); | |
3695 | if (nwpat != wpat) | |
3696 | free (nwpat); | |
3697 | if (len == FNM_NOMATCH) | |
3698 | return (0); | |
3699 | ||
b80f6443 JA |
3700 | switch (mtype) |
3701 | { | |
3702 | case MATCH_ANY: | |
3703 | for (n = 0; n <= wstrlen; n++) | |
3704 | { | |
3705 | if (match_pattern_wchar (wpat, wstring + n)) | |
3706 | { | |
3707 | for (n1 = wstrlen; n1 >= n; n1--) | |
3708 | { | |
3709 | wc = wstring[n1]; wstring[n1] = L'\0'; | |
3710 | if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0) | |
3711 | { | |
3712 | wstring[n1] = wc; | |
3713 | *sp = indices[n]; | |
3714 | *ep = indices[n1]; | |
3715 | return 1; | |
3716 | } | |
3717 | wstring[n1] = wc; | |
3718 | } | |
3719 | } | |
3720 | } | |
3721 | ||
3722 | return (0); | |
3723 | ||
3724 | case MATCH_BEG: | |
3725 | if (match_pattern_wchar (wpat, wstring) == 0) | |
3726 | return (0); | |
3727 | ||
3728 | for (n = wstrlen; n >= 0; n--) | |
3729 | { | |
3730 | wc = wstring[n]; wstring[n] = L'\0'; | |
3731 | if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0) | |
3732 | { | |
3733 | wstring[n] = wc; | |
3734 | *sp = indices[0]; | |
3735 | *ep = indices[n]; | |
3736 | return 1; | |
3737 | } | |
3738 | wstring[n] = wc; | |
3739 | } | |
3740 | ||
3741 | return (0); | |
3742 | ||
3743 | case MATCH_END: | |
3744 | for (n = 0; n <= wstrlen; n++) | |
3745 | { | |
3746 | if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0) | |
3747 | { | |
3748 | *sp = indices[n]; | |
3749 | *ep = indices[wstrlen]; | |
3750 | return 1; | |
3751 | } | |
3752 | } | |
3753 | ||
ccc6cda3 | 3754 | return (0); |
726f6388 | 3755 | } |
ccc6cda3 JA |
3756 | |
3757 | return (0); | |
726f6388 | 3758 | } |
b80f6443 JA |
3759 | #endif /* HANDLE_MULTIBYTE */ |
3760 | ||
3761 | static int | |
3762 | match_pattern (string, pat, mtype, sp, ep) | |
3763 | char *string, *pat; | |
3764 | int mtype; | |
3765 | char **sp, **ep; | |
3766 | { | |
3767 | #if defined (HANDLE_MULTIBYTE) | |
3768 | int ret; | |
3769 | size_t n; | |
3770 | wchar_t *wstring, *wpat; | |
3771 | char **indices; | |
3772 | #endif | |
3773 | ||
3774 | if (string == 0 || *string == 0 || pat == 0 || *pat == 0) | |
3775 | return (0); | |
3776 | ||
3777 | #if defined (HANDLE_MULTIBYTE) | |
3778 | if (MB_CUR_MAX > 1) | |
3779 | { | |
3780 | n = xdupmbstowcs (&wpat, NULL, pat); | |
3781 | if (n == (size_t)-1) | |
3782 | return (match_upattern (string, pat, mtype, sp, ep)); | |
3783 | n = xdupmbstowcs (&wstring, &indices, string); | |
3784 | if (n == (size_t)-1) | |
3785 | { | |
3786 | free (wpat); | |
3787 | return (match_upattern (string, pat, mtype, sp, ep)); | |
3788 | } | |
3789 | ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep); | |
3790 | ||
3791 | free (wpat); | |
3792 | free (wstring); | |
3793 | free (indices); | |
3794 | ||
3795 | return (ret); | |
3796 | } | |
3797 | else | |
3798 | #endif | |
3799 | return (match_upattern (string, pat, mtype, sp, ep)); | |
3800 | } | |
726f6388 | 3801 | |
cce855bc JA |
3802 | static int |
3803 | getpatspec (c, value) | |
3804 | int c; | |
3805 | char *value; | |
3806 | { | |
3807 | if (c == '#') | |
3808 | return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT); | |
3809 | else /* c == '%' */ | |
3810 | return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT); | |
3811 | } | |
3812 | ||
3813 | /* Posix.2 says that the WORD should be run through tilde expansion, | |
3814 | parameter expansion, command substitution and arithmetic expansion. | |
3815 | This leaves the result quoted, so quote_string_for_globbing () has | |
f73dda09 | 3816 | to be called to fix it up for strmatch (). If QUOTED is non-zero, |
cce855bc JA |
3817 | it means that the entire expression was enclosed in double quotes. |
3818 | This means that quoting characters in the pattern do not make any | |
3819 | special pattern characters quoted. For example, the `*' in the | |
3820 | following retains its special meaning: "${foo#'*'}". */ | |
3821 | static char * | |
3822 | getpattern (value, quoted, expandpat) | |
3823 | char *value; | |
3824 | int quoted, expandpat; | |
3825 | { | |
3826 | char *pat, *tword; | |
3827 | WORD_LIST *l; | |
0628567a | 3828 | #if 0 |
cce855bc | 3829 | int i; |
0628567a | 3830 | #endif |
cce855bc | 3831 | |
7117c2d2 JA |
3832 | /* There is a problem here: how to handle single or double quotes in the |
3833 | pattern string when the whole expression is between double quotes? | |
3834 | POSIX.2 says that enclosing double quotes do not cause the pattern to | |
3835 | be quoted, but does that leave us a problem with @ and array[@] and their | |
3836 | expansions inside a pattern? */ | |
3837 | #if 0 | |
cce855bc JA |
3838 | if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword) |
3839 | { | |
3840 | i = 0; | |
3841 | pat = string_extract_double_quoted (tword, &i, 1); | |
3842 | free (tword); | |
3843 | tword = pat; | |
3844 | } | |
7117c2d2 | 3845 | #endif |
cce855bc | 3846 | |
7117c2d2 JA |
3847 | /* expand_string_for_rhs () leaves WORD quoted and does not perform |
3848 | word splitting. */ | |
95732b49 | 3849 | l = *value ? expand_string_for_rhs (value, |
7117c2d2 | 3850 | (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted, |
cce855bc | 3851 | (int *)NULL, (int *)NULL) |
cce855bc | 3852 | : (WORD_LIST *)0; |
cce855bc JA |
3853 | pat = string_list (l); |
3854 | dispose_words (l); | |
3855 | if (pat) | |
3856 | { | |
3857 | tword = quote_string_for_globbing (pat, QGLOB_CVTNULL); | |
3858 | free (pat); | |
3859 | pat = tword; | |
3860 | } | |
3861 | return (pat); | |
3862 | } | |
3863 | ||
7117c2d2 | 3864 | #if 0 |
cce855bc JA |
3865 | /* Handle removing a pattern from a string as a result of ${name%[%]value} |
3866 | or ${name#[#]value}. */ | |
3867 | static char * | |
7117c2d2 JA |
3868 | variable_remove_pattern (value, pattern, patspec, quoted) |
3869 | char *value, *pattern; | |
3870 | int patspec, quoted; | |
cce855bc | 3871 | { |
7117c2d2 | 3872 | char *tword; |
cce855bc | 3873 | |
7117c2d2 | 3874 | tword = remove_pattern (value, pattern, patspec); |
cce855bc | 3875 | |
cce855bc JA |
3876 | return (tword); |
3877 | } | |
7117c2d2 | 3878 | #endif |
cce855bc JA |
3879 | |
3880 | static char * | |
7117c2d2 | 3881 | list_remove_pattern (list, pattern, patspec, itype, quoted) |
cce855bc JA |
3882 | WORD_LIST *list; |
3883 | char *pattern; | |
7117c2d2 | 3884 | int patspec, itype, quoted; |
cce855bc JA |
3885 | { |
3886 | WORD_LIST *new, *l; | |
3887 | WORD_DESC *w; | |
3888 | char *tword; | |
3889 | ||
3890 | for (new = (WORD_LIST *)NULL, l = list; l; l = l->next) | |
3891 | { | |
3892 | tword = remove_pattern (l->word->word, pattern, patspec); | |
95732b49 JA |
3893 | w = alloc_word_desc (); |
3894 | w->word = tword ? tword : savestring (""); | |
cce855bc JA |
3895 | new = make_word_list (w, new); |
3896 | } | |
3897 | ||
3898 | l = REVERSE_LIST (new, WORD_LIST *); | |
7117c2d2 | 3899 | if (itype == '*') |
b80f6443 | 3900 | tword = (quoted & Q_DOUBLE_QUOTES) ? string_list_dollar_star (l) : string_list (l); |
cce855bc JA |
3901 | else |
3902 | tword = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (l) : l); | |
3903 | ||
3904 | dispose_words (l); | |
3905 | return (tword); | |
3906 | } | |
3907 | ||
3908 | static char * | |
7117c2d2 JA |
3909 | parameter_list_remove_pattern (itype, pattern, patspec, quoted) |
3910 | int itype; | |
3911 | char *pattern; | |
3912 | int patspec, quoted; | |
cce855bc | 3913 | { |
7117c2d2 | 3914 | char *ret; |
cce855bc JA |
3915 | WORD_LIST *list; |
3916 | ||
cce855bc | 3917 | list = list_rest_of_args (); |
7117c2d2 JA |
3918 | if (list == 0) |
3919 | return ((char *)NULL); | |
3920 | ret = list_remove_pattern (list, pattern, patspec, itype, quoted); | |
cce855bc | 3921 | dispose_words (list); |
cce855bc JA |
3922 | return (ret); |
3923 | } | |
3924 | ||
3925 | #if defined (ARRAY_VARS) | |
3926 | static char * | |
7117c2d2 JA |
3927 | array_remove_pattern (a, pattern, patspec, varname, quoted) |
3928 | ARRAY *a; | |
3929 | char *pattern; | |
3930 | int patspec; | |
3931 | char *varname; /* so we can figure out how it's indexed */ | |
3932 | int quoted; | |
cce855bc | 3933 | { |
7117c2d2 JA |
3934 | int itype; |
3935 | char *ret; | |
3936 | WORD_LIST *list; | |
3937 | SHELL_VAR *v; | |
cce855bc | 3938 | |
7117c2d2 JA |
3939 | /* compute itype from varname here */ |
3940 | v = array_variable_part (varname, &ret, 0); | |
3941 | itype = ret[0]; | |
3942 | ||
3943 | list = array_to_word_list (a); | |
3944 | if (list == 0) | |
3945 | return ((char *)NULL); | |
3946 | ret = list_remove_pattern (list, pattern, patspec, itype, quoted); | |
3947 | dispose_words (list); | |
3948 | ||
3949 | return ret; | |
3950 | } | |
3951 | #endif /* ARRAY_VARS */ | |
3952 | ||
3953 | static char * | |
3954 | parameter_brace_remove_pattern (varname, value, patstr, rtype, quoted) | |
3955 | char *varname, *value, *patstr; | |
3956 | int rtype, quoted; | |
3957 | { | |
b80f6443 | 3958 | int vtype, patspec, starsub; |
7117c2d2 JA |
3959 | char *temp1, *val, *pattern; |
3960 | SHELL_VAR *v; | |
3961 | ||
3962 | if (value == 0) | |
3963 | return ((char *)NULL); | |
3964 | ||
3965 | this_command_name = varname; | |
3966 | ||
b80f6443 | 3967 | vtype = get_var_and_type (varname, value, quoted, &v, &val); |
7117c2d2 | 3968 | if (vtype == -1) |
cce855bc JA |
3969 | return ((char *)NULL); |
3970 | ||
b80f6443 JA |
3971 | starsub = vtype & VT_STARSUB; |
3972 | vtype &= ~VT_STARSUB; | |
3973 | ||
7117c2d2 | 3974 | patspec = getpatspec (rtype, patstr); |
cce855bc | 3975 | if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT) |
7117c2d2 | 3976 | patstr++; |
cce855bc | 3977 | |
f1be666c JA |
3978 | /* Need to pass getpattern newly-allocated memory in case of expansion -- |
3979 | the expansion code will free the passed string on an error. */ | |
3980 | temp1 = savestring (patstr); | |
3981 | pattern = getpattern (temp1, quoted, 1); | |
3982 | free (temp1); | |
cce855bc | 3983 | |
7117c2d2 JA |
3984 | temp1 = (char *)NULL; /* shut up gcc */ |
3985 | switch (vtype) | |
cce855bc | 3986 | { |
7117c2d2 JA |
3987 | case VT_VARIABLE: |
3988 | case VT_ARRAYMEMBER: | |
3989 | temp1 = remove_pattern (val, pattern, patspec); | |
3990 | if (vtype == VT_VARIABLE) | |
3991 | FREE (val); | |
3992 | if (temp1) | |
28ef6c31 | 3993 | { |
7117c2d2 JA |
3994 | val = quote_escapes (temp1); |
3995 | free (temp1); | |
3996 | temp1 = val; | |
28ef6c31 | 3997 | } |
7117c2d2 JA |
3998 | break; |
3999 | #if defined (ARRAY_VARS) | |
4000 | case VT_ARRAYVAR: | |
4001 | temp1 = array_remove_pattern (array_cell (v), pattern, patspec, varname, quoted); | |
4002 | if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) | |
cce855bc | 4003 | { |
7117c2d2 JA |
4004 | val = quote_escapes (temp1); |
4005 | free (temp1); | |
4006 | temp1 = val; | |
cce855bc | 4007 | } |
7117c2d2 JA |
4008 | break; |
4009 | #endif | |
4010 | case VT_POSPARMS: | |
4011 | temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted); | |
4012 | if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) | |
4013 | { | |
4014 | val = quote_escapes (temp1); | |
4015 | free (temp1); | |
4016 | temp1 = val; | |
4017 | } | |
4018 | break; | |
cce855bc JA |
4019 | } |
4020 | ||
4021 | FREE (pattern); | |
7117c2d2 JA |
4022 | return temp1; |
4023 | } | |
cce855bc | 4024 | |
726f6388 JA |
4025 | /******************************************* |
4026 | * * | |
4027 | * Functions to expand WORD_DESCs * | |
4028 | * * | |
4029 | *******************************************/ | |
4030 | ||
4031 | /* Expand WORD, performing word splitting on the result. This does | |
4032 | parameter expansion, command substitution, arithmetic expansion, | |
4033 | word splitting, and quote removal. */ | |
4034 | ||
4035 | WORD_LIST * | |
4036 | expand_word (word, quoted) | |
4037 | WORD_DESC *word; | |
4038 | int quoted; | |
4039 | { | |
4040 | WORD_LIST *result, *tresult; | |
4041 | ||
b72432fd | 4042 | tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); |
726f6388 JA |
4043 | result = word_list_split (tresult); |
4044 | dispose_words (tresult); | |
ccc6cda3 | 4045 | return (result ? dequote_list (result) : result); |
726f6388 JA |
4046 | } |
4047 | ||
4048 | /* Expand WORD, but do not perform word splitting on the result. This | |
4049 | does parameter expansion, command substitution, arithmetic expansion, | |
4050 | and quote removal. */ | |
4051 | WORD_LIST * | |
28ef6c31 | 4052 | expand_word_unsplit (word, quoted) |
726f6388 JA |
4053 | WORD_DESC *word; |
4054 | int quoted; | |
4055 | { | |
4056 | WORD_LIST *result; | |
4057 | ||
28ef6c31 | 4058 | expand_no_split_dollar_star = 1; |
b72432fd | 4059 | result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); |
28ef6c31 | 4060 | expand_no_split_dollar_star = 0; |
7117c2d2 | 4061 | |
ccc6cda3 | 4062 | return (result ? dequote_list (result) : result); |
726f6388 JA |
4063 | } |
4064 | ||
4065 | /* Perform shell expansions on WORD, but do not perform word splitting or | |
4066 | quote removal on the result. */ | |
4067 | WORD_LIST * | |
4068 | expand_word_leave_quoted (word, quoted) | |
4069 | WORD_DESC *word; | |
4070 | int quoted; | |
4071 | { | |
b72432fd | 4072 | return (call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL)); |
726f6388 JA |
4073 | } |
4074 | ||
726f6388 JA |
4075 | #if defined (PROCESS_SUBSTITUTION) |
4076 | ||
cce855bc JA |
4077 | /*****************************************************************/ |
4078 | /* */ | |
4079 | /* Hacking Process Substitution */ | |
4080 | /* */ | |
4081 | /*****************************************************************/ | |
726f6388 | 4082 | |
726f6388 JA |
4083 | #if !defined (HAVE_DEV_FD) |
4084 | /* Named pipes must be removed explicitly with `unlink'. This keeps a list | |
4085 | of FIFOs the shell has open. unlink_fifo_list will walk the list and | |
4086 | unlink all of them. add_fifo_list adds the name of an open FIFO to the | |
4087 | list. NFIFO is a count of the number of FIFOs in the list. */ | |
4088 | #define FIFO_INCR 20 | |
4089 | ||
f73dda09 JA |
4090 | struct temp_fifo { |
4091 | char *file; | |
4092 | pid_t proc; | |
4093 | }; | |
4094 | ||
4095 | static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL; | |
ccc6cda3 JA |
4096 | static int nfifo; |
4097 | static int fifo_list_size; | |
726f6388 JA |
4098 | |
4099 | static void | |
4100 | add_fifo_list (pathname) | |
4101 | char *pathname; | |
4102 | { | |
4103 | if (nfifo >= fifo_list_size - 1) | |
4104 | { | |
4105 | fifo_list_size += FIFO_INCR; | |
f73dda09 JA |
4106 | fifo_list = (struct temp_fifo *)xrealloc (fifo_list, |
4107 | fifo_list_size * sizeof (struct temp_fifo)); | |
726f6388 JA |
4108 | } |
4109 | ||
f73dda09 JA |
4110 | fifo_list[nfifo].file = savestring (pathname); |
4111 | nfifo++; | |
726f6388 JA |
4112 | } |
4113 | ||
4114 | void | |
4115 | unlink_fifo_list () | |
4116 | { | |
f73dda09 JA |
4117 | int saved, i, j; |
4118 | ||
ccc6cda3 | 4119 | if (nfifo == 0) |
726f6388 JA |
4120 | return; |
4121 | ||
f73dda09 | 4122 | for (i = saved = 0; i < nfifo; i++) |
726f6388 | 4123 | { |
f73dda09 JA |
4124 | if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1)) |
4125 | { | |
7117c2d2 JA |
4126 | unlink (fifo_list[i].file); |
4127 | free (fifo_list[i].file); | |
4128 | fifo_list[i].file = (char *)NULL; | |
4129 | fifo_list[i].proc = -1; | |
f73dda09 JA |
4130 | } |
4131 | else | |
7117c2d2 | 4132 | saved++; |
f73dda09 JA |
4133 | } |
4134 | ||
4135 | /* If we didn't remove some of the FIFOs, compact the list. */ | |
4136 | if (saved) | |
4137 | { | |
4138 | for (i = j = 0; i < nfifo; i++) | |
4139 | if (fifo_list[i].file) | |
4140 | { | |
4141 | fifo_list[j].file = fifo_list[i].file; | |
4142 | fifo_list[j].proc = fifo_list[i].proc; | |
4143 | j++; | |
4144 | } | |
4145 | nfifo = j; | |
726f6388 | 4146 | } |
f73dda09 JA |
4147 | else |
4148 | nfifo = 0; | |
726f6388 JA |
4149 | } |
4150 | ||
f1be666c JA |
4151 | int |
4152 | fifos_pending () | |
4153 | { | |
4154 | return nfifo; | |
4155 | } | |
4156 | ||
726f6388 JA |
4157 | static char * |
4158 | make_named_pipe () | |
4159 | { | |
4160 | char *tname; | |
4161 | ||
28ef6c31 | 4162 | tname = sh_mktmpname ("sh-np", MT_USERANDOM); |
726f6388 JA |
4163 | if (mkfifo (tname, 0600) < 0) |
4164 | { | |
4165 | free (tname); | |
4166 | return ((char *)NULL); | |
4167 | } | |
4168 | ||
4169 | add_fifo_list (tname); | |
4170 | return (tname); | |
4171 | } | |
4172 | ||
726f6388 JA |
4173 | #else /* HAVE_DEV_FD */ |
4174 | ||
4175 | /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell | |
4176 | has open to children. NFDS is a count of the number of bits currently | |
4177 | set in DEV_FD_LIST. TOTFDS is a count of the highest possible number | |
4178 | of open files. */ | |
4179 | static char *dev_fd_list = (char *)NULL; | |
ccc6cda3 | 4180 | static int nfds; |
726f6388 JA |
4181 | static int totfds; /* The highest possible number of open files. */ |
4182 | ||
4183 | static void | |
4184 | add_fifo_list (fd) | |
4185 | int fd; | |
4186 | { | |
4187 | if (!dev_fd_list || fd >= totfds) | |
4188 | { | |
4189 | int ofds; | |
4190 | ||
4191 | ofds = totfds; | |
4192 | totfds = getdtablesize (); | |
4193 | if (totfds < 0 || totfds > 256) | |
4194 | totfds = 256; | |
4195 | if (fd > totfds) | |
4196 | totfds = fd + 2; | |
4197 | ||
f73dda09 | 4198 | dev_fd_list = (char *)xrealloc (dev_fd_list, totfds); |
7117c2d2 | 4199 | memset (dev_fd_list + ofds, '\0', totfds - ofds); |
726f6388 JA |
4200 | } |
4201 | ||
4202 | dev_fd_list[fd] = 1; | |
4203 | nfds++; | |
4204 | } | |
4205 | ||
f1be666c JA |
4206 | int |
4207 | fifos_pending () | |
4208 | { | |
4209 | return 0; /* used for cleanup; not needed with /dev/fd */ | |
4210 | } | |
4211 | ||
726f6388 JA |
4212 | void |
4213 | unlink_fifo_list () | |
4214 | { | |
4215 | register int i; | |
4216 | ||
ccc6cda3 | 4217 | if (nfds == 0) |
726f6388 JA |
4218 | return; |
4219 | ||
4220 | for (i = 0; nfds && i < totfds; i++) | |
4221 | if (dev_fd_list[i]) | |
4222 | { | |
4223 | close (i); | |
4224 | dev_fd_list[i] = 0; | |
4225 | nfds--; | |
4226 | } | |
4227 | ||
4228 | nfds = 0; | |
4229 | } | |
4230 | ||
4231 | #if defined (NOTDEF) | |
4232 | print_dev_fd_list () | |
4233 | { | |
4234 | register int i; | |
4235 | ||
f73dda09 | 4236 | fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ()); |
726f6388 JA |
4237 | fflush (stderr); |
4238 | ||
4239 | for (i = 0; i < totfds; i++) | |
4240 | { | |
4241 | if (dev_fd_list[i]) | |
4242 | fprintf (stderr, " %d", i); | |
4243 | } | |
4244 | fprintf (stderr, "\n"); | |
4245 | } | |
4246 | #endif /* NOTDEF */ | |
4247 | ||
4248 | static char * | |
4249 | make_dev_fd_filename (fd) | |
4250 | int fd; | |
4251 | { | |
f73dda09 | 4252 | char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p; |
726f6388 | 4253 | |
f73dda09 | 4254 | ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 4); |
bb70624e JA |
4255 | |
4256 | strcpy (ret, DEV_FD_PREFIX); | |
4257 | p = inttostr (fd, intbuf, sizeof (intbuf)); | |
4258 | strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p); | |
4259 | ||
726f6388 JA |
4260 | add_fifo_list (fd); |
4261 | return (ret); | |
4262 | } | |
4263 | ||
4264 | #endif /* HAVE_DEV_FD */ | |
4265 | ||
4266 | /* Return a filename that will open a connection to the process defined by | |
4267 | executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return | |
4268 | a filename in /dev/fd corresponding to a descriptor that is one of the | |
4269 | ends of the pipe. If not defined, we use named pipes on systems that have | |
4270 | them. Systems without /dev/fd and named pipes are out of luck. | |
4271 | ||
4272 | OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or | |
4273 | use the read end of the pipe and dup that file descriptor to fd 0 in | |
4274 | the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for | |
4275 | writing or use the write end of the pipe in the child, and dup that | |
4276 | file descriptor to fd 1 in the child. The parent does the opposite. */ | |
4277 | ||
4278 | static char * | |
4279 | process_substitute (string, open_for_read_in_child) | |
4280 | char *string; | |
4281 | int open_for_read_in_child; | |
4282 | { | |
4283 | char *pathname; | |
4284 | int fd, result; | |
4285 | pid_t old_pid, pid; | |
4286 | #if defined (HAVE_DEV_FD) | |
4287 | int parent_pipe_fd, child_pipe_fd; | |
4288 | int fildes[2]; | |
4289 | #endif /* HAVE_DEV_FD */ | |
4290 | #if defined (JOB_CONTROL) | |
4291 | pid_t old_pipeline_pgrp; | |
ccc6cda3 | 4292 | #endif |
726f6388 | 4293 | |
cce855bc | 4294 | if (!string || !*string || wordexp_only) |
726f6388 JA |
4295 | return ((char *)NULL); |
4296 | ||
4297 | #if !defined (HAVE_DEV_FD) | |
4298 | pathname = make_named_pipe (); | |
4299 | #else /* HAVE_DEV_FD */ | |
4300 | if (pipe (fildes) < 0) | |
4301 | { | |
b80f6443 | 4302 | sys_error (_("cannot make pipe for process substitution")); |
726f6388 JA |
4303 | return ((char *)NULL); |
4304 | } | |
4305 | /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of | |
4306 | the pipe in the parent, otherwise the read end. */ | |
4307 | parent_pipe_fd = fildes[open_for_read_in_child]; | |
4308 | child_pipe_fd = fildes[1 - open_for_read_in_child]; | |
d166f048 JA |
4309 | /* Move the parent end of the pipe to some high file descriptor, to |
4310 | avoid clashes with FDs used by the script. */ | |
4311 | parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64); | |
4312 | ||
726f6388 JA |
4313 | pathname = make_dev_fd_filename (parent_pipe_fd); |
4314 | #endif /* HAVE_DEV_FD */ | |
4315 | ||
4316 | if (!pathname) | |
4317 | { | |
b80f6443 | 4318 | sys_error (_("cannot make pipe for process substitution")); |
726f6388 JA |
4319 | return ((char *)NULL); |
4320 | } | |
4321 | ||
4322 | old_pid = last_made_pid; | |
4323 | ||
4324 | #if defined (JOB_CONTROL) | |
4325 | old_pipeline_pgrp = pipeline_pgrp; | |
4326 | pipeline_pgrp = shell_pgrp; | |
ccc6cda3 | 4327 | save_pipeline (1); |
ccc6cda3 JA |
4328 | #endif /* JOB_CONTROL */ |
4329 | ||
726f6388 JA |
4330 | pid = make_child ((char *)NULL, 1); |
4331 | if (pid == 0) | |
4332 | { | |
ccc6cda3 | 4333 | reset_terminating_signals (); /* XXX */ |
b80f6443 | 4334 | free_pushed_string_input (); |
726f6388 JA |
4335 | /* Cancel traps, in trap.c. */ |
4336 | restore_original_signals (); | |
4337 | setup_async_signals (); | |
28ef6c31 | 4338 | subshell_environment |= SUBSHELL_COMSUB; |
726f6388 | 4339 | } |
ccc6cda3 JA |
4340 | |
4341 | #if defined (JOB_CONTROL) | |
726f6388 JA |
4342 | set_sigchld_handler (); |
4343 | stop_making_children (); | |
4344 | pipeline_pgrp = old_pipeline_pgrp; | |
ccc6cda3 | 4345 | #endif /* JOB_CONTROL */ |
726f6388 JA |
4346 | |
4347 | if (pid < 0) | |
4348 | { | |
b80f6443 | 4349 | sys_error (_("cannot make child for process substitution")); |
726f6388 JA |
4350 | free (pathname); |
4351 | #if defined (HAVE_DEV_FD) | |
4352 | close (parent_pipe_fd); | |
4353 | close (child_pipe_fd); | |
4354 | #endif /* HAVE_DEV_FD */ | |
4355 | return ((char *)NULL); | |
4356 | } | |
4357 | ||
4358 | if (pid > 0) | |
4359 | { | |
ccc6cda3 JA |
4360 | #if defined (JOB_CONTROL) |
4361 | restore_pipeline (1); | |
4362 | #endif | |
4363 | ||
f73dda09 JA |
4364 | #if !defined (HAVE_DEV_FD) |
4365 | fifo_list[nfifo-1].proc = pid; | |
4366 | #endif | |
4367 | ||
726f6388 JA |
4368 | last_made_pid = old_pid; |
4369 | ||
4370 | #if defined (JOB_CONTROL) && defined (PGRP_PIPE) | |
4371 | close_pgrp_pipe (); | |
4372 | #endif /* JOB_CONTROL && PGRP_PIPE */ | |
4373 | ||
4374 | #if defined (HAVE_DEV_FD) | |
4375 | close (child_pipe_fd); | |
4376 | #endif /* HAVE_DEV_FD */ | |
4377 | ||
4378 | return (pathname); | |
4379 | } | |
4380 | ||
4381 | set_sigint_handler (); | |
4382 | ||
4383 | #if defined (JOB_CONTROL) | |
4384 | set_job_control (0); | |
4385 | #endif /* JOB_CONTROL */ | |
4386 | ||
4387 | #if !defined (HAVE_DEV_FD) | |
4388 | /* Open the named pipe in the child. */ | |
ccc6cda3 | 4389 | fd = open (pathname, open_for_read_in_child ? O_RDONLY|O_NONBLOCK : O_WRONLY); |
726f6388 JA |
4390 | if (fd < 0) |
4391 | { | |
b80f6443 JA |
4392 | /* Two separate strings for ease of translation. */ |
4393 | if (open_for_read_in_child) | |
4394 | sys_error (_("cannot open named pipe %s for reading"), pathname); | |
4395 | else | |
4396 | sys_error (_("cannot open named pipe %s for writing"), pathname); | |
4397 | ||
726f6388 JA |
4398 | exit (127); |
4399 | } | |
bb70624e JA |
4400 | if (open_for_read_in_child) |
4401 | { | |
28ef6c31 | 4402 | if (sh_unset_nodelay_mode (fd) < 0) |
bb70624e | 4403 | { |
b80f6443 | 4404 | sys_error (_("cannout reset nodelay mode for fd %d"), fd); |
bb70624e JA |
4405 | exit (127); |
4406 | } | |
4407 | } | |
726f6388 JA |
4408 | #else /* HAVE_DEV_FD */ |
4409 | fd = child_pipe_fd; | |
4410 | #endif /* HAVE_DEV_FD */ | |
4411 | ||
4412 | if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0) | |
4413 | { | |
b80f6443 | 4414 | sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname, |
ccc6cda3 | 4415 | open_for_read_in_child ? 0 : 1); |
726f6388 JA |
4416 | exit (127); |
4417 | } | |
4418 | ||
f73dda09 JA |
4419 | if (fd != (open_for_read_in_child ? 0 : 1)) |
4420 | close (fd); | |
726f6388 JA |
4421 | |
4422 | /* Need to close any files that this process has open to pipes inherited | |
4423 | from its parent. */ | |
4424 | if (current_fds_to_close) | |
4425 | { | |
4426 | close_fd_bitmap (current_fds_to_close); | |
4427 | current_fds_to_close = (struct fd_bitmap *)NULL; | |
4428 | } | |
4429 | ||
4430 | #if defined (HAVE_DEV_FD) | |
4431 | /* Make sure we close the parent's end of the pipe and clear the slot | |
4432 | in the fd list so it is not closed later, if reallocated by, for | |
4433 | instance, pipe(2). */ | |
4434 | close (parent_pipe_fd); | |
4435 | dev_fd_list[parent_pipe_fd] = 0; | |
4436 | #endif /* HAVE_DEV_FD */ | |
4437 | ||
d166f048 | 4438 | result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST)); |
726f6388 JA |
4439 | |
4440 | #if !defined (HAVE_DEV_FD) | |
4441 | /* Make sure we close the named pipe in the child before we exit. */ | |
4442 | close (open_for_read_in_child ? 0 : 1); | |
4443 | #endif /* !HAVE_DEV_FD */ | |
4444 | ||
4445 | exit (result); | |
4446 | /*NOTREACHED*/ | |
4447 | } | |
4448 | #endif /* PROCESS_SUBSTITUTION */ | |
4449 | ||
cce855bc JA |
4450 | /***********************************/ |
4451 | /* */ | |
4452 | /* Command Substitution */ | |
4453 | /* */ | |
4454 | /***********************************/ | |
4455 | ||
d166f048 JA |
4456 | static char * |
4457 | read_comsub (fd, quoted) | |
4458 | int fd, quoted; | |
4459 | { | |
4460 | char *istring, buf[128], *bufp; | |
f73dda09 JA |
4461 | int istring_index, istring_size, c; |
4462 | ssize_t bufn; | |
d166f048 JA |
4463 | |
4464 | istring = (char *)NULL; | |
4465 | istring_index = istring_size = bufn = 0; | |
4466 | ||
28ef6c31 JA |
4467 | #ifdef __CYGWIN__ |
4468 | setmode (fd, O_TEXT); /* we don't want CR/LF, we want Unix-style */ | |
4469 | #endif | |
4470 | ||
d166f048 JA |
4471 | /* Read the output of the command through the pipe. */ |
4472 | while (1) | |
4473 | { | |
4474 | if (fd < 0) | |
28ef6c31 | 4475 | break; |
d166f048 JA |
4476 | if (--bufn <= 0) |
4477 | { | |
bb70624e | 4478 | bufn = zread (fd, buf, sizeof (buf)); |
d166f048 JA |
4479 | if (bufn <= 0) |
4480 | break; | |
4481 | bufp = buf; | |
4482 | } | |
4483 | c = *bufp++; | |
4484 | ||
28ef6c31 JA |
4485 | if (c == 0) |
4486 | { | |
4487 | #if 0 | |
4488 | internal_warning ("read_comsub: ignored null byte in input"); | |
4489 | #endif | |
4490 | continue; | |
4491 | } | |
4492 | ||
d166f048 JA |
4493 | /* Add the character to ISTRING, possibly after resizing it. */ |
4494 | RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE); | |
4495 | ||
f1be666c JA |
4496 | /* This is essentially quote_string inline */ |
4497 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */) | |
4498 | istring[istring_index++] = CTLESC; | |
4499 | /* Escape CTLESC and CTLNUL in the output to protect those characters | |
4500 | from the rest of the word expansions (word splitting and globbing.) | |
4501 | This is essentially quote_escapes inline. */ | |
4502 | else if (c == CTLESC) | |
4503 | istring[istring_index++] = CTLESC; | |
4504 | else if (c == CTLNUL || (c == ' ' && (ifs_value && *ifs_value == 0))) | |
d166f048 JA |
4505 | istring[istring_index++] = CTLESC; |
4506 | ||
4507 | istring[istring_index++] = c; | |
28ef6c31 JA |
4508 | |
4509 | #if 0 | |
4510 | #if defined (__CYGWIN__) | |
4511 | if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r') | |
4512 | { | |
4513 | istring_index--; | |
4514 | istring[istring_index - 1] = '\n'; | |
4515 | } | |
4516 | #endif | |
4517 | #endif | |
d166f048 JA |
4518 | } |
4519 | ||
4520 | if (istring) | |
4521 | istring[istring_index] = '\0'; | |
4522 | ||
4523 | /* If we read no output, just return now and save ourselves some | |
4524 | trouble. */ | |
4525 | if (istring_index == 0) | |
4526 | { | |
4527 | FREE (istring); | |
4528 | return (char *)NULL; | |
4529 | } | |
4530 | ||
4531 | /* Strip trailing newlines from the output of the command. */ | |
4532 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
4533 | { | |
4534 | while (istring_index > 0) | |
4535 | { | |
4536 | if (istring[istring_index - 1] == '\n') | |
4537 | { | |
4538 | --istring_index; | |
4539 | ||
4540 | /* If the newline was quoted, remove the quoting char. */ | |
4541 | if (istring[istring_index - 1] == CTLESC) | |
4542 | --istring_index; | |
4543 | } | |
4544 | else | |
4545 | break; | |
4546 | } | |
4547 | istring[istring_index] = '\0'; | |
4548 | } | |
4549 | else | |
4550 | strip_trailing (istring, istring_index - 1, 1); | |
4551 | ||
4552 | return istring; | |
4553 | } | |
4554 | ||
726f6388 JA |
4555 | /* Perform command substitution on STRING. This returns a string, |
4556 | possibly quoted. */ | |
bb70624e | 4557 | char * |
726f6388 JA |
4558 | command_substitute (string, quoted) |
4559 | char *string; | |
4560 | int quoted; | |
4561 | { | |
95732b49 | 4562 | pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid; |
ccc6cda3 | 4563 | char *istring; |
b80f6443 | 4564 | int result, fildes[2], function_value, pflags, rc; |
726f6388 | 4565 | |
ccc6cda3 | 4566 | istring = (char *)NULL; |
726f6388 JA |
4567 | |
4568 | /* Don't fork () if there is no need to. In the case of no command to | |
4569 | run, just return NULL. */ | |
4570 | if (!string || !*string || (string[0] == '\n' && !string[1])) | |
4571 | return ((char *)NULL); | |
4572 | ||
cce855bc JA |
4573 | if (wordexp_only && read_but_dont_execute) |
4574 | { | |
4575 | last_command_exit_value = 125; | |
4576 | jump_to_top_level (EXITPROG); | |
4577 | } | |
4578 | ||
bb70624e JA |
4579 | /* We're making the assumption here that the command substitution will |
4580 | eventually run a command from the file system. Since we'll run | |
4581 | maybe_make_export_env in this subshell before executing that command, | |
4582 | the parent shell and any other shells it starts will have to remake | |
4583 | the environment. If we make it before we fork, other shells won't | |
4584 | have to. Don't bother if we have any temporary variable assignments, | |
4585 | though, because the export environment will be remade after this | |
4586 | command completes anyway, but do it if all the words to be expanded | |
4587 | are variable assignments. */ | |
4588 | if (subst_assign_varlist == 0 || garglist == 0) | |
4589 | maybe_make_export_env (); /* XXX */ | |
4590 | ||
b80f6443 JA |
4591 | /* Flags to pass to parse_and_execute() */ |
4592 | pflags = interactive ? SEVAL_RESETLINE : 0; | |
4593 | ||
726f6388 JA |
4594 | /* Pipe the output of executing STRING into the current shell. */ |
4595 | if (pipe (fildes) < 0) | |
4596 | { | |
b80f6443 | 4597 | sys_error (_("cannot make pipe for command substitution")); |
726f6388 JA |
4598 | goto error_exit; |
4599 | } | |
4600 | ||
4601 | old_pid = last_made_pid; | |
4602 | #if defined (JOB_CONTROL) | |
ccc6cda3 | 4603 | old_pipeline_pgrp = pipeline_pgrp; |
28ef6c31 JA |
4604 | /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */ |
4605 | if ((subshell_environment & SUBSHELL_PIPE) == 0) | |
4606 | pipeline_pgrp = shell_pgrp; | |
ccc6cda3 | 4607 | cleanup_the_pipeline (); |
95732b49 | 4608 | #endif /* JOB_CONTROL */ |
726f6388 | 4609 | |
95732b49 JA |
4610 | old_async_pid = last_asynchronous_pid; |
4611 | #if 0 | |
ccc6cda3 | 4612 | pid = make_child ((char *)NULL, 0); |
95732b49 JA |
4613 | #else |
4614 | pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC); | |
4615 | #endif | |
4616 | last_asynchronous_pid = old_async_pid; | |
4617 | ||
726f6388 JA |
4618 | if (pid == 0) |
4619 | /* Reset the signal handlers in the child, but don't free the | |
4620 | trap strings. */ | |
4621 | reset_signal_handlers (); | |
ccc6cda3 JA |
4622 | |
4623 | #if defined (JOB_CONTROL) | |
4624 | set_sigchld_handler (); | |
4625 | stop_making_children (); | |
f1be666c JA |
4626 | if (pid != 0) |
4627 | pipeline_pgrp = old_pipeline_pgrp; | |
f73dda09 JA |
4628 | #else |
4629 | stop_making_children (); | |
ccc6cda3 | 4630 | #endif /* JOB_CONTROL */ |
726f6388 JA |
4631 | |
4632 | if (pid < 0) | |
4633 | { | |
b80f6443 | 4634 | sys_error (_("cannot make child for command substitution")); |
726f6388 JA |
4635 | error_exit: |
4636 | ||
4637 | FREE (istring); | |
4638 | close (fildes[0]); | |
4639 | close (fildes[1]); | |
4640 | return ((char *)NULL); | |
4641 | } | |
4642 | ||
4643 | if (pid == 0) | |
4644 | { | |
4645 | set_sigint_handler (); /* XXX */ | |
28ef6c31 | 4646 | |
b80f6443 JA |
4647 | free_pushed_string_input (); |
4648 | ||
726f6388 JA |
4649 | if (dup2 (fildes[1], 1) < 0) |
4650 | { | |
b80f6443 | 4651 | sys_error (_("command_substitute: cannot duplicate pipe as fd 1")); |
726f6388 JA |
4652 | exit (EXECUTION_FAILURE); |
4653 | } | |
4654 | ||
4655 | /* If standard output is closed in the parent shell | |
4656 | (such as after `exec >&-'), file descriptor 1 will be | |
4657 | the lowest available file descriptor, and end up in | |
4658 | fildes[0]. This can happen for stdin and stderr as well, | |
4659 | but stdout is more important -- it will cause no output | |
4660 | to be generated from this command. */ | |
4661 | if ((fildes[1] != fileno (stdin)) && | |
4662 | (fildes[1] != fileno (stdout)) && | |
4663 | (fildes[1] != fileno (stderr))) | |
4664 | close (fildes[1]); | |
4665 | ||
4666 | if ((fildes[0] != fileno (stdin)) && | |
4667 | (fildes[0] != fileno (stdout)) && | |
4668 | (fildes[0] != fileno (stderr))) | |
4669 | close (fildes[0]); | |
4670 | ||
4671 | /* The currently executing shell is not interactive. */ | |
4672 | interactive = 0; | |
4673 | ||
ccc6cda3 | 4674 | /* This is a subshell environment. */ |
28ef6c31 | 4675 | subshell_environment |= SUBSHELL_COMSUB; |
ccc6cda3 | 4676 | |
28ef6c31 | 4677 | /* When not in POSIX mode, command substitution does not inherit |
7117c2d2 | 4678 | the -e flag. */ |
28ef6c31 JA |
4679 | if (posixly_correct == 0) |
4680 | exit_immediately_on_error = 0; | |
726f6388 JA |
4681 | |
4682 | remove_quoted_escapes (string); | |
4683 | ||
ccc6cda3 | 4684 | startup_state = 2; /* see if we can avoid a fork */ |
726f6388 JA |
4685 | /* Give command substitution a place to jump back to on failure, |
4686 | so we don't go back up to main (). */ | |
4687 | result = setjmp (top_level); | |
4688 | ||
bb70624e JA |
4689 | /* If we're running a command substitution inside a shell function, |
4690 | trap `return' so we don't return from the function in the subshell | |
4691 | and go off to never-never land. */ | |
4692 | if (result == 0 && return_catch_flag) | |
4693 | function_value = setjmp (return_catch); | |
4694 | else | |
4695 | function_value = 0; | |
4696 | ||
b80f6443 JA |
4697 | if (result == ERREXIT) |
4698 | rc = last_command_exit_value; | |
4699 | else if (result == EXITPROG) | |
4700 | rc = last_command_exit_value; | |
726f6388 | 4701 | else if (result) |
b80f6443 | 4702 | rc = EXECUTION_FAILURE; |
bb70624e | 4703 | else if (function_value) |
b80f6443 | 4704 | rc = return_catch_value; |
726f6388 | 4705 | else |
b80f6443 JA |
4706 | { |
4707 | subshell_level++; | |
4708 | rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST); | |
4709 | subshell_level--; | |
4710 | } | |
4711 | ||
4712 | last_command_exit_value = rc; | |
4713 | rc = run_exit_trap (); | |
f1be666c JA |
4714 | #if defined (PROCESS_SUBSTITUTION) |
4715 | unlink_fifo_list (); | |
4716 | #endif | |
b80f6443 | 4717 | exit (rc); |
726f6388 JA |
4718 | } |
4719 | else | |
4720 | { | |
726f6388 JA |
4721 | #if defined (JOB_CONTROL) && defined (PGRP_PIPE) |
4722 | close_pgrp_pipe (); | |
4723 | #endif /* JOB_CONTROL && PGRP_PIPE */ | |
4724 | ||
4725 | close (fildes[1]); | |
4726 | ||
d166f048 | 4727 | istring = read_comsub (fildes[0], quoted); |
ccc6cda3 | 4728 | |
726f6388 JA |
4729 | close (fildes[0]); |
4730 | ||
b72432fd | 4731 | current_command_subst_pid = pid; |
726f6388 JA |
4732 | last_command_exit_value = wait_for (pid); |
4733 | last_command_subst_pid = pid; | |
4734 | last_made_pid = old_pid; | |
4735 | ||
4736 | #if defined (JOB_CONTROL) | |
4737 | /* If last_command_exit_value > 128, then the substituted command | |
4738 | was terminated by a signal. If that signal was SIGINT, then send | |
4739 | SIGINT to ourselves. This will break out of loops, for instance. */ | |
b80f6443 | 4740 | if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT) |
726f6388 JA |
4741 | kill (getpid (), SIGINT); |
4742 | ||
4743 | /* wait_for gives the terminal back to shell_pgrp. If some other | |
cce855bc JA |
4744 | process group should have it, give it away to that group here. |
4745 | pipeline_pgrp is non-zero only while we are constructing a | |
4746 | pipline, so what we are concerned about is whether or not that | |
4747 | pipeline was started in the background. A pipeline started in | |
4748 | the background should never get the tty back here. */ | |
4749 | #if 0 | |
4750 | if (interactive && pipeline_pgrp != (pid_t)0 && pipeline_pgrp != last_asynchronous_pid) | |
4751 | #else | |
28ef6c31 | 4752 | if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0) |
cce855bc | 4753 | #endif |
28ef6c31 | 4754 | give_terminal_to (pipeline_pgrp, 0); |
726f6388 JA |
4755 | #endif /* JOB_CONTROL */ |
4756 | ||
726f6388 JA |
4757 | return (istring); |
4758 | } | |
4759 | } | |
4760 | ||
4761 | /******************************************************** | |
4762 | * * | |
4763 | * Utility functions for parameter expansion * | |
4764 | * * | |
4765 | ********************************************************/ | |
4766 | ||
ccc6cda3 | 4767 | #if defined (ARRAY_VARS) |
ccc6cda3 | 4768 | |
f73dda09 | 4769 | static arrayind_t |
ccc6cda3 JA |
4770 | array_length_reference (s) |
4771 | char *s; | |
4772 | { | |
f73dda09 JA |
4773 | int len; |
4774 | arrayind_t ind; | |
4775 | char *t, c; | |
ccc6cda3 JA |
4776 | ARRAY *array; |
4777 | SHELL_VAR *var; | |
4778 | ||
4779 | var = array_variable_part (s, &t, &len); | |
726f6388 | 4780 | |
ccc6cda3 JA |
4781 | /* If unbound variables should generate an error, report one and return |
4782 | failure. */ | |
4783 | if ((var == 0 || array_p (var) == 0) && unbound_vars_is_error) | |
726f6388 | 4784 | { |
f73dda09 | 4785 | c = *--t; |
ccc6cda3 | 4786 | *t = '\0'; |
7117c2d2 | 4787 | err_unboundvar (s); |
f73dda09 | 4788 | *t = c; |
ccc6cda3 | 4789 | return (-1); |
726f6388 | 4790 | } |
ccc6cda3 JA |
4791 | else if (var == 0) |
4792 | return 0; | |
726f6388 | 4793 | |
28ef6c31 JA |
4794 | /* We support a couple of expansions for variables that are not arrays. |
4795 | We'll return the length of the value for v[0], and 1 for v[@] or | |
4796 | v[*]. Return 0 for everything else. */ | |
4797 | ||
4798 | array = array_p (var) ? array_cell (var) : (ARRAY *)NULL; | |
726f6388 | 4799 | |
ccc6cda3 | 4800 | if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']') |
28ef6c31 | 4801 | return (array_p (var) ? array_num_elements (array) : 1); |
ccc6cda3 JA |
4802 | |
4803 | ind = array_expand_index (t, len); | |
4804 | if (ind < 0) | |
4805 | { | |
7117c2d2 | 4806 | err_badarraysub (t); |
ccc6cda3 JA |
4807 | return (-1); |
4808 | } | |
ccc6cda3 | 4809 | |
28ef6c31 JA |
4810 | if (array_p (var)) |
4811 | t = array_reference (array, ind); | |
4812 | else | |
4813 | t = (ind == 0) ? value_cell (var) : (char *)NULL; | |
4814 | ||
f1be666c | 4815 | len = MB_STRLEN (t); |
ccc6cda3 | 4816 | return (len); |
726f6388 | 4817 | } |
ccc6cda3 | 4818 | #endif /* ARRAY_VARS */ |
726f6388 JA |
4819 | |
4820 | static int | |
4821 | valid_brace_expansion_word (name, var_is_special) | |
4822 | char *name; | |
4823 | int var_is_special; | |
4824 | { | |
f73dda09 | 4825 | if (DIGIT (*name) && all_digits (name)) |
726f6388 JA |
4826 | return 1; |
4827 | else if (var_is_special) | |
4828 | return 1; | |
ccc6cda3 JA |
4829 | #if defined (ARRAY_VARS) |
4830 | else if (valid_array_reference (name)) | |
4831 | return 1; | |
4832 | #endif /* ARRAY_VARS */ | |
726f6388 JA |
4833 | else if (legal_identifier (name)) |
4834 | return 1; | |
4835 | else | |
4836 | return 0; | |
4837 | } | |
ccc6cda3 | 4838 | |
b80f6443 JA |
4839 | static int |
4840 | chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at) | |
4841 | char *name; | |
4842 | int quoted; | |
4843 | int *quoted_dollar_atp, *contains_dollar_at; | |
4844 | { | |
4845 | char *temp1; | |
4846 | ||
4847 | if (name == 0) | |
4848 | { | |
4849 | if (quoted_dollar_atp) | |
4850 | *quoted_dollar_atp = 0; | |
4851 | if (contains_dollar_at) | |
4852 | *contains_dollar_at = 0; | |
4853 | return 0; | |
4854 | } | |
4855 | ||
4856 | /* check for $@ and $* */ | |
4857 | if (name[0] == '@' && name[1] == 0) | |
4858 | { | |
4859 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
4860 | *quoted_dollar_atp = 1; | |
4861 | if (contains_dollar_at) | |
4862 | *contains_dollar_at = 1; | |
4863 | return 1; | |
4864 | } | |
4865 | else if (name[0] == '*' && name[1] == '\0' && quoted == 0) | |
4866 | { | |
4867 | if (contains_dollar_at) | |
4868 | *contains_dollar_at = 1; | |
4869 | return 1; | |
4870 | } | |
4871 | ||
4872 | /* Now check for ${array[@]} and ${array[*]} */ | |
4873 | #if defined (ARRAY_VARS) | |
4874 | else if (valid_array_reference (name)) | |
4875 | { | |
4876 | temp1 = xstrchr (name, '['); | |
4877 | if (temp1 && temp1[1] == '@' && temp1[2] == ']') | |
4878 | { | |
4879 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
4880 | *quoted_dollar_atp = 1; | |
4881 | if (contains_dollar_at) | |
4882 | *contains_dollar_at = 1; | |
4883 | return 1; | |
4884 | } /* [ */ | |
4885 | /* ${array[*]}, when unquoted, should be treated like ${array[@]}, | |
4886 | which should result in separate words even when IFS is unset. */ | |
4887 | if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0) | |
4888 | { | |
4889 | if (contains_dollar_at) | |
4890 | *contains_dollar_at = 1; | |
4891 | return 1; | |
4892 | } | |
4893 | } | |
4894 | #endif | |
4895 | return 0; | |
4896 | } | |
4897 | ||
726f6388 JA |
4898 | /* Parameter expand NAME, and return a new string which is the expansion, |
4899 | or NULL if there was no expansion. | |
4900 | VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in | |
4901 | the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that | |
4902 | NAME was found inside of a double-quoted expression. */ | |
95732b49 | 4903 | static WORD_DESC * |
726f6388 JA |
4904 | parameter_brace_expand_word (name, var_is_special, quoted) |
4905 | char *name; | |
4906 | int var_is_special, quoted; | |
4907 | { | |
95732b49 | 4908 | WORD_DESC *ret; |
ccc6cda3 | 4909 | char *temp, *tt; |
7117c2d2 | 4910 | intmax_t arg_index; |
ccc6cda3 | 4911 | SHELL_VAR *var; |
f1be666c | 4912 | int atype, rflags; |
726f6388 | 4913 | |
95732b49 JA |
4914 | ret = 0; |
4915 | temp = 0; | |
f1be666c | 4916 | rflags = 0; |
95732b49 JA |
4917 | |
4918 | /* Handle multiple digit arguments, as in ${11}. */ | |
f73dda09 | 4919 | if (legal_number (name, &arg_index)) |
7117c2d2 JA |
4920 | { |
4921 | tt = get_dollar_var_value (arg_index); | |
b80f6443 JA |
4922 | if (tt) |
4923 | temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
4924 | ? quote_string (tt) | |
4925 | : quote_escapes (tt); | |
4926 | else | |
4927 | temp = (char *)NULL; | |
7117c2d2 JA |
4928 | FREE (tt); |
4929 | } | |
726f6388 JA |
4930 | else if (var_is_special) /* ${@} */ |
4931 | { | |
cce855bc | 4932 | int sindex; |
f73dda09 | 4933 | tt = (char *)xmalloc (2 + strlen (name)); |
cce855bc | 4934 | tt[sindex = 0] = '$'; |
726f6388 | 4935 | strcpy (tt + 1, name); |
7117c2d2 | 4936 | |
95732b49 JA |
4937 | ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL, |
4938 | (int *)NULL, (int *)NULL, 0); | |
cce855bc | 4939 | free (tt); |
726f6388 | 4940 | } |
ccc6cda3 JA |
4941 | #if defined (ARRAY_VARS) |
4942 | else if (valid_array_reference (name)) | |
4943 | { | |
7117c2d2 JA |
4944 | temp = array_value (name, quoted, &atype); |
4945 | if (atype == 0 && temp) | |
b80f6443 JA |
4946 | temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
4947 | ? quote_string (temp) | |
4948 | : quote_escapes (temp); | |
f1be666c JA |
4949 | else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
4950 | rflags |= W_HASQUOTEDNULL; | |
ccc6cda3 JA |
4951 | } |
4952 | #endif | |
4953 | else if (var = find_variable (name)) | |
4954 | { | |
7117c2d2 | 4955 | if (var_isset (var) && invisible_p (var) == 0) |
28ef6c31 | 4956 | { |
ccc6cda3 JA |
4957 | #if defined (ARRAY_VARS) |
4958 | temp = array_p (var) ? array_reference (array_cell (var), 0) : value_cell (var); | |
4959 | #else | |
4960 | temp = value_cell (var); | |
4961 | #endif | |
4962 | ||
4963 | if (temp) | |
b80f6443 JA |
4964 | temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
4965 | ? quote_string (temp) | |
4966 | : quote_escapes (temp); | |
28ef6c31 | 4967 | } |
ccc6cda3 JA |
4968 | else |
4969 | temp = (char *)NULL; | |
4970 | } | |
726f6388 | 4971 | else |
ccc6cda3 | 4972 | temp = (char *)NULL; |
726f6388 | 4973 | |
95732b49 JA |
4974 | if (ret == 0) |
4975 | { | |
4976 | ret = alloc_word_desc (); | |
4977 | ret->word = temp; | |
f1be666c | 4978 | ret->flags |= rflags; |
95732b49 JA |
4979 | } |
4980 | return ret; | |
726f6388 JA |
4981 | } |
4982 | ||
ccc6cda3 JA |
4983 | /* Expand an indirect reference to a variable: ${!NAME} expands to the |
4984 | value of the variable whose name is the value of NAME. */ | |
95732b49 | 4985 | static WORD_DESC * |
b80f6443 | 4986 | parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at) |
ccc6cda3 JA |
4987 | char *name; |
4988 | int var_is_special, quoted; | |
b80f6443 | 4989 | int *quoted_dollar_atp, *contains_dollar_at; |
ccc6cda3 JA |
4990 | { |
4991 | char *temp, *t; | |
95732b49 | 4992 | WORD_DESC *w; |
ccc6cda3 | 4993 | |
95732b49 JA |
4994 | w = parameter_brace_expand_word (name, var_is_special, quoted); |
4995 | t = w->word; | |
b80f6443 JA |
4996 | /* Have to dequote here if necessary */ |
4997 | if (t) | |
4998 | { | |
4999 | temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
5000 | ? dequote_string (t) | |
5001 | : dequote_escapes (t); | |
5002 | free (t); | |
5003 | t = temp; | |
5004 | } | |
95732b49 JA |
5005 | dispose_word_desc (w); |
5006 | ||
b80f6443 | 5007 | chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at); |
ccc6cda3 | 5008 | if (t == 0) |
95732b49 JA |
5009 | return (WORD_DESC *)NULL; |
5010 | ||
5011 | w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted); | |
ccc6cda3 | 5012 | free (t); |
95732b49 JA |
5013 | |
5014 | return w; | |
ccc6cda3 JA |
5015 | } |
5016 | ||
726f6388 JA |
5017 | /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE}, |
5018 | depending on the value of C, the separating character. C can be one of | |
ccc6cda3 JA |
5019 | "-", "+", or "=". QUOTED is true if the entire brace expression occurs |
5020 | between double quotes. */ | |
95732b49 | 5021 | static WORD_DESC * |
ccc6cda3 | 5022 | parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat) |
726f6388 | 5023 | char *name, *value; |
ccc6cda3 | 5024 | int c, quoted, *qdollaratp, *hasdollarat; |
726f6388 | 5025 | { |
95732b49 | 5026 | WORD_DESC *w; |
726f6388 JA |
5027 | WORD_LIST *l; |
5028 | char *t, *t1, *temp; | |
ccc6cda3 | 5029 | int hasdol; |
726f6388 | 5030 | |
ccc6cda3 JA |
5031 | /* If the entire expression is between double quotes, we want to treat |
5032 | the value as a double-quoted string, with the exception that we strip | |
5033 | embedded unescaped double quotes. */ | |
95732b49 | 5034 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value) |
726f6388 | 5035 | { |
ccc6cda3 | 5036 | hasdol = 0; |
95732b49 | 5037 | temp = string_extract_double_quoted (value, &hasdol, 1); |
726f6388 | 5038 | } |
95732b49 JA |
5039 | else |
5040 | temp = value; | |
ccc6cda3 | 5041 | |
95732b49 | 5042 | w = alloc_word_desc (); |
726f6388 | 5043 | hasdol = 0; |
ccc6cda3 JA |
5044 | /* XXX was 0 not quoted */ |
5045 | l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL) | |
5046 | : (WORD_LIST *)0; | |
5047 | if (hasdollarat) | |
5048 | *hasdollarat = hasdol || (l && l->next); | |
95732b49 JA |
5049 | if (temp != value) |
5050 | free (temp); | |
726f6388 JA |
5051 | if (l) |
5052 | { | |
ccc6cda3 | 5053 | /* The expansion of TEMP returned something. We need to treat things |
b80f6443 JA |
5054 | slightly differently if HASDOL is non-zero. If we have "$@", the |
5055 | individual words have already been quoted. We need to turn them | |
5056 | into a string with the words separated by the first character of | |
5057 | $IFS without any additional quoting, so string_list_dollar_at won't | |
5058 | do the right thing. We use string_list_dollar_star instead. */ | |
5059 | temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l); | |
5060 | ||
ccc6cda3 JA |
5061 | /* If l->next is not null, we know that TEMP contained "$@", since that |
5062 | is the only expansion that creates more than one word. */ | |
b80f6443 | 5063 | if (qdollaratp && ((hasdol && quoted) || l->next)) |
ccc6cda3 | 5064 | *qdollaratp = 1; |
726f6388 JA |
5065 | dispose_words (l); |
5066 | } | |
ccc6cda3 | 5067 | else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol) |
726f6388 | 5068 | { |
ccc6cda3 JA |
5069 | /* The brace expansion occurred between double quotes and there was |
5070 | a $@ in TEMP. It does not matter if the $@ is quoted, as long as | |
7117c2d2 JA |
5071 | it does not expand to anything. In this case, we want to return |
5072 | a quoted empty string. */ | |
0628567a | 5073 | temp = make_quoted_char ('\0'); |
95732b49 | 5074 | w->flags |= W_HASQUOTEDNULL; |
726f6388 JA |
5075 | } |
5076 | else | |
5077 | temp = (char *)NULL; | |
5078 | ||
5079 | if (c == '-' || c == '+') | |
95732b49 JA |
5080 | { |
5081 | w->word = temp; | |
5082 | return w; | |
5083 | } | |
726f6388 JA |
5084 | |
5085 | /* c == '=' */ | |
ccc6cda3 | 5086 | t = temp ? savestring (temp) : savestring (""); |
726f6388 JA |
5087 | t1 = dequote_string (t); |
5088 | free (t); | |
b80f6443 JA |
5089 | #if defined (ARRAY_VARS) |
5090 | if (valid_array_reference (name)) | |
95732b49 | 5091 | assign_array_element (name, t1, 0); |
b80f6443 JA |
5092 | else |
5093 | #endif /* ARRAY_VARS */ | |
95732b49 | 5094 | bind_variable (name, t1, 0); |
726f6388 | 5095 | free (t1); |
95732b49 JA |
5096 | |
5097 | w->word = temp; | |
5098 | return w; | |
726f6388 JA |
5099 | } |
5100 | ||
5101 | /* Deal with the right hand side of a ${name:?value} expansion in the case | |
5102 | that NAME is null or not set. If VALUE is non-null it is expanded and | |
5103 | used as the error message to print, otherwise a standard message is | |
5104 | printed. */ | |
5105 | static void | |
5106 | parameter_brace_expand_error (name, value) | |
5107 | char *name, *value; | |
5108 | { | |
ccc6cda3 JA |
5109 | WORD_LIST *l; |
5110 | char *temp; | |
5111 | ||
726f6388 JA |
5112 | if (value && *value) |
5113 | { | |
95732b49 | 5114 | l = expand_string (value, 0); |
ccc6cda3 JA |
5115 | temp = string_list (l); |
5116 | report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */ | |
5117 | FREE (temp); | |
726f6388 JA |
5118 | dispose_words (l); |
5119 | } | |
5120 | else | |
b80f6443 | 5121 | report_error (_("%s: parameter null or not set"), name); |
726f6388 JA |
5122 | |
5123 | /* Free the data we have allocated during this expansion, since we | |
5124 | are about to longjmp out. */ | |
5125 | free (name); | |
5126 | FREE (value); | |
5127 | } | |
5128 | ||
5129 | /* Return 1 if NAME is something for which parameter_brace_expand_length is | |
5130 | OK to do. */ | |
5131 | static int | |
5132 | valid_length_expression (name) | |
5133 | char *name; | |
5134 | { | |
28ef6c31 | 5135 | return (name[1] == '\0' || /* ${#} */ |
f73dda09 JA |
5136 | ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */ |
5137 | (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */ | |
ccc6cda3 JA |
5138 | #if defined (ARRAY_VARS) |
5139 | valid_array_reference (name + 1) || /* ${#a[7]} */ | |
5140 | #endif | |
726f6388 JA |
5141 | legal_identifier (name + 1)); /* ${#PS1} */ |
5142 | } | |
5143 | ||
eb873671 JA |
5144 | #if defined (HANDLE_MULTIBYTE) |
5145 | size_t | |
5146 | mbstrlen (s) | |
5147 | const char *s; | |
5148 | { | |
5149 | size_t clen, nc; | |
95732b49 | 5150 | mbstate_t mbs, mbsbak; |
eb873671 JA |
5151 | |
5152 | nc = 0; | |
5153 | memset (&mbs, 0, sizeof (mbs)); | |
95732b49 JA |
5154 | mbsbak = mbs; |
5155 | while ((clen = mbrlen(s, MB_CUR_MAX, &mbs)) != 0) | |
eb873671 | 5156 | { |
95732b49 JA |
5157 | if (MB_INVALIDCH(clen)) |
5158 | { | |
5159 | clen = 1; /* assume single byte */ | |
5160 | mbs = mbsbak; | |
5161 | } | |
5162 | ||
eb873671 JA |
5163 | s += clen; |
5164 | nc++; | |
95732b49 | 5165 | mbsbak = mbs; |
eb873671 JA |
5166 | } |
5167 | return nc; | |
5168 | } | |
5169 | #endif | |
5170 | ||
5171 | ||
726f6388 JA |
5172 | /* Handle the parameter brace expansion that requires us to return the |
5173 | length of a parameter. */ | |
7117c2d2 | 5174 | static intmax_t |
726f6388 JA |
5175 | parameter_brace_expand_length (name) |
5176 | char *name; | |
5177 | { | |
ccc6cda3 | 5178 | char *t, *newname; |
7117c2d2 | 5179 | intmax_t number, arg_index; |
ccc6cda3 JA |
5180 | WORD_LIST *list; |
5181 | #if defined (ARRAY_VARS) | |
5182 | SHELL_VAR *var; | |
5183 | #endif | |
5184 | ||
5185 | if (name[1] == '\0') /* ${#} */ | |
5186 | number = number_of_args (); | |
cce855bc JA |
5187 | else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */ |
5188 | number = number_of_args (); | |
f73dda09 | 5189 | else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') |
cce855bc JA |
5190 | { |
5191 | /* Take the lengths of some of the shell's special parameters. */ | |
5192 | switch (name[1]) | |
5193 | { | |
5194 | case '-': | |
5195 | t = which_set_flags (); | |
5196 | break; | |
5197 | case '?': | |
5198 | t = itos (last_command_exit_value); | |
5199 | break; | |
5200 | case '$': | |
5201 | t = itos (dollar_dollar_pid); | |
5202 | break; | |
5203 | case '!': | |
5204 | if (last_asynchronous_pid == NO_PID) | |
5205 | t = (char *)NULL; | |
5206 | else | |
f73dda09 | 5207 | t = itos (last_asynchronous_pid); |
cce855bc JA |
5208 | break; |
5209 | case '#': | |
5210 | t = itos (number_of_args ()); | |
5211 | break; | |
5212 | } | |
5213 | number = STRLEN (t); | |
5214 | FREE (t); | |
5215 | } | |
ccc6cda3 JA |
5216 | #if defined (ARRAY_VARS) |
5217 | else if (valid_array_reference (name + 1)) | |
5218 | number = array_length_reference (name + 1); | |
5219 | #endif /* ARRAY_VARS */ | |
cce855bc | 5220 | else |
ccc6cda3 JA |
5221 | { |
5222 | number = 0; | |
5223 | ||
f73dda09 | 5224 | if (legal_number (name + 1, &arg_index)) /* ${#1} */ |
ccc6cda3 | 5225 | { |
f73dda09 | 5226 | t = get_dollar_var_value (arg_index); |
eb873671 | 5227 | number = MB_STRLEN (t); |
ccc6cda3 JA |
5228 | FREE (t); |
5229 | } | |
5230 | #if defined (ARRAY_VARS) | |
eb873671 | 5231 | else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && array_p (var)) |
ccc6cda3 JA |
5232 | { |
5233 | t = array_reference (array_cell (var), 0); | |
eb873671 | 5234 | number = MB_STRLEN (t); |
ccc6cda3 JA |
5235 | } |
5236 | #endif | |
5237 | else /* ${#PS1} */ | |
5238 | { | |
5239 | newname = savestring (name); | |
5240 | newname[0] = '$'; | |
5241 | list = expand_string (newname, Q_DOUBLE_QUOTES); | |
5242 | t = list ? string_list (list) : (char *)NULL; | |
5243 | free (newname); | |
5244 | if (list) | |
5245 | dispose_words (list); | |
5246 | ||
eb873671 | 5247 | number = MB_STRLEN (t); |
ccc6cda3 JA |
5248 | FREE (t); |
5249 | } | |
5250 | } | |
ccc6cda3 JA |
5251 | |
5252 | return (number); | |
5253 | } | |
5254 | ||
28ef6c31 JA |
5255 | /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression, |
5256 | so we do some ad-hoc parsing of an arithmetic expression to find | |
5257 | the first DELIM, instead of using strchr(3). Two rules: | |
5258 | 1. If the substring contains a `(', read until closing `)'. | |
5259 | 2. If the substring contains a `?', read past one `:' for each `?'. | |
5260 | */ | |
5261 | ||
5262 | static char * | |
5263 | skiparith (substr, delim) | |
5264 | char *substr; | |
5265 | int delim; | |
5266 | { | |
7117c2d2 JA |
5267 | size_t sublen; |
5268 | int skipcol, pcount, i; | |
5269 | DECLARE_MBSTATE; | |
28ef6c31 | 5270 | |
7117c2d2 JA |
5271 | sublen = strlen (substr); |
5272 | i = skipcol = pcount = 0; | |
5273 | while (substr[i]) | |
28ef6c31 JA |
5274 | { |
5275 | /* Balance parens */ | |
7117c2d2 | 5276 | if (substr[i] == LPAREN) |
28ef6c31 JA |
5277 | { |
5278 | pcount++; | |
7117c2d2 | 5279 | i++; |
28ef6c31 JA |
5280 | continue; |
5281 | } | |
7117c2d2 | 5282 | if (substr[i] == RPAREN && pcount) |
28ef6c31 JA |
5283 | { |
5284 | pcount--; | |
7117c2d2 | 5285 | i++; |
28ef6c31 JA |
5286 | continue; |
5287 | } | |
5288 | if (pcount) | |
7117c2d2 JA |
5289 | { |
5290 | ADVANCE_CHAR (substr, sublen, i); | |
5291 | continue; | |
5292 | } | |
28ef6c31 JA |
5293 | |
5294 | /* Skip one `:' for each `?' */ | |
7117c2d2 | 5295 | if (substr[i] == ':' && skipcol) |
28ef6c31 JA |
5296 | { |
5297 | skipcol--; | |
7117c2d2 | 5298 | i++; |
28ef6c31 JA |
5299 | continue; |
5300 | } | |
7117c2d2 | 5301 | if (substr[i] == delim) |
28ef6c31 | 5302 | break; |
7117c2d2 | 5303 | if (substr[i] == '?') |
28ef6c31 JA |
5304 | { |
5305 | skipcol++; | |
7117c2d2 | 5306 | i++; |
28ef6c31 JA |
5307 | continue; |
5308 | } | |
7117c2d2 | 5309 | ADVANCE_CHAR (substr, sublen, i); |
28ef6c31 | 5310 | } |
7117c2d2 JA |
5311 | |
5312 | return (substr + i); | |
28ef6c31 JA |
5313 | } |
5314 | ||
ccc6cda3 JA |
5315 | /* Verify and limit the start and end of the desired substring. If |
5316 | VTYPE == 0, a regular shell variable is being used; if it is 1, | |
cce855bc | 5317 | then the positional parameters are being used; if it is 2, then |
e8ce775d JA |
5318 | VALUE is really a pointer to an array variable that should be used. |
5319 | Return value is 1 if both values were OK, 0 if there was a problem | |
5320 | with an invalid expression, or -1 if the values were out of range. */ | |
ccc6cda3 JA |
5321 | static int |
5322 | verify_substring_values (value, substr, vtype, e1p, e2p) | |
5323 | char *value, *substr; | |
f73dda09 | 5324 | int vtype; |
7117c2d2 | 5325 | intmax_t *e1p, *e2p; |
ccc6cda3 | 5326 | { |
bb70624e | 5327 | char *t, *temp1, *temp2; |
f73dda09 JA |
5328 | arrayind_t len; |
5329 | int expok; | |
ccc6cda3 JA |
5330 | #if defined (ARRAY_VARS) |
5331 | ARRAY *a; | |
5332 | #endif | |
5333 | ||
28ef6c31 JA |
5334 | /* duplicate behavior of strchr(3) */ |
5335 | t = skiparith (substr, ':'); | |
5336 | if (*t && *t == ':') | |
7117c2d2 | 5337 | *t = '\0'; |
28ef6c31 JA |
5338 | else |
5339 | t = (char *)0; | |
f73dda09 | 5340 | |
0628567a | 5341 | temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES); |
d166f048 | 5342 | *e1p = evalexp (temp1, &expok); |
ccc6cda3 | 5343 | free (temp1); |
d166f048 JA |
5344 | if (expok == 0) |
5345 | return (0); | |
ccc6cda3 | 5346 | |
f73dda09 | 5347 | len = -1; /* paranoia */ |
ccc6cda3 JA |
5348 | switch (vtype) |
5349 | { | |
5350 | case VT_VARIABLE: | |
d166f048 | 5351 | case VT_ARRAYMEMBER: |
eb873671 | 5352 | len = MB_STRLEN (value); |
ccc6cda3 JA |
5353 | break; |
5354 | case VT_POSPARMS: | |
5355 | len = number_of_args () + 1; | |
5356 | break; | |
5357 | #if defined (ARRAY_VARS) | |
5358 | case VT_ARRAYVAR: | |
5359 | a = (ARRAY *)value; | |
eb873671 JA |
5360 | /* For arrays, the first value deals with array indices. Negative |
5361 | offsets count from one past the array's maximum index. */ | |
5362 | len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */ | |
ccc6cda3 JA |
5363 | break; |
5364 | #endif | |
5365 | } | |
5366 | ||
f73dda09 JA |
5367 | if (len == -1) /* paranoia */ |
5368 | return -1; | |
5369 | ||
ccc6cda3 JA |
5370 | if (*e1p < 0) /* negative offsets count from end */ |
5371 | *e1p += len; | |
5372 | ||
eb873671 | 5373 | if (*e1p > len || *e1p < 0) |
e8ce775d | 5374 | return (-1); |
d166f048 | 5375 | |
b80f6443 JA |
5376 | #if defined (ARRAY_VARS) |
5377 | /* For arrays, the second offset deals with the number of elements. */ | |
5378 | if (vtype == VT_ARRAYVAR) | |
5379 | len = array_num_elements (a); | |
5380 | #endif | |
5381 | ||
ccc6cda3 JA |
5382 | if (t) |
5383 | { | |
5384 | t++; | |
bb70624e | 5385 | temp2 = savestring (t); |
0628567a | 5386 | temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES); |
bb70624e | 5387 | free (temp2); |
ccc6cda3 | 5388 | t[-1] = ':'; |
d166f048 | 5389 | *e2p = evalexp (temp1, &expok); |
ccc6cda3 | 5390 | free (temp1); |
d166f048 | 5391 | if (expok == 0) |
28ef6c31 | 5392 | return (0); |
ccc6cda3 | 5393 | if (*e2p < 0) |
28ef6c31 | 5394 | { |
b80f6443 | 5395 | internal_error (_("%s: substring expression < 0"), t); |
ccc6cda3 | 5396 | return (0); |
28ef6c31 | 5397 | } |
b80f6443 JA |
5398 | #if defined (ARRAY_VARS) |
5399 | /* In order to deal with sparse arrays, push the intelligence about how | |
5400 | to deal with the number of elements desired down to the array- | |
5401 | specific functions. */ | |
5402 | if (vtype != VT_ARRAYVAR) | |
5403 | #endif | |
5404 | { | |
5405 | *e2p += *e1p; /* want E2 chars starting at E1 */ | |
5406 | if (*e2p > len) | |
5407 | *e2p = len; | |
5408 | } | |
ccc6cda3 JA |
5409 | } |
5410 | else | |
5411 | *e2p = len; | |
5412 | ||
5413 | return (1); | |
5414 | } | |
5415 | ||
ccc6cda3 | 5416 | /* Return the type of variable specified by VARNAME (simple variable, |
cce855bc | 5417 | positional param, or array variable). Also return the value specified |
7117c2d2 JA |
5418 | by VARNAME (value of a variable or a reference to an array element). |
5419 | If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL | |
5420 | characters in the value are quoted with CTLESC and takes appropriate | |
5421 | steps. For convenience, *VALP is set to the dequoted VALUE. */ | |
ccc6cda3 | 5422 | static int |
b80f6443 | 5423 | get_var_and_type (varname, value, quoted, varp, valp) |
ccc6cda3 | 5424 | char *varname, *value; |
b80f6443 | 5425 | int quoted; |
ccc6cda3 JA |
5426 | SHELL_VAR **varp; |
5427 | char **valp; | |
5428 | { | |
5429 | int vtype; | |
5430 | char *temp; | |
5431 | #if defined (ARRAY_VARS) | |
5432 | SHELL_VAR *v; | |
5433 | #endif | |
5434 | ||
7117c2d2 JA |
5435 | /* This sets vtype to VT_VARIABLE or VT_POSPARMS */ |
5436 | vtype = (varname[0] == '@' || varname[0] == '*') && varname[1] == '\0'; | |
b80f6443 JA |
5437 | if (vtype == VT_POSPARMS && varname[0] == '*') |
5438 | vtype |= VT_STARSUB; | |
ccc6cda3 JA |
5439 | *varp = (SHELL_VAR *)NULL; |
5440 | ||
5441 | #if defined (ARRAY_VARS) | |
5442 | if (valid_array_reference (varname)) | |
5443 | { | |
5444 | v = array_variable_part (varname, &temp, (int *)0); | |
5445 | if (v && array_p (v)) | |
f73dda09 JA |
5446 | { /* [ */ |
5447 | if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']') | |
ccc6cda3 JA |
5448 | { |
5449 | vtype = VT_ARRAYVAR; | |
b80f6443 JA |
5450 | if (temp[0] == '*') |
5451 | vtype |= VT_STARSUB; | |
ccc6cda3 JA |
5452 | *valp = (char *)array_cell (v); |
5453 | } | |
5454 | else | |
5455 | { | |
d166f048 | 5456 | vtype = VT_ARRAYMEMBER; |
7117c2d2 | 5457 | *valp = array_value (varname, 1, (int *)NULL); |
ccc6cda3 JA |
5458 | } |
5459 | *varp = v; | |
5460 | } | |
95732b49 JA |
5461 | else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')) |
5462 | { | |
5463 | vtype = VT_VARIABLE; | |
5464 | *varp = v; | |
5465 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
5466 | *valp = dequote_string (value); | |
5467 | else | |
5468 | *valp = dequote_escapes (value); | |
5469 | } | |
ccc6cda3 JA |
5470 | else |
5471 | return -1; | |
5472 | } | |
eb873671 | 5473 | else if ((v = find_variable (varname)) && (invisible_p (v) == 0) && array_p (v)) |
ccc6cda3 | 5474 | { |
7117c2d2 | 5475 | vtype = VT_ARRAYMEMBER; |
ccc6cda3 JA |
5476 | *varp = v; |
5477 | *valp = array_reference (array_cell (v), 0); | |
5478 | } | |
5479 | else | |
5480 | #endif | |
b80f6443 JA |
5481 | { |
5482 | if (value && vtype == VT_VARIABLE) | |
5483 | { | |
5484 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
5485 | *valp = dequote_string (value); | |
5486 | else | |
5487 | *valp = dequote_escapes (value); | |
5488 | } | |
5489 | else | |
5490 | *valp = value; | |
5491 | } | |
ccc6cda3 JA |
5492 | |
5493 | return vtype; | |
5494 | } | |
5495 | ||
cce855bc JA |
5496 | /******************************************************/ |
5497 | /* */ | |
5498 | /* Functions to extract substrings of variable values */ | |
5499 | /* */ | |
5500 | /******************************************************/ | |
5501 | ||
b80f6443 JA |
5502 | #if defined (HANDLE_MULTIBYTE) |
5503 | /* Character-oriented rather than strictly byte-oriented substrings. S and | |
5504 | E, rather being strict indices into STRING, indicate character (possibly | |
5505 | multibyte character) positions that require calculation. | |
5506 | Used by the ${param:offset[:length]} expansion. */ | |
5507 | static char * | |
5508 | mb_substring (string, s, e) | |
5509 | char *string; | |
5510 | int s, e; | |
5511 | { | |
5512 | char *tt; | |
5513 | int start, stop, i, slen; | |
5514 | DECLARE_MBSTATE; | |
5515 | ||
5516 | start = 0; | |
95732b49 JA |
5517 | /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */ |
5518 | slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0; | |
b80f6443 JA |
5519 | |
5520 | i = s; | |
5521 | while (string[start] && i--) | |
5522 | ADVANCE_CHAR (string, slen, start); | |
5523 | stop = start; | |
5524 | i = e - s; | |
5525 | while (string[stop] && i--) | |
5526 | ADVANCE_CHAR (string, slen, stop); | |
5527 | tt = substring (string, start, stop); | |
5528 | return tt; | |
5529 | } | |
5530 | #endif | |
5531 | ||
ccc6cda3 JA |
5532 | /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME |
5533 | is `@', use the positional parameters; otherwise, use the value of | |
5534 | VARNAME. If VARNAME is an array variable, use the array elements. */ | |
5535 | ||
5536 | static char * | |
5537 | parameter_brace_substring (varname, value, substr, quoted) | |
5538 | char *varname, *value, *substr; | |
5539 | int quoted; | |
5540 | { | |
7117c2d2 | 5541 | intmax_t e1, e2; |
b80f6443 | 5542 | int vtype, r, starsub; |
0628567a | 5543 | char *temp, *val, *tt, *oname; |
ccc6cda3 JA |
5544 | SHELL_VAR *v; |
5545 | ||
5546 | if (value == 0) | |
5547 | return ((char *)NULL); | |
5548 | ||
0628567a | 5549 | oname = this_command_name; |
ccc6cda3 JA |
5550 | this_command_name = varname; |
5551 | ||
b80f6443 | 5552 | vtype = get_var_and_type (varname, value, quoted, &v, &val); |
ccc6cda3 | 5553 | if (vtype == -1) |
0628567a JA |
5554 | { |
5555 | this_command_name = oname; | |
5556 | return ((char *)NULL); | |
5557 | } | |
ccc6cda3 | 5558 | |
b80f6443 JA |
5559 | starsub = vtype & VT_STARSUB; |
5560 | vtype &= ~VT_STARSUB; | |
5561 | ||
e8ce775d | 5562 | r = verify_substring_values (val, substr, vtype, &e1, &e2); |
0628567a | 5563 | this_command_name = oname; |
e8ce775d | 5564 | if (r <= 0) |
7117c2d2 | 5565 | return ((r == 0) ? &expand_param_error : (char *)NULL); |
ccc6cda3 JA |
5566 | |
5567 | switch (vtype) | |
5568 | { | |
5569 | case VT_VARIABLE: | |
d166f048 | 5570 | case VT_ARRAYMEMBER: |
b80f6443 JA |
5571 | #if defined (HANDLE_MULTIBYTE) |
5572 | if (MB_CUR_MAX > 1) | |
5573 | tt = mb_substring (val, e1, e2); | |
5574 | else | |
5575 | #endif | |
7117c2d2 | 5576 | tt = substring (val, e1, e2); |
b80f6443 | 5577 | |
7117c2d2 JA |
5578 | if (vtype == VT_VARIABLE) |
5579 | FREE (val); | |
5580 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
5581 | temp = quote_string (tt); | |
5582 | else | |
5583 | temp = tt ? quote_escapes (tt) : (char *)NULL; | |
5584 | FREE (tt); | |
ccc6cda3 JA |
5585 | break; |
5586 | case VT_POSPARMS: | |
7117c2d2 JA |
5587 | tt = pos_params (varname, e1, e2, quoted); |
5588 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0) | |
5589 | { | |
5590 | temp = tt ? quote_escapes (tt) : (char *)NULL; | |
5591 | FREE (tt); | |
5592 | } | |
5593 | else | |
5594 | temp = tt; | |
ccc6cda3 JA |
5595 | break; |
5596 | #if defined (ARRAY_VARS) | |
5597 | case VT_ARRAYVAR: | |
b80f6443 JA |
5598 | /* We want E2 to be the number of elements desired (arrays can be sparse, |
5599 | so verify_substring_values just returns the numbers specified and we | |
5600 | rely on array_subrange to understand how to deal with them). */ | |
5601 | tt = array_subrange (array_cell (v), e1, e2, starsub, quoted); | |
f1be666c JA |
5602 | #if 0 |
5603 | /* array_subrange now calls array_quote_escapes as appropriate, so the | |
5604 | caller no longer needs to. */ | |
7117c2d2 JA |
5605 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0) |
5606 | { | |
5607 | temp = tt ? quote_escapes (tt) : (char *)NULL; | |
5608 | FREE (tt); | |
5609 | } | |
5610 | else | |
f1be666c | 5611 | #endif |
7117c2d2 | 5612 | temp = tt; |
ccc6cda3 JA |
5613 | break; |
5614 | #endif | |
f73dda09 JA |
5615 | default: |
5616 | temp = (char *)NULL; | |
ccc6cda3 JA |
5617 | } |
5618 | ||
5619 | return temp; | |
5620 | } | |
5621 | ||
cce855bc JA |
5622 | /****************************************************************/ |
5623 | /* */ | |
5624 | /* Functions to perform pattern substitution on variable values */ | |
5625 | /* */ | |
5626 | /****************************************************************/ | |
5627 | ||
ccc6cda3 JA |
5628 | char * |
5629 | pat_subst (string, pat, rep, mflags) | |
5630 | char *string, *pat, *rep; | |
5631 | int mflags; | |
5632 | { | |
5633 | char *ret, *s, *e, *str; | |
5634 | int rsize, rptr, l, replen, mtype; | |
5635 | ||
b72432fd JA |
5636 | mtype = mflags & MATCH_TYPEMASK; |
5637 | ||
5638 | /* Special cases: | |
5639 | * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING | |
5640 | * with REP and return the result. | |
5641 | * 2. A null pattern with mtype == MATCH_END means to append REP to | |
5642 | * STRING and return the result. | |
5643 | */ | |
5644 | if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END)) | |
5645 | { | |
5646 | replen = STRLEN (rep); | |
5647 | l = strlen (string); | |
f73dda09 | 5648 | ret = (char *)xmalloc (replen + l + 2); |
bb70624e JA |
5649 | if (replen == 0) |
5650 | strcpy (ret, string); | |
5651 | else if (mtype == MATCH_BEG) | |
b72432fd JA |
5652 | { |
5653 | strcpy (ret, rep); | |
5654 | strcpy (ret + replen, string); | |
5655 | } | |
5656 | else | |
5657 | { | |
5658 | strcpy (ret, string); | |
5659 | strcpy (ret + l, rep); | |
5660 | } | |
5661 | return (ret); | |
5662 | } | |
5663 | ||
f73dda09 | 5664 | ret = (char *)xmalloc (rsize = 64); |
ccc6cda3 JA |
5665 | ret[0] = '\0'; |
5666 | ||
ccc6cda3 JA |
5667 | for (replen = STRLEN (rep), rptr = 0, str = string;;) |
5668 | { | |
5669 | if (match_pattern (str, pat, mtype, &s, &e) == 0) | |
5670 | break; | |
5671 | l = s - str; | |
5672 | RESIZE_MALLOCED_BUFFER (ret, rptr, (l + replen), rsize, 64); | |
5673 | ||
5674 | /* OK, now copy the leading unmatched portion of the string (from | |
5675 | str to s) to ret starting at rptr (the current offset). Then copy | |
28ef6c31 JA |
5676 | the replacement string at ret + rptr + (s - str). Increment |
5677 | rptr (if necessary) and str and go on. */ | |
ccc6cda3 JA |
5678 | if (l) |
5679 | { | |
5680 | strncpy (ret + rptr, str, l); | |
5681 | rptr += l; | |
5682 | } | |
5683 | if (replen) | |
5684 | { | |
5685 | strncpy (ret + rptr, rep, replen); | |
5686 | rptr += replen; | |
5687 | } | |
5688 | str = e; /* e == end of match */ | |
b80f6443 | 5689 | |
ccc6cda3 | 5690 | if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY) |
28ef6c31 | 5691 | break; |
b80f6443 JA |
5692 | |
5693 | if (s == e) | |
5694 | e++, str++; /* avoid infinite recursion on zero-length match */ | |
ccc6cda3 JA |
5695 | } |
5696 | ||
5697 | /* Now copy the unmatched portion of the input string */ | |
5698 | if (*str) | |
d166f048 JA |
5699 | { |
5700 | RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64); | |
5701 | strcpy (ret + rptr, str); | |
5702 | } | |
ccc6cda3 JA |
5703 | else |
5704 | ret[rptr] = '\0'; | |
5705 | ||
5706 | return ret; | |
5707 | } | |
5708 | ||
5709 | /* Do pattern match and replacement on the positional parameters. */ | |
5710 | static char * | |
5711 | pos_params_pat_subst (string, pat, rep, mflags) | |
5712 | char *string, *pat, *rep; | |
5713 | int mflags; | |
5714 | { | |
5715 | WORD_LIST *save, *params; | |
5716 | WORD_DESC *w; | |
0628567a | 5717 | char *ret; |
ccc6cda3 JA |
5718 | |
5719 | save = params = list_rest_of_args (); | |
5720 | if (save == 0) | |
5721 | return ((char *)NULL); | |
5722 | ||
5723 | for ( ; params; params = params->next) | |
5724 | { | |
5725 | ret = pat_subst (params->word->word, pat, rep, mflags); | |
95732b49 JA |
5726 | w = alloc_word_desc (); |
5727 | w->word = ret ? ret : savestring (""); | |
ccc6cda3 JA |
5728 | dispose_word (params->word); |
5729 | params->word = w; | |
ccc6cda3 JA |
5730 | } |
5731 | ||
b80f6443 JA |
5732 | if ((mflags & (MATCH_QUOTED|MATCH_STARSUB)) == (MATCH_QUOTED|MATCH_STARSUB)) |
5733 | ret = string_list_dollar_star (quote_list (save)); | |
5734 | else | |
5735 | ret = string_list ((mflags & MATCH_QUOTED) ? quote_list (save) : save); | |
ccc6cda3 JA |
5736 | dispose_words (save); |
5737 | ||
5738 | return (ret); | |
5739 | } | |
5740 | ||
cce855bc JA |
5741 | /* Perform pattern substitution on VALUE, which is the expansion of |
5742 | VARNAME. PATSUB is an expression supplying the pattern to match | |
5743 | and the string to substitute. QUOTED is a flags word containing | |
5744 | the type of quoting currently in effect. */ | |
ccc6cda3 JA |
5745 | static char * |
5746 | parameter_brace_patsub (varname, value, patsub, quoted) | |
5747 | char *varname, *value, *patsub; | |
5748 | int quoted; | |
5749 | { | |
b80f6443 | 5750 | int vtype, mflags, starsub; |
7117c2d2 | 5751 | char *val, *temp, *pat, *rep, *p, *lpatsub, *tt; |
ccc6cda3 JA |
5752 | SHELL_VAR *v; |
5753 | ||
5754 | if (value == 0) | |
5755 | return ((char *)NULL); | |
5756 | ||
5757 | this_command_name = varname; | |
5758 | ||
b80f6443 | 5759 | vtype = get_var_and_type (varname, value, quoted, &v, &val); |
ccc6cda3 JA |
5760 | if (vtype == -1) |
5761 | return ((char *)NULL); | |
5762 | ||
b80f6443 JA |
5763 | starsub = vtype & VT_STARSUB; |
5764 | vtype &= ~VT_STARSUB; | |
5765 | ||
ccc6cda3 | 5766 | mflags = 0; |
f1be666c JA |
5767 | if (patsub && *patsub == '/') |
5768 | { | |
5769 | mflags |= MATCH_GLOBREP; | |
5770 | patsub++; | |
5771 | } | |
7117c2d2 JA |
5772 | |
5773 | /* Malloc this because expand_string_if_necessary or one of the expansion | |
5774 | functions in its call chain may free it on a substitution error. */ | |
bb70624e | 5775 | lpatsub = savestring (patsub); |
ccc6cda3 JA |
5776 | |
5777 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
5778 | mflags |= MATCH_QUOTED; | |
5779 | ||
b80f6443 JA |
5780 | if (starsub) |
5781 | mflags |= MATCH_STARSUB; | |
5782 | ||
0628567a JA |
5783 | /* If the pattern starts with a `/', make sure we skip over it when looking |
5784 | for the replacement delimiter. */ | |
5785 | if (rep = quoted_strchr ((*patsub == '/') ? lpatsub+1 : lpatsub, '/', ST_BACKSL)) | |
ccc6cda3 JA |
5786 | *rep++ = '\0'; |
5787 | else | |
5788 | rep = (char *)NULL; | |
5789 | ||
5790 | if (rep && *rep == '\0') | |
5791 | rep = (char *)NULL; | |
5792 | ||
b80f6443 JA |
5793 | /* Perform the same expansions on the pattern as performed by the |
5794 | pattern removal expansions. */ | |
5795 | pat = getpattern (lpatsub, quoted, 1); | |
bb70624e | 5796 | |
ccc6cda3 | 5797 | if (rep) |
d166f048 JA |
5798 | { |
5799 | if ((mflags & MATCH_QUOTED) == 0) | |
f73dda09 | 5800 | rep = expand_string_if_necessary (rep, quoted, expand_string_unsplit); |
d166f048 | 5801 | else |
f73dda09 | 5802 | rep = expand_string_to_string_internal (rep, quoted, expand_string_unsplit); |
d166f048 | 5803 | } |
ccc6cda3 | 5804 | |
0628567a | 5805 | /* ksh93 doesn't allow the match specifier to be a part of the expanded |
f1be666c JA |
5806 | pattern. This is an extension. Make sure we don't anchor the pattern |
5807 | at the beginning or end of the string if we're doing global replacement, | |
5808 | though. */ | |
ccc6cda3 | 5809 | p = pat; |
f1be666c JA |
5810 | if (mflags & MATCH_GLOBREP) |
5811 | mflags |= MATCH_ANY; | |
0628567a | 5812 | else if (pat && pat[0] == '#') |
ccc6cda3 JA |
5813 | { |
5814 | mflags |= MATCH_BEG; | |
5815 | p++; | |
5816 | } | |
d166f048 | 5817 | else if (pat && pat[0] == '%') |
ccc6cda3 JA |
5818 | { |
5819 | mflags |= MATCH_END; | |
5820 | p++; | |
5821 | } | |
5822 | else | |
5823 | mflags |= MATCH_ANY; | |
5824 | ||
cce855bc JA |
5825 | /* OK, we now want to substitute REP for PAT in VAL. If |
5826 | flags & MATCH_GLOBREP is non-zero, the substitution is done | |
5827 | everywhere, otherwise only the first occurrence of PAT is | |
7117c2d2 JA |
5828 | replaced. The pattern matching code doesn't understand |
5829 | CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable | |
5830 | values passed in (VT_VARIABLE) so the pattern substitution | |
5831 | code works right. We need to requote special chars after | |
5832 | we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the | |
5833 | other cases if QUOTED == 0, since the posparams and arrays | |
5834 | indexed by * or @ do special things when QUOTED != 0. */ | |
5835 | ||
ccc6cda3 JA |
5836 | switch (vtype) |
5837 | { | |
5838 | case VT_VARIABLE: | |
d166f048 | 5839 | case VT_ARRAYMEMBER: |
ccc6cda3 | 5840 | temp = pat_subst (val, p, rep, mflags); |
7117c2d2 JA |
5841 | if (vtype == VT_VARIABLE) |
5842 | FREE (val); | |
5843 | if (temp) | |
5844 | { | |
5845 | tt = quote_escapes (temp); | |
5846 | free (temp); | |
5847 | temp = tt; | |
5848 | } | |
ccc6cda3 JA |
5849 | break; |
5850 | case VT_POSPARMS: | |
5851 | temp = pos_params_pat_subst (val, p, rep, mflags); | |
7117c2d2 JA |
5852 | if (temp && (mflags & MATCH_QUOTED) == 0) |
5853 | { | |
5854 | tt = quote_escapes (temp); | |
5855 | free (temp); | |
5856 | temp = tt; | |
5857 | } | |
ccc6cda3 JA |
5858 | break; |
5859 | #if defined (ARRAY_VARS) | |
5860 | case VT_ARRAYVAR: | |
7117c2d2 | 5861 | temp = array_patsub (array_cell (v), p, rep, mflags); |
f1be666c JA |
5862 | #if 0 |
5863 | /* Don't need to do this anymore; array_patsub calls array_quote_escapes | |
5864 | as appropriate before adding the space separators. */ | |
7117c2d2 JA |
5865 | if (temp && (mflags & MATCH_QUOTED) == 0) |
5866 | { | |
5867 | tt = quote_escapes (temp); | |
5868 | free (temp); | |
5869 | temp = tt; | |
5870 | } | |
f1be666c | 5871 | #endif |
ccc6cda3 JA |
5872 | break; |
5873 | #endif | |
5874 | } | |
5875 | ||
5876 | FREE (pat); | |
5877 | FREE (rep); | |
bb70624e | 5878 | free (lpatsub); |
ccc6cda3 JA |
5879 | |
5880 | return temp; | |
5881 | } | |
5882 | ||
0628567a JA |
5883 | /* Check for unbalanced parens in S, which is the contents of $(( ... )). If |
5884 | any occur, this must be a nested command substitution, so return 0. | |
5885 | Otherwise, return 1. A valid arithmetic expression must always have a | |
5886 | ( before a matching ), so any cases where there are more right parens | |
5887 | means that this must not be an arithmetic expression, though the parser | |
5888 | will not accept it without a balanced total number of parens. */ | |
5889 | static int | |
5890 | chk_arithsub (s, len) | |
5891 | const char *s; | |
5892 | int len; | |
5893 | { | |
5894 | int i, count; | |
5895 | DECLARE_MBSTATE; | |
5896 | ||
5897 | i = count = 0; | |
5898 | while (i < len) | |
5899 | { | |
5900 | if (s[i] == '(') | |
5901 | count++; | |
5902 | else if (s[i] == ')') | |
5903 | { | |
5904 | count--; | |
5905 | if (count < 0) | |
5906 | return 0; | |
5907 | } | |
5908 | ||
5909 | switch (s[i]) | |
5910 | { | |
5911 | default: | |
5912 | ADVANCE_CHAR (s, len, i); | |
5913 | break; | |
5914 | ||
5915 | case '\\': | |
5916 | i++; | |
5917 | if (s[i]) | |
5918 | ADVANCE_CHAR (s, len, i); | |
5919 | break; | |
5920 | ||
5921 | case '\'': | |
5922 | i = skip_single_quoted (s, len, ++i); | |
5923 | break; | |
5924 | ||
5925 | case '"': | |
5926 | i = skip_double_quoted ((char *)s, len, ++i); | |
5927 | break; | |
5928 | } | |
5929 | } | |
5930 | ||
5931 | return (count == 0); | |
5932 | } | |
5933 | ||
cce855bc JA |
5934 | /****************************************************************/ |
5935 | /* */ | |
5936 | /* Functions to perform parameter expansion on a string */ | |
5937 | /* */ | |
5938 | /****************************************************************/ | |
5939 | ||
ccc6cda3 | 5940 | /* ${[#][!]name[[:]#[#]%[%]-=?+[word][:e1[:e2]]]} */ |
95732b49 | 5941 | static WORD_DESC * |
ccc6cda3 JA |
5942 | parameter_brace_expand (string, indexp, quoted, quoted_dollar_atp, contains_dollar_at) |
5943 | char *string; | |
5944 | int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at; | |
5945 | { | |
5946 | int check_nullness, var_is_set, var_is_null, var_is_special; | |
5947 | int want_substring, want_indir, want_patsub; | |
5948 | char *name, *value, *temp, *temp1; | |
95732b49 JA |
5949 | WORD_DESC *tdesc, *ret; |
5950 | int t_index, sindex, c, tflag; | |
7117c2d2 | 5951 | intmax_t number; |
ccc6cda3 | 5952 | |
ccc6cda3 JA |
5953 | value = (char *)NULL; |
5954 | var_is_set = var_is_null = var_is_special = check_nullness = 0; | |
5955 | want_substring = want_indir = want_patsub = 0; | |
5956 | ||
cce855bc JA |
5957 | sindex = *indexp; |
5958 | t_index = ++sindex; | |
0628567a JA |
5959 | /* ${#var} doesn't have any of the other parameter expansions on it. */ |
5960 | if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */ | |
5961 | name = string_extract (string, &t_index, "}", EX_VARNAME); | |
5962 | else | |
5963 | name = string_extract (string, &t_index, "#%:-=?+/}", EX_VARNAME); | |
cce855bc | 5964 | |
95732b49 JA |
5965 | ret = 0; |
5966 | tflag = 0; | |
5967 | ||
cce855bc JA |
5968 | /* If the name really consists of a special variable, then make sure |
5969 | that we have the entire name. We don't allow indirect references | |
5970 | to special variables except `#', `?', `@' and `*'. */ | |
5971 | if ((sindex == t_index && | |
ccc6cda3 JA |
5972 | (string[t_index] == '-' || |
5973 | string[t_index] == '?' || | |
cce855bc JA |
5974 | string[t_index] == '#')) || |
5975 | (sindex == t_index - 1 && string[sindex] == '!' && | |
28ef6c31 JA |
5976 | (string[t_index] == '#' || |
5977 | string[t_index] == '?' || | |
5978 | string[t_index] == '@' || | |
5979 | string[t_index] == '*'))) | |
ccc6cda3 JA |
5980 | { |
5981 | t_index++; | |
5982 | free (name); | |
5983 | temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0); | |
f73dda09 | 5984 | name = (char *)xmalloc (3 + (strlen (temp1))); |
ccc6cda3 JA |
5985 | *name = string[sindex]; |
5986 | if (string[sindex] == '!') | |
5987 | { | |
28ef6c31 JA |
5988 | /* indirect reference of $#, $?, $@, or $* */ |
5989 | name[1] = string[sindex + 1]; | |
5990 | strcpy (name + 2, temp1); | |
ccc6cda3 | 5991 | } |
cce855bc | 5992 | else |
ccc6cda3 JA |
5993 | strcpy (name + 1, temp1); |
5994 | free (temp1); | |
5995 | } | |
5996 | sindex = t_index; | |
5997 | ||
5998 | /* Find out what character ended the variable name. Then | |
5999 | do the appropriate thing. */ | |
6000 | if (c = string[sindex]) | |
6001 | sindex++; | |
6002 | ||
6003 | /* If c is followed by one of the valid parameter expansion | |
6004 | characters, move past it as normal. If not, assume that | |
6005 | a substring specification is being given, and do not move | |
6006 | past it. */ | |
28ef6c31 | 6007 | if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex])) |
ccc6cda3 JA |
6008 | { |
6009 | check_nullness++; | |
6010 | if (c = string[sindex]) | |
6011 | sindex++; | |
6012 | } | |
cce855bc | 6013 | else if (c == ':' && string[sindex] != RBRACE) |
ccc6cda3 | 6014 | want_substring = 1; |
cce855bc | 6015 | else if (c == '/' && string[sindex] != RBRACE) |
ccc6cda3 JA |
6016 | want_patsub = 1; |
6017 | ||
cce855bc JA |
6018 | /* Catch the valid and invalid brace expressions that made it through the |
6019 | tests above. */ | |
6020 | /* ${#-} is a valid expansion and means to take the length of $-. | |
6021 | Similarly for ${#?} and ${##}... */ | |
6022 | if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 && | |
28ef6c31 | 6023 | VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE) |
cce855bc | 6024 | { |
f73dda09 | 6025 | name = (char *)xrealloc (name, 3); |
cce855bc JA |
6026 | name[1] = c; |
6027 | name[2] = '\0'; | |
6028 | c = string[sindex++]; | |
6029 | } | |
6030 | ||
6031 | /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */ | |
6032 | if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 && | |
6033 | member (c, "%:=+/") && string[sindex] == RBRACE) | |
6034 | { | |
6035 | temp = (char *)NULL; | |
6036 | goto bad_substitution; | |
6037 | } | |
6038 | ||
6039 | /* Indirect expansion begins with a `!'. A valid indirect expansion is | |
6040 | either a variable name, one of the positional parameters or a special | |
6041 | variable that expands to one of the positional parameters. */ | |
6042 | want_indir = *name == '!' && | |
f73dda09 | 6043 | (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) |
7117c2d2 | 6044 | || VALID_INDIR_PARAM (name[1])); |
ccc6cda3 JA |
6045 | |
6046 | /* Determine the value of this variable. */ | |
6047 | ||
cce855bc | 6048 | /* Check for special variables, directly referenced. */ |
bb70624e | 6049 | if (SPECIAL_VAR (name, want_indir)) |
ccc6cda3 JA |
6050 | var_is_special++; |
6051 | ||
cce855bc JA |
6052 | /* Check for special expansion things, like the length of a parameter */ |
6053 | if (*name == '#' && name[1]) | |
ccc6cda3 | 6054 | { |
cce855bc | 6055 | /* If we are not pointing at the character just after the |
28ef6c31 JA |
6056 | closing brace, then we haven't gotten all of the name. |
6057 | Since it begins with a special character, this is a bad | |
6058 | substitution. Also check NAME for validity before trying | |
6059 | to go on. */ | |
cce855bc | 6060 | if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0)) |
ccc6cda3 JA |
6061 | { |
6062 | temp = (char *)NULL; | |
6063 | goto bad_substitution; | |
6064 | } | |
6065 | ||
6066 | number = parameter_brace_expand_length (name); | |
6067 | free (name); | |
6068 | ||
6069 | *indexp = sindex; | |
95732b49 JA |
6070 | if (number < 0) |
6071 | return (&expand_wdesc_error); | |
6072 | else | |
6073 | { | |
6074 | ret = alloc_word_desc (); | |
6075 | ret->word = itos (number); | |
6076 | return ret; | |
6077 | } | |
ccc6cda3 JA |
6078 | } |
6079 | ||
6080 | /* ${@} is identical to $@. */ | |
6081 | if (name[0] == '@' && name[1] == '\0') | |
6082 | { | |
6083 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
6084 | *quoted_dollar_atp = 1; | |
6085 | ||
6086 | if (contains_dollar_at) | |
6087 | *contains_dollar_at = 1; | |
6088 | } | |
6089 | ||
b80f6443 | 6090 | /* Process ${!PREFIX*} expansion. */ |
bb70624e JA |
6091 | if (want_indir && string[sindex - 1] == RBRACE && |
6092 | (string[sindex - 2] == '*' || string[sindex - 2] == '@') && | |
f73dda09 | 6093 | legal_variable_starter ((unsigned char) name[1])) |
bb70624e JA |
6094 | { |
6095 | char **x; | |
6096 | WORD_LIST *xlist; | |
6097 | ||
6098 | temp1 = savestring (name + 1); | |
6099 | number = strlen (temp1); | |
6100 | temp1[number - 1] = '\0'; | |
6101 | x = all_variables_matching_prefix (temp1); | |
7117c2d2 | 6102 | xlist = strvec_to_word_list (x, 0, 0); |
28ef6c31 JA |
6103 | if (string[sindex - 2] == '*') |
6104 | temp = string_list_dollar_star (xlist); | |
6105 | else | |
6106 | { | |
6107 | temp = string_list_dollar_at (xlist, quoted); | |
6108 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
6109 | *quoted_dollar_atp = 1; | |
6110 | if (contains_dollar_at) | |
6111 | *contains_dollar_at = 1; | |
6112 | } | |
bb70624e JA |
6113 | free (x); |
6114 | free (xlist); | |
6115 | free (temp1); | |
6116 | *indexp = sindex; | |
95732b49 JA |
6117 | |
6118 | ret = alloc_word_desc (); | |
6119 | ret->word = temp; | |
6120 | return ret; | |
bb70624e | 6121 | } |
b80f6443 JA |
6122 | |
6123 | #if defined (ARRAY_VARS) | |
6124 | /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */ | |
6125 | if (want_indir && string[sindex - 1] == RBRACE && | |
6126 | string[sindex - 2] == ']' && valid_array_reference (name+1)) | |
6127 | { | |
6128 | char *x, *x1; | |
6129 | ||
6130 | temp1 = savestring (name + 1); | |
6131 | x = array_variable_name (temp1, &x1, (int *)0); /* [ */ | |
6132 | FREE (x); | |
6133 | if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']') | |
6134 | { | |
6135 | temp = array_keys (temp1, quoted); | |
6136 | if (x1[0] == '@') | |
6137 | { | |
6138 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
6139 | *quoted_dollar_atp = 1; | |
6140 | if (contains_dollar_at) | |
6141 | *contains_dollar_at = 1; | |
6142 | } | |
6143 | ||
6144 | free (temp1); | |
6145 | *indexp = sindex; | |
95732b49 JA |
6146 | |
6147 | ret = alloc_word_desc (); | |
6148 | ret->word = temp; | |
6149 | return ret; | |
b80f6443 JA |
6150 | } |
6151 | ||
6152 | free (temp1); | |
6153 | } | |
6154 | #endif /* ARRAY_VARS */ | |
bb70624e | 6155 | |
ccc6cda3 JA |
6156 | /* Make sure that NAME is valid before trying to go on. */ |
6157 | if (valid_brace_expansion_word (want_indir ? name + 1 : name, | |
6158 | var_is_special) == 0) | |
6159 | { | |
6160 | temp = (char *)NULL; | |
6161 | goto bad_substitution; | |
6162 | } | |
6163 | ||
6164 | if (want_indir) | |
95732b49 JA |
6165 | tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at); |
6166 | else | |
6167 | tdesc = parameter_brace_expand_word (name, var_is_special, quoted); | |
6168 | ||
6169 | if (tdesc) | |
6170 | { | |
6171 | temp = tdesc->word; | |
6172 | tflag = tdesc->flags; | |
6173 | dispose_word_desc (tdesc); | |
6174 | } | |
ccc6cda3 | 6175 | else |
95732b49 | 6176 | temp = (char *)0; |
ccc6cda3 JA |
6177 | |
6178 | #if defined (ARRAY_VARS) | |
cce855bc | 6179 | if (valid_array_reference (name)) |
b80f6443 | 6180 | chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at); |
ccc6cda3 JA |
6181 | #endif |
6182 | ||
6183 | var_is_set = temp != (char *)0; | |
6184 | var_is_null = check_nullness && (var_is_set == 0 || *temp == 0); | |
6185 | ||
6186 | /* Get the rest of the stuff inside the braces. */ | |
cce855bc | 6187 | if (c && c != RBRACE) |
ccc6cda3 JA |
6188 | { |
6189 | /* Extract the contents of the ${ ... } expansion | |
28ef6c31 | 6190 | according to the Posix.2 rules. */ |
7117c2d2 | 6191 | value = extract_dollar_brace_string (string, &sindex, quoted, 0); |
cce855bc | 6192 | if (string[sindex] == RBRACE) |
28ef6c31 | 6193 | sindex++; |
ccc6cda3 JA |
6194 | else |
6195 | goto bad_substitution; | |
6196 | } | |
6197 | else | |
6198 | value = (char *)NULL; | |
726f6388 | 6199 | |
ccc6cda3 JA |
6200 | *indexp = sindex; |
6201 | ||
6202 | /* If this is a substring spec, process it and add the result. */ | |
6203 | if (want_substring) | |
726f6388 | 6204 | { |
ccc6cda3 JA |
6205 | temp1 = parameter_brace_substring (name, temp, value, quoted); |
6206 | FREE (name); | |
6207 | FREE (value); | |
6208 | FREE (temp); | |
95732b49 JA |
6209 | |
6210 | if (temp1 == &expand_param_error) | |
6211 | return (&expand_wdesc_error); | |
6212 | else if (temp1 == &expand_param_fatal) | |
6213 | return (&expand_wdesc_fatal); | |
6214 | ||
6215 | ret = alloc_word_desc (); | |
6216 | ret->word = temp1; | |
0628567a JA |
6217 | if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
6218 | ret->flags |= W_QUOTED|W_HASQUOTEDNULL; | |
95732b49 | 6219 | return ret; |
726f6388 | 6220 | } |
ccc6cda3 | 6221 | else if (want_patsub) |
726f6388 | 6222 | { |
ccc6cda3 JA |
6223 | temp1 = parameter_brace_patsub (name, temp, value, quoted); |
6224 | FREE (name); | |
6225 | FREE (value); | |
6226 | FREE (temp); | |
95732b49 JA |
6227 | |
6228 | if (temp1 == &expand_param_error) | |
6229 | return (&expand_wdesc_error); | |
6230 | else if (temp1 == &expand_param_fatal) | |
6231 | return (&expand_wdesc_fatal); | |
6232 | ||
6233 | ret = alloc_word_desc (); | |
6234 | ret->word = temp1; | |
6235 | return ret; | |
ccc6cda3 | 6236 | } |
726f6388 | 6237 | |
ccc6cda3 JA |
6238 | /* Do the right thing based on which character ended the variable name. */ |
6239 | switch (c) | |
6240 | { | |
6241 | default: | |
6242 | case '\0': | |
6243 | bad_substitution: | |
b80f6443 | 6244 | report_error (_("%s: bad substitution"), string ? string : "??"); |
ccc6cda3 JA |
6245 | FREE (value); |
6246 | FREE (temp); | |
6247 | free (name); | |
95732b49 | 6248 | return &expand_wdesc_error; |
ccc6cda3 | 6249 | |
cce855bc | 6250 | case RBRACE: |
ccc6cda3 | 6251 | if (var_is_set == 0 && unbound_vars_is_error) |
28ef6c31 | 6252 | { |
7117c2d2 | 6253 | err_unboundvar (name); |
ccc6cda3 JA |
6254 | FREE (value); |
6255 | FREE (temp); | |
6256 | free (name); | |
6257 | last_command_exit_value = EXECUTION_FAILURE; | |
95732b49 | 6258 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
726f6388 | 6259 | } |
ccc6cda3 | 6260 | break; |
726f6388 | 6261 | |
ccc6cda3 JA |
6262 | case '#': /* ${param#[#]pattern} */ |
6263 | case '%': /* ${param%[%]pattern} */ | |
6264 | if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0') | |
28ef6c31 JA |
6265 | { |
6266 | FREE (value); | |
ccc6cda3 | 6267 | break; |
28ef6c31 | 6268 | } |
7117c2d2 | 6269 | temp1 = parameter_brace_remove_pattern (name, temp, value, c, quoted); |
ccc6cda3 JA |
6270 | free (temp); |
6271 | free (value); | |
6272 | temp = temp1; | |
6273 | break; | |
6274 | ||
6275 | case '-': | |
6276 | case '=': | |
6277 | case '?': | |
6278 | case '+': | |
6279 | if (var_is_set && var_is_null == 0) | |
28ef6c31 JA |
6280 | { |
6281 | /* If the operator is `+', we don't want the value of the named | |
6282 | variable for anything, just the value of the right hand side. */ | |
bb70624e | 6283 | |
ccc6cda3 JA |
6284 | if (c == '+') |
6285 | { | |
28ef6c31 JA |
6286 | /* XXX -- if we're double-quoted and the named variable is "$@", |
6287 | we want to turn off any special handling of "$@" -- | |
6288 | we're not using it, so whatever is on the rhs applies. */ | |
6289 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
6290 | *quoted_dollar_atp = 0; | |
6291 | if (contains_dollar_at) | |
6292 | *contains_dollar_at = 0; | |
6293 | ||
ccc6cda3 JA |
6294 | FREE (temp); |
6295 | if (value) | |
28ef6c31 | 6296 | { |
95732b49 JA |
6297 | ret = parameter_brace_expand_rhs (name, value, c, |
6298 | quoted, | |
6299 | quoted_dollar_atp, | |
6300 | contains_dollar_at); | |
6301 | /* XXX - fix up later, esp. noting presence of | |
6302 | W_HASQUOTEDNULL in ret->flags */ | |
ccc6cda3 JA |
6303 | free (value); |
6304 | } | |
6305 | else | |
28ef6c31 | 6306 | temp = (char *)NULL; |
ccc6cda3 JA |
6307 | } |
6308 | else | |
6309 | { | |
6310 | FREE (value); | |
6311 | } | |
6312 | /* Otherwise do nothing; just use the value in TEMP. */ | |
726f6388 | 6313 | } |
ccc6cda3 | 6314 | else /* VAR not set or VAR is NULL. */ |
28ef6c31 | 6315 | { |
ccc6cda3 JA |
6316 | FREE (temp); |
6317 | temp = (char *)NULL; | |
6318 | if (c == '=' && var_is_special) | |
6319 | { | |
b80f6443 | 6320 | report_error (_("$%s: cannot assign in this way"), name); |
ccc6cda3 JA |
6321 | free (name); |
6322 | free (value); | |
95732b49 | 6323 | return &expand_wdesc_error; |
ccc6cda3 JA |
6324 | } |
6325 | else if (c == '?') | |
6326 | { | |
6327 | parameter_brace_expand_error (name, value); | |
95732b49 | 6328 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
ccc6cda3 JA |
6329 | } |
6330 | else if (c != '+') | |
28ef6c31 JA |
6331 | { |
6332 | /* XXX -- if we're double-quoted and the named variable is "$@", | |
6333 | we want to turn off any special handling of "$@" -- | |
6334 | we're not using it, so whatever is on the rhs applies. */ | |
6335 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
6336 | *quoted_dollar_atp = 0; | |
6337 | if (contains_dollar_at) | |
6338 | *contains_dollar_at = 0; | |
6339 | ||
95732b49 JA |
6340 | ret = parameter_brace_expand_rhs (name, value, c, quoted, |
6341 | quoted_dollar_atp, | |
6342 | contains_dollar_at); | |
6343 | /* XXX - fix up later, esp. noting presence of | |
6344 | W_HASQUOTEDNULL in tdesc->flags */ | |
28ef6c31 | 6345 | } |
ccc6cda3 | 6346 | free (value); |
726f6388 | 6347 | } |
28ef6c31 | 6348 | |
ccc6cda3 | 6349 | break; |
726f6388 | 6350 | } |
ccc6cda3 | 6351 | free (name); |
95732b49 JA |
6352 | |
6353 | if (ret == 0) | |
6354 | { | |
6355 | ret = alloc_word_desc (); | |
6356 | ret->flags = tflag; | |
6357 | ret->word = temp; | |
6358 | } | |
6359 | return (ret); | |
726f6388 JA |
6360 | } |
6361 | ||
cce855bc JA |
6362 | /* Expand a single ${xxx} expansion. The braces are optional. When |
6363 | the braces are used, parameter_brace_expand() does the work, | |
6364 | possibly calling param_expand recursively. */ | |
95732b49 | 6365 | static WORD_DESC * |
cce855bc JA |
6366 | param_expand (string, sindex, quoted, expanded_something, |
6367 | contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p, | |
6368 | pflags) | |
6369 | char *string; | |
6370 | int *sindex, quoted, *expanded_something, *contains_dollar_at; | |
6371 | int *quoted_dollar_at_p, *had_quoted_null_p, pflags; | |
6372 | { | |
7117c2d2 | 6373 | char *temp, *temp1, uerror[3]; |
f73dda09 JA |
6374 | int zindex, t_index, expok; |
6375 | unsigned char c; | |
7117c2d2 | 6376 | intmax_t number; |
cce855bc | 6377 | SHELL_VAR *var; |
f73dda09 | 6378 | WORD_LIST *list; |
95732b49 JA |
6379 | WORD_DESC *tdesc, *ret; |
6380 | int tflag; | |
cce855bc JA |
6381 | |
6382 | zindex = *sindex; | |
6383 | c = string[++zindex]; | |
6384 | ||
6385 | temp = (char *)NULL; | |
95732b49 JA |
6386 | ret = tdesc = (WORD_DESC *)NULL; |
6387 | tflag = 0; | |
cce855bc JA |
6388 | |
6389 | /* Do simple cases first. Switch on what follows '$'. */ | |
6390 | switch (c) | |
6391 | { | |
6392 | /* $0 .. $9? */ | |
6393 | case '0': | |
6394 | case '1': | |
6395 | case '2': | |
6396 | case '3': | |
6397 | case '4': | |
6398 | case '5': | |
6399 | case '6': | |
6400 | case '7': | |
6401 | case '8': | |
6402 | case '9': | |
f73dda09 | 6403 | temp1 = dollar_vars[TODIGIT (c)]; |
cce855bc JA |
6404 | if (unbound_vars_is_error && temp1 == (char *)NULL) |
6405 | { | |
7117c2d2 JA |
6406 | uerror[0] = '$'; |
6407 | uerror[1] = c; | |
6408 | uerror[2] = '\0'; | |
6409 | err_unboundvar (uerror); | |
cce855bc | 6410 | last_command_exit_value = EXECUTION_FAILURE; |
95732b49 | 6411 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
cce855bc | 6412 | } |
b80f6443 JA |
6413 | if (temp1) |
6414 | temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
6415 | ? quote_string (temp1) | |
6416 | : quote_escapes (temp1); | |
6417 | else | |
6418 | temp = (char *)NULL; | |
95732b49 | 6419 | |
cce855bc JA |
6420 | break; |
6421 | ||
6422 | /* $$ -- pid of the invoking shell. */ | |
6423 | case '$': | |
6424 | temp = itos (dollar_dollar_pid); | |
6425 | break; | |
6426 | ||
6427 | /* $# -- number of positional parameters. */ | |
6428 | case '#': | |
6429 | temp = itos (number_of_args ()); | |
6430 | break; | |
6431 | ||
6432 | /* $? -- return value of the last synchronous command. */ | |
6433 | case '?': | |
6434 | temp = itos (last_command_exit_value); | |
6435 | break; | |
6436 | ||
6437 | /* $- -- flags supplied to the shell on invocation or by `set'. */ | |
6438 | case '-': | |
6439 | temp = which_set_flags (); | |
6440 | break; | |
6441 | ||
6442 | /* $! -- Pid of the last asynchronous command. */ | |
6443 | case '!': | |
6444 | /* If no asynchronous pids have been created, expand to nothing. | |
6445 | If `set -u' has been executed, and no async processes have | |
6446 | been created, this is an expansion error. */ | |
6447 | if (last_asynchronous_pid == NO_PID) | |
6448 | { | |
6449 | if (expanded_something) | |
6450 | *expanded_something = 0; | |
6451 | temp = (char *)NULL; | |
6452 | if (unbound_vars_is_error) | |
6453 | { | |
7117c2d2 JA |
6454 | uerror[0] = '$'; |
6455 | uerror[1] = c; | |
6456 | uerror[2] = '\0'; | |
6457 | err_unboundvar (uerror); | |
cce855bc | 6458 | last_command_exit_value = EXECUTION_FAILURE; |
95732b49 | 6459 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
cce855bc JA |
6460 | } |
6461 | } | |
6462 | else | |
f73dda09 | 6463 | temp = itos (last_asynchronous_pid); |
cce855bc JA |
6464 | break; |
6465 | ||
6466 | /* The only difference between this and $@ is when the arg is quoted. */ | |
6467 | case '*': /* `$*' */ | |
6468 | list = list_rest_of_args (); | |
6469 | ||
6470 | /* If there are no command-line arguments, this should just | |
6471 | disappear if there are other characters in the expansion, | |
6472 | even if it's quoted. */ | |
6473 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0) | |
6474 | temp = (char *)NULL; | |
6475 | else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
6476 | { | |
6477 | /* If we have "$*" we want to make a string of the positional | |
6478 | parameters, separated by the first character of $IFS, and | |
6479 | quote the whole string, including the separators. If IFS | |
6480 | is unset, the parameters are separated by ' '; if $IFS is | |
6481 | null, the parameters are concatenated. */ | |
b80f6443 | 6482 | temp = (quoted & Q_DOUBLE_QUOTES) ? string_list_dollar_star (list) : string_list (list); |
cce855bc | 6483 | temp1 = quote_string (temp); |
95732b49 JA |
6484 | if (*temp == 0) |
6485 | tflag |= W_HASQUOTEDNULL; | |
cce855bc JA |
6486 | free (temp); |
6487 | temp = temp1; | |
6488 | } | |
6489 | else | |
28ef6c31 | 6490 | { |
95732b49 JA |
6491 | /* We check whether or not we're eventually going to split $* here, |
6492 | for example when IFS is empty and we are processing the rhs of | |
6493 | an assignment statement. In that case, we don't separate the | |
6494 | arguments at all. Otherwise, if the $* is not quoted it is | |
6495 | identical to $@ */ | |
6496 | #if 1 | |
6497 | # if defined (HANDLE_MULTIBYTE) | |
6498 | if (expand_no_split_dollar_star && ifs_firstc[0] == 0) | |
6499 | # else | |
6500 | if (expand_no_split_dollar_star && ifs_firstc == 0) | |
6501 | # endif | |
6502 | temp = string_list_dollar_star (list); | |
6503 | else | |
6504 | temp = string_list_dollar_at (list, quoted); | |
6505 | #else | |
28ef6c31 | 6506 | temp = string_list_dollar_at (list, quoted); |
95732b49 | 6507 | #endif |
28ef6c31 JA |
6508 | if (expand_no_split_dollar_star == 0 && contains_dollar_at) |
6509 | *contains_dollar_at = 1; | |
6510 | } | |
cce855bc JA |
6511 | |
6512 | dispose_words (list); | |
6513 | break; | |
6514 | ||
6515 | /* When we have "$@" what we want is "$1" "$2" "$3" ... This | |
6516 | means that we have to turn quoting off after we split into | |
6517 | the individually quoted arguments so that the final split | |
6518 | on the first character of $IFS is still done. */ | |
6519 | case '@': /* `$@' */ | |
6520 | list = list_rest_of_args (); | |
6521 | ||
6522 | /* We want to flag the fact that we saw this. We can't turn | |
6523 | off quoting entirely, because other characters in the | |
6524 | string might need it (consider "\"$@\""), but we need some | |
6525 | way to signal that the final split on the first character | |
6526 | of $IFS should be done, even though QUOTED is 1. */ | |
6527 | if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
6528 | *quoted_dollar_at_p = 1; | |
6529 | if (contains_dollar_at) | |
6530 | *contains_dollar_at = 1; | |
6531 | ||
6532 | /* We want to separate the positional parameters with the first | |
6533 | character of $IFS in case $IFS is something other than a space. | |
6534 | We also want to make sure that splitting is done no matter what -- | |
6535 | according to POSIX.2, this expands to a list of the positional | |
6536 | parameters no matter what IFS is set to. */ | |
6537 | temp = string_list_dollar_at (list, quoted); | |
6538 | ||
6539 | dispose_words (list); | |
6540 | break; | |
6541 | ||
6542 | case LBRACE: | |
95732b49 JA |
6543 | tdesc = parameter_brace_expand (string, &zindex, quoted, |
6544 | quoted_dollar_at_p, | |
6545 | contains_dollar_at); | |
6546 | ||
95732b49 JA |
6547 | if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal) |
6548 | return (tdesc); | |
6549 | temp = tdesc ? tdesc->word : (char *)0; | |
cce855bc JA |
6550 | |
6551 | /* XXX */ | |
bb70624e | 6552 | /* Quoted nulls should be removed if there is anything else |
cce855bc JA |
6553 | in the string. */ |
6554 | /* Note that we saw the quoted null so we can add one back at | |
6555 | the end of this function if there are no other characters | |
28ef6c31 JA |
6556 | in the string, discard TEMP, and go on. The exception to |
6557 | this is when we have "${@}" and $1 is '', since $@ needs | |
6558 | special handling. */ | |
95732b49 | 6559 | if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp)) |
cce855bc JA |
6560 | { |
6561 | if (had_quoted_null_p) | |
6562 | *had_quoted_null_p = 1; | |
28ef6c31 JA |
6563 | if (*quoted_dollar_at_p == 0) |
6564 | { | |
6565 | free (temp); | |
95732b49 | 6566 | tdesc->word = temp = (char *)NULL; |
28ef6c31 JA |
6567 | } |
6568 | ||
cce855bc JA |
6569 | } |
6570 | ||
95732b49 | 6571 | ret = tdesc; |
cce855bc JA |
6572 | goto return0; |
6573 | ||
6574 | /* Do command or arithmetic substitution. */ | |
6575 | case LPAREN: | |
6576 | /* We have to extract the contents of this paren substitution. */ | |
6577 | t_index = zindex + 1; | |
6578 | temp = extract_command_subst (string, &t_index); | |
6579 | zindex = t_index; | |
6580 | ||
6581 | /* For Posix.2-style `$(( ))' arithmetic substitution, | |
28ef6c31 | 6582 | extract the expression and pass it to the evaluator. */ |
cce855bc JA |
6583 | if (temp && *temp == LPAREN) |
6584 | { | |
6585 | char *temp2; | |
6586 | temp1 = temp + 1; | |
6587 | temp2 = savestring (temp1); | |
6588 | t_index = strlen (temp2) - 1; | |
6589 | ||
6590 | if (temp2[t_index] != RPAREN) | |
6591 | { | |
6592 | free (temp2); | |
6593 | goto comsub; | |
6594 | } | |
6595 | ||
6596 | /* Cut off ending `)' */ | |
6597 | temp2[t_index] = '\0'; | |
6598 | ||
0628567a JA |
6599 | if (chk_arithsub (temp2, t_index) == 0) |
6600 | { | |
6601 | free (temp2); | |
6602 | goto comsub; | |
6603 | } | |
6604 | ||
cce855bc | 6605 | /* Expand variables found inside the expression. */ |
0628567a | 6606 | temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES); |
cce855bc JA |
6607 | free (temp2); |
6608 | ||
6609 | arithsub: | |
6610 | /* No error messages. */ | |
6611 | this_command_name = (char *)NULL; | |
6612 | number = evalexp (temp1, &expok); | |
6613 | free (temp); | |
6614 | free (temp1); | |
6615 | if (expok == 0) | |
6616 | { | |
6617 | if (interactive_shell == 0 && posixly_correct) | |
6618 | { | |
6619 | last_command_exit_value = EXECUTION_FAILURE; | |
95732b49 | 6620 | return (&expand_wdesc_fatal); |
cce855bc JA |
6621 | } |
6622 | else | |
95732b49 | 6623 | return (&expand_wdesc_error); |
cce855bc JA |
6624 | } |
6625 | temp = itos (number); | |
6626 | break; | |
6627 | } | |
6628 | ||
6629 | comsub: | |
b80f6443 JA |
6630 | if (pflags & PF_NOCOMSUB) |
6631 | /* we need zindex+1 because string[zindex] == RPAREN */ | |
6632 | temp1 = substring (string, *sindex, zindex+1); | |
6633 | else | |
6634 | temp1 = command_substitute (temp, quoted); | |
cce855bc JA |
6635 | FREE (temp); |
6636 | temp = temp1; | |
6637 | break; | |
6638 | ||
6639 | /* Do POSIX.2d9-style arithmetic substitution. This will probably go | |
6640 | away in a future bash release. */ | |
6641 | case '[': | |
bb70624e | 6642 | /* Extract the contents of this arithmetic substitution. */ |
cce855bc JA |
6643 | t_index = zindex + 1; |
6644 | temp = extract_arithmetic_subst (string, &t_index); | |
6645 | zindex = t_index; | |
6646 | ||
6647 | /* Do initial variable expansion. */ | |
0628567a | 6648 | temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES); |
cce855bc JA |
6649 | |
6650 | goto arithsub; | |
6651 | ||
6652 | default: | |
6653 | /* Find the variable in VARIABLE_LIST. */ | |
6654 | temp = (char *)NULL; | |
6655 | ||
6656 | for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++) | |
6657 | ; | |
6658 | temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL; | |
6659 | ||
6660 | /* If this isn't a variable name, then just output the `$'. */ | |
6661 | if (temp1 == 0 || *temp1 == '\0') | |
6662 | { | |
6663 | FREE (temp1); | |
f73dda09 | 6664 | temp = (char *)xmalloc (2); |
cce855bc JA |
6665 | temp[0] = '$'; |
6666 | temp[1] = '\0'; | |
6667 | if (expanded_something) | |
6668 | *expanded_something = 0; | |
6669 | goto return0; | |
6670 | } | |
6671 | ||
6672 | /* If the variable exists, return its value cell. */ | |
6673 | var = find_variable (temp1); | |
6674 | ||
7117c2d2 | 6675 | if (var && invisible_p (var) == 0 && var_isset (var)) |
cce855bc JA |
6676 | { |
6677 | #if defined (ARRAY_VARS) | |
6678 | if (array_p (var)) | |
6679 | { | |
6680 | temp = array_reference (array_cell (var), 0); | |
6681 | if (temp) | |
b80f6443 JA |
6682 | temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
6683 | ? quote_string (temp) | |
6684 | : quote_escapes (temp); | |
6685 | else if (unbound_vars_is_error) | |
6686 | goto unbound_variable; | |
cce855bc JA |
6687 | } |
6688 | else | |
6689 | #endif | |
b80f6443 JA |
6690 | { |
6691 | temp = value_cell (var); | |
6692 | ||
6693 | temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
6694 | ? quote_string (temp) | |
6695 | : quote_escapes (temp); | |
6696 | } | |
6697 | ||
cce855bc | 6698 | free (temp1); |
7117c2d2 | 6699 | |
cce855bc JA |
6700 | goto return0; |
6701 | } | |
6702 | ||
6703 | temp = (char *)NULL; | |
6704 | ||
b80f6443 | 6705 | unbound_variable: |
cce855bc | 6706 | if (unbound_vars_is_error) |
7117c2d2 | 6707 | err_unboundvar (temp1); |
cce855bc JA |
6708 | else |
6709 | { | |
6710 | free (temp1); | |
6711 | goto return0; | |
6712 | } | |
6713 | ||
6714 | free (temp1); | |
6715 | last_command_exit_value = EXECUTION_FAILURE; | |
6716 | return ((unbound_vars_is_error && interactive_shell == 0) | |
95732b49 JA |
6717 | ? &expand_wdesc_fatal |
6718 | : &expand_wdesc_error); | |
cce855bc JA |
6719 | } |
6720 | ||
6721 | if (string[zindex]) | |
6722 | zindex++; | |
6723 | ||
6724 | return0: | |
6725 | *sindex = zindex; | |
95732b49 JA |
6726 | |
6727 | if (ret == 0) | |
6728 | { | |
6729 | ret = alloc_word_desc (); | |
6730 | ret->flags = tflag; /* XXX */ | |
6731 | ret->word = temp; | |
6732 | } | |
6733 | return ret; | |
cce855bc JA |
6734 | } |
6735 | ||
6736 | /* Make a word list which is the result of parameter and variable | |
6737 | expansion, command substitution, arithmetic substitution, and | |
6738 | quote removal of WORD. Return a pointer to a WORD_LIST which is | |
6739 | the result of the expansion. If WORD contains a null word, the | |
6740 | word list returned is also null. | |
726f6388 | 6741 | |
ccc6cda3 JA |
6742 | QUOTED contains flag values defined in shell.h. |
6743 | ||
b72432fd JA |
6744 | ISEXP is used to tell expand_word_internal that the word should be |
6745 | treated as the result of an expansion. This has implications for | |
6746 | how IFS characters in the word are treated. | |
6747 | ||
726f6388 JA |
6748 | CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null |
6749 | they point to an integer value which receives information about expansion. | |
6750 | CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero. | |
6751 | EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions, | |
6752 | else zero. | |
6753 | ||
6754 | This only does word splitting in the case of $@ expansion. In that | |
6755 | case, we split on ' '. */ | |
6756 | ||
6757 | /* Values for the local variable quoted_state. */ | |
6758 | #define UNQUOTED 0 | |
6759 | #define PARTIALLY_QUOTED 1 | |
6760 | #define WHOLLY_QUOTED 2 | |
6761 | ||
6762 | static WORD_LIST * | |
b72432fd | 6763 | expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something) |
726f6388 | 6764 | WORD_DESC *word; |
b72432fd | 6765 | int quoted, isexp; |
726f6388 JA |
6766 | int *contains_dollar_at; |
6767 | int *expanded_something; | |
6768 | { | |
ccc6cda3 JA |
6769 | WORD_LIST *list; |
6770 | WORD_DESC *tword; | |
726f6388 JA |
6771 | |
6772 | /* The intermediate string that we build while expanding. */ | |
ccc6cda3 | 6773 | char *istring; |
726f6388 JA |
6774 | |
6775 | /* The current size of the above object. */ | |
ccc6cda3 | 6776 | int istring_size; |
726f6388 JA |
6777 | |
6778 | /* Index into ISTRING. */ | |
ccc6cda3 | 6779 | int istring_index; |
726f6388 JA |
6780 | |
6781 | /* Temporary string storage. */ | |
ccc6cda3 | 6782 | char *temp, *temp1; |
726f6388 JA |
6783 | |
6784 | /* The text of WORD. */ | |
ccc6cda3 | 6785 | register char *string; |
726f6388 | 6786 | |
7117c2d2 JA |
6787 | /* The size of STRING. */ |
6788 | size_t string_size; | |
6789 | ||
726f6388 | 6790 | /* The index into STRING. */ |
ccc6cda3 | 6791 | int sindex; |
726f6388 JA |
6792 | |
6793 | /* This gets 1 if we see a $@ while quoted. */ | |
ccc6cda3 | 6794 | int quoted_dollar_at; |
726f6388 JA |
6795 | |
6796 | /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on | |
6797 | whether WORD contains no quoting characters, a partially quoted | |
6798 | string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */ | |
ccc6cda3 JA |
6799 | int quoted_state; |
6800 | ||
95732b49 | 6801 | /* State flags */ |
ccc6cda3 | 6802 | int had_quoted_null; |
cce855bc | 6803 | int has_dollar_at; |
28ef6c31 | 6804 | int tflag; |
726f6388 | 6805 | |
95732b49 JA |
6806 | int assignoff; /* If assignment, offset of `=' */ |
6807 | ||
f73dda09 | 6808 | register unsigned char c; /* Current character. */ |
726f6388 | 6809 | int t_index; /* For calls to string_extract_xxx. */ |
726f6388 | 6810 | |
bb70624e | 6811 | char twochars[2]; |
b72432fd | 6812 | |
7117c2d2 JA |
6813 | DECLARE_MBSTATE; |
6814 | ||
f73dda09 | 6815 | istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE); |
ccc6cda3 | 6816 | istring[istring_index = 0] = '\0'; |
cce855bc | 6817 | quoted_dollar_at = had_quoted_null = has_dollar_at = 0; |
ccc6cda3 JA |
6818 | quoted_state = UNQUOTED; |
6819 | ||
6820 | string = word->word; | |
6821 | if (string == 0) | |
6822 | goto finished_with_string; | |
95732b49 JA |
6823 | /* Don't need the string length for the SADD... and COPY_ macros unless |
6824 | multibyte characters are possible. */ | |
6825 | string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1; | |
726f6388 JA |
6826 | |
6827 | if (contains_dollar_at) | |
6828 | *contains_dollar_at = 0; | |
6829 | ||
95732b49 JA |
6830 | assignoff = -1; |
6831 | ||
726f6388 JA |
6832 | /* Begin the expansion. */ |
6833 | ||
ccc6cda3 | 6834 | for (sindex = 0; ;) |
726f6388 JA |
6835 | { |
6836 | c = string[sindex]; | |
6837 | ||
6838 | /* Case on toplevel character. */ | |
6839 | switch (c) | |
6840 | { | |
6841 | case '\0': | |
6842 | goto finished_with_string; | |
6843 | ||
6844 | case CTLESC: | |
7117c2d2 JA |
6845 | sindex++; |
6846 | #if HANDLE_MULTIBYTE | |
6847 | if (MB_CUR_MAX > 1 && string[sindex]) | |
6848 | { | |
b80f6443 | 6849 | SADD_MBQCHAR_BODY(temp, string, sindex, string_size); |
7117c2d2 JA |
6850 | } |
6851 | else | |
6852 | #endif | |
6853 | { | |
6854 | temp = (char *)xmalloc (3); | |
6855 | temp[0] = CTLESC; | |
6856 | temp[1] = c = string[sindex]; | |
6857 | temp[2] = '\0'; | |
6858 | } | |
726f6388 | 6859 | |
cce855bc | 6860 | dollar_add_string: |
726f6388 JA |
6861 | if (string[sindex]) |
6862 | sindex++; | |
6863 | ||
cce855bc JA |
6864 | add_string: |
6865 | if (temp) | |
6866 | { | |
6867 | istring = sub_append_string (temp, istring, &istring_index, &istring_size); | |
6868 | temp = (char *)0; | |
6869 | } | |
6870 | ||
6871 | break; | |
726f6388 JA |
6872 | |
6873 | #if defined (PROCESS_SUBSTITUTION) | |
6874 | /* Process substitution. */ | |
6875 | case '<': | |
6876 | case '>': | |
6877 | { | |
0628567a | 6878 | if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct) |
726f6388 | 6879 | { |
bb70624e | 6880 | sindex--; /* add_character: label increments sindex */ |
726f6388 JA |
6881 | goto add_character; |
6882 | } | |
6883 | else | |
cce855bc | 6884 | t_index = sindex + 1; /* skip past both '<' and LPAREN */ |
726f6388 | 6885 | |
cce855bc | 6886 | temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index); /*))*/ |
ccc6cda3 | 6887 | sindex = t_index; |
726f6388 JA |
6888 | |
6889 | /* If the process substitution specification is `<()', we want to | |
6890 | open the pipe for writing in the child and produce output; if | |
6891 | it is `>()', we want to open the pipe for reading in the child | |
6892 | and consume input. */ | |
ccc6cda3 | 6893 | temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0; |
726f6388 JA |
6894 | |
6895 | FREE (temp1); | |
6896 | ||
6897 | goto dollar_add_string; | |
6898 | } | |
6899 | #endif /* PROCESS_SUBSTITUTION */ | |
6900 | ||
95732b49 JA |
6901 | case '=': |
6902 | /* Posix.2 section 3.6.1 says that tildes following `=' in words | |
6903 | which are not assignment statements are not expanded. If the | |
6904 | shell isn't in posix mode, though, we perform tilde expansion | |
6905 | on `likely candidate' unquoted assignment statements (flags | |
6906 | include W_ASSIGNMENT but not W_QUOTED). A likely candidate | |
6907 | contains an unquoted :~ or =~. Something to think about: we | |
6908 | now have a flag that says to perform tilde expansion on arguments | |
6909 | to `assignment builtins' like declare and export that look like | |
6910 | assignment statements. We now do tilde expansion on such words | |
6911 | even in POSIX mode. */ | |
6912 | if (word->flags & (W_ASSIGNRHS|W_NOTILDE)) | |
6913 | goto add_character; | |
6914 | /* If we're not in posix mode or forcing assignment-statement tilde | |
6915 | expansion, note where the `=' appears in the word and prepare to | |
6916 | do tilde expansion following the first `='. */ | |
6917 | if ((word->flags & W_ASSIGNMENT) && | |
6918 | (posixly_correct == 0 || (word->flags & W_TILDEEXP)) && | |
6919 | assignoff == -1 && sindex > 0) | |
6920 | assignoff = sindex; | |
6921 | if (sindex == assignoff && string[sindex+1] == '~') /* XXX */ | |
6922 | word->flags |= W_ITILDE; | |
6923 | #if 0 | |
6924 | else if ((word->flags & W_ASSIGNMENT) && | |
6925 | (posixly_correct == 0 || (word->flags & W_TILDEEXP)) && | |
6926 | string[sindex+1] == '~') | |
6927 | word->flags |= W_ITILDE; | |
6928 | #endif | |
6929 | goto add_character; | |
6930 | ||
6931 | case ':': | |
6932 | if (word->flags & W_NOTILDE) | |
6933 | goto add_character; | |
6934 | ||
6935 | if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) && | |
6936 | string[sindex+1] == '~') | |
6937 | word->flags |= W_ITILDE; | |
6938 | goto add_character; | |
6939 | ||
6940 | case '~': | |
6941 | /* If the word isn't supposed to be tilde expanded, or we're not | |
6942 | at the start of a word or after an unquoted : or = in an | |
6943 | assignment statement, we don't do tilde expansion. */ | |
6944 | if ((word->flags & (W_NOTILDE|W_DQUOTE)) || | |
6945 | (sindex > 0 && ((word->flags & W_ITILDE) == 0)) || | |
6946 | (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
6947 | { | |
6948 | word->flags &= ~W_ITILDE; | |
6949 | goto add_character; | |
6950 | } | |
6951 | ||
6952 | if (word->flags & W_ASSIGNRHS) | |
6953 | tflag = 2; | |
6954 | else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP)) | |
6955 | tflag = 1; | |
6956 | else | |
6957 | tflag = 0; | |
6958 | ||
6959 | temp = bash_tilde_find_word (string + sindex, tflag, &t_index); | |
6960 | ||
6961 | word->flags &= ~W_ITILDE; | |
6962 | ||
6963 | if (temp && *temp && t_index > 0) | |
6964 | { | |
6965 | temp1 = bash_tilde_expand (temp, tflag); | |
0628567a JA |
6966 | if (temp1 && *temp1 == '~' && STREQ (temp, temp1)) |
6967 | { | |
6968 | FREE (temp); | |
6969 | FREE (temp1); | |
6970 | goto add_character; /* tilde expansion failed */ | |
6971 | } | |
95732b49 JA |
6972 | free (temp); |
6973 | temp = temp1; | |
6974 | sindex += t_index; | |
6975 | goto add_string; | |
6976 | } | |
6977 | else | |
6978 | { | |
6979 | FREE (temp); | |
6980 | goto add_character; | |
6981 | } | |
6982 | ||
726f6388 | 6983 | case '$': |
726f6388 JA |
6984 | if (expanded_something) |
6985 | *expanded_something = 1; | |
6986 | ||
cce855bc | 6987 | has_dollar_at = 0; |
95732b49 | 6988 | tword = param_expand (string, &sindex, quoted, expanded_something, |
cce855bc | 6989 | &has_dollar_at, "ed_dollar_at, |
b80f6443 JA |
6990 | &had_quoted_null, |
6991 | (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0); | |
726f6388 | 6992 | |
95732b49 | 6993 | if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal) |
726f6388 | 6994 | { |
cce855bc JA |
6995 | free (string); |
6996 | free (istring); | |
95732b49 JA |
6997 | return ((tword == &expand_wdesc_error) ? &expand_word_error |
6998 | : &expand_word_fatal); | |
cce855bc JA |
6999 | } |
7000 | if (contains_dollar_at && has_dollar_at) | |
7001 | *contains_dollar_at = 1; | |
95732b49 JA |
7002 | |
7003 | if (tword && (tword->flags & W_HASQUOTEDNULL)) | |
7004 | had_quoted_null = 1; | |
7005 | ||
7006 | temp = tword->word; | |
7007 | dispose_word_desc (tword); | |
7008 | ||
cce855bc JA |
7009 | goto add_string; |
7010 | break; | |
726f6388 | 7011 | |
cce855bc JA |
7012 | case '`': /* Backquoted command substitution. */ |
7013 | { | |
b80f6443 | 7014 | t_index = sindex++; |
726f6388 | 7015 | |
95732b49 JA |
7016 | temp = string_extract (string, &sindex, "`", EX_REQMATCH); |
7017 | /* The test of sindex against t_index is to allow bare instances of | |
7018 | ` to pass through, for backwards compatibility. */ | |
7019 | if (temp == &extract_string_error || temp == &extract_string_fatal) | |
7020 | { | |
7021 | if (sindex - 1 == t_index) | |
7022 | { | |
7023 | sindex = t_index; | |
7024 | goto add_character; | |
7025 | } | |
7026 | report_error ("bad substitution: no closing \"`\" in %s", string+t_index); | |
7027 | free (string); | |
7028 | free (istring); | |
7029 | return ((temp == &extract_string_error) ? &expand_word_error | |
7030 | : &expand_word_fatal); | |
7031 | } | |
7032 | ||
cce855bc JA |
7033 | if (expanded_something) |
7034 | *expanded_something = 1; | |
726f6388 | 7035 | |
b80f6443 JA |
7036 | if (word->flags & W_NOCOMSUB) |
7037 | /* sindex + 1 because string[sindex] == '`' */ | |
7038 | temp1 = substring (string, t_index, sindex + 1); | |
7039 | else | |
7040 | { | |
7041 | de_backslash (temp); | |
7042 | temp1 = command_substitute (temp, quoted); | |
7043 | } | |
cce855bc JA |
7044 | FREE (temp); |
7045 | temp = temp1; | |
7046 | goto dollar_add_string; | |
7047 | } | |
ccc6cda3 | 7048 | |
cce855bc JA |
7049 | case '\\': |
7050 | if (string[sindex + 1] == '\n') | |
7051 | { | |
7052 | sindex += 2; | |
7053 | continue; | |
7054 | } | |
726f6388 | 7055 | |
cce855bc | 7056 | c = string[++sindex]; |
726f6388 | 7057 | |
cce855bc | 7058 | if (quoted & Q_HERE_DOCUMENT) |
28ef6c31 | 7059 | tflag = CBSHDOC; |
cce855bc | 7060 | else if (quoted & Q_DOUBLE_QUOTES) |
28ef6c31 | 7061 | tflag = CBSDQUOTE; |
cce855bc | 7062 | else |
28ef6c31 JA |
7063 | tflag = 0; |
7064 | ||
28ef6c31 | 7065 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0)) |
cce855bc | 7066 | { |
7117c2d2 | 7067 | SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size); |
bb70624e JA |
7068 | } |
7069 | else if (c == 0) | |
7070 | { | |
7071 | c = CTLNUL; | |
7072 | sindex--; /* add_character: label increments sindex */ | |
7073 | goto add_character; | |
cce855bc JA |
7074 | } |
7075 | else | |
bb70624e | 7076 | { |
7117c2d2 | 7077 | SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size); |
bb70624e | 7078 | } |
726f6388 | 7079 | |
bb70624e JA |
7080 | sindex++; |
7081 | add_twochars: | |
7082 | /* BEFORE jumping here, we need to increment sindex if appropriate */ | |
7083 | RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, | |
7084 | DEFAULT_ARRAY_SIZE); | |
7085 | istring[istring_index++] = twochars[0]; | |
7086 | istring[istring_index++] = twochars[1]; | |
7087 | istring[istring_index] = '\0'; | |
7088 | ||
7089 | break; | |
726f6388 | 7090 | |
cce855bc | 7091 | case '"': |
7117c2d2 | 7092 | #if 0 |
95732b49 | 7093 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE)) |
7117c2d2 | 7094 | #else |
95732b49 | 7095 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
7117c2d2 | 7096 | #endif |
cce855bc | 7097 | goto add_character; |
ccc6cda3 JA |
7098 | |
7099 | t_index = ++sindex; | |
7100 | temp = string_extract_double_quoted (string, &sindex, 0); | |
7101 | ||
7102 | /* If the quotes surrounded the entire string, then the | |
7103 | whole word was quoted. */ | |
7104 | quoted_state = (t_index == 1 && string[sindex] == '\0') | |
7105 | ? WHOLLY_QUOTED | |
7117c2d2 | 7106 | : PARTIALLY_QUOTED; |
ccc6cda3 JA |
7107 | |
7108 | if (temp && *temp) | |
726f6388 | 7109 | { |
95732b49 JA |
7110 | tword = alloc_word_desc (); |
7111 | tword->word = temp; | |
7112 | ||
ccc6cda3 JA |
7113 | temp = (char *)NULL; |
7114 | ||
cce855bc | 7115 | has_dollar_at = 0; |
95732b49 | 7116 | /* Need to get W_HASQUOTEDNULL flag through this function. */ |
b72432fd | 7117 | list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &has_dollar_at, (int *)NULL); |
726f6388 | 7118 | |
ccc6cda3 JA |
7119 | if (list == &expand_word_error || list == &expand_word_fatal) |
7120 | { | |
7121 | free (istring); | |
7122 | free (string); | |
7123 | /* expand_word_internal has already freed temp_word->word | |
7124 | for us because of the way it prints error messages. */ | |
7125 | tword->word = (char *)NULL; | |
7126 | dispose_word (tword); | |
7127 | return list; | |
7128 | } | |
726f6388 | 7129 | |
ccc6cda3 | 7130 | dispose_word (tword); |
726f6388 | 7131 | |
ccc6cda3 JA |
7132 | /* "$@" (a double-quoted dollar-at) expands into nothing, |
7133 | not even a NULL word, when there are no positional | |
7134 | parameters. */ | |
cce855bc | 7135 | if (list == 0 && has_dollar_at) |
726f6388 | 7136 | { |
ccc6cda3 JA |
7137 | quoted_dollar_at++; |
7138 | break; | |
7139 | } | |
7140 | ||
7141 | /* If we get "$@", we know we have expanded something, so we | |
7142 | need to remember it for the final split on $IFS. This is | |
7143 | a special case; it's the only case where a quoted string | |
7144 | can expand into more than one word. It's going to come back | |
7145 | from the above call to expand_word_internal as a list with | |
7146 | a single word, in which all characters are quoted and | |
7147 | separated by blanks. What we want to do is to turn it back | |
7148 | into a list for the next piece of code. */ | |
7149 | if (list) | |
7150 | dequote_list (list); | |
7151 | ||
95732b49 JA |
7152 | if (list && list->word && (list->word->flags & W_HASQUOTEDNULL)) |
7153 | had_quoted_null = 1; | |
7154 | ||
cce855bc | 7155 | if (has_dollar_at) |
ccc6cda3 JA |
7156 | { |
7157 | quoted_dollar_at++; | |
7158 | if (contains_dollar_at) | |
7159 | *contains_dollar_at = 1; | |
7160 | if (expanded_something) | |
7161 | *expanded_something = 1; | |
7162 | } | |
7163 | } | |
7164 | else | |
7165 | { | |
7166 | /* What we have is "". This is a minor optimization. */ | |
f73dda09 | 7167 | FREE (temp); |
ccc6cda3 JA |
7168 | list = (WORD_LIST *)NULL; |
7169 | } | |
7170 | ||
7171 | /* The code above *might* return a list (consider the case of "$@", | |
7172 | where it returns "$1", "$2", etc.). We can't throw away the | |
7173 | rest of the list, and we have to make sure each word gets added | |
7174 | as quoted. We test on tresult->next: if it is non-NULL, we | |
7175 | quote the whole list, save it to a string with string_list, and | |
7176 | add that string. We don't need to quote the results of this | |
7177 | (and it would be wrong, since that would quote the separators | |
7178 | as well), so we go directly to add_string. */ | |
7179 | if (list) | |
7180 | { | |
7181 | if (list->next) | |
7182 | { | |
bc4cd23c JA |
7183 | /* Testing quoted_dollar_at makes sure that "$@" is |
7184 | split correctly when $IFS does not contain a space. */ | |
7185 | temp = quoted_dollar_at | |
7186 | ? string_list_dollar_at (list, Q_DOUBLE_QUOTES) | |
7187 | : string_list (quote_list (list)); | |
ccc6cda3 | 7188 | dispose_words (list); |
726f6388 JA |
7189 | goto add_string; |
7190 | } | |
7191 | else | |
7192 | { | |
ccc6cda3 | 7193 | temp = savestring (list->word->word); |
95732b49 | 7194 | tflag = list->word->flags; |
ccc6cda3 | 7195 | dispose_words (list); |
95732b49 | 7196 | |
cce855bc JA |
7197 | /* If the string is not a quoted null string, we want |
7198 | to remove any embedded unquoted CTLNUL characters. | |
7199 | We do not want to turn quoted null strings back into | |
7200 | the empty string, though. We do this because we | |
7201 | want to remove any quoted nulls from expansions that | |
7202 | contain other characters. For example, if we have | |
7203 | x"$*"y or "x$*y" and there are no positional parameters, | |
7117c2d2 | 7204 | the $* should expand into nothing. */ |
95732b49 JA |
7205 | /* We use the W_HASQUOTEDNULL flag to differentiate the |
7206 | cases: a quoted null character as above and when | |
7207 | CTLNUL is contained in the (non-null) expansion | |
7208 | of some variable. We use the had_quoted_null flag to | |
7209 | pass the value through this function to its caller. */ | |
7210 | if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0) | |
cce855bc | 7211 | remove_quoted_nulls (temp); /* XXX */ |
726f6388 JA |
7212 | } |
7213 | } | |
ccc6cda3 JA |
7214 | else |
7215 | temp = (char *)NULL; | |
726f6388 | 7216 | |
ccc6cda3 JA |
7217 | /* We do not want to add quoted nulls to strings that are only |
7218 | partially quoted; we can throw them away. */ | |
7219 | if (temp == 0 && quoted_state == PARTIALLY_QUOTED) | |
cce855bc | 7220 | continue; |
726f6388 | 7221 | |
ccc6cda3 | 7222 | add_quoted_string: |
726f6388 | 7223 | |
ccc6cda3 JA |
7224 | if (temp) |
7225 | { | |
7226 | temp1 = temp; | |
7227 | temp = quote_string (temp); | |
7228 | free (temp1); | |
bb70624e | 7229 | goto add_string; |
ccc6cda3 JA |
7230 | } |
7231 | else | |
7232 | { | |
7233 | /* Add NULL arg. */ | |
bb70624e JA |
7234 | c = CTLNUL; |
7235 | sindex--; /* add_character: label increments sindex */ | |
7236 | goto add_character; | |
ccc6cda3 | 7237 | } |
bb70624e | 7238 | |
ccc6cda3 | 7239 | /* break; */ |
726f6388 | 7240 | |
ccc6cda3 | 7241 | case '\'': |
7117c2d2 | 7242 | #if 0 |
95732b49 | 7243 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE)) |
7117c2d2 | 7244 | #else |
95732b49 | 7245 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
7117c2d2 | 7246 | #endif |
ccc6cda3 | 7247 | goto add_character; |
726f6388 | 7248 | |
ccc6cda3 JA |
7249 | t_index = ++sindex; |
7250 | temp = string_extract_single_quoted (string, &sindex); | |
726f6388 | 7251 | |
ccc6cda3 JA |
7252 | /* If the entire STRING was surrounded by single quotes, |
7253 | then the string is wholly quoted. */ | |
7254 | quoted_state = (t_index == 1 && string[sindex] == '\0') | |
7255 | ? WHOLLY_QUOTED | |
7117c2d2 | 7256 | : PARTIALLY_QUOTED; |
726f6388 | 7257 | |
ccc6cda3 JA |
7258 | /* If all we had was '', it is a null expansion. */ |
7259 | if (*temp == '\0') | |
7260 | { | |
7261 | free (temp); | |
7262 | temp = (char *)NULL; | |
7263 | } | |
7264 | else | |
7117c2d2 | 7265 | remove_quoted_escapes (temp); /* ??? */ |
726f6388 | 7266 | |
ccc6cda3 JA |
7267 | /* We do not want to add quoted nulls to strings that are only |
7268 | partially quoted; such nulls are discarded. */ | |
7269 | if (temp == 0 && (quoted_state == PARTIALLY_QUOTED)) | |
7270 | continue; | |
726f6388 | 7271 | |
bb70624e JA |
7272 | /* If we have a quoted null expansion, add a quoted NULL to istring. */ |
7273 | if (temp == 0) | |
7274 | { | |
7275 | c = CTLNUL; | |
7276 | sindex--; /* add_character: label increments sindex */ | |
7277 | goto add_character; | |
7278 | } | |
7279 | else | |
7280 | goto add_quoted_string; | |
7281 | ||
ccc6cda3 | 7282 | /* break; */ |
726f6388 JA |
7283 | |
7284 | default: | |
726f6388 | 7285 | /* This is the fix for " $@ " */ |
7117c2d2 | 7286 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c))) |
726f6388 | 7287 | { |
bb70624e JA |
7288 | if (string[sindex]) /* from old goto dollar_add_string */ |
7289 | sindex++; | |
7290 | if (c == 0) | |
7291 | { | |
7292 | c = CTLNUL; | |
7293 | goto add_character; | |
7294 | } | |
7295 | else | |
7296 | { | |
7117c2d2 | 7297 | #if HANDLE_MULTIBYTE |
b80f6443 JA |
7298 | if (MB_CUR_MAX > 1) |
7299 | sindex--; | |
7300 | ||
7117c2d2 JA |
7301 | if (MB_CUR_MAX > 1) |
7302 | { | |
b80f6443 | 7303 | SADD_MBQCHAR_BODY(temp, string, sindex, string_size); |
7117c2d2 JA |
7304 | } |
7305 | else | |
7306 | #endif | |
7307 | { | |
7308 | twochars[0] = CTLESC; | |
7309 | twochars[1] = c; | |
7310 | goto add_twochars; | |
7311 | } | |
bb70624e | 7312 | } |
726f6388 JA |
7313 | } |
7314 | ||
7117c2d2 JA |
7315 | SADD_MBCHAR (temp, string, sindex, string_size); |
7316 | ||
726f6388 | 7317 | add_character: |
ccc6cda3 JA |
7318 | RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size, |
7319 | DEFAULT_ARRAY_SIZE); | |
726f6388 JA |
7320 | istring[istring_index++] = c; |
7321 | istring[istring_index] = '\0'; | |
7322 | ||
7323 | /* Next character. */ | |
7324 | sindex++; | |
7325 | } | |
7326 | } | |
7327 | ||
7328 | finished_with_string: | |
726f6388 JA |
7329 | /* OK, we're ready to return. If we have a quoted string, and |
7330 | quoted_dollar_at is not set, we do no splitting at all; otherwise | |
7331 | we split on ' '. The routines that call this will handle what to | |
7332 | do if nothing has been expanded. */ | |
ccc6cda3 JA |
7333 | |
7334 | /* Partially and wholly quoted strings which expand to the empty | |
7335 | string are retained as an empty arguments. Unquoted strings | |
7336 | which expand to the empty string are discarded. The single | |
7337 | exception is the case of expanding "$@" when there are no | |
7338 | positional parameters. In that case, we discard the expansion. */ | |
7339 | ||
7340 | /* Because of how the code that handles "" and '' in partially | |
7341 | quoted strings works, we need to make ISTRING into a QUOTED_NULL | |
7342 | if we saw quoting characters, but the expansion was empty. | |
7343 | "" and '' are tossed away before we get to this point when | |
7344 | processing partially quoted strings. This makes "" and $xxx"" | |
7345 | equivalent when xxx is unset. We also look to see whether we | |
7346 | saw a quoted null from a ${} expansion and add one back if we | |
7347 | need to. */ | |
7348 | ||
7349 | /* If we expand to nothing and there were no single or double quotes | |
7350 | in the word, we throw it away. Otherwise, we return a NULL word. | |
7351 | The single exception is for $@ surrounded by double quotes when | |
7352 | there are no positional parameters. In that case, we also throw | |
7353 | the word away. */ | |
7354 | ||
7355 | if (*istring == '\0') | |
7356 | { | |
7357 | if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED)) | |
726f6388 | 7358 | { |
726f6388 JA |
7359 | istring[0] = CTLNUL; |
7360 | istring[1] = '\0'; | |
ccc6cda3 | 7361 | tword = make_bare_word (istring); |
95732b49 | 7362 | tword->flags |= W_HASQUOTEDNULL; /* XXX */ |
ccc6cda3 JA |
7363 | list = make_word_list (tword, (WORD_LIST *)NULL); |
7364 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
7365 | tword->flags |= W_QUOTED; | |
726f6388 | 7366 | } |
ccc6cda3 JA |
7367 | /* According to sh, ksh, and Posix.2, if a word expands into nothing |
7368 | and a double-quoted "$@" appears anywhere in it, then the entire | |
7369 | word is removed. */ | |
7370 | else if (quoted_state == UNQUOTED || quoted_dollar_at) | |
7371 | list = (WORD_LIST *)NULL; | |
7372 | #if 0 | |
7373 | else | |
726f6388 | 7374 | { |
ccc6cda3 | 7375 | tword = make_bare_word (istring); |
ccc6cda3 JA |
7376 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
7377 | tword->flags |= W_QUOTED; | |
95732b49 | 7378 | list = make_word_list (tword, (WORD_LIST *)NULL); |
726f6388 | 7379 | } |
f73dda09 JA |
7380 | #else |
7381 | else | |
7382 | list = (WORD_LIST *)NULL; | |
ccc6cda3 JA |
7383 | #endif |
7384 | } | |
7385 | else if (word->flags & W_NOSPLIT) | |
7386 | { | |
7387 | tword = make_bare_word (istring); | |
ccc6cda3 JA |
7388 | if (word->flags & W_ASSIGNMENT) |
7389 | tword->flags |= W_ASSIGNMENT; /* XXX */ | |
95732b49 JA |
7390 | if (word->flags & W_COMPASSIGN) |
7391 | tword->flags |= W_COMPASSIGN; /* XXX */ | |
b72432fd JA |
7392 | if (word->flags & W_NOGLOB) |
7393 | tword->flags |= W_NOGLOB; /* XXX */ | |
95732b49 JA |
7394 | if (word->flags & W_NOEXPAND) |
7395 | tword->flags |= W_NOEXPAND; /* XXX */ | |
ccc6cda3 | 7396 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
28ef6c31 | 7397 | tword->flags |= W_QUOTED; |
95732b49 JA |
7398 | if (had_quoted_null) |
7399 | tword->flags |= W_HASQUOTEDNULL; | |
7400 | list = make_word_list (tword, (WORD_LIST *)NULL); | |
ccc6cda3 JA |
7401 | } |
7402 | else | |
7403 | { | |
7404 | char *ifs_chars; | |
7405 | ||
7117c2d2 | 7406 | ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL; |
726f6388 | 7407 | |
cce855bc JA |
7408 | /* If we have $@, we need to split the results no matter what. If |
7409 | IFS is unset or NULL, string_list_dollar_at has separated the | |
7410 | positional parameters with a space, so we split on space (we have | |
7411 | set ifs_chars to " \t\n" above if ifs is unset). If IFS is set, | |
7412 | string_list_dollar_at has separated the positional parameters | |
7413 | with the first character of $IFS, so we split on $IFS. */ | |
7414 | if (has_dollar_at && ifs_chars) | |
7415 | list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1); | |
ccc6cda3 JA |
7416 | else |
7417 | { | |
7418 | tword = make_bare_word (istring); | |
ccc6cda3 JA |
7419 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED)) |
7420 | tword->flags |= W_QUOTED; | |
7421 | if (word->flags & W_ASSIGNMENT) | |
7422 | tword->flags |= W_ASSIGNMENT; | |
95732b49 JA |
7423 | if (word->flags & W_COMPASSIGN) |
7424 | tword->flags |= W_COMPASSIGN; | |
b72432fd JA |
7425 | if (word->flags & W_NOGLOB) |
7426 | tword->flags |= W_NOGLOB; | |
95732b49 JA |
7427 | if (word->flags & W_NOEXPAND) |
7428 | tword->flags |= W_NOEXPAND; | |
7429 | if (had_quoted_null) | |
7430 | tword->flags |= W_HASQUOTEDNULL; /* XXX */ | |
7431 | list = make_word_list (tword, (WORD_LIST *)NULL); | |
726f6388 | 7432 | } |
726f6388 | 7433 | } |
726f6388 | 7434 | |
ccc6cda3 JA |
7435 | free (istring); |
7436 | return (list); | |
726f6388 JA |
7437 | } |
7438 | ||
7439 | /* **************************************************************** */ | |
7440 | /* */ | |
7441 | /* Functions for Quote Removal */ | |
7442 | /* */ | |
7443 | /* **************************************************************** */ | |
7444 | ||
7445 | /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the | |
7117c2d2 | 7446 | backslash quoting rules for within double quotes or a here document. */ |
726f6388 JA |
7447 | char * |
7448 | string_quote_removal (string, quoted) | |
7449 | char *string; | |
7450 | int quoted; | |
7451 | { | |
7117c2d2 JA |
7452 | size_t slen; |
7453 | char *r, *result_string, *temp, *send; | |
f73dda09 JA |
7454 | int sindex, tindex, dquote; |
7455 | unsigned char c; | |
7117c2d2 | 7456 | DECLARE_MBSTATE; |
726f6388 JA |
7457 | |
7458 | /* The result can be no longer than the original string. */ | |
7117c2d2 JA |
7459 | slen = strlen (string); |
7460 | send = string + slen; | |
7461 | ||
7462 | r = result_string = (char *)xmalloc (slen + 1); | |
726f6388 | 7463 | |
ccc6cda3 | 7464 | for (dquote = sindex = 0; c = string[sindex];) |
726f6388 JA |
7465 | { |
7466 | switch (c) | |
7467 | { | |
7468 | case '\\': | |
7469 | c = string[++sindex]; | |
28ef6c31 | 7470 | if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0) |
726f6388 | 7471 | *r++ = '\\'; |
ccc6cda3 | 7472 | /* FALLTHROUGH */ |
726f6388 JA |
7473 | |
7474 | default: | |
7117c2d2 | 7475 | SCOPY_CHAR_M (r, string, send, sindex); |
726f6388 JA |
7476 | break; |
7477 | ||
7478 | case '\'': | |
ccc6cda3 | 7479 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) |
726f6388 JA |
7480 | { |
7481 | *r++ = c; | |
7482 | sindex++; | |
ccc6cda3 | 7483 | break; |
726f6388 | 7484 | } |
ccc6cda3 JA |
7485 | tindex = sindex + 1; |
7486 | temp = string_extract_single_quoted (string, &tindex); | |
7487 | if (temp) | |
726f6388 | 7488 | { |
ccc6cda3 JA |
7489 | strcpy (r, temp); |
7490 | r += strlen (r); | |
7491 | free (temp); | |
726f6388 | 7492 | } |
ccc6cda3 | 7493 | sindex = tindex; |
726f6388 JA |
7494 | break; |
7495 | ||
7496 | case '"': | |
7497 | dquote = 1 - dquote; | |
7498 | sindex++; | |
7499 | break; | |
7500 | } | |
7501 | } | |
7502 | *r = '\0'; | |
7503 | return (result_string); | |
7504 | } | |
7505 | ||
ccc6cda3 JA |
7506 | #if 0 |
7507 | /* UNUSED */ | |
726f6388 JA |
7508 | /* Perform quote removal on word WORD. This allocates and returns a new |
7509 | WORD_DESC *. */ | |
7510 | WORD_DESC * | |
7511 | word_quote_removal (word, quoted) | |
7512 | WORD_DESC *word; | |
7513 | int quoted; | |
7514 | { | |
7515 | WORD_DESC *w; | |
7516 | char *t; | |
7517 | ||
7518 | t = string_quote_removal (word->word, quoted); | |
95732b49 JA |
7519 | w = alloc_word_desc (); |
7520 | w->word = t ? t : savestring (""); | |
726f6388 JA |
7521 | return (w); |
7522 | } | |
7523 | ||
7524 | /* Perform quote removal on all words in LIST. If QUOTED is non-zero, | |
7525 | the members of the list are treated as if they are surrounded by | |
7526 | double quotes. Return a new list, or NULL if LIST is NULL. */ | |
7527 | WORD_LIST * | |
7528 | word_list_quote_removal (list, quoted) | |
7529 | WORD_LIST *list; | |
7530 | int quoted; | |
7531 | { | |
95732b49 | 7532 | WORD_LIST *result, *t, *tresult, *e; |
726f6388 | 7533 | |
ccc6cda3 | 7534 | for (t = list, result = (WORD_LIST *)NULL; t; t = t->next) |
726f6388 | 7535 | { |
7117c2d2 | 7536 | tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL); |
95732b49 | 7537 | #if 0 |
726f6388 | 7538 | result = (WORD_LIST *) list_append (result, tresult); |
95732b49 JA |
7539 | #else |
7540 | if (result == 0) | |
7541 | result = e = tresult; | |
7542 | else | |
7543 | { | |
7544 | e->next = tresult; | |
7545 | while (e->next) | |
7546 | e = e->next; | |
7547 | } | |
7548 | #endif | |
726f6388 JA |
7549 | } |
7550 | return (result); | |
7551 | } | |
ccc6cda3 | 7552 | #endif |
726f6388 | 7553 | |
726f6388 JA |
7554 | /******************************************* |
7555 | * * | |
7556 | * Functions to perform word splitting * | |
7557 | * * | |
7558 | *******************************************/ | |
7559 | ||
7117c2d2 JA |
7560 | void |
7561 | setifs (v) | |
7562 | SHELL_VAR *v; | |
b72432fd | 7563 | { |
7117c2d2 JA |
7564 | char *t; |
7565 | unsigned char uc; | |
7566 | ||
7567 | ifs_var = v; | |
95732b49 | 7568 | #if 0 |
7117c2d2 | 7569 | ifs_value = v ? value_cell (v) : " \t\n"; |
95732b49 JA |
7570 | #else |
7571 | ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n"; | |
7572 | #endif | |
b72432fd | 7573 | |
95732b49 JA |
7574 | /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet |
7575 | handle multibyte chars in IFS */ | |
7117c2d2 JA |
7576 | memset (ifs_cmap, '\0', sizeof (ifs_cmap)); |
7577 | for (t = ifs_value ; t && *t; t++) | |
7578 | { | |
7579 | uc = *t; | |
7580 | ifs_cmap[uc] = 1; | |
7581 | } | |
7582 | ||
95732b49 JA |
7583 | #if defined (HANDLE_MULTIBYTE) |
7584 | if (ifs_value == 0) | |
7585 | { | |
7586 | ifs_firstc[0] = '\0'; | |
7587 | ifs_firstc_len = 1; | |
7588 | } | |
7589 | else | |
7590 | { | |
7591 | size_t ifs_len; | |
7592 | ifs_len = strnlen (ifs_value, MB_CUR_MAX); | |
7593 | ifs_firstc_len = MBLEN (ifs_value, ifs_len); | |
7594 | if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len)) | |
7595 | { | |
7596 | ifs_firstc[0] = ifs_value[0]; | |
7597 | ifs_firstc[1] = '\0'; | |
7598 | ifs_firstc_len = 1; | |
7599 | } | |
7600 | else | |
7601 | memcpy (ifs_firstc, ifs_value, ifs_firstc_len); | |
7602 | } | |
7603 | #else | |
7117c2d2 | 7604 | ifs_firstc = ifs_value ? *ifs_value : 0; |
95732b49 | 7605 | #endif |
7117c2d2 JA |
7606 | } |
7607 | ||
7608 | char * | |
7609 | getifs () | |
7610 | { | |
7611 | return ifs_value; | |
b72432fd JA |
7612 | } |
7613 | ||
726f6388 JA |
7614 | /* This splits a single word into a WORD LIST on $IFS, but only if the word |
7615 | is not quoted. list_string () performs quote removal for us, even if we | |
7616 | don't do any splitting. */ | |
7617 | WORD_LIST * | |
7117c2d2 | 7618 | word_split (w, ifs_chars) |
726f6388 | 7619 | WORD_DESC *w; |
7117c2d2 | 7620 | char *ifs_chars; |
726f6388 JA |
7621 | { |
7622 | WORD_LIST *result; | |
7623 | ||
7624 | if (w) | |
7625 | { | |
7117c2d2 | 7626 | char *xifs; |
726f6388 | 7627 | |
7117c2d2 JA |
7628 | xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars; |
7629 | result = list_string (w->word, xifs, w->flags & W_QUOTED); | |
726f6388 JA |
7630 | } |
7631 | else | |
7632 | result = (WORD_LIST *)NULL; | |
ccc6cda3 | 7633 | |
726f6388 JA |
7634 | return (result); |
7635 | } | |
7636 | ||
7637 | /* Perform word splitting on LIST and return the RESULT. It is possible | |
7638 | to return (WORD_LIST *)NULL. */ | |
7639 | static WORD_LIST * | |
7640 | word_list_split (list) | |
7641 | WORD_LIST *list; | |
7642 | { | |
95732b49 | 7643 | WORD_LIST *result, *t, *tresult, *e; |
726f6388 | 7644 | |
ccc6cda3 | 7645 | for (t = list, result = (WORD_LIST *)NULL; t; t = t->next) |
726f6388 | 7646 | { |
7117c2d2 | 7647 | tresult = word_split (t->word, ifs_value); |
95732b49 JA |
7648 | if (result == 0) |
7649 | result = e = tresult; | |
7650 | else | |
7651 | { | |
7652 | e->next = tresult; | |
7653 | while (e->next) | |
7654 | e = e->next; | |
7655 | } | |
726f6388 JA |
7656 | } |
7657 | return (result); | |
7658 | } | |
7659 | ||
7660 | /************************************************** | |
7661 | * * | |
cce855bc | 7662 | * Functions to expand an entire WORD_LIST * |
726f6388 JA |
7663 | * * |
7664 | **************************************************/ | |
7665 | ||
b80f6443 JA |
7666 | /* Do any word-expansion-specific cleanup and jump to top_level */ |
7667 | static void | |
7668 | exp_jump_to_top_level (v) | |
7669 | int v; | |
7670 | { | |
7671 | /* Cleanup code goes here. */ | |
7672 | expand_no_split_dollar_star = 0; /* XXX */ | |
7673 | expanding_redir = 0; | |
7674 | ||
f1be666c JA |
7675 | if (parse_and_execute_level == 0) |
7676 | top_level_cleanup (); /* from sig.c */ | |
7677 | ||
7678 | ||
b80f6443 JA |
7679 | jump_to_top_level (v); |
7680 | } | |
7681 | ||
cce855bc JA |
7682 | /* Put NLIST (which is a WORD_LIST * of only one element) at the front of |
7683 | ELIST, and set ELIST to the new list. */ | |
7684 | #define PREPEND_LIST(nlist, elist) \ | |
7685 | do { nlist->next = elist; elist = nlist; } while (0) | |
7686 | ||
726f6388 JA |
7687 | /* Separate out any initial variable assignments from TLIST. If set -k has |
7688 | been executed, remove all assignment statements from TLIST. Initial | |
7689 | variable assignments and other environment assignments are placed | |
bb70624e | 7690 | on SUBST_ASSIGN_VARLIST. */ |
726f6388 JA |
7691 | static WORD_LIST * |
7692 | separate_out_assignments (tlist) | |
7693 | WORD_LIST *tlist; | |
7694 | { | |
7695 | register WORD_LIST *vp, *lp; | |
7696 | ||
7697 | if (!tlist) | |
7698 | return ((WORD_LIST *)NULL); | |
7699 | ||
bb70624e JA |
7700 | if (subst_assign_varlist) |
7701 | dispose_words (subst_assign_varlist); /* Clean up after previous error */ | |
b72432fd | 7702 | |
bb70624e | 7703 | subst_assign_varlist = (WORD_LIST *)NULL; |
726f6388 JA |
7704 | vp = lp = tlist; |
7705 | ||
7706 | /* Separate out variable assignments at the start of the command. | |
7707 | Loop invariant: vp->next == lp | |
7708 | Loop postcondition: | |
7117c2d2 JA |
7709 | lp = list of words left after assignment statements skipped |
7710 | tlist = original list of words | |
726f6388 | 7711 | */ |
ccc6cda3 | 7712 | while (lp && (lp->word->flags & W_ASSIGNMENT)) |
726f6388 JA |
7713 | { |
7714 | vp = lp; | |
7715 | lp = lp->next; | |
7716 | } | |
7717 | ||
bb70624e JA |
7718 | /* If lp != tlist, we have some initial assignment statements. |
7719 | We make SUBST_ASSIGN_VARLIST point to the list of assignment | |
7720 | words and TLIST point to the remaining words. */ | |
726f6388 JA |
7721 | if (lp != tlist) |
7722 | { | |
bb70624e | 7723 | subst_assign_varlist = tlist; |
726f6388 JA |
7724 | /* ASSERT(vp->next == lp); */ |
7725 | vp->next = (WORD_LIST *)NULL; /* terminate variable list */ | |
7726 | tlist = lp; /* remainder of word list */ | |
7727 | } | |
7728 | ||
7729 | /* vp == end of variable list */ | |
7730 | /* tlist == remainder of original word list without variable assignments */ | |
7731 | if (!tlist) | |
7732 | /* All the words in tlist were assignment statements */ | |
7733 | return ((WORD_LIST *)NULL); | |
7734 | ||
7735 | /* ASSERT(tlist != NULL); */ | |
ccc6cda3 | 7736 | /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */ |
726f6388 JA |
7737 | |
7738 | /* If the -k option is in effect, we need to go through the remaining | |
bb70624e JA |
7739 | words, separate out the assignment words, and place them on |
7740 | SUBST_ASSIGN_VARLIST. */ | |
726f6388 JA |
7741 | if (place_keywords_in_env) |
7742 | { | |
7743 | WORD_LIST *tp; /* tp == running pointer into tlist */ | |
7744 | ||
7745 | tp = tlist; | |
7746 | lp = tlist->next; | |
7747 | ||
7748 | /* Loop Invariant: tp->next == lp */ | |
7749 | /* Loop postcondition: tlist == word list without assignment statements */ | |
7750 | while (lp) | |
7751 | { | |
ccc6cda3 | 7752 | if (lp->word->flags & W_ASSIGNMENT) |
726f6388 JA |
7753 | { |
7754 | /* Found an assignment statement, add this word to end of | |
bb70624e JA |
7755 | subst_assign_varlist (vp). */ |
7756 | if (!subst_assign_varlist) | |
7757 | subst_assign_varlist = vp = lp; | |
726f6388 JA |
7758 | else |
7759 | { | |
7760 | vp->next = lp; | |
7761 | vp = lp; | |
7762 | } | |
7763 | ||
7764 | /* Remove the word pointed to by LP from TLIST. */ | |
7765 | tp->next = lp->next; | |
7766 | /* ASSERT(vp == lp); */ | |
7767 | lp->next = (WORD_LIST *)NULL; | |
7768 | lp = tp->next; | |
7769 | } | |
7770 | else | |
7771 | { | |
7772 | tp = lp; | |
7773 | lp = lp->next; | |
7774 | } | |
7775 | } | |
7776 | } | |
7777 | return (tlist); | |
7778 | } | |
7779 | ||
cce855bc JA |
7780 | #define WEXP_VARASSIGN 0x001 |
7781 | #define WEXP_BRACEEXP 0x002 | |
7782 | #define WEXP_TILDEEXP 0x004 | |
7783 | #define WEXP_PARAMEXP 0x008 | |
7784 | #define WEXP_PATHEXP 0x010 | |
7785 | ||
7786 | /* All of the expansions, including variable assignments at the start of | |
7787 | the list. */ | |
7788 | #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP) | |
7789 | ||
7790 | /* All of the expansions except variable assignments at the start of | |
7791 | the list. */ | |
7792 | #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP) | |
7793 | ||
7794 | /* All of the `shell expansions': brace expansion, tilde expansion, parameter | |
7795 | expansion, command substitution, arithmetic expansion, word splitting, and | |
7796 | quote removal. */ | |
7797 | #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP) | |
7798 | ||
726f6388 JA |
7799 | /* Take the list of words in LIST and do the various substitutions. Return |
7800 | a new list of words which is the expanded list, and without things like | |
7801 | variable assignments. */ | |
7802 | ||
7803 | WORD_LIST * | |
7804 | expand_words (list) | |
7805 | WORD_LIST *list; | |
7806 | { | |
cce855bc | 7807 | return (expand_word_list_internal (list, WEXP_ALL)); |
726f6388 JA |
7808 | } |
7809 | ||
7810 | /* Same as expand_words (), but doesn't hack variable or environment | |
7811 | variables. */ | |
7812 | WORD_LIST * | |
7813 | expand_words_no_vars (list) | |
7814 | WORD_LIST *list; | |
7815 | { | |
cce855bc | 7816 | return (expand_word_list_internal (list, WEXP_NOVARS)); |
726f6388 JA |
7817 | } |
7818 | ||
cce855bc JA |
7819 | WORD_LIST * |
7820 | expand_words_shellexp (list) | |
726f6388 | 7821 | WORD_LIST *list; |
726f6388 | 7822 | { |
cce855bc JA |
7823 | return (expand_word_list_internal (list, WEXP_SHELLEXP)); |
7824 | } | |
726f6388 | 7825 | |
cce855bc JA |
7826 | static WORD_LIST * |
7827 | glob_expand_word_list (tlist, eflags) | |
7828 | WORD_LIST *tlist; | |
7829 | int eflags; | |
7830 | { | |
7831 | char **glob_array, *temp_string; | |
7832 | register int glob_index; | |
7833 | WORD_LIST *glob_list, *output_list, *disposables, *next; | |
7834 | WORD_DESC *tword; | |
726f6388 | 7835 | |
cce855bc JA |
7836 | output_list = disposables = (WORD_LIST *)NULL; |
7837 | glob_array = (char **)NULL; | |
7838 | while (tlist) | |
7839 | { | |
7840 | /* For each word, either globbing is attempted or the word is | |
7841 | added to orig_list. If globbing succeeds, the results are | |
7842 | added to orig_list and the word (tlist) is added to the list | |
7843 | of disposable words. If globbing fails and failed glob | |
7844 | expansions are left unchanged (the shell default), the | |
7845 | original word is added to orig_list. If globbing fails and | |
7846 | failed glob expansions are removed, the original word is | |
7847 | added to the list of disposable words. orig_list ends up | |
7117c2d2 | 7848 | in reverse order and requires a call to REVERSE_LIST to |
cce855bc JA |
7849 | be set right. After all words are examined, the disposable |
7850 | words are freed. */ | |
7851 | next = tlist->next; | |
726f6388 | 7852 | |
cce855bc | 7853 | /* If the word isn't an assignment and contains an unquoted |
28ef6c31 | 7854 | pattern matching character, then glob it. */ |
b72432fd | 7855 | if ((tlist->word->flags & W_NOGLOB) == 0 && |
cce855bc | 7856 | unquoted_glob_pattern_p (tlist->word->word)) |
726f6388 | 7857 | { |
cce855bc JA |
7858 | glob_array = shell_glob_filename (tlist->word->word); |
7859 | ||
7860 | /* Handle error cases. | |
7861 | I don't think we should report errors like "No such file | |
7862 | or directory". However, I would like to report errors | |
7863 | like "Read failed". */ | |
7864 | ||
b80f6443 | 7865 | if (glob_array == 0 || GLOB_FAILED (glob_array)) |
726f6388 | 7866 | { |
bb70624e | 7867 | glob_array = (char **)xmalloc (sizeof (char *)); |
cce855bc JA |
7868 | glob_array[0] = (char *)NULL; |
7869 | } | |
7870 | ||
7871 | /* Dequote the current word in case we have to use it. */ | |
7872 | if (glob_array[0] == NULL) | |
7873 | { | |
7874 | temp_string = dequote_string (tlist->word->word); | |
7875 | free (tlist->word->word); | |
7876 | tlist->word->word = temp_string; | |
7877 | } | |
7878 | ||
7879 | /* Make the array into a word list. */ | |
7880 | glob_list = (WORD_LIST *)NULL; | |
7881 | for (glob_index = 0; glob_array[glob_index]; glob_index++) | |
7882 | { | |
7883 | tword = make_bare_word (glob_array[glob_index]); | |
7884 | tword->flags |= W_GLOBEXP; /* XXX */ | |
7885 | glob_list = make_word_list (tword, glob_list); | |
7886 | } | |
7887 | ||
7888 | if (glob_list) | |
7889 | { | |
7890 | output_list = (WORD_LIST *)list_append (glob_list, output_list); | |
7891 | PREPEND_LIST (tlist, disposables); | |
7892 | } | |
b80f6443 JA |
7893 | else if (fail_glob_expansion != 0) |
7894 | { | |
7895 | report_error (_("no match: %s"), tlist->word->word); | |
f1be666c | 7896 | exp_jump_to_top_level (DISCARD); |
b80f6443 | 7897 | } |
cce855bc JA |
7898 | else if (allow_null_glob_expansion == 0) |
7899 | { | |
7900 | /* Failed glob expressions are left unchanged. */ | |
7901 | PREPEND_LIST (tlist, output_list); | |
7902 | } | |
7903 | else | |
7904 | { | |
7905 | /* Failed glob expressions are removed. */ | |
7906 | PREPEND_LIST (tlist, disposables); | |
726f6388 | 7907 | } |
726f6388 | 7908 | } |
cce855bc JA |
7909 | else |
7910 | { | |
7911 | /* Dequote the string. */ | |
7912 | temp_string = dequote_string (tlist->word->word); | |
7913 | free (tlist->word->word); | |
7914 | tlist->word->word = temp_string; | |
7915 | PREPEND_LIST (tlist, output_list); | |
7916 | } | |
7917 | ||
7117c2d2 | 7918 | strvec_dispose (glob_array); |
cce855bc JA |
7919 | glob_array = (char **)NULL; |
7920 | ||
7921 | tlist = next; | |
726f6388 JA |
7922 | } |
7923 | ||
cce855bc JA |
7924 | if (disposables) |
7925 | dispose_words (disposables); | |
7926 | ||
7927 | if (output_list) | |
7928 | output_list = REVERSE_LIST (output_list, WORD_LIST *); | |
7929 | ||
7930 | return (output_list); | |
7931 | } | |
726f6388 JA |
7932 | |
7933 | #if defined (BRACE_EXPANSION) | |
cce855bc JA |
7934 | static WORD_LIST * |
7935 | brace_expand_word_list (tlist, eflags) | |
7936 | WORD_LIST *tlist; | |
7937 | int eflags; | |
7938 | { | |
7939 | register char **expansions; | |
7940 | char *temp_string; | |
7941 | WORD_LIST *disposables, *output_list, *next; | |
7942 | WORD_DESC *w; | |
7943 | int eindex; | |
7944 | ||
7945 | for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next) | |
726f6388 | 7946 | { |
cce855bc | 7947 | next = tlist->next; |
726f6388 | 7948 | |
cce855bc JA |
7949 | /* Only do brace expansion if the word has a brace character. If |
7950 | not, just add the word list element to BRACES and continue. In | |
7951 | the common case, at least when running shell scripts, this will | |
7117c2d2 | 7952 | degenerate to a bunch of calls to `xstrchr', and then what is |
cce855bc | 7953 | basically a reversal of TLIST into BRACES, which is corrected |
7117c2d2 | 7954 | by a call to REVERSE_LIST () on BRACES when the end of TLIST |
cce855bc | 7955 | is reached. */ |
7117c2d2 | 7956 | if (xstrchr (tlist->word->word, LBRACE)) |
726f6388 | 7957 | { |
cce855bc | 7958 | expansions = brace_expand (tlist->word->word); |
726f6388 | 7959 | |
cce855bc | 7960 | for (eindex = 0; temp_string = expansions[eindex]; eindex++) |
726f6388 | 7961 | { |
cce855bc JA |
7962 | w = make_word (temp_string); |
7963 | /* If brace expansion didn't change the word, preserve | |
7964 | the flags. We may want to preserve the flags | |
7965 | unconditionally someday -- XXX */ | |
7966 | if (STREQ (temp_string, tlist->word->word)) | |
7967 | w->flags = tlist->word->flags; | |
7968 | output_list = make_word_list (w, output_list); | |
7969 | free (expansions[eindex]); | |
726f6388 | 7970 | } |
cce855bc | 7971 | free (expansions); |
726f6388 | 7972 | |
cce855bc JA |
7973 | /* Add TLIST to the list of words to be freed after brace |
7974 | expansion has been performed. */ | |
7975 | PREPEND_LIST (tlist, disposables); | |
7976 | } | |
7977 | else | |
7978 | PREPEND_LIST (tlist, output_list); | |
726f6388 | 7979 | } |
cce855bc JA |
7980 | |
7981 | if (disposables) | |
7982 | dispose_words (disposables); | |
7983 | ||
7984 | if (output_list) | |
7985 | output_list = REVERSE_LIST (output_list, WORD_LIST *); | |
7986 | ||
7987 | return (output_list); | |
7988 | } | |
7989 | #endif | |
7990 | ||
7991 | static WORD_LIST * | |
7992 | shell_expand_word_list (tlist, eflags) | |
7993 | WORD_LIST *tlist; | |
7994 | int eflags; | |
7995 | { | |
7996 | WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list; | |
7997 | int expanded_something, has_dollar_at; | |
7998 | char *temp_string; | |
726f6388 | 7999 | |
726f6388 | 8000 | /* We do tilde expansion all the time. This is what 1003.2 says. */ |
cce855bc JA |
8001 | new_list = (WORD_LIST *)NULL; |
8002 | for (orig_list = tlist; tlist; tlist = next) | |
726f6388 | 8003 | { |
ccc6cda3 | 8004 | temp_string = tlist->word->word; |
726f6388 JA |
8005 | |
8006 | next = tlist->next; | |
8007 | ||
95732b49 JA |
8008 | #if defined (ARRAY_VARS) |
8009 | /* If this is a compound array assignment to a builtin that accepts | |
8010 | such assignments (e.g., `declare'), take the assignment and perform | |
8011 | it separately, handling the semantics of declarations inside shell | |
8012 | functions. This avoids the double-evaluation of such arguments, | |
8013 | because `declare' does some evaluation of compound assignments on | |
8014 | its own. */ | |
8015 | if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG)) | |
8016 | { | |
8017 | int t; | |
8018 | ||
8019 | t = do_word_assignment (tlist->word); | |
8020 | if (t == 0) | |
8021 | { | |
8022 | last_command_exit_value = EXECUTION_FAILURE; | |
8023 | exp_jump_to_top_level (DISCARD); | |
8024 | } | |
8025 | ||
8026 | /* Now transform the word as ksh93 appears to do and go on */ | |
8027 | t = assignment (tlist->word->word, 0); | |
8028 | tlist->word->word[t] = '\0'; | |
8029 | tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG); | |
726f6388 | 8030 | } |
95732b49 | 8031 | #endif |
726f6388 | 8032 | |
ccc6cda3 | 8033 | expanded_something = 0; |
726f6388 | 8034 | expanded = expand_word_internal |
b72432fd | 8035 | (tlist->word, 0, 0, &has_dollar_at, &expanded_something); |
726f6388 JA |
8036 | |
8037 | if (expanded == &expand_word_error || expanded == &expand_word_fatal) | |
8038 | { | |
8039 | /* By convention, each time this error is returned, | |
8040 | tlist->word->word has already been freed. */ | |
8041 | tlist->word->word = (char *)NULL; | |
ccc6cda3 | 8042 | |
726f6388 JA |
8043 | /* Dispose our copy of the original list. */ |
8044 | dispose_words (orig_list); | |
d166f048 | 8045 | /* Dispose the new list we're building. */ |
726f6388 JA |
8046 | dispose_words (new_list); |
8047 | ||
28ef6c31 | 8048 | last_command_exit_value = EXECUTION_FAILURE; |
726f6388 | 8049 | if (expanded == &expand_word_error) |
b80f6443 | 8050 | exp_jump_to_top_level (DISCARD); |
726f6388 | 8051 | else |
b80f6443 | 8052 | exp_jump_to_top_level (FORCE_EOF); |
726f6388 JA |
8053 | } |
8054 | ||
ccc6cda3 JA |
8055 | /* Don't split words marked W_NOSPLIT. */ |
8056 | if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0) | |
726f6388 | 8057 | { |
ccc6cda3 | 8058 | temp_list = word_list_split (expanded); |
726f6388 JA |
8059 | dispose_words (expanded); |
8060 | } | |
8061 | else | |
8062 | { | |
8063 | /* If no parameter expansion, command substitution, process | |
8064 | substitution, or arithmetic substitution took place, then | |
8065 | do not do word splitting. We still have to remove quoted | |
8066 | null characters from the result. */ | |
8067 | word_list_remove_quoted_nulls (expanded); | |
ccc6cda3 | 8068 | temp_list = expanded; |
726f6388 JA |
8069 | } |
8070 | ||
ccc6cda3 JA |
8071 | expanded = REVERSE_LIST (temp_list, WORD_LIST *); |
8072 | new_list = (WORD_LIST *)list_append (expanded, new_list); | |
726f6388 JA |
8073 | } |
8074 | ||
cce855bc JA |
8075 | if (orig_list) |
8076 | dispose_words (orig_list); | |
726f6388 | 8077 | |
726f6388 | 8078 | if (new_list) |
cce855bc | 8079 | new_list = REVERSE_LIST (new_list, WORD_LIST *); |
726f6388 | 8080 | |
cce855bc JA |
8081 | return (new_list); |
8082 | } | |
726f6388 | 8083 | |
cce855bc JA |
8084 | /* The workhorse for expand_words () and expand_words_no_vars (). |
8085 | First arg is LIST, a WORD_LIST of words. | |
b72432fd JA |
8086 | Second arg EFLAGS is a flags word controlling which expansions are |
8087 | performed. | |
726f6388 | 8088 | |
cce855bc JA |
8089 | This does all of the substitutions: brace expansion, tilde expansion, |
8090 | parameter expansion, command substitution, arithmetic expansion, | |
8091 | process substitution, word splitting, and pathname expansion, according | |
8092 | to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits | |
8093 | set, or for which no expansion is done, do not undergo word splitting. | |
b72432fd | 8094 | Words with the W_NOGLOB bit set do not undergo pathname expansion. */ |
cce855bc JA |
8095 | static WORD_LIST * |
8096 | expand_word_list_internal (list, eflags) | |
8097 | WORD_LIST *list; | |
8098 | int eflags; | |
8099 | { | |
8100 | WORD_LIST *new_list, *temp_list; | |
8101 | int tint; | |
726f6388 | 8102 | |
cce855bc JA |
8103 | if (list == 0) |
8104 | return ((WORD_LIST *)NULL); | |
726f6388 | 8105 | |
bb70624e | 8106 | garglist = new_list = copy_word_list (list); |
cce855bc JA |
8107 | if (eflags & WEXP_VARASSIGN) |
8108 | { | |
bb70624e | 8109 | garglist = new_list = separate_out_assignments (new_list); |
cce855bc JA |
8110 | if (new_list == 0) |
8111 | { | |
bb70624e | 8112 | if (subst_assign_varlist) |
cce855bc JA |
8113 | { |
8114 | /* All the words were variable assignments, so they are placed | |
8115 | into the shell's environment. */ | |
bb70624e | 8116 | for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next) |
cce855bc JA |
8117 | { |
8118 | this_command_name = (char *)NULL; /* no arithmetic errors */ | |
95732b49 | 8119 | tint = do_word_assignment (temp_list->word); |
cce855bc JA |
8120 | /* Variable assignment errors in non-interactive shells |
8121 | running in Posix.2 mode cause the shell to exit. */ | |
28ef6c31 | 8122 | if (tint == 0) |
ccc6cda3 | 8123 | { |
cce855bc | 8124 | last_command_exit_value = EXECUTION_FAILURE; |
28ef6c31 | 8125 | if (interactive_shell == 0 && posixly_correct) |
b80f6443 | 8126 | exp_jump_to_top_level (FORCE_EOF); |
28ef6c31 | 8127 | else |
b80f6443 | 8128 | exp_jump_to_top_level (DISCARD); |
ccc6cda3 | 8129 | } |
726f6388 | 8130 | } |
bb70624e JA |
8131 | dispose_words (subst_assign_varlist); |
8132 | subst_assign_varlist = (WORD_LIST *)NULL; | |
cce855bc JA |
8133 | } |
8134 | return ((WORD_LIST *)NULL); | |
8135 | } | |
8136 | } | |
726f6388 | 8137 | |
cce855bc JA |
8138 | /* Begin expanding the words that remain. The expansions take place on |
8139 | things that aren't really variable assignments. */ | |
726f6388 | 8140 | |
cce855bc JA |
8141 | #if defined (BRACE_EXPANSION) |
8142 | /* Do brace expansion on this word if there are any brace characters | |
8143 | in the string. */ | |
8144 | if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list) | |
8145 | new_list = brace_expand_word_list (new_list, eflags); | |
8146 | #endif /* BRACE_EXPANSION */ | |
726f6388 | 8147 | |
cce855bc JA |
8148 | /* Perform the `normal' shell expansions: tilde expansion, parameter and |
8149 | variable substitution, command substitution, arithmetic expansion, | |
8150 | and word splitting. */ | |
8151 | new_list = shell_expand_word_list (new_list, eflags); | |
726f6388 | 8152 | |
cce855bc JA |
8153 | /* Okay, we're almost done. Now let's just do some filename |
8154 | globbing. */ | |
8155 | if (new_list) | |
8156 | { | |
8157 | if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0) | |
8158 | /* Glob expand the word list unless globbing has been disabled. */ | |
8159 | new_list = glob_expand_word_list (new_list, eflags); | |
726f6388 | 8160 | else |
cce855bc JA |
8161 | /* Dequote the words, because we're not performing globbing. */ |
8162 | new_list = dequote_list (new_list); | |
726f6388 JA |
8163 | } |
8164 | ||
bb70624e | 8165 | if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist) |
726f6388 | 8166 | { |
95732b49 | 8167 | sh_wassign_func_t *assign_func; |
726f6388 JA |
8168 | |
8169 | /* If the remainder of the words expand to nothing, Posix.2 requires | |
8170 | that the variable and environment assignments affect the shell's | |
8171 | environment. */ | |
95732b49 | 8172 | assign_func = new_list ? assign_in_env : do_word_assignment; |
b80f6443 | 8173 | tempenv_assign_error = 0; |
726f6388 | 8174 | |
bb70624e | 8175 | for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next) |
726f6388 | 8176 | { |
ccc6cda3 | 8177 | this_command_name = (char *)NULL; |
95732b49 | 8178 | tint = (*assign_func) (temp_list->word); |
ccc6cda3 JA |
8179 | /* Variable assignment errors in non-interactive shells running |
8180 | in Posix.2 mode cause the shell to exit. */ | |
b80f6443 | 8181 | if (tint == 0) |
ccc6cda3 | 8182 | { |
95732b49 | 8183 | if (assign_func == do_word_assignment) |
b80f6443 JA |
8184 | { |
8185 | last_command_exit_value = EXECUTION_FAILURE; | |
8186 | if (interactive_shell == 0 && posixly_correct) | |
8187 | exp_jump_to_top_level (FORCE_EOF); | |
8188 | else | |
8189 | exp_jump_to_top_level (DISCARD); | |
8190 | } | |
28ef6c31 | 8191 | else |
b80f6443 | 8192 | tempenv_assign_error++; |
ccc6cda3 | 8193 | } |
726f6388 | 8194 | } |
726f6388 | 8195 | |
bb70624e JA |
8196 | dispose_words (subst_assign_varlist); |
8197 | subst_assign_varlist = (WORD_LIST *)NULL; | |
726f6388 JA |
8198 | } |
8199 | ||
cce855bc | 8200 | #if 0 |
ccc6cda3 JA |
8201 | tint = list_length (new_list) + 1; |
8202 | RESIZE_MALLOCED_BUFFER (glob_argv_flags, 0, tint, glob_argv_flags_size, 16); | |
cce855bc JA |
8203 | for (tint = 0, temp_list = new_list; temp_list; temp_list = temp_list->next) |
8204 | glob_argv_flags[tint++] = (temp_list->word->flags & W_GLOBEXP) ? '1' : '0'; | |
ccc6cda3 | 8205 | glob_argv_flags[tint] = '\0'; |
ccc6cda3 | 8206 | #endif |
726f6388 | 8207 | |
cce855bc | 8208 | return (new_list); |
ccc6cda3 | 8209 | } |