]>
Commit | Line | Data |
---|---|---|
cc87ba64 CR |
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 | ||
1f6ec1a8 | 7 | /* Copyright (C) 1987-2012 Free Software Foundation, Inc. |
726f6388 JA |
8 | |
9 | This file is part of GNU Bash, the Bourne Again SHell. | |
10 | ||
2e4498b3 CR |
11 | Bash is free software: you can redistribute it and/or modify |
12 | it under the terms of the GNU General Public License as published by | |
13 | the Free Software Foundation, either version 3 of the License, or | |
14 | (at your option) any later version. | |
15 | ||
16 | Bash is distributed in the hope that it will be useful, | |
17 | but WITHOUT ANY WARRANTY; without even the implied warranty of | |
18 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | |
19 | GNU General Public License for more details. | |
20 | ||
21 | You should have received a copy of the GNU General Public License | |
22 | along with Bash. If not, see <http://www.gnu.org/licenses/>. | |
23 | */ | |
726f6388 | 24 | |
ccc6cda3 JA |
25 | #include "config.h" |
26 | ||
726f6388 JA |
27 | #include "bashtypes.h" |
28 | #include <stdio.h> | |
f73dda09 | 29 | #include "chartypes.h" |
e77a3058 CR |
30 | #if defined (HAVE_PWD_H) |
31 | # include <pwd.h> | |
32 | #endif | |
726f6388 JA |
33 | #include <signal.h> |
34 | #include <errno.h> | |
ccc6cda3 JA |
35 | |
36 | #if defined (HAVE_UNISTD_H) | |
37 | # include <unistd.h> | |
38 | #endif | |
726f6388 JA |
39 | |
40 | #include "bashansi.h" | |
41 | #include "posixstat.h" | |
5e13499c | 42 | #include "bashintl.h" |
726f6388 JA |
43 | |
44 | #include "shell.h" | |
9e51a74d | 45 | #include "parser.h" |
726f6388 JA |
46 | #include "flags.h" |
47 | #include "jobs.h" | |
48 | #include "execute_cmd.h" | |
49 | #include "filecntl.h" | |
ccc6cda3 JA |
50 | #include "trap.h" |
51 | #include "pathexp.h" | |
52 | #include "mailcheck.h" | |
53 | ||
7117c2d2 | 54 | #include "shmbutil.h" |
06dff54a | 55 | #include "typemax.h" |
7117c2d2 | 56 | |
ccc6cda3 JA |
57 | #include "builtins/getopt.h" |
58 | #include "builtins/common.h" | |
726f6388 | 59 | |
48ff5447 CR |
60 | #include "builtins/builtext.h" |
61 | ||
cce855bc | 62 | #include <tilde/tilde.h> |
f73dda09 | 63 | #include <glob/strmatch.h> |
ccc6cda3 JA |
64 | |
65 | #if !defined (errno) | |
66 | extern int errno; | |
67 | #endif /* !errno */ | |
726f6388 JA |
68 | |
69 | /* The size that strings change by. */ | |
d166f048 | 70 | #define DEFAULT_INITIAL_ARRAY_SIZE 112 |
ccc6cda3 JA |
71 | #define DEFAULT_ARRAY_SIZE 128 |
72 | ||
73 | /* Variable types. */ | |
74 | #define VT_VARIABLE 0 | |
75 | #define VT_POSPARMS 1 | |
76 | #define VT_ARRAYVAR 2 | |
d166f048 | 77 | #define VT_ARRAYMEMBER 3 |
e33f2203 | 78 | #define VT_ASSOCVAR 4 |
726f6388 | 79 | |
d3a24ed2 CR |
80 | #define VT_STARSUB 128 /* $* or ${array[*]} -- used to split */ |
81 | ||
ccc6cda3 JA |
82 | /* Flags for quoted_strchr */ |
83 | #define ST_BACKSL 0x01 | |
84 | #define ST_CTLESC 0x02 | |
7117c2d2 JA |
85 | #define ST_SQUOTE 0x04 /* unused yet */ |
86 | #define ST_DQUOTE 0x08 /* unused yet */ | |
87 | ||
d3a24ed2 CR |
88 | /* Flags for the `pflags' argument to param_expand() */ |
89 | #define PF_NOCOMSUB 0x01 /* Do not perform command substitution */ | |
1231ac47 | 90 | #define PF_IGNUNBOUND 0x02 /* ignore unbound vars even if -u set */ |
e1e48bba | 91 | #define PF_NOSPLIT2 0x04 /* same as W_NOSPLIT2 */ |
40647963 | 92 | #define PF_ASSIGNRHS 0x08 /* same as W_ASSIGNRHS */ |
d3a24ed2 | 93 | |
cce855bc JA |
94 | /* These defs make it easier to use the editor. */ |
95 | #define LBRACE '{' | |
96 | #define RBRACE '}' | |
97 | #define LPAREN '(' | |
98 | #define RPAREN ')' | |
726f6388 | 99 | |
176b12ee CR |
100 | #if defined (HANDLE_MULTIBYTE) |
101 | #define WLPAREN L'(' | |
102 | #define WRPAREN L')' | |
103 | #endif | |
104 | ||
28ef6c31 JA |
105 | /* Evaluates to 1 if C is one of the shell's special parameters whose length |
106 | can be taken, but is also one of the special expansion characters. */ | |
107 | #define VALID_SPECIAL_LENGTH_PARAM(c) \ | |
108 | ((c) == '-' || (c) == '?' || (c) == '#') | |
109 | ||
110 | /* Evaluates to 1 if C is one of the shell's special parameters for which an | |
111 | indirect variable reference may be made. */ | |
112 | #define VALID_INDIR_PARAM(c) \ | |
e05be32d | 113 | ((posixly_correct == 0 && (c) == '#') || (posixly_correct == 0 && (c) == '?') || (c) == '@' || (c) == '*') |
28ef6c31 JA |
114 | |
115 | /* Evaluates to 1 if C is one of the OP characters that follows the parameter | |
116 | in ${parameter[:]OPword}. */ | |
7117c2d2 | 117 | #define VALID_PARAM_EXPAND_CHAR(c) (sh_syntaxtab[(unsigned char)c] & CSUBSTOP) |
28ef6c31 | 118 | |
bb70624e JA |
119 | /* Evaluates to 1 if this is one of the shell's special variables. */ |
120 | #define SPECIAL_VAR(name, wi) \ | |
f73dda09 JA |
121 | ((DIGIT (*name) && all_digits (name)) || \ |
122 | (name[1] == '\0' && (sh_syntaxtab[(unsigned char)*name] & CSPECVAR)) || \ | |
28ef6c31 | 123 | (wi && name[2] == '\0' && VALID_INDIR_PARAM (name[1]))) |
bb70624e | 124 | |
f73dda09 JA |
125 | /* An expansion function that takes a string and a quoted flag and returns |
126 | a WORD_LIST *. Used as the type of the third argument to | |
127 | expand_string_if_necessary(). */ | |
128 | typedef WORD_LIST *EXPFUNC __P((char *, int)); | |
129 | ||
726f6388 JA |
130 | /* Process ID of the last command executed within command substitution. */ |
131 | pid_t last_command_subst_pid = NO_PID; | |
b72432fd | 132 | pid_t current_command_subst_pid = NO_PID; |
726f6388 | 133 | |
7117c2d2 JA |
134 | /* Variables used to keep track of the characters in IFS. */ |
135 | SHELL_VAR *ifs_var; | |
136 | char *ifs_value; | |
137 | unsigned char ifs_cmap[UCHAR_MAX + 1]; | |
633e5c6d CR |
138 | |
139 | #if defined (HANDLE_MULTIBYTE) | |
140 | unsigned char ifs_firstc[MB_LEN_MAX]; | |
141 | size_t ifs_firstc_len; | |
142 | #else | |
7117c2d2 | 143 | unsigned char ifs_firstc; |
633e5c6d | 144 | #endif |
7117c2d2 | 145 | |
0527c903 CR |
146 | /* Sentinel to tell when we are performing variable assignments preceding a |
147 | command name and putting them into the environment. Used to make sure | |
148 | we use the temporary environment when looking up variable values. */ | |
d7f49990 CR |
149 | int assigning_in_environment; |
150 | ||
0527c903 CR |
151 | /* Used to hold a list of variable assignments preceding a command. Global |
152 | so the SIGCHLD handler in jobs.c can unwind-protect it when it runs a | |
153 | SIGCHLD trap and so it can be saved and restored by the trap handlers. */ | |
154 | WORD_LIST *subst_assign_varlist = (WORD_LIST *)NULL; | |
155 | ||
726f6388 | 156 | /* Extern functions and variables from different files. */ |
d3a24ed2 | 157 | extern int last_command_exit_value, last_command_exit_signal; |
f486d0a1 CR |
158 | extern int subshell_environment, line_number; |
159 | extern int subshell_level, parse_and_execute_level, sourcelevel; | |
7117c2d2 | 160 | extern int eof_encountered; |
bb70624e | 161 | extern int return_catch_flag, return_catch_value; |
f73dda09 | 162 | extern pid_t dollar_dollar_pid; |
726f6388 | 163 | extern int posixly_correct; |
726f6388 | 164 | extern char *this_command_name; |
ccc6cda3 | 165 | extern struct fd_bitmap *current_fds_to_close; |
cce855bc | 166 | extern int wordexp_only; |
d3a24ed2 | 167 | extern int expanding_redir; |
56299fa5 | 168 | extern int tempenv_assign_error; |
726f6388 | 169 | |
d3ad40de CR |
170 | #if !defined (HAVE_WCSDUP) && defined (HANDLE_MULTIBYTE) |
171 | extern wchar_t *wcsdup __P((const wchar_t *)); | |
172 | #endif | |
173 | ||
ccc6cda3 JA |
174 | /* Non-zero means to allow unmatched globbed filenames to expand to |
175 | a null file. */ | |
176 | int allow_null_glob_expansion; | |
177 | ||
d3a24ed2 CR |
178 | /* Non-zero means to throw an error when globbing fails to match anything. */ |
179 | int fail_glob_expansion; | |
180 | ||
f73dda09 | 181 | #if 0 |
ccc6cda3 JA |
182 | /* Variables to keep track of which words in an expanded word list (the |
183 | output of expand_word_list_internal) are the result of globbing | |
f73dda09 JA |
184 | expansions. GLOB_ARGV_FLAGS is used by execute_cmd.c. |
185 | (CURRENTLY UNUSED). */ | |
ccc6cda3 JA |
186 | char *glob_argv_flags; |
187 | static int glob_argv_flags_size; | |
f73dda09 | 188 | #endif |
726f6388 JA |
189 | |
190 | static WORD_LIST expand_word_error, expand_word_fatal; | |
227f982e | 191 | static WORD_DESC expand_wdesc_error, expand_wdesc_fatal; |
726f6388 | 192 | static char expand_param_error, expand_param_fatal; |
7027abcb | 193 | static char extract_string_error, extract_string_fatal; |
726f6388 | 194 | |
28ef6c31 JA |
195 | /* Tell the expansion functions to not longjmp back to top_level on fatal |
196 | errors. Enabled when doing completion and prompt string expansion. */ | |
197 | static int no_longjmp_on_fatal_error = 0; | |
198 | ||
199 | /* Set by expand_word_unsplit; used to inhibit splitting and re-joining | |
200 | $* on $IFS, primarily when doing assignment statements. */ | |
201 | static int expand_no_split_dollar_star = 0; | |
bb70624e | 202 | |
bb70624e JA |
203 | /* A WORD_LIST of words to be expanded by expand_word_list_internal, |
204 | without any leading variable assignments. */ | |
205 | static WORD_LIST *garglist = (WORD_LIST *)NULL; | |
b72432fd | 206 | |
f73dda09 | 207 | static char *quoted_substring __P((char *, int, int)); |
7117c2d2 JA |
208 | static int quoted_strlen __P((char *)); |
209 | static char *quoted_strchr __P((char *, int, int)); | |
f73dda09 JA |
210 | |
211 | static char *expand_string_if_necessary __P((char *, int, EXPFUNC *)); | |
212 | static inline char *expand_string_to_string_internal __P((char *, int, EXPFUNC *)); | |
213 | static WORD_LIST *call_expand_word_internal __P((WORD_DESC *, int, int, int *, int *)); | |
214 | static WORD_LIST *expand_string_internal __P((char *, int)); | |
215 | static WORD_LIST *expand_string_leave_quoted __P((char *, int)); | |
216 | static WORD_LIST *expand_string_for_rhs __P((char *, int, int *, int *)); | |
217 | ||
f73dda09 JA |
218 | static WORD_LIST *list_quote_escapes __P((WORD_LIST *)); |
219 | static char *make_quoted_char __P((int)); | |
220 | static WORD_LIST *quote_list __P((WORD_LIST *)); | |
f73dda09 JA |
221 | |
222 | static int unquoted_substring __P((char *, char *)); | |
223 | static int unquoted_member __P((int, char *)); | |
224 | ||
43df7bbb CR |
225 | #if defined (ARRAY_VARS) |
226 | static SHELL_VAR *do_compound_assignment __P((char *, char *, int)); | |
227 | #endif | |
228 | static int do_assignment_internal __P((const WORD_DESC *, int)); | |
f73dda09 | 229 | |
d3ad40de | 230 | static char *string_extract_verbatim __P((char *, size_t, int *, char *, int)); |
f73dda09 JA |
231 | static char *string_extract __P((char *, int *, char *, int)); |
232 | static char *string_extract_double_quoted __P((char *, int *, int)); | |
7117c2d2 | 233 | static inline char *string_extract_single_quoted __P((char *, int *)); |
d3ad40de | 234 | static inline int skip_single_quoted __P((const char *, size_t, int)); |
7117c2d2 JA |
235 | static int skip_double_quoted __P((char *, size_t, int)); |
236 | static char *extract_delimited_string __P((char *, int *, char *, char *, char *, int)); | |
237 | static char *extract_dollar_brace_string __P((char *, int *, int, int)); | |
6932f7f5 | 238 | static int skip_matched_pair __P((const char *, int, int, int, int)); |
f73dda09 | 239 | |
f73dda09 JA |
240 | static char *pos_params __P((char *, int, int, int)); |
241 | ||
545f34cf CR |
242 | static unsigned char *mb_getcharlens __P((char *, int)); |
243 | ||
704a1a2a | 244 | static char *remove_upattern __P((char *, char *, int)); |
d3ad40de | 245 | #if defined (HANDLE_MULTIBYTE) |
704a1a2a CR |
246 | static wchar_t *remove_wpattern __P((wchar_t *, size_t, wchar_t *, int)); |
247 | #endif | |
f73dda09 | 248 | static char *remove_pattern __P((char *, char *, int)); |
704a1a2a | 249 | |
704a1a2a CR |
250 | static int match_upattern __P((char *, char *, int, char **, char **)); |
251 | #if defined (HANDLE_MULTIBYTE) | |
704a1a2a CR |
252 | static int match_wpattern __P((wchar_t *, char **, size_t, wchar_t *, int, char **, char **)); |
253 | #endif | |
f73dda09 JA |
254 | static int match_pattern __P((char *, char *, int, char **, char **)); |
255 | static int getpatspec __P((int, char *)); | |
256 | static char *getpattern __P((char *, int, int)); | |
7117c2d2 | 257 | static char *variable_remove_pattern __P((char *, char *, int, int)); |
f73dda09 | 258 | static char *list_remove_pattern __P((WORD_LIST *, char *, int, int, int)); |
7117c2d2 | 259 | static char *parameter_list_remove_pattern __P((int, char *, int, int)); |
f73dda09 | 260 | #ifdef ARRAY_VARS |
fdf670ea | 261 | static char *array_remove_pattern __P((SHELL_VAR *, char *, int, char *, int)); |
f73dda09 | 262 | #endif |
5f8cde23 | 263 | static char *parameter_brace_remove_pattern __P((char *, char *, int, char *, int, int, int)); |
f73dda09 JA |
264 | |
265 | static char *process_substitute __P((char *, int)); | |
266 | ||
d3ad40de | 267 | static char *read_comsub __P((int, int, int *)); |
f73dda09 JA |
268 | |
269 | #ifdef ARRAY_VARS | |
270 | static arrayind_t array_length_reference __P((char *)); | |
271 | #endif | |
272 | ||
273 | static int valid_brace_expansion_word __P((char *, int)); | |
d3a24ed2 | 274 | static int chk_atstar __P((char *, int, int *, int *)); |
d3ad40de | 275 | static int chk_arithsub __P((const char *, int)); |
d3a24ed2 | 276 | |
5f8cde23 | 277 | static WORD_DESC *parameter_brace_expand_word __P((char *, int, int, int, arrayind_t *)); |
227f982e CR |
278 | static WORD_DESC *parameter_brace_expand_indir __P((char *, int, int, int *, int *)); |
279 | static WORD_DESC *parameter_brace_expand_rhs __P((char *, char *, int, int, int *, int *)); | |
f73dda09 JA |
280 | static void parameter_brace_expand_error __P((char *, char *)); |
281 | ||
282 | static int valid_length_expression __P((char *)); | |
7117c2d2 | 283 | static intmax_t parameter_brace_expand_length __P((char *)); |
f73dda09 JA |
284 | |
285 | static char *skiparith __P((char *, int)); | |
09767ff0 | 286 | static int verify_substring_values __P((SHELL_VAR *, char *, char *, int, intmax_t *, intmax_t *)); |
5f8cde23 | 287 | static int get_var_and_type __P((char *, char *, arrayind_t, int, int, SHELL_VAR **, char **)); |
545f34cf | 288 | static char *mb_substring __P((char *, int, int)); |
5f8cde23 | 289 | static char *parameter_brace_substring __P((char *, char *, int, char *, int, int)); |
f73dda09 | 290 | |
ee56a865 CR |
291 | static int shouldexp_replacement __P((char *)); |
292 | ||
f73dda09 JA |
293 | static char *pos_params_pat_subst __P((char *, char *, char *, int)); |
294 | ||
5f8cde23 | 295 | static char *parameter_brace_patsub __P((char *, char *, int, char *, int, int)); |
f73dda09 | 296 | |
09767ff0 | 297 | static char *pos_params_casemod __P((char *, char *, int, int)); |
5f8cde23 | 298 | static char *parameter_brace_casemod __P((char *, char *, int, int, char *, int, int)); |
09767ff0 | 299 | |
e1e48bba | 300 | static WORD_DESC *parameter_brace_expand __P((char *, int *, int, int, int *, int *)); |
227f982e | 301 | static WORD_DESC *param_expand __P((char *, int *, int, int *, int *, int *, int *, int)); |
f73dda09 JA |
302 | |
303 | static WORD_LIST *expand_word_internal __P((WORD_DESC *, int, int, int *, int *)); | |
304 | ||
f73dda09 JA |
305 | static WORD_LIST *word_list_split __P((WORD_LIST *)); |
306 | ||
d3a24ed2 CR |
307 | static void exp_jump_to_top_level __P((int)); |
308 | ||
f73dda09 JA |
309 | static WORD_LIST *separate_out_assignments __P((WORD_LIST *)); |
310 | static WORD_LIST *glob_expand_word_list __P((WORD_LIST *, int)); | |
311 | #ifdef BRACE_EXPANSION | |
312 | static WORD_LIST *brace_expand_word_list __P((WORD_LIST *, int)); | |
313 | #endif | |
fdf670ea CR |
314 | #if defined (ARRAY_VARS) |
315 | static int make_internal_declare __P((char *, char *)); | |
316 | #endif | |
f73dda09 JA |
317 | static WORD_LIST *shell_expand_word_list __P((WORD_LIST *, int)); |
318 | static WORD_LIST *expand_word_list_internal __P((WORD_LIST *, int)); | |
726f6388 JA |
319 | |
320 | /* **************************************************************** */ | |
321 | /* */ | |
322 | /* Utility Functions */ | |
323 | /* */ | |
324 | /* **************************************************************** */ | |
325 | ||
9dd88db7 CR |
326 | #if defined (DEBUG) |
327 | void | |
328 | dump_word_flags (flags) | |
329 | int flags; | |
330 | { | |
331 | int f; | |
332 | ||
333 | f = flags; | |
334 | fprintf (stderr, "%d -> ", f); | |
335 | if (f & W_ASSIGNASSOC) | |
336 | { | |
337 | f &= ~W_ASSIGNASSOC; | |
338 | fprintf (stderr, "W_ASSIGNASSOC%s", f ? "|" : ""); | |
339 | } | |
340 | if (f & W_HASCTLESC) | |
341 | { | |
342 | f &= ~W_HASCTLESC; | |
343 | fprintf (stderr, "W_HASCTLESC%s", f ? "|" : ""); | |
344 | } | |
345 | if (f & W_NOPROCSUB) | |
346 | { | |
347 | f &= ~W_NOPROCSUB; | |
348 | fprintf (stderr, "W_NOPROCSUB%s", f ? "|" : ""); | |
349 | } | |
350 | if (f & W_DQUOTE) | |
351 | { | |
352 | f &= ~W_DQUOTE; | |
353 | fprintf (stderr, "W_DQUOTE%s", f ? "|" : ""); | |
354 | } | |
355 | if (f & W_HASQUOTEDNULL) | |
356 | { | |
357 | f &= ~W_HASQUOTEDNULL; | |
358 | fprintf (stderr, "W_HASQUOTEDNULL%s", f ? "|" : ""); | |
359 | } | |
360 | if (f & W_ASSIGNARG) | |
361 | { | |
362 | f &= ~W_ASSIGNARG; | |
363 | fprintf (stderr, "W_ASSIGNARG%s", f ? "|" : ""); | |
364 | } | |
365 | if (f & W_ASSNBLTIN) | |
366 | { | |
367 | f &= ~W_ASSNBLTIN; | |
368 | fprintf (stderr, "W_ASSNBLTIN%s", f ? "|" : ""); | |
369 | } | |
ec860d76 CR |
370 | if (f & W_ASSNGLOBAL) |
371 | { | |
372 | f &= ~W_ASSNGLOBAL; | |
373 | fprintf (stderr, "W_ASSNGLOBAL%s", f ? "|" : ""); | |
374 | } | |
9dd88db7 CR |
375 | if (f & W_COMPASSIGN) |
376 | { | |
377 | f &= ~W_COMPASSIGN; | |
378 | fprintf (stderr, "W_COMPASSIGN%s", f ? "|" : ""); | |
379 | } | |
380 | if (f & W_NOEXPAND) | |
381 | { | |
382 | f &= ~W_NOEXPAND; | |
383 | fprintf (stderr, "W_NOEXPAND%s", f ? "|" : ""); | |
384 | } | |
385 | if (f & W_ITILDE) | |
386 | { | |
387 | f &= ~W_ITILDE; | |
388 | fprintf (stderr, "W_ITILDE%s", f ? "|" : ""); | |
389 | } | |
390 | if (f & W_NOTILDE) | |
391 | { | |
392 | f &= ~W_NOTILDE; | |
393 | fprintf (stderr, "W_NOTILDE%s", f ? "|" : ""); | |
394 | } | |
395 | if (f & W_ASSIGNRHS) | |
396 | { | |
397 | f &= ~W_ASSIGNRHS; | |
398 | fprintf (stderr, "W_ASSIGNRHS%s", f ? "|" : ""); | |
399 | } | |
400 | if (f & W_NOCOMSUB) | |
401 | { | |
402 | f &= ~W_NOCOMSUB; | |
403 | fprintf (stderr, "W_NOCOMSUB%s", f ? "|" : ""); | |
404 | } | |
405 | if (f & W_DOLLARSTAR) | |
406 | { | |
407 | f &= ~W_DOLLARSTAR; | |
408 | fprintf (stderr, "W_DOLLARSTAR%s", f ? "|" : ""); | |
409 | } | |
410 | if (f & W_DOLLARAT) | |
411 | { | |
412 | f &= ~W_DOLLARAT; | |
413 | fprintf (stderr, "W_DOLLARAT%s", f ? "|" : ""); | |
414 | } | |
415 | if (f & W_TILDEEXP) | |
416 | { | |
417 | f &= ~W_TILDEEXP; | |
418 | fprintf (stderr, "W_TILDEEXP%s", f ? "|" : ""); | |
419 | } | |
420 | if (f & W_NOSPLIT2) | |
421 | { | |
422 | f &= ~W_NOSPLIT2; | |
423 | fprintf (stderr, "W_NOSPLIT2%s", f ? "|" : ""); | |
424 | } | |
425 | if (f & W_NOGLOB) | |
426 | { | |
427 | f &= ~W_NOGLOB; | |
428 | fprintf (stderr, "W_NOGLOB%s", f ? "|" : ""); | |
429 | } | |
430 | if (f & W_NOSPLIT) | |
431 | { | |
432 | f &= ~W_NOSPLIT; | |
433 | fprintf (stderr, "W_NOSPLIT%s", f ? "|" : ""); | |
434 | } | |
435 | if (f & W_GLOBEXP) | |
436 | { | |
437 | f &= ~W_GLOBEXP; | |
438 | fprintf (stderr, "W_GLOBEXP%s", f ? "|" : ""); | |
439 | } | |
440 | if (f & W_ASSIGNMENT) | |
441 | { | |
442 | f &= ~W_ASSIGNMENT; | |
443 | fprintf (stderr, "W_ASSIGNMENT%s", f ? "|" : ""); | |
444 | } | |
445 | if (f & W_QUOTED) | |
446 | { | |
447 | f &= ~W_QUOTED; | |
448 | fprintf (stderr, "W_QUOTED%s", f ? "|" : ""); | |
449 | } | |
450 | if (f & W_HASDOLLAR) | |
451 | { | |
452 | f &= ~W_HASDOLLAR; | |
453 | fprintf (stderr, "W_HASDOLLAR%s", f ? "|" : ""); | |
454 | } | |
455 | fprintf (stderr, "\n"); | |
456 | fflush (stderr); | |
457 | } | |
458 | #endif | |
459 | ||
7117c2d2 | 460 | #ifdef INCLUDE_UNUSED |
ccc6cda3 JA |
461 | static char * |
462 | quoted_substring (string, start, end) | |
463 | char *string; | |
464 | int start, end; | |
465 | { | |
466 | register int len, l; | |
467 | register char *result, *s, *r; | |
468 | ||
469 | len = end - start; | |
470 | ||
471 | /* Move to string[start], skipping quoted characters. */ | |
472 | for (s = string, l = 0; *s && l < start; ) | |
473 | { | |
474 | if (*s == CTLESC) | |
475 | { | |
28ef6c31 JA |
476 | s++; |
477 | continue; | |
ccc6cda3 JA |
478 | } |
479 | l++; | |
480 | if (*s == 0) | |
28ef6c31 | 481 | break; |
ccc6cda3 JA |
482 | } |
483 | ||
f73dda09 | 484 | r = result = (char *)xmalloc (2*len + 1); /* save room for quotes */ |
ccc6cda3 JA |
485 | |
486 | /* Copy LEN characters, including quote characters. */ | |
487 | s = string + l; | |
488 | for (l = 0; l < len; s++) | |
489 | { | |
490 | if (*s == CTLESC) | |
28ef6c31 | 491 | *r++ = *s++; |
ccc6cda3 JA |
492 | *r++ = *s; |
493 | l++; | |
494 | if (*s == 0) | |
28ef6c31 | 495 | break; |
ccc6cda3 JA |
496 | } |
497 | *r = '\0'; | |
498 | return result; | |
499 | } | |
7117c2d2 JA |
500 | #endif |
501 | ||
502 | #ifdef INCLUDE_UNUSED | |
503 | /* Return the length of S, skipping over quoted characters */ | |
504 | static int | |
505 | quoted_strlen (s) | |
506 | char *s; | |
507 | { | |
508 | register char *p; | |
509 | int i; | |
510 | ||
511 | i = 0; | |
512 | for (p = s; *p; p++) | |
513 | { | |
514 | if (*p == CTLESC) | |
515 | { | |
516 | p++; | |
517 | if (*p == 0) | |
518 | return (i + 1); | |
519 | } | |
520 | i++; | |
521 | } | |
522 | ||
523 | return i; | |
524 | } | |
525 | #endif | |
ccc6cda3 JA |
526 | |
527 | /* Find the first occurrence of character C in string S, obeying shell | |
528 | quoting rules. If (FLAGS & ST_BACKSL) is non-zero, backslash-escaped | |
529 | characters are skipped. If (FLAGS & ST_CTLESC) is non-zero, characters | |
530 | escaped with CTLESC are skipped. */ | |
7117c2d2 | 531 | static char * |
ccc6cda3 JA |
532 | quoted_strchr (s, c, flags) |
533 | char *s; | |
534 | int c, flags; | |
535 | { | |
536 | register char *p; | |
537 | ||
538 | for (p = s; *p; p++) | |
539 | { | |
540 | if (((flags & ST_BACKSL) && *p == '\\') | |
541 | || ((flags & ST_CTLESC) && *p == CTLESC)) | |
542 | { | |
543 | p++; | |
544 | if (*p == '\0') | |
545 | return ((char *)NULL); | |
546 | continue; | |
547 | } | |
548 | else if (*p == c) | |
549 | return p; | |
550 | } | |
551 | return ((char *)NULL); | |
552 | } | |
553 | ||
cce855bc | 554 | /* Return 1 if CHARACTER appears in an unquoted portion of |
7117c2d2 | 555 | STRING. Return 0 otherwise. CHARACTER must be a single-byte character. */ |
cce855bc JA |
556 | static int |
557 | unquoted_member (character, string) | |
558 | int character; | |
726f6388 JA |
559 | char *string; |
560 | { | |
7117c2d2 | 561 | size_t slen; |
cce855bc | 562 | int sindex, c; |
7117c2d2 | 563 | DECLARE_MBSTATE; |
726f6388 | 564 | |
7117c2d2 JA |
565 | slen = strlen (string); |
566 | sindex = 0; | |
567 | while (c = string[sindex]) | |
726f6388 | 568 | { |
cce855bc JA |
569 | if (c == character) |
570 | return (1); | |
571 | ||
572 | switch (c) | |
ccc6cda3 | 573 | { |
cce855bc | 574 | default: |
7117c2d2 | 575 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
576 | break; |
577 | ||
578 | case '\\': | |
579 | sindex++; | |
580 | if (string[sindex]) | |
7117c2d2 | 581 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
582 | break; |
583 | ||
584 | case '\'': | |
7117c2d2 | 585 | sindex = skip_single_quoted (string, slen, ++sindex); |
cce855bc JA |
586 | break; |
587 | ||
588 | case '"': | |
7117c2d2 | 589 | sindex = skip_double_quoted (string, slen, ++sindex); |
cce855bc | 590 | break; |
ccc6cda3 | 591 | } |
726f6388 | 592 | } |
cce855bc | 593 | return (0); |
726f6388 JA |
594 | } |
595 | ||
cce855bc JA |
596 | /* Return 1 if SUBSTR appears in an unquoted portion of STRING. */ |
597 | static int | |
598 | unquoted_substring (substr, string) | |
599 | char *substr, *string; | |
726f6388 | 600 | { |
7117c2d2 | 601 | size_t slen; |
cce855bc | 602 | int sindex, c, sublen; |
7117c2d2 | 603 | DECLARE_MBSTATE; |
726f6388 | 604 | |
cce855bc JA |
605 | if (substr == 0 || *substr == '\0') |
606 | return (0); | |
607 | ||
7117c2d2 | 608 | slen = strlen (string); |
cce855bc JA |
609 | sublen = strlen (substr); |
610 | for (sindex = 0; c = string[sindex]; ) | |
726f6388 | 611 | { |
cce855bc JA |
612 | if (STREQN (string + sindex, substr, sublen)) |
613 | return (1); | |
614 | ||
615 | switch (c) | |
616 | { | |
617 | case '\\': | |
618 | sindex++; | |
cce855bc | 619 | if (string[sindex]) |
7117c2d2 | 620 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
621 | break; |
622 | ||
623 | case '\'': | |
7117c2d2 | 624 | sindex = skip_single_quoted (string, slen, ++sindex); |
cce855bc JA |
625 | break; |
626 | ||
627 | case '"': | |
7117c2d2 | 628 | sindex = skip_double_quoted (string, slen, ++sindex); |
cce855bc JA |
629 | break; |
630 | ||
631 | default: | |
7117c2d2 | 632 | ADVANCE_CHAR (string, slen, sindex); |
cce855bc JA |
633 | break; |
634 | } | |
726f6388 | 635 | } |
cce855bc | 636 | return (0); |
ccc6cda3 | 637 | } |
726f6388 | 638 | |
cce855bc JA |
639 | /* Most of the substitutions must be done in parallel. In order |
640 | to avoid using tons of unclear goto's, I have some functions | |
641 | for manipulating malloc'ed strings. They all take INDX, a | |
642 | pointer to an integer which is the offset into the string | |
643 | where manipulation is taking place. They also take SIZE, a | |
644 | pointer to an integer which is the current length of the | |
645 | character array for this string. */ | |
726f6388 | 646 | |
cce855bc JA |
647 | /* Append SOURCE to TARGET at INDEX. SIZE is the current amount |
648 | of space allocated to TARGET. SOURCE can be NULL, in which | |
649 | case nothing happens. Gets rid of SOURCE by freeing it. | |
650 | Returns TARGET in case the location has changed. */ | |
7117c2d2 | 651 | INLINE char * |
cce855bc JA |
652 | sub_append_string (source, target, indx, size) |
653 | char *source, *target; | |
654 | int *indx, *size; | |
655 | { | |
656 | if (source) | |
726f6388 | 657 | { |
cce855bc JA |
658 | int srclen, n; |
659 | ||
660 | srclen = STRLEN (source); | |
661 | if (srclen >= (int)(*size - *indx)) | |
726f6388 | 662 | { |
cce855bc JA |
663 | n = srclen + *indx; |
664 | n = (n + DEFAULT_ARRAY_SIZE) - (n % DEFAULT_ARRAY_SIZE); | |
f73dda09 | 665 | target = (char *)xrealloc (target, (*size = n)); |
726f6388 | 666 | } |
cce855bc JA |
667 | |
668 | FASTCOPY (source, target + *indx, srclen); | |
669 | *indx += srclen; | |
670 | target[*indx] = '\0'; | |
671 | ||
672 | free (source); | |
726f6388 | 673 | } |
cce855bc JA |
674 | return (target); |
675 | } | |
676 | ||
677 | #if 0 | |
678 | /* UNUSED */ | |
679 | /* Append the textual representation of NUMBER to TARGET. | |
680 | INDX and SIZE are as in SUB_APPEND_STRING. */ | |
681 | char * | |
682 | sub_append_number (number, target, indx, size) | |
7117c2d2 | 683 | intmax_t number; |
f73dda09 | 684 | int *indx, *size; |
cce855bc JA |
685 | char *target; |
686 | { | |
687 | char *temp; | |
688 | ||
689 | temp = itos (number); | |
690 | return (sub_append_string (temp, target, indx, size)); | |
726f6388 | 691 | } |
d166f048 | 692 | #endif |
726f6388 JA |
693 | |
694 | /* Extract a substring from STRING, starting at SINDEX and ending with | |
695 | one of the characters in CHARLIST. Don't make the ending character | |
696 | part of the string. Leave SINDEX pointing at the ending character. | |
e6598ba4 | 697 | Understand about backslashes in the string. If (flags & SX_VARNAME) |
7117c2d2 JA |
698 | is non-zero, and array variables have been compiled into the shell, |
699 | everything between a `[' and a corresponding `]' is skipped over. | |
e6598ba4 CR |
700 | If (flags & SX_NOALLOC) is non-zero, don't return the substring, just |
701 | update SINDEX. If (flags & SX_REQMATCH) is non-zero, the string must | |
7027abcb | 702 | contain a closing character from CHARLIST. */ |
726f6388 | 703 | static char * |
7117c2d2 | 704 | string_extract (string, sindex, charlist, flags) |
f73dda09 JA |
705 | char *string; |
706 | int *sindex; | |
707 | char *charlist; | |
7117c2d2 | 708 | int flags; |
726f6388 | 709 | { |
ccc6cda3 | 710 | register int c, i; |
7027abcb | 711 | int found; |
7117c2d2 | 712 | size_t slen; |
726f6388 | 713 | char *temp; |
7117c2d2 | 714 | DECLARE_MBSTATE; |
726f6388 | 715 | |
da713c23 | 716 | slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0; |
7117c2d2 | 717 | i = *sindex; |
7027abcb | 718 | found = 0; |
7117c2d2 | 719 | while (c = string[i]) |
726f6388 JA |
720 | { |
721 | if (c == '\\') | |
7117c2d2 JA |
722 | { |
723 | if (string[i + 1]) | |
724 | i++; | |
725 | else | |
726 | break; | |
727 | } | |
ccc6cda3 | 728 | #if defined (ARRAY_VARS) |
e6598ba4 | 729 | else if ((flags & SX_VARNAME) && c == '[') |
ccc6cda3 JA |
730 | { |
731 | int ni; | |
732 | /* If this is an array subscript, skip over it and continue. */ | |
3eb2d94a | 733 | ni = skipsubscript (string, i, 0); |
ccc6cda3 JA |
734 | if (string[ni] == ']') |
735 | i = ni; | |
736 | } | |
737 | #endif | |
738 | else if (MEMBER (c, charlist)) | |
7027abcb CR |
739 | { |
740 | found = 1; | |
726f6388 | 741 | break; |
7027abcb | 742 | } |
7117c2d2 JA |
743 | |
744 | ADVANCE_CHAR (string, slen, i); | |
726f6388 | 745 | } |
bb70624e | 746 | |
7027abcb CR |
747 | /* If we had to have a matching delimiter and didn't find one, return an |
748 | error and let the caller deal with it. */ | |
e6598ba4 | 749 | if ((flags & SX_REQMATCH) && found == 0) |
7027abcb CR |
750 | { |
751 | *sindex = i; | |
752 | return (&extract_string_error); | |
753 | } | |
754 | ||
e6598ba4 | 755 | temp = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i); |
726f6388 | 756 | *sindex = i; |
7027abcb | 757 | |
726f6388 JA |
758 | return (temp); |
759 | } | |
760 | ||
ccc6cda3 JA |
761 | /* Extract the contents of STRING as if it is enclosed in double quotes. |
762 | SINDEX, when passed in, is the offset of the character immediately | |
763 | following the opening double quote; on exit, SINDEX is left pointing after | |
764 | the closing double quote. If STRIPDQ is non-zero, unquoted double | |
765 | quotes are stripped and the string is terminated by a null byte. | |
766 | Backslashes between the embedded double quotes are processed. If STRIPDQ | |
767 | is zero, an unquoted `"' terminates the string. */ | |
7117c2d2 | 768 | static char * |
ccc6cda3 | 769 | string_extract_double_quoted (string, sindex, stripdq) |
726f6388 | 770 | char *string; |
ccc6cda3 | 771 | int *sindex, stripdq; |
726f6388 | 772 | { |
7117c2d2 JA |
773 | size_t slen; |
774 | char *send; | |
f73dda09 JA |
775 | int j, i, t; |
776 | unsigned char c; | |
ccc6cda3 JA |
777 | char *temp, *ret; /* The new string we return. */ |
778 | int pass_next, backquote, si; /* State variables for the machine. */ | |
779 | int dquote; | |
7117c2d2 JA |
780 | DECLARE_MBSTATE; |
781 | ||
782 | slen = strlen (string + *sindex) + *sindex; | |
783 | send = string + slen; | |
726f6388 | 784 | |
ccc6cda3 | 785 | pass_next = backquote = dquote = 0; |
7117c2d2 | 786 | temp = (char *)xmalloc (1 + slen - *sindex); |
726f6388 | 787 | |
7117c2d2 JA |
788 | j = 0; |
789 | i = *sindex; | |
790 | while (c = string[i]) | |
726f6388 | 791 | { |
ccc6cda3 JA |
792 | /* Process a character that was quoted by a backslash. */ |
793 | if (pass_next) | |
726f6388 | 794 | { |
9e51a74d | 795 | /* XXX - take another look at this in light of Interp 221 */ |
ccc6cda3 | 796 | /* Posix.2 sez: |
726f6388 | 797 | |
ccc6cda3 JA |
798 | ``The backslash shall retain its special meaning as an escape |
799 | character only when followed by one of the characters: | |
7117c2d2 | 800 | $ ` " \ <newline>''. |
726f6388 | 801 | |
ccc6cda3 JA |
802 | If STRIPDQ is zero, we handle the double quotes here and let |
803 | expand_word_internal handle the rest. If STRIPDQ is non-zero, | |
804 | we have already been through one round of backslash stripping, | |
805 | and want to strip these backslashes only if DQUOTE is non-zero, | |
806 | indicating that we are inside an embedded double-quoted string. */ | |
807 | ||
808 | /* If we are in an embedded quoted string, then don't strip | |
809 | backslashes before characters for which the backslash | |
810 | retains its special meaning, but remove backslashes in | |
811 | front of other characters. If we are not in an | |
812 | embedded quoted string, don't strip backslashes at all. | |
813 | This mess is necessary because the string was already | |
814 | surrounded by double quotes (and sh has some really weird | |
815 | quoting rules). | |
816 | The returned string will be run through expansion as if | |
817 | it were double-quoted. */ | |
818 | if ((stripdq == 0 && c != '"') || | |
28ef6c31 | 819 | (stripdq && ((dquote && (sh_syntaxtab[c] & CBSDQUOTE)) || dquote == 0))) |
ccc6cda3 | 820 | temp[j++] = '\\'; |
ccc6cda3 | 821 | pass_next = 0; |
7117c2d2 JA |
822 | |
823 | add_one_character: | |
824 | COPY_CHAR_I (temp, j, string, send, i); | |
ccc6cda3 JA |
825 | continue; |
826 | } | |
726f6388 | 827 | |
ccc6cda3 JA |
828 | /* A backslash protects the next character. The code just above |
829 | handles preserving the backslash in front of any character but | |
830 | a double quote. */ | |
831 | if (c == '\\') | |
726f6388 | 832 | { |
ccc6cda3 | 833 | pass_next++; |
7117c2d2 | 834 | i++; |
726f6388 JA |
835 | continue; |
836 | } | |
837 | ||
ccc6cda3 JA |
838 | /* Inside backquotes, ``the portion of the quoted string from the |
839 | initial backquote and the characters up to the next backquote | |
840 | that is not preceded by a backslash, having escape characters | |
841 | removed, defines that command''. */ | |
842 | if (backquote) | |
726f6388 | 843 | { |
ccc6cda3 JA |
844 | if (c == '`') |
845 | backquote = 0; | |
846 | temp[j++] = c; | |
7117c2d2 | 847 | i++; |
726f6388 JA |
848 | continue; |
849 | } | |
850 | ||
ccc6cda3 | 851 | if (c == '`') |
726f6388 | 852 | { |
ccc6cda3 JA |
853 | temp[j++] = c; |
854 | backquote++; | |
7117c2d2 | 855 | i++; |
ccc6cda3 | 856 | continue; |
726f6388 JA |
857 | } |
858 | ||
ccc6cda3 JA |
859 | /* Pass everything between `$(' and the matching `)' or a quoted |
860 | ${ ... } pair through according to the Posix.2 specification. */ | |
cce855bc | 861 | if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE))) |
726f6388 | 862 | { |
61deeb13 CR |
863 | int free_ret = 1; |
864 | ||
ccc6cda3 | 865 | si = i + 2; |
cce855bc | 866 | if (string[i + 1] == LPAREN) |
e33f2203 | 867 | ret = extract_command_subst (string, &si, 0); |
ccc6cda3 | 868 | else |
64419627 | 869 | ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, 0); |
726f6388 | 870 | |
ccc6cda3 JA |
871 | temp[j++] = '$'; |
872 | temp[j++] = string[i + 1]; | |
726f6388 | 873 | |
61deeb13 CR |
874 | /* Just paranoia; ret will not be 0 unless no_longjmp_on_fatal_error |
875 | is set. */ | |
876 | if (ret == 0 && no_longjmp_on_fatal_error) | |
877 | { | |
878 | free_ret = 0; | |
879 | ret = string + i + 2; | |
880 | } | |
881 | ||
ccc6cda3 JA |
882 | for (t = 0; ret[t]; t++, j++) |
883 | temp[j] = ret[t]; | |
56299fa5 CR |
884 | temp[j] = string[si]; |
885 | ||
886 | if (string[si]) | |
887 | { | |
888 | j++; | |
889 | i = si + 1; | |
890 | } | |
891 | else | |
892 | i = si; | |
726f6388 | 893 | |
61deeb13 CR |
894 | if (free_ret) |
895 | free (ret); | |
ccc6cda3 | 896 | continue; |
726f6388 JA |
897 | } |
898 | ||
ccc6cda3 | 899 | /* Add any character but a double quote to the quoted string we're |
28ef6c31 | 900 | accumulating. */ |
ccc6cda3 | 901 | if (c != '"') |
7117c2d2 | 902 | goto add_one_character; |
ccc6cda3 JA |
903 | |
904 | /* c == '"' */ | |
905 | if (stripdq) | |
726f6388 | 906 | { |
ccc6cda3 | 907 | dquote ^= 1; |
7117c2d2 | 908 | i++; |
ccc6cda3 | 909 | continue; |
726f6388 | 910 | } |
ccc6cda3 JA |
911 | |
912 | break; | |
726f6388 | 913 | } |
ccc6cda3 | 914 | temp[j] = '\0'; |
726f6388 | 915 | |
ccc6cda3 JA |
916 | /* Point to after the closing quote. */ |
917 | if (c) | |
918 | i++; | |
726f6388 JA |
919 | *sindex = i; |
920 | ||
ccc6cda3 JA |
921 | return (temp); |
922 | } | |
923 | ||
924 | /* This should really be another option to string_extract_double_quoted. */ | |
f73dda09 | 925 | static int |
7117c2d2 | 926 | skip_double_quoted (string, slen, sind) |
ccc6cda3 | 927 | char *string; |
7117c2d2 | 928 | size_t slen; |
ccc6cda3 JA |
929 | int sind; |
930 | { | |
f73dda09 | 931 | int c, i; |
ccc6cda3 JA |
932 | char *ret; |
933 | int pass_next, backquote, si; | |
7117c2d2 | 934 | DECLARE_MBSTATE; |
ccc6cda3 JA |
935 | |
936 | pass_next = backquote = 0; | |
7117c2d2 JA |
937 | i = sind; |
938 | while (c = string[i]) | |
726f6388 | 939 | { |
ccc6cda3 JA |
940 | if (pass_next) |
941 | { | |
942 | pass_next = 0; | |
7117c2d2 | 943 | ADVANCE_CHAR (string, slen, i); |
ccc6cda3 JA |
944 | continue; |
945 | } | |
946 | else if (c == '\\') | |
947 | { | |
948 | pass_next++; | |
7117c2d2 | 949 | i++; |
ccc6cda3 JA |
950 | continue; |
951 | } | |
952 | else if (backquote) | |
953 | { | |
954 | if (c == '`') | |
955 | backquote = 0; | |
7117c2d2 | 956 | ADVANCE_CHAR (string, slen, i); |
ccc6cda3 JA |
957 | continue; |
958 | } | |
959 | else if (c == '`') | |
960 | { | |
961 | backquote++; | |
7117c2d2 | 962 | i++; |
ccc6cda3 JA |
963 | continue; |
964 | } | |
cce855bc | 965 | else if (c == '$' && ((string[i + 1] == LPAREN) || (string[i + 1] == LBRACE))) |
ccc6cda3 JA |
966 | { |
967 | si = i + 2; | |
cce855bc | 968 | if (string[i + 1] == LPAREN) |
e33f2203 | 969 | ret = extract_command_subst (string, &si, SX_NOALLOC); |
ccc6cda3 | 970 | else |
9e51a74d | 971 | ret = extract_dollar_brace_string (string, &si, Q_DOUBLE_QUOTES, SX_NOALLOC); |
ccc6cda3 | 972 | |
7117c2d2 | 973 | i = si + 1; |
ccc6cda3 JA |
974 | continue; |
975 | } | |
976 | else if (c != '"') | |
7117c2d2 JA |
977 | { |
978 | ADVANCE_CHAR (string, slen, i); | |
979 | continue; | |
980 | } | |
ccc6cda3 JA |
981 | else |
982 | break; | |
726f6388 | 983 | } |
ccc6cda3 JA |
984 | |
985 | if (c) | |
986 | i++; | |
987 | ||
988 | return (i); | |
726f6388 JA |
989 | } |
990 | ||
ccc6cda3 JA |
991 | /* Extract the contents of STRING as if it is enclosed in single quotes. |
992 | SINDEX, when passed in, is the offset of the character immediately | |
993 | following the opening single quote; on exit, SINDEX is left pointing after | |
994 | the closing single quote. */ | |
995 | static inline char * | |
996 | string_extract_single_quoted (string, sindex) | |
997 | char *string; | |
998 | int *sindex; | |
999 | { | |
f73dda09 | 1000 | register int i; |
7117c2d2 | 1001 | size_t slen; |
ccc6cda3 | 1002 | char *t; |
7117c2d2 | 1003 | DECLARE_MBSTATE; |
ccc6cda3 | 1004 | |
da713c23 CR |
1005 | /* Don't need slen for ADVANCE_CHAR unless multibyte chars possible. */ |
1006 | slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 0; | |
7117c2d2 JA |
1007 | i = *sindex; |
1008 | while (string[i] && string[i] != '\'') | |
1009 | ADVANCE_CHAR (string, slen, i); | |
ccc6cda3 | 1010 | |
bb70624e | 1011 | t = substring (string, *sindex, i); |
ccc6cda3 JA |
1012 | |
1013 | if (string[i]) | |
1014 | i++; | |
1015 | *sindex = i; | |
1016 | ||
1017 | return (t); | |
1018 | } | |
1019 | ||
1020 | static inline int | |
7117c2d2 | 1021 | skip_single_quoted (string, slen, sind) |
d3ad40de | 1022 | const char *string; |
7117c2d2 | 1023 | size_t slen; |
ccc6cda3 JA |
1024 | int sind; |
1025 | { | |
28ef6c31 | 1026 | register int c; |
7117c2d2 JA |
1027 | DECLARE_MBSTATE; |
1028 | ||
1029 | c = sind; | |
1030 | while (string[c] && string[c] != '\'') | |
1031 | ADVANCE_CHAR (string, slen, c); | |
ccc6cda3 | 1032 | |
28ef6c31 JA |
1033 | if (string[c]) |
1034 | c++; | |
1035 | return c; | |
ccc6cda3 JA |
1036 | } |
1037 | ||
1038 | /* Just like string_extract, but doesn't hack backslashes or any of | |
bb70624e | 1039 | that other stuff. Obeys CTLESC quoting. Used to do splitting on $IFS. */ |
726f6388 | 1040 | static char * |
d3ad40de | 1041 | string_extract_verbatim (string, slen, sindex, charlist, flags) |
f73dda09 | 1042 | char *string; |
da713c23 | 1043 | size_t slen; |
ccc6cda3 | 1044 | int *sindex; |
f73dda09 | 1045 | char *charlist; |
d3ad40de | 1046 | int flags; |
ccc6cda3 | 1047 | { |
824dfe68 | 1048 | register int i; |
633e5c6d CR |
1049 | #if defined (HANDLE_MULTIBYTE) |
1050 | size_t clen; | |
1051 | wchar_t *wcharlist; | |
1052 | #endif | |
ccc6cda3 JA |
1053 | int c; |
1054 | char *temp; | |
633e5c6d | 1055 | DECLARE_MBSTATE; |
ccc6cda3 JA |
1056 | |
1057 | if (charlist[0] == '\'' && charlist[1] == '\0') | |
1058 | { | |
1059 | temp = string_extract_single_quoted (string, sindex); | |
1060 | --*sindex; /* leave *sindex at separator character */ | |
1061 | return temp; | |
1062 | } | |
1063 | ||
633e5c6d | 1064 | i = *sindex; |
da713c23 CR |
1065 | #if 0 |
1066 | /* See how the MBLEN and ADVANCE_CHAR macros work to understand why we need | |
1067 | this only if MB_CUR_MAX > 1. */ | |
1068 | slen = (MB_CUR_MAX > 1) ? strlen (string + *sindex) + *sindex : 1; | |
1069 | #endif | |
633e5c6d CR |
1070 | #if defined (HANDLE_MULTIBYTE) |
1071 | clen = strlen (charlist); | |
1072 | wcharlist = 0; | |
1073 | #endif | |
1074 | while (c = string[i]) | |
ccc6cda3 | 1075 | { |
633e5c6d CR |
1076 | #if defined (HANDLE_MULTIBYTE) |
1077 | size_t mblength; | |
1078 | #endif | |
e6598ba4 CR |
1079 | if ((flags & SX_NOCTLESC) == 0 && c == CTLESC) |
1080 | { | |
1081 | i += 2; | |
1082 | continue; | |
1083 | } | |
1084 | /* Even if flags contains SX_NOCTLESC, we let CTLESC quoting CTLNUL | |
1085 | through, to protect the CTLNULs from later calls to | |
1086 | remove_quoted_nulls. */ | |
1087 | else if ((flags & SX_NOESCCTLNUL) == 0 && c == CTLESC && string[i+1] == CTLNUL) | |
ccc6cda3 | 1088 | { |
633e5c6d | 1089 | i += 2; |
ccc6cda3 JA |
1090 | continue; |
1091 | } | |
e33f2203 | 1092 | |
633e5c6d | 1093 | #if defined (HANDLE_MULTIBYTE) |
da713c23 | 1094 | mblength = MBLEN (string + i, slen - i); |
633e5c6d CR |
1095 | if (mblength > 1) |
1096 | { | |
1097 | wchar_t wc; | |
1098 | mblength = mbtowc (&wc, string + i, slen - i); | |
1099 | if (MB_INVALIDCH (mblength)) | |
1100 | { | |
1101 | if (MEMBER (c, charlist)) | |
1102 | break; | |
1103 | } | |
1104 | else | |
1105 | { | |
1106 | if (wcharlist == 0) | |
1107 | { | |
1108 | size_t len; | |
1109 | len = mbstowcs (wcharlist, charlist, 0); | |
1110 | if (len == -1) | |
1111 | len = 0; | |
d3ad40de CR |
1112 | wcharlist = (wchar_t *)xmalloc (sizeof (wchar_t) * (len + 1)); |
1113 | mbstowcs (wcharlist, charlist, len + 1); | |
633e5c6d CR |
1114 | } |
1115 | ||
1116 | if (wcschr (wcharlist, wc)) | |
1117 | break; | |
1118 | } | |
1119 | } | |
1120 | else | |
1121 | #endif | |
ccc6cda3 JA |
1122 | if (MEMBER (c, charlist)) |
1123 | break; | |
633e5c6d CR |
1124 | |
1125 | ADVANCE_CHAR (string, slen, i); | |
ccc6cda3 JA |
1126 | } |
1127 | ||
633e5c6d CR |
1128 | #if defined (HANDLE_MULTIBYTE) |
1129 | FREE (wcharlist); | |
1130 | #endif | |
1131 | ||
bb70624e | 1132 | temp = substring (string, *sindex, i); |
ccc6cda3 JA |
1133 | *sindex = i; |
1134 | ||
1135 | return (temp); | |
1136 | } | |
1137 | ||
1138 | /* Extract the $( construct in STRING, and return a new string. | |
1139 | Start extracting at (SINDEX) as if we had just seen "$(". | |
e33f2203 | 1140 | Make (SINDEX) get the position of the matching ")". ) |
ecf57862 | 1141 | XFLAGS is additional flags to pass to other extraction functions. */ |
ccc6cda3 | 1142 | char * |
e33f2203 | 1143 | extract_command_subst (string, sindex, xflags) |
726f6388 JA |
1144 | char *string; |
1145 | int *sindex; | |
e33f2203 | 1146 | int xflags; |
726f6388 | 1147 | { |
176b12ee | 1148 | if (string[*sindex] == LPAREN) |
e33f2203 CR |
1149 | return (extract_delimited_string (string, sindex, "$(", "(", ")", xflags|SX_COMMAND)); /*)*/ |
1150 | else | |
1151 | { | |
1152 | xflags |= (no_longjmp_on_fatal_error ? SX_NOLONGJMP : 0); | |
1153 | return (xparse_dolparen (string, string+*sindex, sindex, xflags)); | |
1154 | } | |
ccc6cda3 JA |
1155 | } |
1156 | ||
28ef6c31 | 1157 | /* Extract the $[ construct in STRING, and return a new string. (]) |
ccc6cda3 JA |
1158 | Start extracting at (SINDEX) as if we had just seen "$[". |
1159 | Make (SINDEX) get the position of the matching "]". */ | |
1160 | char * | |
1161 | extract_arithmetic_subst (string, sindex) | |
1162 | char *string; | |
1163 | int *sindex; | |
1164 | { | |
7117c2d2 | 1165 | return (extract_delimited_string (string, sindex, "$[", "[", "]", 0)); /*]*/ |
ccc6cda3 JA |
1166 | } |
1167 | ||
1168 | #if defined (PROCESS_SUBSTITUTION) | |
1169 | /* Extract the <( or >( construct in STRING, and return a new string. | |
1170 | Start extracting at (SINDEX) as if we had just seen "<(". | |
cce855bc | 1171 | Make (SINDEX) get the position of the matching ")". */ /*))*/ |
ccc6cda3 JA |
1172 | char * |
1173 | extract_process_subst (string, starter, sindex) | |
1174 | char *string; | |
1175 | char *starter; | |
1176 | int *sindex; | |
1177 | { | |
f4f5e1c2 | 1178 | return (extract_delimited_string (string, sindex, starter, "(", ")", SX_COMMAND)); |
ccc6cda3 JA |
1179 | } |
1180 | #endif /* PROCESS_SUBSTITUTION */ | |
1181 | ||
1182 | #if defined (ARRAY_VARS) | |
43df7bbb CR |
1183 | /* This can be fooled by unquoted right parens in the passed string. If |
1184 | each caller verifies that the last character in STRING is a right paren, | |
1185 | we don't even need to call extract_delimited_string. */ | |
ccc6cda3 JA |
1186 | char * |
1187 | extract_array_assignment_list (string, sindex) | |
1188 | char *string; | |
1189 | int *sindex; | |
1190 | { | |
43df7bbb CR |
1191 | int slen; |
1192 | char *ret; | |
1193 | ||
1194 | slen = strlen (string); /* ( */ | |
1195 | if (string[slen - 1] == ')') | |
1196 | { | |
1197 | ret = substring (string, *sindex, slen - 1); | |
1198 | *sindex = slen - 1; | |
1199 | return ret; | |
1200 | } | |
e225d5a9 | 1201 | return 0; |
ccc6cda3 JA |
1202 | } |
1203 | #endif | |
1204 | ||
1205 | /* Extract and create a new string from the contents of STRING, a | |
1206 | character string delimited with OPENER and CLOSER. SINDEX is | |
1207 | the address of an int describing the current offset in STRING; | |
1208 | it should point to just after the first OPENER found. On exit, | |
1209 | SINDEX gets the position of the last character of the matching CLOSER. | |
1210 | If OPENER is more than a single character, ALT_OPENER, if non-null, | |
1211 | contains a character string that can also match CLOSER and thus | |
1212 | needs to be skipped. */ | |
1213 | static char * | |
7117c2d2 | 1214 | extract_delimited_string (string, sindex, opener, alt_opener, closer, flags) |
ccc6cda3 JA |
1215 | char *string; |
1216 | int *sindex; | |
1217 | char *opener, *alt_opener, *closer; | |
7117c2d2 | 1218 | int flags; |
ccc6cda3 JA |
1219 | { |
1220 | int i, c, si; | |
7117c2d2 | 1221 | size_t slen; |
ccc6cda3 | 1222 | char *t, *result; |
d3ad40de | 1223 | int pass_character, nesting_level, in_comment; |
ccc6cda3 | 1224 | int len_closer, len_opener, len_alt_opener; |
7117c2d2 | 1225 | DECLARE_MBSTATE; |
ccc6cda3 | 1226 | |
7117c2d2 | 1227 | slen = strlen (string + *sindex) + *sindex; |
ccc6cda3 JA |
1228 | len_opener = STRLEN (opener); |
1229 | len_alt_opener = STRLEN (alt_opener); | |
1230 | len_closer = STRLEN (closer); | |
726f6388 | 1231 | |
d3ad40de | 1232 | pass_character = in_comment = 0; |
726f6388 JA |
1233 | |
1234 | nesting_level = 1; | |
ccc6cda3 | 1235 | i = *sindex; |
726f6388 | 1236 | |
ccc6cda3 | 1237 | while (nesting_level) |
726f6388 | 1238 | { |
ccc6cda3 JA |
1239 | c = string[i]; |
1240 | ||
1241 | if (c == 0) | |
28ef6c31 | 1242 | break; |
ccc6cda3 | 1243 | |
d3ad40de CR |
1244 | if (in_comment) |
1245 | { | |
1246 | if (c == '\n') | |
1247 | in_comment = 0; | |
1248 | ADVANCE_CHAR (string, slen, i); | |
1249 | continue; | |
1250 | } | |
1251 | ||
ccc6cda3 | 1252 | if (pass_character) /* previous char was backslash */ |
726f6388 JA |
1253 | { |
1254 | pass_character = 0; | |
7117c2d2 | 1255 | ADVANCE_CHAR (string, slen, i); |
726f6388 JA |
1256 | continue; |
1257 | } | |
1258 | ||
d3ad40de | 1259 | /* Not exactly right yet; should handle shell metacharacters and |
602bb739 | 1260 | multibyte characters, too. See COMMENT_BEGIN define in parse.y */ |
e33f2203 | 1261 | if ((flags & SX_COMMAND) && c == '#' && (i == 0 || string[i - 1] == '\n' || shellblank (string[i - 1]))) |
d3ad40de CR |
1262 | { |
1263 | in_comment = 1; | |
1264 | ADVANCE_CHAR (string, slen, i); | |
1265 | continue; | |
1266 | } | |
1267 | ||
7117c2d2 | 1268 | if (c == CTLESC || c == '\\') |
726f6388 | 1269 | { |
ccc6cda3 JA |
1270 | pass_character++; |
1271 | i++; | |
1272 | continue; | |
726f6388 JA |
1273 | } |
1274 | ||
6faad625 CR |
1275 | /* Process a nested command substitution, but only if we're parsing an |
1276 | arithmetic substitution. */ | |
ecf57862 CR |
1277 | if ((flags & SX_COMMAND) && string[i] == '$' && string[i+1] == LPAREN) |
1278 | { | |
1279 | si = i + 2; | |
3d8cce26 | 1280 | t = extract_command_subst (string, &si, flags|SX_NOALLOC); |
ecf57862 CR |
1281 | i = si + 1; |
1282 | continue; | |
1283 | } | |
ecf57862 | 1284 | |
ccc6cda3 JA |
1285 | /* Process a nested OPENER. */ |
1286 | if (STREQN (string + i, opener, len_opener)) | |
726f6388 | 1287 | { |
ccc6cda3 | 1288 | si = i + len_opener; |
e6598ba4 | 1289 | t = extract_delimited_string (string, &si, opener, alt_opener, closer, flags|SX_NOALLOC); |
ccc6cda3 | 1290 | i = si + 1; |
ccc6cda3 | 1291 | continue; |
726f6388 JA |
1292 | } |
1293 | ||
ccc6cda3 JA |
1294 | /* Process a nested ALT_OPENER */ |
1295 | if (len_alt_opener && STREQN (string + i, alt_opener, len_alt_opener)) | |
726f6388 | 1296 | { |
ccc6cda3 | 1297 | si = i + len_alt_opener; |
e6598ba4 | 1298 | t = extract_delimited_string (string, &si, alt_opener, alt_opener, closer, flags|SX_NOALLOC); |
ccc6cda3 | 1299 | i = si + 1; |
726f6388 JA |
1300 | continue; |
1301 | } | |
ccc6cda3 JA |
1302 | |
1303 | /* If the current substring terminates the delimited string, decrement | |
1304 | the nesting level. */ | |
1305 | if (STREQN (string + i, closer, len_closer)) | |
726f6388 | 1306 | { |
7117c2d2 | 1307 | i += len_closer - 1; /* move to last byte of the closer */ |
ccc6cda3 JA |
1308 | nesting_level--; |
1309 | if (nesting_level == 0) | |
1310 | break; | |
726f6388 | 1311 | } |
ccc6cda3 JA |
1312 | |
1313 | /* Pass old-style command substitution through verbatim. */ | |
1314 | if (c == '`') | |
28ef6c31 JA |
1315 | { |
1316 | si = i + 1; | |
e6598ba4 | 1317 | t = string_extract (string, &si, "`", flags|SX_NOALLOC); |
28ef6c31 | 1318 | i = si + 1; |
28ef6c31 JA |
1319 | continue; |
1320 | } | |
ccc6cda3 | 1321 | |
7117c2d2 JA |
1322 | /* Pass single-quoted and double-quoted strings through verbatim. */ |
1323 | if (c == '\'' || c == '"') | |
28ef6c31 JA |
1324 | { |
1325 | si = i + 1; | |
7117c2d2 JA |
1326 | i = (c == '\'') ? skip_single_quoted (string, slen, si) |
1327 | : skip_double_quoted (string, slen, si); | |
28ef6c31 JA |
1328 | continue; |
1329 | } | |
ccc6cda3 | 1330 | |
7117c2d2 JA |
1331 | /* move past this character, which was not special. */ |
1332 | ADVANCE_CHAR (string, slen, i); | |
726f6388 JA |
1333 | } |
1334 | ||
d3a24ed2 | 1335 | if (c == 0 && nesting_level) |
726f6388 | 1336 | { |
d3a24ed2 CR |
1337 | if (no_longjmp_on_fatal_error == 0) |
1338 | { | |
d3a24ed2 | 1339 | last_command_exit_value = EXECUTION_FAILURE; |
7f947b68 | 1340 | report_error (_("bad substitution: no closing `%s' in %s"), closer, string); |
d3a24ed2 CR |
1341 | exp_jump_to_top_level (DISCARD); |
1342 | } | |
1343 | else | |
1344 | { | |
1345 | *sindex = i; | |
1346 | return (char *)NULL; | |
1347 | } | |
726f6388 | 1348 | } |
ccc6cda3 | 1349 | |
cce855bc | 1350 | si = i - *sindex - len_closer + 1; |
e6598ba4 | 1351 | if (flags & SX_NOALLOC) |
7117c2d2 JA |
1352 | result = (char *)NULL; |
1353 | else | |
1354 | { | |
1355 | result = (char *)xmalloc (1 + si); | |
1356 | strncpy (result, string + *sindex, si); | |
1357 | result[si] = '\0'; | |
1358 | } | |
cce855bc JA |
1359 | *sindex = i; |
1360 | ||
726f6388 JA |
1361 | return (result); |
1362 | } | |
1363 | ||
ccc6cda3 JA |
1364 | /* Extract a parameter expansion expression within ${ and } from STRING. |
1365 | Obey the Posix.2 rules for finding the ending `}': count braces while | |
1366 | skipping over enclosed quoted strings and command substitutions. | |
1367 | SINDEX is the address of an int describing the current offset in STRING; | |
1368 | it should point to just after the first `{' found. On exit, SINDEX | |
1369 | gets the position of the matching `}'. QUOTED is non-zero if this | |
1370 | occurs inside double quotes. */ | |
1371 | /* XXX -- this is very similar to extract_delimited_string -- XXX */ | |
726f6388 | 1372 | static char * |
7117c2d2 | 1373 | extract_dollar_brace_string (string, sindex, quoted, flags) |
726f6388 | 1374 | char *string; |
7117c2d2 | 1375 | int *sindex, quoted, flags; |
726f6388 | 1376 | { |
f73dda09 | 1377 | register int i, c; |
7117c2d2 | 1378 | size_t slen; |
9e51a74d | 1379 | int pass_character, nesting_level, si, dolbrace_state; |
ccc6cda3 | 1380 | char *result, *t; |
7117c2d2 | 1381 | DECLARE_MBSTATE; |
726f6388 | 1382 | |
ccc6cda3 | 1383 | pass_character = 0; |
ccc6cda3 | 1384 | nesting_level = 1; |
7117c2d2 | 1385 | slen = strlen (string + *sindex) + *sindex; |
ccc6cda3 | 1386 | |
9e51a74d | 1387 | /* The handling of dolbrace_state needs to agree with the code in parse.y: |
e192f341 CR |
1388 | parse_matched_pair(). The different initial value is to handle the |
1389 | case where this function is called to parse the word in | |
1390 | ${param op word} (SX_WORD). */ | |
1391 | dolbrace_state = (flags & SX_WORD) ? DOLBRACE_WORD : DOLBRACE_PARAM; | |
1392 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && (flags & SX_POSIXEXP)) | |
1393 | dolbrace_state = DOLBRACE_QUOTE; | |
9e51a74d | 1394 | |
7117c2d2 JA |
1395 | i = *sindex; |
1396 | while (c = string[i]) | |
726f6388 | 1397 | { |
ccc6cda3 | 1398 | if (pass_character) |
726f6388 | 1399 | { |
ccc6cda3 | 1400 | pass_character = 0; |
7117c2d2 | 1401 | ADVANCE_CHAR (string, slen, i); |
ccc6cda3 JA |
1402 | continue; |
1403 | } | |
726f6388 | 1404 | |
cce855bc JA |
1405 | /* CTLESCs and backslashes quote the next character. */ |
1406 | if (c == CTLESC || c == '\\') | |
726f6388 | 1407 | { |
ccc6cda3 | 1408 | pass_character++; |
7117c2d2 | 1409 | i++; |
726f6388 JA |
1410 | continue; |
1411 | } | |
1412 | ||
cce855bc | 1413 | if (string[i] == '$' && string[i+1] == LBRACE) |
726f6388 | 1414 | { |
ccc6cda3 | 1415 | nesting_level++; |
7117c2d2 | 1416 | i += 2; |
726f6388 JA |
1417 | continue; |
1418 | } | |
1419 | ||
cce855bc | 1420 | if (c == RBRACE) |
726f6388 | 1421 | { |
ccc6cda3 JA |
1422 | nesting_level--; |
1423 | if (nesting_level == 0) | |
1424 | break; | |
7117c2d2 | 1425 | i++; |
726f6388 JA |
1426 | continue; |
1427 | } | |
1428 | ||
ccc6cda3 JA |
1429 | /* Pass the contents of old-style command substitutions through |
1430 | verbatim. */ | |
1431 | if (c == '`') | |
726f6388 | 1432 | { |
ccc6cda3 | 1433 | si = i + 1; |
e6598ba4 | 1434 | t = string_extract (string, &si, "`", flags|SX_NOALLOC); |
7117c2d2 | 1435 | i = si + 1; |
ccc6cda3 JA |
1436 | continue; |
1437 | } | |
726f6388 | 1438 | |
cce855bc JA |
1439 | /* Pass the contents of new-style command substitutions and |
1440 | arithmetic substitutions through verbatim. */ | |
1441 | if (string[i] == '$' && string[i+1] == LPAREN) | |
ccc6cda3 | 1442 | { |
726f6388 | 1443 | si = i + 2; |
e33f2203 | 1444 | t = extract_command_subst (string, &si, flags|SX_NOALLOC); |
7117c2d2 | 1445 | i = si + 1; |
726f6388 JA |
1446 | continue; |
1447 | } | |
1448 | ||
6faad625 | 1449 | #if 0 |
cce855bc JA |
1450 | /* Pass the contents of single-quoted and double-quoted strings |
1451 | through verbatim. */ | |
1452 | if (c == '\'' || c == '"') | |
ccc6cda3 JA |
1453 | { |
1454 | si = i + 1; | |
7117c2d2 JA |
1455 | i = (c == '\'') ? skip_single_quoted (string, slen, si) |
1456 | : skip_double_quoted (string, slen, si); | |
cce855bc | 1457 | /* skip_XXX_quoted leaves index one past close quote */ |
ccc6cda3 JA |
1458 | continue; |
1459 | } | |
9e51a74d CR |
1460 | #else /* XXX - bash-4.2 */ |
1461 | /* Pass the contents of double-quoted strings through verbatim. */ | |
1462 | if (c == '"') | |
1463 | { | |
1464 | si = i + 1; | |
1465 | i = skip_double_quoted (string, slen, si); | |
1466 | /* skip_XXX_quoted leaves index one past close quote */ | |
1467 | continue; | |
1468 | } | |
1469 | ||
1470 | if (c == '\'') | |
1471 | { | |
1472 | /*itrace("extract_dollar_brace_string: c == single quote flags = %d quoted = %d dolbrace_state = %d", flags, quoted, dolbrace_state);*/ | |
048b249e | 1473 | if (posixly_correct && shell_compatibility_level > 41 && dolbrace_state != DOLBRACE_QUOTE && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
9e51a74d CR |
1474 | ADVANCE_CHAR (string, slen, i); |
1475 | else | |
1476 | { | |
1477 | si = i + 1; | |
1478 | i = skip_single_quoted (string, slen, si); | |
1479 | } | |
1480 | ||
1481 | continue; | |
1482 | } | |
1483 | #endif | |
7117c2d2 JA |
1484 | |
1485 | /* move past this character, which was not special. */ | |
1486 | ADVANCE_CHAR (string, slen, i); | |
9e51a74d CR |
1487 | |
1488 | /* This logic must agree with parse.y:parse_matched_pair, since they | |
1489 | share the same defines. */ | |
3d35553a | 1490 | if (dolbrace_state == DOLBRACE_PARAM && c == '%' && (i - *sindex) > 1) |
9e51a74d CR |
1491 | dolbrace_state = DOLBRACE_QUOTE; |
1492 | else if (dolbrace_state == DOLBRACE_PARAM && c == '#' && (i - *sindex) > 1) | |
1493 | dolbrace_state = DOLBRACE_QUOTE; | |
3d35553a CR |
1494 | else if (dolbrace_state == DOLBRACE_PARAM && c == '/' && (i - *sindex) > 1) |
1495 | dolbrace_state = DOLBRACE_QUOTE; | |
1496 | else if (dolbrace_state == DOLBRACE_PARAM && c == '^' && (i - *sindex) > 1) | |
1497 | dolbrace_state = DOLBRACE_QUOTE; | |
1498 | else if (dolbrace_state == DOLBRACE_PARAM && c == ',' && (i - *sindex) > 1) | |
1499 | dolbrace_state = DOLBRACE_QUOTE; | |
9e51a74d CR |
1500 | else if (dolbrace_state == DOLBRACE_PARAM && strchr ("#%^,~:-=?+/", c) != 0) |
1501 | dolbrace_state = DOLBRACE_OP; | |
1502 | else if (dolbrace_state == DOLBRACE_OP && strchr ("#%^,~:-=?+/", c) == 0) | |
1503 | dolbrace_state = DOLBRACE_WORD; | |
cce855bc | 1504 | } |
726f6388 | 1505 | |
d3a24ed2 | 1506 | if (c == 0 && nesting_level) |
cce855bc | 1507 | { |
d3a24ed2 CR |
1508 | if (no_longjmp_on_fatal_error == 0) |
1509 | { /* { */ | |
d3a24ed2 | 1510 | last_command_exit_value = EXECUTION_FAILURE; |
7f947b68 | 1511 | report_error (_("bad substitution: no closing `%s' in %s"), "}", string); |
d3a24ed2 CR |
1512 | exp_jump_to_top_level (DISCARD); |
1513 | } | |
1514 | else | |
1515 | { | |
1516 | *sindex = i; | |
1517 | return ((char *)NULL); | |
1518 | } | |
726f6388 | 1519 | } |
726f6388 | 1520 | |
e6598ba4 | 1521 | result = (flags & SX_NOALLOC) ? (char *)NULL : substring (string, *sindex, i); |
726f6388 JA |
1522 | *sindex = i; |
1523 | ||
ccc6cda3 | 1524 | return (result); |
726f6388 JA |
1525 | } |
1526 | ||
ccc6cda3 JA |
1527 | /* Remove backslashes which are quoting backquotes from STRING. Modifies |
1528 | STRING, and returns a pointer to it. */ | |
1529 | char * | |
1530 | de_backslash (string) | |
726f6388 | 1531 | char *string; |
ccc6cda3 | 1532 | { |
7117c2d2 JA |
1533 | register size_t slen; |
1534 | register int i, j, prev_i; | |
1535 | DECLARE_MBSTATE; | |
726f6388 | 1536 | |
7117c2d2 JA |
1537 | slen = strlen (string); |
1538 | i = j = 0; | |
1539 | ||
1540 | /* Loop copying string[i] to string[j], i >= j. */ | |
1541 | while (i < slen) | |
1542 | { | |
1543 | if (string[i] == '\\' && (string[i + 1] == '`' || string[i + 1] == '\\' || | |
ccc6cda3 | 1544 | string[i + 1] == '$')) |
7117c2d2 JA |
1545 | i++; |
1546 | prev_i = i; | |
1547 | ADVANCE_CHAR (string, slen, i); | |
1548 | if (j < prev_i) | |
704a1a2a | 1549 | do string[j++] = string[prev_i++]; while (prev_i < i); |
7117c2d2 | 1550 | else |
704a1a2a | 1551 | j = i; |
7117c2d2 JA |
1552 | } |
1553 | string[j] = '\0'; | |
1554 | ||
ccc6cda3 JA |
1555 | return (string); |
1556 | } | |
726f6388 | 1557 | |
ccc6cda3 | 1558 | #if 0 |
cce855bc | 1559 | /*UNUSED*/ |
ccc6cda3 JA |
1560 | /* Replace instances of \! in a string with !. */ |
1561 | void | |
1562 | unquote_bang (string) | |
1563 | char *string; | |
1564 | { | |
1565 | register int i, j; | |
1566 | register char *temp; | |
726f6388 | 1567 | |
f73dda09 | 1568 | temp = (char *)xmalloc (1 + strlen (string)); |
726f6388 | 1569 | |
ccc6cda3 JA |
1570 | for (i = 0, j = 0; (temp[j] = string[i]); i++, j++) |
1571 | { | |
1572 | if (string[i] == '\\' && string[i + 1] == '!') | |
1573 | { | |
1574 | temp[j] = '!'; | |
1575 | i++; | |
1576 | } | |
1577 | } | |
1578 | strcpy (string, temp); | |
1579 | free (temp); | |
726f6388 | 1580 | } |
ccc6cda3 | 1581 | #endif |
726f6388 | 1582 | |
4d8d005b CR |
1583 | #define CQ_RETURN(x) do { no_longjmp_on_fatal_error = 0; return (x); } while (0) |
1584 | ||
6932f7f5 | 1585 | /* This function assumes s[i] == open; returns with s[ret] == close; used to |
3eb2d94a CR |
1586 | parse array subscripts. FLAGS & 1 means to not attempt to skip over |
1587 | matched pairs of quotes or backquotes, or skip word expansions; it is | |
1588 | intended to be used after expansion has been performed and during final | |
1589 | assignment parsing (see arrayfunc.c:assign_compound_array_list()). */ | |
6932f7f5 CR |
1590 | static int |
1591 | skip_matched_pair (string, start, open, close, flags) | |
76af2125 | 1592 | const char *string; |
6932f7f5 | 1593 | int start, open, close, flags; |
76af2125 CR |
1594 | { |
1595 | int i, pass_next, backq, si, c, count; | |
1596 | size_t slen; | |
1597 | char *temp, *ss; | |
1598 | DECLARE_MBSTATE; | |
1599 | ||
1600 | slen = strlen (string + start) + start; | |
1601 | no_longjmp_on_fatal_error = 1; | |
1602 | ||
1603 | i = start + 1; /* skip over leading bracket */ | |
1604 | count = 1; | |
1605 | pass_next = backq = 0; | |
1606 | ss = (char *)string; | |
1607 | while (c = string[i]) | |
1608 | { | |
1609 | if (pass_next) | |
1610 | { | |
1611 | pass_next = 0; | |
1612 | if (c == 0) | |
1613 | CQ_RETURN(i); | |
1614 | ADVANCE_CHAR (string, slen, i); | |
1615 | continue; | |
1616 | } | |
1617 | else if (c == '\\') | |
1618 | { | |
1619 | pass_next = 1; | |
1620 | i++; | |
1621 | continue; | |
1622 | } | |
1623 | else if (backq) | |
1624 | { | |
1625 | if (c == '`') | |
1626 | backq = 0; | |
1627 | ADVANCE_CHAR (string, slen, i); | |
1628 | continue; | |
1629 | } | |
3eb2d94a | 1630 | else if ((flags & 1) == 0 && c == '`') |
76af2125 CR |
1631 | { |
1632 | backq = 1; | |
1633 | i++; | |
1634 | continue; | |
1635 | } | |
3eb2d94a | 1636 | else if ((flags & 1) == 0 && c == open) |
76af2125 CR |
1637 | { |
1638 | count++; | |
1639 | i++; | |
1640 | continue; | |
1641 | } | |
6932f7f5 | 1642 | else if (c == close) |
76af2125 CR |
1643 | { |
1644 | count--; | |
1645 | if (count == 0) | |
1646 | break; | |
1647 | i++; | |
1648 | continue; | |
1649 | } | |
3eb2d94a | 1650 | else if ((flags & 1) == 0 && (c == '\'' || c == '"')) |
76af2125 CR |
1651 | { |
1652 | i = (c == '\'') ? skip_single_quoted (ss, slen, ++i) | |
1653 | : skip_double_quoted (ss, slen, ++i); | |
1654 | /* no increment, the skip functions increment past the closing quote. */ | |
1655 | } | |
3eb2d94a | 1656 | else if ((flags&1) == 0 && c == '$' && (string[i+1] == LPAREN || string[i+1] == LBRACE)) |
76af2125 CR |
1657 | { |
1658 | si = i + 2; | |
1659 | if (string[si] == '\0') | |
1660 | CQ_RETURN(si); | |
1661 | ||
1662 | if (string[i+1] == LPAREN) | |
1663 | temp = extract_delimited_string (ss, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */ | |
1664 | else | |
1665 | temp = extract_dollar_brace_string (ss, &si, 0, SX_NOALLOC); | |
1666 | i = si; | |
1667 | if (string[i] == '\0') /* don't increment i past EOS in loop */ | |
1668 | break; | |
1669 | i++; | |
1670 | continue; | |
1671 | } | |
1672 | else | |
1673 | ADVANCE_CHAR (string, slen, i); | |
1674 | } | |
1675 | ||
1676 | CQ_RETURN(i); | |
1677 | } | |
1678 | ||
6932f7f5 CR |
1679 | #if defined (ARRAY_VARS) |
1680 | int | |
3eb2d94a | 1681 | skipsubscript (string, start, flags) |
6932f7f5 | 1682 | const char *string; |
3eb2d94a | 1683 | int start, flags; |
6932f7f5 | 1684 | { |
3eb2d94a | 1685 | return (skip_matched_pair (string, start, '[', ']', flags)); |
6932f7f5 CR |
1686 | } |
1687 | #endif | |
1688 | ||
c2a47ea9 CR |
1689 | /* Skip characters in STRING until we find a character in DELIMS, and return |
1690 | the index of that character. START is the index into string at which we | |
1691 | begin. This is similar in spirit to strpbrk, but it returns an index into | |
1692 | STRING and takes a starting index. This little piece of code knows quite | |
1693 | a lot of shell syntax. It's very similar to skip_double_quoted and other | |
1694 | functions of that ilk. */ | |
1695 | int | |
4d8d005b | 1696 | skip_to_delim (string, start, delims, flags) |
c2a47ea9 CR |
1697 | char *string; |
1698 | int start; | |
1699 | char *delims; | |
4d8d005b | 1700 | int flags; |
c2a47ea9 | 1701 | { |
176b12ee | 1702 | int i, pass_next, backq, si, c, invert, skipquote, skipcmd; |
c2a47ea9 | 1703 | size_t slen; |
5f8cde23 | 1704 | char *temp, open[3]; |
c2a47ea9 CR |
1705 | DECLARE_MBSTATE; |
1706 | ||
1707 | slen = strlen (string + start) + start; | |
4d8d005b CR |
1708 | if (flags & SD_NOJMP) |
1709 | no_longjmp_on_fatal_error = 1; | |
8c2fef19 | 1710 | invert = (flags & SD_INVERT); |
176b12ee | 1711 | skipcmd = (flags & SD_NOSKIPCMD) == 0; |
8c2fef19 | 1712 | |
c2a47ea9 CR |
1713 | i = start; |
1714 | pass_next = backq = 0; | |
1715 | while (c = string[i]) | |
1716 | { | |
a8fd3f3e CR |
1717 | /* If this is non-zero, we should not let quote characters be delimiters |
1718 | and the current character is a single or double quote. We should not | |
1719 | test whether or not it's a delimiter until after we skip single- or | |
1720 | double-quoted strings. */ | |
1721 | skipquote = ((flags & SD_NOQUOTEDELIM) && (c == '\'' || c =='"')); | |
c2a47ea9 CR |
1722 | if (pass_next) |
1723 | { | |
1724 | pass_next = 0; | |
1725 | if (c == 0) | |
1726 | CQ_RETURN(i); | |
1727 | ADVANCE_CHAR (string, slen, i); | |
1728 | continue; | |
1729 | } | |
1730 | else if (c == '\\') | |
1731 | { | |
1732 | pass_next = 1; | |
1733 | i++; | |
1734 | continue; | |
1735 | } | |
1736 | else if (backq) | |
1737 | { | |
1738 | if (c == '`') | |
1739 | backq = 0; | |
1740 | ADVANCE_CHAR (string, slen, i); | |
1741 | continue; | |
1742 | } | |
1743 | else if (c == '`') | |
1744 | { | |
1745 | backq = 1; | |
1746 | i++; | |
1747 | continue; | |
1748 | } | |
a8fd3f3e | 1749 | else if (skipquote == 0 && invert == 0 && member (c, delims)) |
8c2fef19 | 1750 | break; |
c2a47ea9 CR |
1751 | else if (c == '\'' || c == '"') |
1752 | { | |
1753 | i = (c == '\'') ? skip_single_quoted (string, slen, ++i) | |
1754 | : skip_double_quoted (string, slen, ++i); | |
1755 | /* no increment, the skip functions increment past the closing quote. */ | |
1756 | } | |
176b12ee | 1757 | else if (c == '$' && ((skipcmd && string[i+1] == LPAREN) || string[i+1] == LBRACE)) |
c2a47ea9 CR |
1758 | { |
1759 | si = i + 2; | |
1760 | if (string[si] == '\0') | |
1761 | CQ_RETURN(si); | |
1762 | ||
1763 | if (string[i+1] == LPAREN) | |
1764 | temp = extract_delimited_string (string, &si, "$(", "(", ")", SX_NOALLOC|SX_COMMAND); /* ) */ | |
1765 | else | |
1766 | temp = extract_dollar_brace_string (string, &si, 0, SX_NOALLOC); | |
1767 | i = si; | |
1768 | if (string[i] == '\0') /* don't increment i past EOS in loop */ | |
1769 | break; | |
1770 | i++; | |
1771 | continue; | |
1772 | } | |
176b12ee CR |
1773 | #if defined (PROCESS_SUBSTITUTION) |
1774 | else if (skipcmd && (c == '<' || c == '>') && string[i+1] == LPAREN) | |
1775 | { | |
1776 | si = i + 2; | |
1777 | if (string[si] == '\0') | |
1778 | CQ_RETURN(si); | |
1779 | temp = extract_process_subst (string, (c == '<') ? "<(" : ">(", &si); | |
631b20c6 | 1780 | free (temp); /* no SX_ALLOC here */ |
176b12ee CR |
1781 | i = si; |
1782 | if (string[i] == '\0') | |
1783 | break; | |
1784 | i++; | |
1785 | continue; | |
1786 | } | |
1787 | #endif /* PROCESS_SUBSTITUTION */ | |
5f8cde23 CR |
1788 | #if defined (EXTENDED_GLOB) |
1789 | else if ((flags & SD_EXTGLOB) && extended_glob && string[i+1] == LPAREN && member (c, "?*+!@")) | |
1790 | { | |
1791 | si = i + 2; | |
1792 | if (string[si] == '\0') | |
1793 | CQ_RETURN(si); | |
1794 | ||
1795 | open[0] = c; | |
1796 | open[1] = LPAREN; | |
1797 | open[2] = '\0'; | |
1798 | temp = extract_delimited_string (string, &si, open, "(", ")", SX_NOALLOC); /* ) */ | |
1799 | ||
1800 | i = si; | |
1801 | if (string[i] == '\0') /* don't increment i past EOS in loop */ | |
1802 | break; | |
1803 | i++; | |
1804 | continue; | |
1805 | } | |
1806 | #endif | |
a8fd3f3e | 1807 | else if ((skipquote || invert) && (member (c, delims) == 0)) |
c2a47ea9 CR |
1808 | break; |
1809 | else | |
1810 | ADVANCE_CHAR (string, slen, i); | |
1811 | } | |
1812 | ||
1813 | CQ_RETURN(i); | |
1814 | } | |
1815 | ||
ccc6cda3 | 1816 | #if defined (READLINE) |
726f6388 JA |
1817 | /* Return 1 if the portion of STRING ending at EINDEX is quoted (there is |
1818 | an unclosed quoted string), or if the character at EINDEX is quoted | |
28ef6c31 | 1819 | by a backslash. NO_LONGJMP_ON_FATAL_ERROR is used to flag that the various |
b72432fd | 1820 | single and double-quoted string parsing functions should not return an |
7117c2d2 JA |
1821 | error if there are unclosed quotes or braces. The characters that this |
1822 | recognizes need to be the same as the contents of | |
1823 | rl_completer_quote_characters. */ | |
b72432fd | 1824 | |
726f6388 JA |
1825 | int |
1826 | char_is_quoted (string, eindex) | |
1827 | char *string; | |
1828 | int eindex; | |
1829 | { | |
7117c2d2 JA |
1830 | int i, pass_next, c; |
1831 | size_t slen; | |
1832 | DECLARE_MBSTATE; | |
726f6388 | 1833 | |
7117c2d2 | 1834 | slen = strlen (string); |
28ef6c31 | 1835 | no_longjmp_on_fatal_error = 1; |
7117c2d2 JA |
1836 | i = pass_next = 0; |
1837 | while (i <= eindex) | |
726f6388 | 1838 | { |
7117c2d2 JA |
1839 | c = string[i]; |
1840 | ||
726f6388 JA |
1841 | if (pass_next) |
1842 | { | |
1843 | pass_next = 0; | |
1844 | if (i >= eindex) /* XXX was if (i >= eindex - 1) */ | |
b72432fd | 1845 | CQ_RETURN(1); |
7117c2d2 | 1846 | ADVANCE_CHAR (string, slen, i); |
726f6388 JA |
1847 | continue; |
1848 | } | |
7117c2d2 | 1849 | else if (c == '\\') |
ccc6cda3 JA |
1850 | { |
1851 | pass_next = 1; | |
7117c2d2 | 1852 | i++; |
ccc6cda3 JA |
1853 | continue; |
1854 | } | |
7117c2d2 JA |
1855 | else if (c == '\'' || c == '"') |
1856 | { | |
1857 | i = (c == '\'') ? skip_single_quoted (string, slen, ++i) | |
1858 | : skip_double_quoted (string, slen, ++i); | |
1859 | if (i > eindex) | |
1860 | CQ_RETURN(1); | |
1861 | /* no increment, the skip_xxx functions go one past end */ | |
1862 | } | |
1863 | else | |
1864 | ADVANCE_CHAR (string, slen, i); | |
726f6388 | 1865 | } |
7117c2d2 | 1866 | |
b72432fd | 1867 | CQ_RETURN(0); |
726f6388 JA |
1868 | } |
1869 | ||
726f6388 JA |
1870 | int |
1871 | unclosed_pair (string, eindex, openstr) | |
1872 | char *string; | |
1873 | int eindex; | |
1874 | char *openstr; | |
1875 | { | |
ccc6cda3 | 1876 | int i, pass_next, openc, olen; |
7117c2d2 JA |
1877 | size_t slen; |
1878 | DECLARE_MBSTATE; | |
726f6388 | 1879 | |
7117c2d2 | 1880 | slen = strlen (string); |
726f6388 | 1881 | olen = strlen (openstr); |
7117c2d2 JA |
1882 | i = pass_next = openc = 0; |
1883 | while (i <= eindex) | |
726f6388 JA |
1884 | { |
1885 | if (pass_next) | |
1886 | { | |
1887 | pass_next = 0; | |
1888 | if (i >= eindex) /* XXX was if (i >= eindex - 1) */ | |
1889 | return 0; | |
7117c2d2 JA |
1890 | ADVANCE_CHAR (string, slen, i); |
1891 | continue; | |
1892 | } | |
1893 | else if (string[i] == '\\') | |
1894 | { | |
1895 | pass_next = 1; | |
1896 | i++; | |
726f6388 JA |
1897 | continue; |
1898 | } | |
1899 | else if (STREQN (string + i, openstr, olen)) | |
1900 | { | |
1901 | openc = 1 - openc; | |
7117c2d2 | 1902 | i += olen; |
726f6388 | 1903 | } |
ccc6cda3 | 1904 | else if (string[i] == '\'' || string[i] == '"') |
726f6388 | 1905 | { |
7117c2d2 JA |
1906 | i = (string[i] == '\'') ? skip_single_quoted (string, slen, i) |
1907 | : skip_double_quoted (string, slen, i); | |
726f6388 JA |
1908 | if (i > eindex) |
1909 | return 0; | |
1910 | } | |
7117c2d2 JA |
1911 | else |
1912 | ADVANCE_CHAR (string, slen, i); | |
726f6388 JA |
1913 | } |
1914 | return (openc); | |
1915 | } | |
bb70624e | 1916 | |
bb70624e JA |
1917 | /* Split STRING (length SLEN) at DELIMS, and return a WORD_LIST with the |
1918 | individual words. If DELIMS is NULL, the current value of $IFS is used | |
d3a24ed2 CR |
1919 | to split the string, and the function follows the shell field splitting |
1920 | rules. SENTINEL is an index to look for. NWP, if non-NULL, | |
bb70624e JA |
1921 | gets the number of words in the returned list. CWP, if non-NULL, gets |
1922 | the index of the word containing SENTINEL. Non-whitespace chars in | |
1923 | DELIMS delimit separate fields. */ | |
1924 | WORD_LIST * | |
a8fd3f3e | 1925 | split_at_delims (string, slen, delims, sentinel, flags, nwp, cwp) |
bb70624e JA |
1926 | char *string; |
1927 | int slen; | |
1928 | char *delims; | |
a8fd3f3e | 1929 | int sentinel, flags; |
bb70624e JA |
1930 | int *nwp, *cwp; |
1931 | { | |
a8fd3f3e | 1932 | int ts, te, i, nw, cw, ifs_split, dflags; |
f73dda09 | 1933 | char *token, *d, *d2; |
bb70624e JA |
1934 | WORD_LIST *ret, *tl; |
1935 | ||
1936 | if (string == 0 || *string == '\0') | |
1937 | { | |
1938 | if (nwp) | |
1939 | *nwp = 0; | |
1940 | if (cwp) | |
1941 | *cwp = 0; | |
1942 | return ((WORD_LIST *)NULL); | |
1943 | } | |
1944 | ||
7117c2d2 | 1945 | d = (delims == 0) ? ifs_value : delims; |
d3a24ed2 | 1946 | ifs_split = delims == 0; |
bb70624e JA |
1947 | |
1948 | /* Make d2 the non-whitespace characters in delims */ | |
1949 | d2 = 0; | |
1950 | if (delims) | |
1951 | { | |
633e5c6d CR |
1952 | size_t slength; |
1953 | #if defined (HANDLE_MULTIBYTE) | |
1954 | size_t mblength = 1; | |
1955 | #endif | |
1956 | DECLARE_MBSTATE; | |
1957 | ||
1958 | slength = strlen (delims); | |
1959 | d2 = (char *)xmalloc (slength + 1); | |
1960 | i = ts = 0; | |
1961 | while (delims[i]) | |
bb70624e | 1962 | { |
633e5c6d | 1963 | #if defined (HANDLE_MULTIBYTE) |
d3ad40de CR |
1964 | mbstate_t state_bak; |
1965 | state_bak = state; | |
633e5c6d CR |
1966 | mblength = MBRLEN (delims + i, slength, &state); |
1967 | if (MB_INVALIDCH (mblength)) | |
1968 | state = state_bak; | |
22e63b05 | 1969 | else if (mblength > 1) |
633e5c6d CR |
1970 | { |
1971 | memcpy (d2 + ts, delims + i, mblength); | |
1972 | ts += mblength; | |
1973 | i += mblength; | |
1974 | slength -= mblength; | |
1975 | continue; | |
1976 | } | |
1977 | #endif | |
1978 | if (whitespace (delims[i]) == 0) | |
bb70624e | 1979 | d2[ts++] = delims[i]; |
633e5c6d CR |
1980 | |
1981 | i++; | |
1982 | slength--; | |
bb70624e JA |
1983 | } |
1984 | d2[ts] = '\0'; | |
1985 | } | |
1986 | ||
1987 | ret = (WORD_LIST *)NULL; | |
1988 | ||
dfc21851 | 1989 | /* Remove sequences of whitespace characters at the start of the string, as |
d3a24ed2 CR |
1990 | long as those characters are delimiters. */ |
1991 | for (i = 0; member (string[i], d) && spctabnl (string[i]); i++) | |
bb70624e JA |
1992 | ; |
1993 | if (string[i] == '\0') | |
1994 | return (ret); | |
1995 | ||
1996 | ts = i; | |
1997 | nw = 0; | |
1998 | cw = -1; | |
a8fd3f3e | 1999 | dflags = flags|SD_NOJMP; |
bb70624e JA |
2000 | while (1) |
2001 | { | |
a8fd3f3e | 2002 | te = skip_to_delim (string, ts, d, dflags); |
bb70624e JA |
2003 | |
2004 | /* If we have a non-whitespace delimiter character, use it to make a | |
2005 | separate field. This is just about what $IFS splitting does and | |
2006 | is closer to the behavior of the shell parser. */ | |
28ef6c31 | 2007 | if (ts == te && d2 && member (string[ts], d2)) |
bb70624e JA |
2008 | { |
2009 | te = ts + 1; | |
d3a24ed2 CR |
2010 | /* If we're using IFS splitting, the non-whitespace delimiter char |
2011 | and any additional IFS whitespace delimits a field. */ | |
2012 | if (ifs_split) | |
2013 | while (member (string[te], d) && spctabnl (string[te])) | |
2014 | te++; | |
2015 | else | |
2016 | while (member (string[te], d2)) | |
2017 | te++; | |
bb70624e JA |
2018 | } |
2019 | ||
2020 | token = substring (string, ts, te); | |
2021 | ||
2022 | ret = add_string_to_list (token, ret); | |
2023 | free (token); | |
2024 | nw++; | |
2025 | ||
2026 | if (sentinel >= ts && sentinel <= te) | |
2027 | cw = nw; | |
2028 | ||
2029 | /* If the cursor is at whitespace just before word start, set the | |
28ef6c31 | 2030 | sentinel word to the current word. */ |
bb70624e JA |
2031 | if (cwp && cw == -1 && sentinel == ts-1) |
2032 | cw = nw; | |
2033 | ||
2034 | /* If the cursor is at whitespace between two words, make a new, empty | |
28ef6c31 JA |
2035 | word, add it before (well, after, since the list is in reverse order) |
2036 | the word we just added, and set the current word to that one. */ | |
bb70624e | 2037 | if (cwp && cw == -1 && sentinel < ts) |
28ef6c31 | 2038 | { |
7117c2d2 | 2039 | tl = make_word_list (make_word (""), ret->next); |
28ef6c31 JA |
2040 | ret->next = tl; |
2041 | cw = nw; | |
2042 | nw++; | |
2043 | } | |
bb70624e JA |
2044 | |
2045 | if (string[te] == 0) | |
2046 | break; | |
2047 | ||
d3a24ed2 CR |
2048 | i = te; |
2049 | while (member (string[i], d) && (ifs_split || spctabnl(string[i]))) | |
bb70624e JA |
2050 | i++; |
2051 | ||
2052 | if (string[i]) | |
2053 | ts = i; | |
2054 | else | |
2055 | break; | |
2056 | } | |
2057 | ||
2058 | /* Special case for SENTINEL at the end of STRING. If we haven't found | |
2059 | the word containing SENTINEL yet, and the index we're looking for is at | |
dfc21851 CR |
2060 | the end of STRING (or past the end of the previously-found token, |
2061 | possible if the end of the line is composed solely of IFS whitespace) | |
2062 | add an additional null argument and set the current word pointer to that. */ | |
2063 | if (cwp && cw == -1 && (sentinel >= slen || sentinel >= te)) | |
bb70624e JA |
2064 | { |
2065 | if (whitespace (string[sentinel - 1])) | |
28ef6c31 JA |
2066 | { |
2067 | token = ""; | |
2068 | ret = add_string_to_list (token, ret); | |
2069 | nw++; | |
2070 | } | |
bb70624e JA |
2071 | cw = nw; |
2072 | } | |
2073 | ||
2074 | if (nwp) | |
2075 | *nwp = nw; | |
2076 | if (cwp) | |
2077 | *cwp = cw; | |
2078 | ||
631b20c6 CR |
2079 | FREE (d2); |
2080 | ||
bb70624e JA |
2081 | return (REVERSE_LIST (ret, WORD_LIST *)); |
2082 | } | |
726f6388 JA |
2083 | #endif /* READLINE */ |
2084 | ||
ccc6cda3 JA |
2085 | #if 0 |
2086 | /* UNUSED */ | |
726f6388 JA |
2087 | /* Extract the name of the variable to bind to from the assignment string. */ |
2088 | char * | |
2089 | assignment_name (string) | |
2090 | char *string; | |
2091 | { | |
ccc6cda3 | 2092 | int offset; |
726f6388 JA |
2093 | char *temp; |
2094 | ||
5e13499c | 2095 | offset = assignment (string, 0); |
ccc6cda3 | 2096 | if (offset == 0) |
726f6388 | 2097 | return (char *)NULL; |
bb70624e | 2098 | temp = substring (string, 0, offset); |
726f6388 JA |
2099 | return (temp); |
2100 | } | |
ccc6cda3 | 2101 | #endif |
726f6388 | 2102 | |
cce855bc JA |
2103 | /* **************************************************************** */ |
2104 | /* */ | |
2105 | /* Functions to convert strings to WORD_LISTs and vice versa */ | |
2106 | /* */ | |
2107 | /* **************************************************************** */ | |
2108 | ||
726f6388 JA |
2109 | /* Return a single string of all the words in LIST. SEP is the separator |
2110 | to put between individual elements of LIST in the output string. */ | |
7117c2d2 | 2111 | char * |
726f6388 JA |
2112 | string_list_internal (list, sep) |
2113 | WORD_LIST *list; | |
2114 | char *sep; | |
2115 | { | |
2116 | register WORD_LIST *t; | |
2117 | char *result, *r; | |
2118 | int word_len, sep_len, result_size; | |
2119 | ||
ccc6cda3 | 2120 | if (list == 0) |
726f6388 JA |
2121 | return ((char *)NULL); |
2122 | ||
12d937f9 CR |
2123 | /* Short-circuit quickly if we don't need to separate anything. */ |
2124 | if (list->next == 0) | |
2125 | return (savestring (list->word->word)); | |
2126 | ||
726f6388 JA |
2127 | /* This is nearly always called with either sep[0] == 0 or sep[1] == 0. */ |
2128 | sep_len = STRLEN (sep); | |
2129 | result_size = 0; | |
2130 | ||
2131 | for (t = list; t; t = t->next) | |
2132 | { | |
2133 | if (t != list) | |
2134 | result_size += sep_len; | |
2135 | result_size += strlen (t->word->word); | |
2136 | } | |
2137 | ||
f73dda09 | 2138 | r = result = (char *)xmalloc (result_size + 1); |
726f6388 JA |
2139 | |
2140 | for (t = list; t; t = t->next) | |
2141 | { | |
2142 | if (t != list && sep_len) | |
2143 | { | |
ccc6cda3 JA |
2144 | if (sep_len > 1) |
2145 | { | |
2146 | FASTCOPY (sep, r, sep_len); | |
2147 | r += sep_len; | |
2148 | } | |
2149 | else | |
2150 | *r++ = sep[0]; | |
726f6388 JA |
2151 | } |
2152 | ||
2153 | word_len = strlen (t->word->word); | |
2154 | FASTCOPY (t->word->word, r, word_len); | |
2155 | r += word_len; | |
2156 | } | |
2157 | ||
ccc6cda3 | 2158 | *r = '\0'; |
726f6388 JA |
2159 | return (result); |
2160 | } | |
2161 | ||
2162 | /* Return a single string of all the words present in LIST, separating | |
2163 | each word with a space. */ | |
2164 | char * | |
2165 | string_list (list) | |
2166 | WORD_LIST *list; | |
2167 | { | |
2168 | return (string_list_internal (list, " ")); | |
2169 | } | |
2170 | ||
c2a47ea9 CR |
2171 | /* An external interface that can be used by the rest of the shell to |
2172 | obtain a string containing the first character in $IFS. Handles all | |
2173 | the multibyte complications. If LENP is non-null, it is set to the | |
2174 | length of the returned string. */ | |
2175 | char * | |
2176 | ifs_firstchar (lenp) | |
2177 | int *lenp; | |
2178 | { | |
2179 | char *ret; | |
2180 | int len; | |
2181 | ||
2182 | ret = xmalloc (MB_LEN_MAX + 1); | |
2183 | #if defined (HANDLE_MULTIBYTE) | |
2184 | if (ifs_firstc_len == 1) | |
2185 | { | |
2186 | ret[0] = ifs_firstc[0]; | |
2187 | ret[1] = '\0'; | |
2188 | len = ret[0] ? 1 : 0; | |
2189 | } | |
2190 | else | |
2191 | { | |
2192 | memcpy (ret, ifs_firstc, ifs_firstc_len); | |
2193 | ret[len = ifs_firstc_len] = '\0'; | |
2194 | } | |
2195 | #else | |
2196 | ret[0] = ifs_firstc; | |
2197 | ret[1] = '\0'; | |
2198 | len = ret[0] ? 0 : 1; | |
2199 | #endif | |
2200 | ||
2201 | if (lenp) | |
2202 | *lenp = len; | |
2203 | ||
2204 | return ret; | |
2205 | } | |
2206 | ||
726f6388 JA |
2207 | /* Return a single string of all the words present in LIST, obeying the |
2208 | quoting rules for "$*", to wit: (P1003.2, draft 11, 3.5.2) "If the | |
2209 | expansion [of $*] appears within a double quoted string, it expands | |
2210 | to a single field with the value of each parameter separated by the | |
2211 | first character of the IFS variable, or by a <space> if IFS is unset." */ | |
f73dda09 | 2212 | char * |
726f6388 JA |
2213 | string_list_dollar_star (list) |
2214 | WORD_LIST *list; | |
2215 | { | |
dc8fbaf9 | 2216 | char *ret; |
633e5c6d | 2217 | #if defined (HANDLE_MULTIBYTE) |
dc8fbaf9 | 2218 | # if defined (__GNUC__) |
633e5c6d | 2219 | char sep[MB_CUR_MAX + 1]; |
dc8fbaf9 CR |
2220 | # else |
2221 | char *sep = 0; | |
2222 | # endif | |
633e5c6d | 2223 | #else |
7117c2d2 | 2224 | char sep[2]; |
633e5c6d CR |
2225 | #endif |
2226 | ||
633e5c6d | 2227 | #if defined (HANDLE_MULTIBYTE) |
dc8fbaf9 CR |
2228 | # if !defined (__GNUC__) |
2229 | sep = (char *)xmalloc (MB_CUR_MAX + 1); | |
2230 | # endif /* !__GNUC__ */ | |
633e5c6d CR |
2231 | if (ifs_firstc_len == 1) |
2232 | { | |
2233 | sep[0] = ifs_firstc[0]; | |
2234 | sep[1] = '\0'; | |
2235 | } | |
2236 | else | |
da713c23 CR |
2237 | { |
2238 | memcpy (sep, ifs_firstc, ifs_firstc_len); | |
2239 | sep[ifs_firstc_len] = '\0'; | |
2240 | } | |
633e5c6d | 2241 | #else |
7117c2d2 | 2242 | sep[0] = ifs_firstc; |
726f6388 | 2243 | sep[1] = '\0'; |
633e5c6d | 2244 | #endif |
726f6388 | 2245 | |
dc8fbaf9 CR |
2246 | ret = string_list_internal (list, sep); |
2247 | #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__) | |
2248 | free (sep); | |
2249 | #endif | |
2250 | return ret; | |
726f6388 JA |
2251 | } |
2252 | ||
cce855bc JA |
2253 | /* Turn $@ into a string. If (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
2254 | is non-zero, the $@ appears within double quotes, and we should quote | |
2255 | the list before converting it into a string. If IFS is unset, and the | |
2256 | word is not quoted, we just need to quote CTLESC and CTLNUL characters | |
2257 | in the words in the list, because the default value of $IFS is | |
2258 | <space><tab><newline>, IFS characters in the words in the list should | |
2259 | also be split. If IFS is null, and the word is not quoted, we need | |
2260 | to quote the words in the list to preserve the positional parameters | |
2261 | exactly. */ | |
f73dda09 | 2262 | char * |
cce855bc JA |
2263 | string_list_dollar_at (list, quoted) |
2264 | WORD_LIST *list; | |
2265 | int quoted; | |
2266 | { | |
227f982e | 2267 | char *ifs, *ret; |
633e5c6d | 2268 | #if defined (HANDLE_MULTIBYTE) |
dc8fbaf9 | 2269 | # if defined (__GNUC__) |
633e5c6d | 2270 | char sep[MB_CUR_MAX + 1]; |
dc8fbaf9 CR |
2271 | # else |
2272 | char *sep = 0; | |
2273 | # endif /* !__GNUC__ */ | |
633e5c6d CR |
2274 | #else |
2275 | char sep[2]; | |
2276 | #endif | |
cce855bc JA |
2277 | WORD_LIST *tlist; |
2278 | ||
7117c2d2 JA |
2279 | /* XXX this could just be ifs = ifs_value; */ |
2280 | ifs = ifs_var ? value_cell (ifs_var) : (char *)0; | |
cce855bc | 2281 | |
633e5c6d | 2282 | #if defined (HANDLE_MULTIBYTE) |
dc8fbaf9 CR |
2283 | # if !defined (__GNUC__) |
2284 | sep = (char *)xmalloc (MB_CUR_MAX + 1); | |
2285 | # endif /* !__GNUC__ */ | |
633e5c6d CR |
2286 | if (ifs && *ifs) |
2287 | { | |
227f982e | 2288 | if (ifs_firstc_len == 1) |
633e5c6d | 2289 | { |
227f982e | 2290 | sep[0] = ifs_firstc[0]; |
633e5c6d CR |
2291 | sep[1] = '\0'; |
2292 | } | |
2293 | else | |
2294 | { | |
227f982e CR |
2295 | memcpy (sep, ifs_firstc, ifs_firstc_len); |
2296 | sep[ifs_firstc_len] = '\0'; | |
633e5c6d CR |
2297 | } |
2298 | } | |
2299 | else | |
2300 | { | |
2301 | sep[0] = ' '; | |
2302 | sep[1] = '\0'; | |
2303 | } | |
2304 | #else | |
cce855bc JA |
2305 | sep[0] = (ifs == 0 || *ifs == 0) ? ' ' : *ifs; |
2306 | sep[1] = '\0'; | |
633e5c6d | 2307 | #endif |
cce855bc | 2308 | |
d3ad40de CR |
2309 | /* XXX -- why call quote_list if ifs == 0? we can get away without doing |
2310 | it now that quote_escapes quotes spaces */ | |
1231ac47 | 2311 | tlist = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE)) |
cce855bc JA |
2312 | ? quote_list (list) |
2313 | : list_quote_escapes (list); | |
dc8fbaf9 CR |
2314 | |
2315 | ret = string_list_internal (tlist, sep); | |
2316 | #if defined (HANDLE_MULTIBYTE) && !defined (__GNUC__) | |
2317 | free (sep); | |
2318 | #endif | |
2319 | return ret; | |
cce855bc JA |
2320 | } |
2321 | ||
e33f2203 CR |
2322 | /* Turn the positional paramters into a string, understanding quoting and |
2323 | the various subtleties of using the first character of $IFS as the | |
2324 | separator. Calls string_list_dollar_at, string_list_dollar_star, and | |
2325 | string_list as appropriate. */ | |
2326 | char * | |
2327 | string_list_pos_params (pchar, list, quoted) | |
2328 | int pchar; | |
2329 | WORD_LIST *list; | |
2330 | int quoted; | |
2331 | { | |
2332 | char *ret; | |
2333 | WORD_LIST *tlist; | |
2334 | ||
2335 | if (pchar == '*' && (quoted & Q_DOUBLE_QUOTES)) | |
2336 | { | |
2337 | tlist = quote_list (list); | |
2338 | word_list_remove_quoted_nulls (tlist); | |
2339 | ret = string_list_dollar_star (tlist); | |
2340 | } | |
2341 | else if (pchar == '*' && (quoted & Q_HERE_DOCUMENT)) | |
2342 | { | |
2343 | tlist = quote_list (list); | |
2344 | word_list_remove_quoted_nulls (tlist); | |
2345 | ret = string_list (tlist); | |
2346 | } | |
2347 | else if (pchar == '*') | |
2348 | { | |
2349 | /* Even when unquoted, string_list_dollar_star does the right thing | |
2350 | making sure that the first character of $IFS is used as the | |
2351 | separator. */ | |
2352 | ret = string_list_dollar_star (list); | |
2353 | } | |
2354 | else if (pchar == '@' && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
2355 | /* We use string_list_dollar_at, but only if the string is quoted, since | |
2356 | that quotes the escapes if it's not, which we don't want. We could | |
2357 | use string_list (the old code did), but that doesn't do the right | |
2358 | thing if the first character of $IFS is not a space. We use | |
2359 | string_list_dollar_star if the string is unquoted so we make sure that | |
2360 | the elements of $@ are separated by the first character of $IFS for | |
2361 | later splitting. */ | |
2362 | ret = string_list_dollar_at (list, quoted); | |
2363 | else if (pchar == '@') | |
2364 | ret = string_list_dollar_star (list); | |
2365 | else | |
2366 | ret = string_list ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? quote_list (list) : list); | |
2367 | ||
2368 | return ret; | |
2369 | } | |
2370 | ||
726f6388 JA |
2371 | /* Return the list of words present in STRING. Separate the string into |
2372 | words at any of the characters found in SEPARATORS. If QUOTED is | |
2373 | non-zero then word in the list will have its quoted flag set, otherwise | |
2374 | the quoted flag is left as make_word () deemed fit. | |
2375 | ||
2376 | This obeys the P1003.2 word splitting semantics. If `separators' is | |
2377 | exactly <space><tab><newline>, then the splitting algorithm is that of | |
2378 | the Bourne shell, which treats any sequence of characters from `separators' | |
2379 | as a delimiter. If IFS is unset, which results in `separators' being set | |
2380 | to "", no splitting occurs. If separators has some other value, the | |
2381 | following rules are applied (`IFS white space' means zero or more | |
2382 | occurrences of <space>, <tab>, or <newline>, as long as those characters | |
2383 | are in `separators'): | |
2384 | ||
2385 | 1) IFS white space is ignored at the start and the end of the | |
2386 | string. | |
2387 | 2) Each occurrence of a character in `separators' that is not | |
2388 | IFS white space, along with any adjacent occurrences of | |
2389 | IFS white space delimits a field. | |
2390 | 3) Any nonzero-length sequence of IFS white space delimits a field. | |
2391 | */ | |
2392 | ||
2393 | /* BEWARE! list_string strips null arguments. Don't call it twice and | |
2394 | expect to have "" preserved! */ | |
2395 | ||
726f6388 JA |
2396 | /* This performs word splitting and quoted null character removal on |
2397 | STRING. */ | |
d3a24ed2 CR |
2398 | #define issep(c) \ |
2399 | (((separators)[0]) ? ((separators)[1] ? isifs(c) \ | |
2400 | : (c) == (separators)[0]) \ | |
2401 | : 0) | |
726f6388 JA |
2402 | |
2403 | WORD_LIST * | |
2404 | list_string (string, separators, quoted) | |
2405 | register char *string, *separators; | |
2406 | int quoted; | |
2407 | { | |
ccc6cda3 JA |
2408 | WORD_LIST *result; |
2409 | WORD_DESC *t; | |
2410 | char *current_word, *s; | |
e6598ba4 | 2411 | int sindex, sh_style_split, whitesep, xflags; |
633e5c6d | 2412 | size_t slen; |
726f6388 JA |
2413 | |
2414 | if (!string || !*string) | |
2415 | return ((WORD_LIST *)NULL); | |
2416 | ||
7117c2d2 JA |
2417 | sh_style_split = separators && separators[0] == ' ' && |
2418 | separators[1] == '\t' && | |
2419 | separators[2] == '\n' && | |
2420 | separators[3] == '\0'; | |
e6598ba4 CR |
2421 | for (xflags = 0, s = ifs_value; s && *s; s++) |
2422 | { | |
2423 | if (*s == CTLESC) xflags |= SX_NOCTLESC; | |
2424 | else if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL; | |
2425 | } | |
726f6388 | 2426 | |
633e5c6d | 2427 | slen = 0; |
726f6388 JA |
2428 | /* Remove sequences of whitespace at the beginning of STRING, as |
2429 | long as those characters appear in IFS. Do not do this if | |
2430 | STRING is quoted or if there are no separator characters. */ | |
2431 | if (!quoted || !separators || !*separators) | |
2432 | { | |
2433 | for (s = string; *s && spctabnl (*s) && issep (*s); s++); | |
2434 | ||
2435 | if (!*s) | |
2436 | return ((WORD_LIST *)NULL); | |
2437 | ||
2438 | string = s; | |
2439 | } | |
2440 | ||
2441 | /* OK, now STRING points to a word that does not begin with white space. | |
2442 | The splitting algorithm is: | |
7117c2d2 JA |
2443 | extract a word, stopping at a separator |
2444 | skip sequences of spc, tab, or nl as long as they are separators | |
726f6388 | 2445 | This obeys the field splitting rules in Posix.2. */ |
da713c23 | 2446 | slen = (MB_CUR_MAX > 1) ? strlen (string) : 1; |
ccc6cda3 | 2447 | for (result = (WORD_LIST *)NULL, sindex = 0; string[sindex]; ) |
726f6388 | 2448 | { |
da713c23 CR |
2449 | /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim |
2450 | unless multibyte chars are possible. */ | |
e6598ba4 | 2451 | current_word = string_extract_verbatim (string, slen, &sindex, separators, xflags); |
ccc6cda3 | 2452 | if (current_word == 0) |
726f6388 JA |
2453 | break; |
2454 | ||
2455 | /* If we have a quoted empty string, add a quoted null argument. We | |
2456 | want to preserve the quoted null character iff this is a quoted | |
2457 | empty string; otherwise the quoted null characters are removed | |
2458 | below. */ | |
2459 | if (QUOTED_NULL (current_word)) | |
2460 | { | |
227f982e | 2461 | t = alloc_word_desc (); |
726f6388 | 2462 | t->word = make_quoted_char ('\0'); |
227f982e | 2463 | t->flags |= W_QUOTED|W_HASQUOTEDNULL; |
726f6388 JA |
2464 | result = make_word_list (t, result); |
2465 | } | |
ccc6cda3 | 2466 | else if (current_word[0] != '\0') |
726f6388 JA |
2467 | { |
2468 | /* If we have something, then add it regardless. However, | |
2469 | perform quoted null character removal on the current word. */ | |
2470 | remove_quoted_nulls (current_word); | |
cce855bc | 2471 | result = add_string_to_list (current_word, result); |
227f982e | 2472 | result->word->flags &= ~W_HASQUOTEDNULL; /* just to be sure */ |
ccc6cda3 JA |
2473 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) |
2474 | result->word->flags |= W_QUOTED; | |
726f6388 JA |
2475 | } |
2476 | ||
2477 | /* If we're not doing sequences of separators in the traditional | |
2478 | Bourne shell style, then add a quoted null argument. */ | |
726f6388 JA |
2479 | else if (!sh_style_split && !spctabnl (string[sindex])) |
2480 | { | |
227f982e | 2481 | t = alloc_word_desc (); |
ccc6cda3 | 2482 | t->word = make_quoted_char ('\0'); |
227f982e | 2483 | t->flags |= W_QUOTED|W_HASQUOTEDNULL; |
ccc6cda3 | 2484 | result = make_word_list (t, result); |
726f6388 JA |
2485 | } |
2486 | ||
2487 | free (current_word); | |
2488 | ||
28ef6c31 JA |
2489 | /* Note whether or not the separator is IFS whitespace, used later. */ |
2490 | whitesep = string[sindex] && spctabnl (string[sindex]); | |
2491 | ||
726f6388 JA |
2492 | /* Move past the current separator character. */ |
2493 | if (string[sindex]) | |
633e5c6d CR |
2494 | { |
2495 | DECLARE_MBSTATE; | |
633e5c6d CR |
2496 | ADVANCE_CHAR (string, slen, sindex); |
2497 | } | |
726f6388 JA |
2498 | |
2499 | /* Now skip sequences of space, tab, or newline characters if they are | |
2500 | in the list of separators. */ | |
2501 | while (string[sindex] && spctabnl (string[sindex]) && issep (string[sindex])) | |
2502 | sindex++; | |
28ef6c31 | 2503 | |
7117c2d2 JA |
2504 | /* If the first separator was IFS whitespace and the current character |
2505 | is a non-whitespace IFS character, it should be part of the current | |
2506 | field delimiter, not a separate delimiter that would result in an | |
2507 | empty field. Look at POSIX.2, 3.6.5, (3)(b). */ | |
28ef6c31 | 2508 | if (string[sindex] && whitesep && issep (string[sindex]) && !spctabnl (string[sindex])) |
9d2b70f0 CR |
2509 | { |
2510 | sindex++; | |
2511 | /* An IFS character that is not IFS white space, along with any | |
2512 | adjacent IFS white space, shall delimit a field. (SUSv3) */ | |
d3ad40de | 2513 | while (string[sindex] && spctabnl (string[sindex]) && isifs (string[sindex])) |
9d2b70f0 CR |
2514 | sindex++; |
2515 | } | |
726f6388 JA |
2516 | } |
2517 | return (REVERSE_LIST (result, WORD_LIST *)); | |
2518 | } | |
2519 | ||
2520 | /* Parse a single word from STRING, using SEPARATORS to separate fields. | |
2521 | ENDPTR is set to the first character after the word. This is used by | |
7117c2d2 JA |
2522 | the `read' builtin. This is never called with SEPARATORS != $IFS; |
2523 | it should be simplified. | |
2524 | ||
726f6388 JA |
2525 | XXX - this function is very similar to list_string; they should be |
2526 | combined - XXX */ | |
2527 | char * | |
2528 | get_word_from_string (stringp, separators, endptr) | |
2529 | char **stringp, *separators, **endptr; | |
2530 | { | |
2531 | register char *s; | |
2532 | char *current_word; | |
e6598ba4 | 2533 | int sindex, sh_style_split, whitesep, xflags; |
633e5c6d | 2534 | size_t slen; |
726f6388 JA |
2535 | |
2536 | if (!stringp || !*stringp || !**stringp) | |
2537 | return ((char *)NULL); | |
ccc6cda3 | 2538 | |
7117c2d2 JA |
2539 | sh_style_split = separators && separators[0] == ' ' && |
2540 | separators[1] == '\t' && | |
2541 | separators[2] == '\n' && | |
2542 | separators[3] == '\0'; | |
e6598ba4 CR |
2543 | for (xflags = 0, s = ifs_value; s && *s; s++) |
2544 | { | |
2545 | if (*s == CTLESC) xflags |= SX_NOCTLESC; | |
2546 | if (*s == CTLNUL) xflags |= SX_NOESCCTLNUL; | |
2547 | } | |
726f6388 | 2548 | |
e6598ba4 | 2549 | s = *stringp; |
633e5c6d CR |
2550 | slen = 0; |
2551 | ||
726f6388 JA |
2552 | /* Remove sequences of whitespace at the beginning of STRING, as |
2553 | long as those characters appear in IFS. */ | |
2554 | if (sh_style_split || !separators || !*separators) | |
2555 | { | |
7117c2d2 | 2556 | for (; *s && spctabnl (*s) && isifs (*s); s++); |
726f6388 JA |
2557 | |
2558 | /* If the string is nothing but whitespace, update it and return. */ | |
2559 | if (!*s) | |
2560 | { | |
2561 | *stringp = s; | |
2562 | if (endptr) | |
2563 | *endptr = s; | |
2564 | return ((char *)NULL); | |
2565 | } | |
2566 | } | |
2567 | ||
2568 | /* OK, S points to a word that does not begin with white space. | |
2569 | Now extract a word, stopping at a separator, save a pointer to | |
2570 | the first character after the word, then skip sequences of spc, | |
2571 | tab, or nl as long as they are separators. | |
ccc6cda3 | 2572 | |
726f6388 JA |
2573 | This obeys the field splitting rules in Posix.2. */ |
2574 | sindex = 0; | |
da713c23 CR |
2575 | /* Don't need string length in ADVANCE_CHAR or string_extract_verbatim |
2576 | unless multibyte chars are possible. */ | |
2577 | slen = (MB_CUR_MAX > 1) ? strlen (s) : 1; | |
e6598ba4 | 2578 | current_word = string_extract_verbatim (s, slen, &sindex, separators, xflags); |
726f6388 JA |
2579 | |
2580 | /* Set ENDPTR to the first character after the end of the word. */ | |
2581 | if (endptr) | |
2582 | *endptr = s + sindex; | |
2583 | ||
28ef6c31 JA |
2584 | /* Note whether or not the separator is IFS whitespace, used later. */ |
2585 | whitesep = s[sindex] && spctabnl (s[sindex]); | |
2586 | ||
726f6388 JA |
2587 | /* Move past the current separator character. */ |
2588 | if (s[sindex]) | |
633e5c6d CR |
2589 | { |
2590 | DECLARE_MBSTATE; | |
633e5c6d CR |
2591 | ADVANCE_CHAR (s, slen, sindex); |
2592 | } | |
726f6388 JA |
2593 | |
2594 | /* Now skip sequences of space, tab, or newline characters if they are | |
2595 | in the list of separators. */ | |
7117c2d2 | 2596 | while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex])) |
726f6388 JA |
2597 | sindex++; |
2598 | ||
28ef6c31 JA |
2599 | /* If the first separator was IFS whitespace and the current character is |
2600 | a non-whitespace IFS character, it should be part of the current field | |
2601 | delimiter, not a separate delimiter that would result in an empty field. | |
2602 | Look at POSIX.2, 3.6.5, (3)(b). */ | |
7117c2d2 | 2603 | if (s[sindex] && whitesep && isifs (s[sindex]) && !spctabnl (s[sindex])) |
9d2b70f0 CR |
2604 | { |
2605 | sindex++; | |
2606 | /* An IFS character that is not IFS white space, along with any adjacent | |
2607 | IFS white space, shall delimit a field. */ | |
2608 | while (s[sindex] && spctabnl (s[sindex]) && isifs (s[sindex])) | |
2609 | sindex++; | |
2610 | } | |
28ef6c31 | 2611 | |
726f6388 JA |
2612 | /* Update STRING to point to the next field. */ |
2613 | *stringp = s + sindex; | |
2614 | return (current_word); | |
2615 | } | |
2616 | ||
2617 | /* Remove IFS white space at the end of STRING. Start at the end | |
2618 | of the string and walk backwards until the beginning of the string | |
2619 | or we find a character that's not IFS white space and not CTLESC. | |
2620 | Only let CTLESC escape a white space character if SAW_ESCAPE is | |
2621 | non-zero. */ | |
2622 | char * | |
2623 | strip_trailing_ifs_whitespace (string, separators, saw_escape) | |
2624 | char *string, *separators; | |
2625 | int saw_escape; | |
2626 | { | |
2627 | char *s; | |
ccc6cda3 | 2628 | |
726f6388 | 2629 | s = string + STRLEN (string) - 1; |
7117c2d2 | 2630 | while (s > string && ((spctabnl (*s) && isifs (*s)) || |
726f6388 JA |
2631 | (saw_escape && *s == CTLESC && spctabnl (s[1])))) |
2632 | s--; | |
2633 | *++s = '\0'; | |
2634 | return string; | |
2635 | } | |
2636 | ||
bb70624e JA |
2637 | #if 0 |
2638 | /* UNUSED */ | |
2639 | /* Split STRING into words at whitespace. Obeys shell-style quoting with | |
2640 | backslashes, single and double quotes. */ | |
ccc6cda3 JA |
2641 | WORD_LIST * |
2642 | list_string_with_quotes (string) | |
2643 | char *string; | |
2644 | { | |
2645 | WORD_LIST *list; | |
2646 | char *token, *s; | |
7117c2d2 | 2647 | size_t s_len; |
ccc6cda3 JA |
2648 | int c, i, tokstart, len; |
2649 | ||
2650 | for (s = string; s && *s && spctabnl (*s); s++) | |
2651 | ; | |
2652 | if (s == 0 || *s == 0) | |
2653 | return ((WORD_LIST *)NULL); | |
2654 | ||
7117c2d2 | 2655 | s_len = strlen (s); |
ccc6cda3 JA |
2656 | tokstart = i = 0; |
2657 | list = (WORD_LIST *)NULL; | |
2658 | while (1) | |
2659 | { | |
2660 | c = s[i]; | |
2661 | if (c == '\\') | |
2662 | { | |
2663 | i++; | |
2664 | if (s[i]) | |
2665 | i++; | |
2666 | } | |
2667 | else if (c == '\'') | |
7117c2d2 | 2668 | i = skip_single_quoted (s, s_len, ++i); |
ccc6cda3 | 2669 | else if (c == '"') |
7117c2d2 | 2670 | i = skip_double_quoted (s, s_len, ++i); |
ccc6cda3 JA |
2671 | else if (c == 0 || spctabnl (c)) |
2672 | { | |
2673 | /* We have found the end of a token. Make a word out of it and | |
2674 | add it to the word list. */ | |
bb70624e | 2675 | token = substring (s, tokstart, i); |
cce855bc | 2676 | list = add_string_to_list (token, list); |
ccc6cda3 JA |
2677 | free (token); |
2678 | while (spctabnl (s[i])) | |
2679 | i++; | |
2680 | if (s[i]) | |
2681 | tokstart = i; | |
2682 | else | |
2683 | break; | |
2684 | } | |
2685 | else | |
2686 | i++; /* normal character */ | |
2687 | } | |
2688 | return (REVERSE_LIST (list, WORD_LIST *)); | |
2689 | } | |
bb70624e | 2690 | #endif |
d166f048 | 2691 | |
cce855bc JA |
2692 | /********************************************************/ |
2693 | /* */ | |
2694 | /* Functions to perform assignment statements */ | |
2695 | /* */ | |
2696 | /********************************************************/ | |
d166f048 | 2697 | |
43df7bbb CR |
2698 | #if defined (ARRAY_VARS) |
2699 | static SHELL_VAR * | |
d11b8b46 | 2700 | do_compound_assignment (name, value, flags) |
43df7bbb | 2701 | char *name, *value; |
d11b8b46 | 2702 | int flags; |
43df7bbb CR |
2703 | { |
2704 | SHELL_VAR *v; | |
09767ff0 | 2705 | int mklocal, mkassoc; |
d3ad40de | 2706 | WORD_LIST *list; |
d11b8b46 CR |
2707 | |
2708 | mklocal = flags & ASS_MKLOCAL; | |
09767ff0 | 2709 | mkassoc = flags & ASS_MKASSOC; |
43df7bbb CR |
2710 | |
2711 | if (mklocal && variable_context) | |
2712 | { | |
2713 | v = find_variable (name); | |
fdf670ea | 2714 | list = expand_compound_array_assignment (v, value, flags); |
09767ff0 CR |
2715 | if (mkassoc) |
2716 | v = make_local_assoc_variable (name); | |
2717 | else if (v == 0 || (array_p (v) == 0 && assoc_p (v) == 0) || v->context != variable_context) | |
7f947b68 | 2718 | v = make_local_array_variable (name, 0); |
d3ad40de | 2719 | assign_compound_array_list (v, list, flags); |
43df7bbb CR |
2720 | } |
2721 | else | |
d11b8b46 | 2722 | v = assign_array_from_string (name, value, flags); |
43df7bbb CR |
2723 | |
2724 | return (v); | |
2725 | } | |
2726 | #endif | |
2727 | ||
726f6388 JA |
2728 | /* Given STRING, an assignment string, get the value of the right side |
2729 | of the `=', and bind it to the left side. If EXPAND is true, then | |
2730 | perform parameter expansion, command substitution, and arithmetic | |
2731 | expansion on the right-hand side. Perform tilde expansion in any | |
2732 | case. Do not perform word splitting on the result of expansion. */ | |
2733 | static int | |
43df7bbb CR |
2734 | do_assignment_internal (word, expand) |
2735 | const WORD_DESC *word; | |
726f6388 JA |
2736 | int expand; |
2737 | { | |
3d8cce26 CR |
2738 | int offset, appendop, assign_list, aflags, retval; |
2739 | char *name, *value, *temp; | |
ccc6cda3 JA |
2740 | SHELL_VAR *entry; |
2741 | #if defined (ARRAY_VARS) | |
2742 | char *t; | |
1d7ecd77 | 2743 | int ni; |
ccc6cda3 | 2744 | #endif |
43df7bbb | 2745 | const char *string; |
ccc6cda3 | 2746 | |
43df7bbb CR |
2747 | if (word == 0 || word->word == 0) |
2748 | return 0; | |
2749 | ||
d11b8b46 | 2750 | appendop = assign_list = aflags = 0; |
43df7bbb | 2751 | string = word->word; |
5e13499c | 2752 | offset = assignment (string, 0); |
ccc6cda3 JA |
2753 | name = savestring (string); |
2754 | value = (char *)NULL; | |
726f6388 JA |
2755 | |
2756 | if (name[offset] == '=') | |
2757 | { | |
d11b8b46 CR |
2758 | if (name[offset - 1] == '+') |
2759 | { | |
2760 | appendop = 1; | |
2761 | name[offset - 1] = '\0'; | |
2762 | } | |
2763 | ||
2764 | name[offset] = 0; /* might need this set later */ | |
726f6388 JA |
2765 | temp = name + offset + 1; |
2766 | ||
ccc6cda3 | 2767 | #if defined (ARRAY_VARS) |
43df7bbb | 2768 | if (expand && (word->flags & W_COMPASSIGN)) |
726f6388 | 2769 | { |
ccc6cda3 | 2770 | assign_list = ni = 1; |
43df7bbb | 2771 | value = extract_array_assignment_list (temp, &ni); |
ccc6cda3 JA |
2772 | } |
2773 | else | |
2774 | #endif | |
ccc6cda3 | 2775 | if (expand && temp[0]) |
22e63b05 | 2776 | value = expand_string_if_necessary (temp, 0, expand_string_assignment); |
726f6388 JA |
2777 | else |
2778 | value = savestring (temp); | |
2779 | } | |
2780 | ||
2781 | if (value == 0) | |
d166f048 | 2782 | { |
f73dda09 | 2783 | value = (char *)xmalloc (1); |
d166f048 JA |
2784 | value[0] = '\0'; |
2785 | } | |
726f6388 | 2786 | |
726f6388 | 2787 | if (echo_command_at_execute) |
d11b8b46 CR |
2788 | { |
2789 | if (appendop) | |
2790 | name[offset - 1] = '+'; | |
2791 | xtrace_print_assignment (name, value, assign_list, 1); | |
2792 | if (appendop) | |
2793 | name[offset - 1] = '\0'; | |
2794 | } | |
726f6388 | 2795 | |
d166f048 | 2796 | #define ASSIGN_RETURN(r) do { FREE (value); free (name); return (r); } while (0) |
ccc6cda3 | 2797 | |
d11b8b46 CR |
2798 | if (appendop) |
2799 | aflags |= ASS_APPEND; | |
2800 | ||
ccc6cda3 | 2801 | #if defined (ARRAY_VARS) |
d0ca3503 | 2802 | if (t = mbschr (name, '[')) /*]*/ |
ccc6cda3 JA |
2803 | { |
2804 | if (assign_list) | |
2805 | { | |
5e13499c | 2806 | report_error (_("%s: cannot assign list to array member"), name); |
ccc6cda3 JA |
2807 | ASSIGN_RETURN (0); |
2808 | } | |
d11b8b46 | 2809 | entry = assign_array_element (name, value, aflags); |
ccc6cda3 | 2810 | if (entry == 0) |
28ef6c31 | 2811 | ASSIGN_RETURN (0); |
ccc6cda3 JA |
2812 | } |
2813 | else if (assign_list) | |
d11b8b46 | 2814 | { |
ec860d76 | 2815 | if ((word->flags & W_ASSIGNARG) && (word->flags & W_ASSNGLOBAL) == 0) |
d11b8b46 | 2816 | aflags |= ASS_MKLOCAL; |
09767ff0 CR |
2817 | if (word->flags & W_ASSIGNASSOC) |
2818 | aflags |= ASS_MKASSOC; | |
d11b8b46 CR |
2819 | entry = do_compound_assignment (name, value, aflags); |
2820 | } | |
ccc6cda3 JA |
2821 | else |
2822 | #endif /* ARRAY_VARS */ | |
d11b8b46 | 2823 | entry = bind_variable (name, value, aflags); |
ccc6cda3 | 2824 | |
726f6388 JA |
2825 | stupidly_hack_special_variables (name); |
2826 | ||
cb603128 CR |
2827 | #if 1 |
2828 | /* Return 1 if the assignment seems to have been performed correctly. */ | |
2829 | if (entry == 0 || readonly_p (entry)) | |
2830 | retval = 0; /* assignment failure */ | |
2831 | else if (noassign_p (entry)) | |
2832 | { | |
2833 | last_command_exit_value = EXECUTION_FAILURE; | |
2834 | retval = 1; /* error status, but not assignment failure */ | |
2835 | } | |
2836 | else | |
2837 | retval = 1; | |
cb603128 | 2838 | |
09767ff0 | 2839 | if (entry && retval != 0 && noassign_p (entry) == 0) |
cb603128 | 2840 | VUNSETATTR (entry, att_invisible); |
09767ff0 CR |
2841 | |
2842 | ASSIGN_RETURN (retval); | |
cb603128 | 2843 | #else |
726f6388 | 2844 | if (entry) |
bb70624e | 2845 | VUNSETATTR (entry, att_invisible); |
726f6388 | 2846 | |
28ef6c31 | 2847 | ASSIGN_RETURN (entry ? ((readonly_p (entry) == 0) && noassign_p (entry) == 0) : 0); |
cb603128 | 2848 | #endif |
726f6388 JA |
2849 | } |
2850 | ||
2851 | /* Perform the assignment statement in STRING, and expand the | |
43df7bbb | 2852 | right side by doing tilde, command and parameter expansion. */ |
ccc6cda3 | 2853 | int |
726f6388 | 2854 | do_assignment (string) |
43df7bbb | 2855 | char *string; |
726f6388 | 2856 | { |
43df7bbb CR |
2857 | WORD_DESC td; |
2858 | ||
2859 | td.flags = W_ASSIGNMENT; | |
2860 | td.word = string; | |
2861 | ||
2862 | return do_assignment_internal (&td, 1); | |
2863 | } | |
2864 | ||
2865 | int | |
adc6cff5 | 2866 | do_word_assignment (word, flags) |
43df7bbb | 2867 | WORD_DESC *word; |
adc6cff5 | 2868 | int flags; |
43df7bbb CR |
2869 | { |
2870 | return do_assignment_internal (word, 1); | |
726f6388 JA |
2871 | } |
2872 | ||
2873 | /* Given STRING, an assignment string, get the value of the right side | |
43df7bbb CR |
2874 | of the `=', and bind it to the left side. Do not perform any word |
2875 | expansions on the right hand side. */ | |
ccc6cda3 | 2876 | int |
726f6388 | 2877 | do_assignment_no_expand (string) |
43df7bbb | 2878 | char *string; |
726f6388 | 2879 | { |
43df7bbb CR |
2880 | WORD_DESC td; |
2881 | ||
2882 | td.flags = W_ASSIGNMENT; | |
2883 | td.word = string; | |
2884 | ||
e225d5a9 | 2885 | return (do_assignment_internal (&td, 0)); |
726f6388 JA |
2886 | } |
2887 | ||
cce855bc JA |
2888 | /*************************************************** |
2889 | * * | |
2890 | * Functions to manage the positional parameters * | |
2891 | * * | |
2892 | ***************************************************/ | |
726f6388 JA |
2893 | |
2894 | /* Return the word list that corresponds to `$*'. */ | |
2895 | WORD_LIST * | |
2896 | list_rest_of_args () | |
2897 | { | |
e3af9370 | 2898 | register WORD_LIST *list, *args; |
726f6388 JA |
2899 | int i; |
2900 | ||
2901 | /* Break out of the loop as soon as one of the dollar variables is null. */ | |
ccc6cda3 | 2902 | for (i = 1, list = (WORD_LIST *)NULL; i < 10 && dollar_vars[i]; i++) |
e3af9370 | 2903 | list = make_word_list (make_bare_word (dollar_vars[i]), list); |
ccc6cda3 JA |
2904 | |
2905 | for (args = rest_of_args; args; args = args->next) | |
e3af9370 | 2906 | list = make_word_list (make_bare_word (args->word->word), list); |
726f6388 | 2907 | |
e3af9370 | 2908 | return (REVERSE_LIST (list, WORD_LIST *)); |
726f6388 JA |
2909 | } |
2910 | ||
ccc6cda3 JA |
2911 | int |
2912 | number_of_args () | |
2913 | { | |
2914 | register WORD_LIST *list; | |
2915 | int n; | |
2916 | ||
2917 | for (n = 0; n < 9 && dollar_vars[n+1]; n++) | |
2918 | ; | |
2919 | for (list = rest_of_args; list; list = list->next) | |
2920 | n++; | |
2921 | return n; | |
2922 | } | |
2923 | ||
cce855bc JA |
2924 | /* Return the value of a positional parameter. This handles values > 10. */ |
2925 | char * | |
2926 | get_dollar_var_value (ind) | |
7117c2d2 | 2927 | intmax_t ind; |
cce855bc JA |
2928 | { |
2929 | char *temp; | |
2930 | WORD_LIST *p; | |
2931 | ||
2932 | if (ind < 10) | |
2933 | temp = dollar_vars[ind] ? savestring (dollar_vars[ind]) : (char *)NULL; | |
2934 | else /* We want something like ${11} */ | |
2935 | { | |
2936 | ind -= 10; | |
2937 | for (p = rest_of_args; p && ind--; p = p->next) | |
28ef6c31 | 2938 | ; |
cce855bc JA |
2939 | temp = p ? savestring (p->word->word) : (char *)NULL; |
2940 | } | |
2941 | return (temp); | |
2942 | } | |
2943 | ||
726f6388 JA |
2944 | /* Make a single large string out of the dollar digit variables, |
2945 | and the rest_of_args. If DOLLAR_STAR is 1, then obey the special | |
2946 | case of "$*" with respect to IFS. */ | |
2947 | char * | |
2948 | string_rest_of_args (dollar_star) | |
2949 | int dollar_star; | |
2950 | { | |
ccc6cda3 | 2951 | register WORD_LIST *list; |
726f6388 JA |
2952 | char *string; |
2953 | ||
ccc6cda3 | 2954 | list = list_rest_of_args (); |
726f6388 JA |
2955 | string = dollar_star ? string_list_dollar_star (list) : string_list (list); |
2956 | dispose_words (list); | |
2957 | return (string); | |
2958 | } | |
2959 | ||
cce855bc JA |
2960 | /* Return a string containing the positional parameters from START to |
2961 | END, inclusive. If STRING[0] == '*', we obey the rules for $*, | |
7117c2d2 JA |
2962 | which only makes a difference if QUOTED is non-zero. If QUOTED includes |
2963 | Q_HERE_DOCUMENT or Q_DOUBLE_QUOTES, this returns a quoted list, otherwise | |
2964 | no quoting chars are added. */ | |
cce855bc JA |
2965 | static char * |
2966 | pos_params (string, start, end, quoted) | |
2967 | char *string; | |
2968 | int start, end, quoted; | |
726f6388 | 2969 | { |
cce855bc JA |
2970 | WORD_LIST *save, *params, *h, *t; |
2971 | char *ret; | |
2972 | int i; | |
726f6388 | 2973 | |
bb70624e JA |
2974 | /* see if we can short-circuit. if start == end, we want 0 parameters. */ |
2975 | if (start == end) | |
2976 | return ((char *)NULL); | |
2977 | ||
cce855bc JA |
2978 | save = params = list_rest_of_args (); |
2979 | if (save == 0) | |
2980 | return ((char *)NULL); | |
2981 | ||
d3ad40de CR |
2982 | if (start == 0) /* handle ${@:0[:x]} specially */ |
2983 | { | |
2984 | t = make_word_list (make_word (dollar_vars[0]), params); | |
2985 | save = params = t; | |
2986 | } | |
2987 | ||
0527c903 | 2988 | for (i = start ? 1 : 0; params && i < start; i++) |
cce855bc JA |
2989 | params = params->next; |
2990 | if (params == 0) | |
2991 | return ((char *)NULL); | |
2992 | for (h = t = params; params && i < end; i++) | |
d166f048 | 2993 | { |
cce855bc JA |
2994 | t = params; |
2995 | params = params->next; | |
d166f048 | 2996 | } |
726f6388 | 2997 | |
cce855bc | 2998 | t->next = (WORD_LIST *)NULL; |
e33f2203 CR |
2999 | |
3000 | ret = string_list_pos_params (string[0], h, quoted); | |
3001 | ||
bb70624e JA |
3002 | if (t != params) |
3003 | t->next = params; | |
726f6388 | 3004 | |
cce855bc JA |
3005 | dispose_words (save); |
3006 | return (ret); | |
3007 | } | |
3008 | ||
3009 | /******************************************************************/ | |
3010 | /* */ | |
3011 | /* Functions to expand strings to strings or WORD_LISTs */ | |
3012 | /* */ | |
3013 | /******************************************************************/ | |
3014 | ||
3015 | #if defined (PROCESS_SUBSTITUTION) | |
22e63b05 | 3016 | #define EXP_CHAR(s) (s == '$' || s == '`' || s == '<' || s == '>' || s == CTLESC || s == '~') |
cce855bc | 3017 | #else |
22e63b05 | 3018 | #define EXP_CHAR(s) (s == '$' || s == '`' || s == CTLESC || s == '~') |
cce855bc JA |
3019 | #endif |
3020 | ||
3021 | /* If there are any characters in STRING that require full expansion, | |
3022 | then call FUNC to expand STRING; otherwise just perform quote | |
3023 | removal if necessary. This returns a new string. */ | |
3024 | static char * | |
f73dda09 | 3025 | expand_string_if_necessary (string, quoted, func) |
cce855bc JA |
3026 | char *string; |
3027 | int quoted; | |
f73dda09 | 3028 | EXPFUNC *func; |
cce855bc JA |
3029 | { |
3030 | WORD_LIST *list; | |
7117c2d2 | 3031 | size_t slen; |
cce855bc JA |
3032 | int i, saw_quote; |
3033 | char *ret; | |
7117c2d2 | 3034 | DECLARE_MBSTATE; |
cce855bc | 3035 | |
da713c23 CR |
3036 | /* Don't need string length for ADVANCE_CHAR unless multibyte chars possible. */ |
3037 | slen = (MB_CUR_MAX > 1) ? strlen (string) : 0; | |
7117c2d2 JA |
3038 | i = saw_quote = 0; |
3039 | while (string[i]) | |
cce855bc JA |
3040 | { |
3041 | if (EXP_CHAR (string[i])) | |
3042 | break; | |
3043 | else if (string[i] == '\'' || string[i] == '\\' || string[i] == '"') | |
3044 | saw_quote = 1; | |
7117c2d2 | 3045 | ADVANCE_CHAR (string, slen, i); |
cce855bc JA |
3046 | } |
3047 | ||
3048 | if (string[i]) | |
3049 | { | |
3050 | list = (*func) (string, quoted); | |
3051 | if (list) | |
3052 | { | |
3053 | ret = string_list (list); | |
3054 | dispose_words (list); | |
3055 | } | |
3056 | else | |
3057 | ret = (char *)NULL; | |
3058 | } | |
3059 | else if (saw_quote && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) | |
3060 | ret = string_quote_removal (string, quoted); | |
3061 | else | |
3062 | ret = savestring (string); | |
7117c2d2 | 3063 | |
cce855bc JA |
3064 | return ret; |
3065 | } | |
3066 | ||
3067 | static inline char * | |
f73dda09 | 3068 | expand_string_to_string_internal (string, quoted, func) |
cce855bc JA |
3069 | char *string; |
3070 | int quoted; | |
f73dda09 | 3071 | EXPFUNC *func; |
cce855bc JA |
3072 | { |
3073 | WORD_LIST *list; | |
3074 | char *ret; | |
3075 | ||
3076 | if (string == 0 || *string == '\0') | |
3077 | return ((char *)NULL); | |
3078 | ||
3079 | list = (*func) (string, quoted); | |
3080 | if (list) | |
3081 | { | |
3082 | ret = string_list (list); | |
3083 | dispose_words (list); | |
3084 | } | |
3085 | else | |
3086 | ret = (char *)NULL; | |
3087 | ||
3088 | return (ret); | |
3089 | } | |
3090 | ||
f73dda09 JA |
3091 | char * |
3092 | expand_string_to_string (string, quoted) | |
3093 | char *string; | |
3094 | int quoted; | |
3095 | { | |
3096 | return (expand_string_to_string_internal (string, quoted, expand_string)); | |
3097 | } | |
3098 | ||
3099 | char * | |
3100 | expand_string_unsplit_to_string (string, quoted) | |
3101 | char *string; | |
3102 | int quoted; | |
3103 | { | |
3104 | return (expand_string_to_string_internal (string, quoted, expand_string_unsplit)); | |
3105 | } | |
3106 | ||
43df7bbb CR |
3107 | char * |
3108 | expand_assignment_string_to_string (string, quoted) | |
3109 | char *string; | |
3110 | int quoted; | |
3111 | { | |
3112 | return (expand_string_to_string_internal (string, quoted, expand_string_assignment)); | |
3113 | } | |
3114 | ||
dc8fbaf9 | 3115 | char * |
d3ad40de | 3116 | expand_arith_string (string, quoted) |
dc8fbaf9 | 3117 | char *string; |
48ff5447 | 3118 | int quoted; |
dc8fbaf9 | 3119 | { |
d3ad40de | 3120 | return (expand_string_if_necessary (string, quoted, expand_string)); |
dc8fbaf9 CR |
3121 | } |
3122 | ||
cce855bc JA |
3123 | #if defined (COND_COMMAND) |
3124 | /* Just remove backslashes in STRING. Returns a new string. */ | |
3125 | char * | |
3126 | remove_backslashes (string) | |
3127 | char *string; | |
3128 | { | |
3129 | char *r, *ret, *s; | |
3130 | ||
f73dda09 | 3131 | r = ret = (char *)xmalloc (strlen (string) + 1); |
cce855bc JA |
3132 | for (s = string; s && *s; ) |
3133 | { | |
3134 | if (*s == '\\') | |
28ef6c31 | 3135 | s++; |
cce855bc | 3136 | if (*s == 0) |
28ef6c31 | 3137 | break; |
cce855bc JA |
3138 | *r++ = *s++; |
3139 | } | |
3140 | *r = '\0'; | |
3141 | return ret; | |
3142 | } | |
3143 | ||
3144 | /* This needs better error handling. */ | |
3145 | /* Expand W for use as an argument to a unary or binary operator in a | |
d3ad40de | 3146 | [[...]] expression. If SPECIAL is 1, this is the rhs argument |
cce855bc | 3147 | to the != or == operator, and should be treated as a pattern. In |
d3ad40de CR |
3148 | this case, we quote the string specially for the globbing code. If |
3149 | SPECIAL is 2, this is an rhs argument for the =~ operator, and should | |
3150 | be quoted appropriately for regcomp/regexec. The caller is responsible | |
3151 | for removing the backslashes if the unquoted word is needed later. */ | |
cce855bc JA |
3152 | char * |
3153 | cond_expand_word (w, special) | |
3154 | WORD_DESC *w; | |
3155 | int special; | |
3156 | { | |
3157 | char *r, *p; | |
3158 | WORD_LIST *l; | |
d3ad40de | 3159 | int qflags; |
cce855bc JA |
3160 | |
3161 | if (w->word == 0 || w->word[0] == '\0') | |
3162 | return ((char *)NULL); | |
3163 | ||
e1e48bba | 3164 | w->flags |= W_NOSPLIT2; |
b72432fd | 3165 | l = call_expand_word_internal (w, 0, 0, (int *)0, (int *)0); |
cce855bc JA |
3166 | if (l) |
3167 | { | |
3168 | if (special == 0) | |
3169 | { | |
3170 | dequote_list (l); | |
3171 | r = string_list (l); | |
3172 | } | |
3173 | else | |
28ef6c31 | 3174 | { |
d3ad40de CR |
3175 | qflags = QGLOB_CVTNULL; |
3176 | if (special == 2) | |
3177 | qflags |= QGLOB_REGEXP; | |
28ef6c31 | 3178 | p = string_list (l); |
d3ad40de | 3179 | r = quote_string_for_globbing (p, qflags); |
28ef6c31 JA |
3180 | free (p); |
3181 | } | |
cce855bc JA |
3182 | dispose_words (l); |
3183 | } | |
3184 | else | |
3185 | r = (char *)NULL; | |
3186 | ||
3187 | return r; | |
3188 | } | |
3189 | #endif | |
3190 | ||
3191 | /* Call expand_word_internal to expand W and handle error returns. | |
3192 | A convenience function for functions that don't want to handle | |
3193 | any errors or free any memory before aborting. */ | |
3194 | static WORD_LIST * | |
b72432fd | 3195 | call_expand_word_internal (w, q, i, c, e) |
cce855bc | 3196 | WORD_DESC *w; |
b72432fd | 3197 | int q, i, *c, *e; |
cce855bc JA |
3198 | { |
3199 | WORD_LIST *result; | |
3200 | ||
b72432fd | 3201 | result = expand_word_internal (w, q, i, c, e); |
bb70624e | 3202 | if (result == &expand_word_error || result == &expand_word_fatal) |
cce855bc JA |
3203 | { |
3204 | /* By convention, each time this error is returned, w->word has | |
bb70624e JA |
3205 | already been freed (it sometimes may not be in the fatal case, |
3206 | but that doesn't result in a memory leak because we're going | |
3207 | to exit in most cases). */ | |
cce855bc | 3208 | w->word = (char *)NULL; |
28ef6c31 | 3209 | last_command_exit_value = EXECUTION_FAILURE; |
d3a24ed2 | 3210 | exp_jump_to_top_level ((result == &expand_word_error) ? DISCARD : FORCE_EOF); |
cce855bc JA |
3211 | /* NOTREACHED */ |
3212 | } | |
cce855bc JA |
3213 | else |
3214 | return (result); | |
3215 | } | |
3216 | ||
3217 | /* Perform parameter expansion, command substitution, and arithmetic | |
9c2db999 CR |
3218 | expansion on STRING, as if it were a word. Leave the result quoted. |
3219 | Since this does not perform word splitting, it leaves quoted nulls | |
3220 | in the result. */ | |
cce855bc JA |
3221 | static WORD_LIST * |
3222 | expand_string_internal (string, quoted) | |
3223 | char *string; | |
3224 | int quoted; | |
3225 | { | |
3226 | WORD_DESC td; | |
3227 | WORD_LIST *tresult; | |
3228 | ||
3229 | if (string == 0 || *string == 0) | |
3230 | return ((WORD_LIST *)NULL); | |
3231 | ||
28ef6c31 JA |
3232 | td.flags = 0; |
3233 | td.word = savestring (string); | |
3234 | ||
b72432fd | 3235 | tresult = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); |
28ef6c31 JA |
3236 | |
3237 | FREE (td.word); | |
cce855bc | 3238 | return (tresult); |
726f6388 JA |
3239 | } |
3240 | ||
3241 | /* Expand STRING by performing parameter expansion, command substitution, | |
3242 | and arithmetic expansion. Dequote the resulting WORD_LIST before | |
3243 | returning it, but do not perform word splitting. The call to | |
3244 | remove_quoted_nulls () is in here because word splitting normally | |
3245 | takes care of quote removal. */ | |
3246 | WORD_LIST * | |
3247 | expand_string_unsplit (string, quoted) | |
3248 | char *string; | |
3249 | int quoted; | |
3250 | { | |
3251 | WORD_LIST *value; | |
3252 | ||
28ef6c31 | 3253 | if (string == 0 || *string == '\0') |
726f6388 JA |
3254 | return ((WORD_LIST *)NULL); |
3255 | ||
28ef6c31 | 3256 | expand_no_split_dollar_star = 1; |
726f6388 | 3257 | value = expand_string_internal (string, quoted); |
28ef6c31 JA |
3258 | expand_no_split_dollar_star = 0; |
3259 | ||
726f6388 JA |
3260 | if (value) |
3261 | { | |
3262 | if (value->word) | |
227f982e CR |
3263 | { |
3264 | remove_quoted_nulls (value->word->word); | |
3265 | value->word->flags &= ~W_HASQUOTEDNULL; | |
3266 | } | |
726f6388 JA |
3267 | dequote_list (value); |
3268 | } | |
3269 | return (value); | |
3270 | } | |
3271 | ||
22e63b05 CR |
3272 | /* Expand the rhs of an assignment statement */ |
3273 | WORD_LIST * | |
3274 | expand_string_assignment (string, quoted) | |
3275 | char *string; | |
3276 | int quoted; | |
3277 | { | |
3278 | WORD_DESC td; | |
3279 | WORD_LIST *value; | |
3280 | ||
3281 | if (string == 0 || *string == '\0') | |
3282 | return ((WORD_LIST *)NULL); | |
3283 | ||
3284 | expand_no_split_dollar_star = 1; | |
3285 | ||
3286 | td.flags = W_ASSIGNRHS; | |
3287 | td.word = savestring (string); | |
3288 | value = call_expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); | |
3289 | FREE (td.word); | |
3290 | ||
3291 | expand_no_split_dollar_star = 0; | |
3292 | ||
3293 | if (value) | |
3294 | { | |
3295 | if (value->word) | |
227f982e CR |
3296 | { |
3297 | remove_quoted_nulls (value->word->word); | |
3298 | value->word->flags &= ~W_HASQUOTEDNULL; | |
3299 | } | |
22e63b05 CR |
3300 | dequote_list (value); |
3301 | } | |
3302 | return (value); | |
3303 | } | |
3304 | ||
bb70624e JA |
3305 | |
3306 | /* Expand one of the PS? prompt strings. This is a sort of combination of | |
3307 | expand_string_unsplit and expand_string_internal, but returns the | |
3308 | passed string when an error occurs. Might want to trap other calls | |
3309 | to jump_to_top_level here so we don't endlessly loop. */ | |
3310 | WORD_LIST * | |
4d8d005b | 3311 | expand_prompt_string (string, quoted, wflags) |
bb70624e JA |
3312 | char *string; |
3313 | int quoted; | |
4d8d005b | 3314 | int wflags; |
bb70624e JA |
3315 | { |
3316 | WORD_LIST *value; | |
3317 | WORD_DESC td; | |
3318 | ||
3319 | if (string == 0 || *string == 0) | |
3320 | return ((WORD_LIST *)NULL); | |
3321 | ||
4d8d005b | 3322 | td.flags = wflags; |
bb70624e | 3323 | td.word = savestring (string); |
28ef6c31 JA |
3324 | |
3325 | no_longjmp_on_fatal_error = 1; | |
bb70624e | 3326 | value = expand_word_internal (&td, quoted, 0, (int *)NULL, (int *)NULL); |
28ef6c31 JA |
3327 | no_longjmp_on_fatal_error = 0; |
3328 | ||
bb70624e JA |
3329 | if (value == &expand_word_error || value == &expand_word_fatal) |
3330 | { | |
3331 | value = make_word_list (make_bare_word (string), (WORD_LIST *)NULL); | |
3332 | return value; | |
3333 | } | |
3334 | FREE (td.word); | |
3335 | if (value) | |
3336 | { | |
3337 | if (value->word) | |
227f982e CR |
3338 | { |
3339 | remove_quoted_nulls (value->word->word); | |
3340 | value->word->flags &= ~W_HASQUOTEDNULL; | |
3341 | } | |
bb70624e JA |
3342 | dequote_list (value); |
3343 | } | |
3344 | return (value); | |
3345 | } | |
3346 | ||
726f6388 JA |
3347 | /* Expand STRING just as if you were expanding a word, but do not dequote |
3348 | the resultant WORD_LIST. This is called only from within this file, | |
3349 | and is used to correctly preserve quoted characters when expanding | |
3350 | things like ${1+"$@"}. This does parameter expansion, command | |
b72432fd | 3351 | substitution, arithmetic expansion, and word splitting. */ |
726f6388 JA |
3352 | static WORD_LIST * |
3353 | expand_string_leave_quoted (string, quoted) | |
3354 | char *string; | |
3355 | int quoted; | |
3356 | { | |
3357 | WORD_LIST *tlist; | |
3358 | WORD_LIST *tresult; | |
3359 | ||
ccc6cda3 | 3360 | if (string == 0 || *string == '\0') |
726f6388 JA |
3361 | return ((WORD_LIST *)NULL); |
3362 | ||
3363 | tlist = expand_string_internal (string, quoted); | |
3364 | ||
3365 | if (tlist) | |
3366 | { | |
3367 | tresult = word_list_split (tlist); | |
3368 | dispose_words (tlist); | |
3369 | return (tresult); | |
3370 | } | |
3371 | return ((WORD_LIST *)NULL); | |
3372 | } | |
3373 | ||
ccc6cda3 JA |
3374 | /* This does not perform word splitting or dequote the WORD_LIST |
3375 | it returns. */ | |
3376 | static WORD_LIST * | |
3377 | expand_string_for_rhs (string, quoted, dollar_at_p, has_dollar_at) | |
3378 | char *string; | |
3379 | int quoted, *dollar_at_p, *has_dollar_at; | |
3380 | { | |
3381 | WORD_DESC td; | |
3382 | WORD_LIST *tresult; | |
3383 | ||
3384 | if (string == 0 || *string == '\0') | |
3385 | return (WORD_LIST *)NULL; | |
3386 | ||
9c2db999 | 3387 | td.flags = W_NOSPLIT2; /* no splitting, remove "" and '' */ |
ccc6cda3 | 3388 | td.word = string; |
b72432fd | 3389 | tresult = call_expand_word_internal (&td, quoted, 1, dollar_at_p, has_dollar_at); |
ccc6cda3 JA |
3390 | return (tresult); |
3391 | } | |
3392 | ||
726f6388 JA |
3393 | /* Expand STRING just as if you were expanding a word. This also returns |
3394 | a list of words. Note that filename globbing is *NOT* done for word | |
3395 | or string expansion, just when the shell is expanding a command. This | |
3396 | does parameter expansion, command substitution, arithmetic expansion, | |
3397 | and word splitting. Dequote the resultant WORD_LIST before returning. */ | |
3398 | WORD_LIST * | |
3399 | expand_string (string, quoted) | |
3400 | char *string; | |
3401 | int quoted; | |
3402 | { | |
3403 | WORD_LIST *result; | |
3404 | ||
28ef6c31 | 3405 | if (string == 0 || *string == '\0') |
726f6388 JA |
3406 | return ((WORD_LIST *)NULL); |
3407 | ||
3408 | result = expand_string_leave_quoted (string, quoted); | |
ccc6cda3 | 3409 | return (result ? dequote_list (result) : result); |
726f6388 JA |
3410 | } |
3411 | ||
3412 | /*************************************************** | |
3413 | * * | |
3414 | * Functions to handle quoting chars * | |
3415 | * * | |
3416 | ***************************************************/ | |
3417 | ||
cce855bc JA |
3418 | /* Conventions: |
3419 | ||
3420 | A string with s[0] == CTLNUL && s[1] == 0 is a quoted null string. | |
3421 | The parser passes CTLNUL as CTLESC CTLNUL. */ | |
3422 | ||
cce855bc JA |
3423 | /* Quote escape characters in string s, but no other characters. This is |
3424 | used to protect CTLESC and CTLNUL in variable values from the rest of | |
e6598ba4 CR |
3425 | the word expansion process after the variable is expanded (word splitting |
3426 | and filename generation). If IFS is null, we quote spaces as well, just | |
3427 | in case we split on spaces later (in the case of unquoted $@, we will | |
3428 | eventually attempt to split the entire word on spaces). Corresponding | |
3429 | code exists in dequote_escapes. Even if we don't end up splitting on | |
3430 | spaces, quoting spaces is not a problem. This should never be called on | |
3431 | a string that is quoted with single or double quotes or part of a here | |
3432 | document (effectively double-quoted). */ | |
f73dda09 | 3433 | char * |
cce855bc JA |
3434 | quote_escapes (string) |
3435 | char *string; | |
3436 | { | |
3437 | register char *s, *t; | |
7117c2d2 JA |
3438 | size_t slen; |
3439 | char *result, *send; | |
e6598ba4 | 3440 | int quote_spaces, skip_ctlesc, skip_ctlnul; |
7117c2d2 | 3441 | DECLARE_MBSTATE; |
cce855bc | 3442 | |
7117c2d2 JA |
3443 | slen = strlen (string); |
3444 | send = string + slen; | |
3445 | ||
d3ad40de | 3446 | quote_spaces = (ifs_value && *ifs_value == 0); |
e6598ba4 CR |
3447 | |
3448 | for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++) | |
3449 | skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL; | |
3450 | ||
7117c2d2 JA |
3451 | t = result = (char *)xmalloc ((slen * 2) + 1); |
3452 | s = string; | |
3453 | ||
3454 | while (*s) | |
cce855bc | 3455 | { |
e6598ba4 | 3456 | if ((skip_ctlesc == 0 && *s == CTLESC) || (skip_ctlnul == 0 && *s == CTLNUL) || (quote_spaces && *s == ' ')) |
cce855bc | 3457 | *t++ = CTLESC; |
7117c2d2 | 3458 | COPY_CHAR_P (t, s, send); |
cce855bc JA |
3459 | } |
3460 | *t = '\0'; | |
3461 | return (result); | |
3462 | } | |
3463 | ||
3464 | static WORD_LIST * | |
3465 | list_quote_escapes (list) | |
3466 | WORD_LIST *list; | |
3467 | { | |
3468 | register WORD_LIST *w; | |
3469 | char *t; | |
3470 | ||
3471 | for (w = list; w; w = w->next) | |
3472 | { | |
3473 | t = w->word->word; | |
3474 | w->word->word = quote_escapes (t); | |
3475 | free (t); | |
3476 | } | |
3477 | return list; | |
3478 | } | |
3479 | ||
7117c2d2 JA |
3480 | /* Inverse of quote_escapes; remove CTLESC protecting CTLESC or CTLNUL. |
3481 | ||
3482 | The parser passes us CTLESC as CTLESC CTLESC and CTLNUL as CTLESC CTLNUL. | |
3483 | This is necessary to make unquoted CTLESC and CTLNUL characters in the | |
3484 | data stream pass through properly. | |
3485 | ||
3486 | We need to remove doubled CTLESC characters inside quoted strings before | |
3487 | quoting the entire string, so we do not double the number of CTLESC | |
3488 | characters. | |
3489 | ||
3490 | Also used by parts of the pattern substitution code. */ | |
09767ff0 | 3491 | char * |
cce855bc JA |
3492 | dequote_escapes (string) |
3493 | char *string; | |
3494 | { | |
e6598ba4 | 3495 | register char *s, *t, *s1; |
7117c2d2 JA |
3496 | size_t slen; |
3497 | char *result, *send; | |
d3ad40de | 3498 | int quote_spaces; |
7117c2d2 | 3499 | DECLARE_MBSTATE; |
cce855bc | 3500 | |
7117c2d2 JA |
3501 | if (string == 0) |
3502 | return string; | |
3503 | ||
3504 | slen = strlen (string); | |
3505 | send = string + slen; | |
3506 | ||
3507 | t = result = (char *)xmalloc (slen + 1); | |
7117c2d2 JA |
3508 | |
3509 | if (strchr (string, CTLESC) == 0) | |
e6598ba4 | 3510 | return (strcpy (result, string)); |
7117c2d2 | 3511 | |
d3ad40de | 3512 | quote_spaces = (ifs_value && *ifs_value == 0); |
e6598ba4 CR |
3513 | |
3514 | s = string; | |
7117c2d2 | 3515 | while (*s) |
cce855bc | 3516 | { |
d3ad40de | 3517 | if (*s == CTLESC && (s[1] == CTLESC || s[1] == CTLNUL || (quote_spaces && s[1] == ' '))) |
cce855bc JA |
3518 | { |
3519 | s++; | |
3520 | if (*s == '\0') | |
3521 | break; | |
3522 | } | |
7117c2d2 | 3523 | COPY_CHAR_P (t, s, send); |
cce855bc JA |
3524 | } |
3525 | *t = '\0'; | |
3526 | return result; | |
3527 | } | |
726f6388 | 3528 | |
d3ad40de CR |
3529 | /* Return a new string with the quoted representation of character C. |
3530 | This turns "" into QUOTED_NULL, so the W_HASQUOTEDNULL flag needs to be | |
3531 | set in any resultant WORD_DESC where this value is the word. */ | |
726f6388 JA |
3532 | static char * |
3533 | make_quoted_char (c) | |
3534 | int c; | |
3535 | { | |
3536 | char *temp; | |
3537 | ||
f73dda09 | 3538 | temp = (char *)xmalloc (3); |
726f6388 JA |
3539 | if (c == 0) |
3540 | { | |
3541 | temp[0] = CTLNUL; | |
3542 | temp[1] = '\0'; | |
3543 | } | |
3544 | else | |
3545 | { | |
3546 | temp[0] = CTLESC; | |
3547 | temp[1] = c; | |
3548 | temp[2] = '\0'; | |
3549 | } | |
3550 | return (temp); | |
3551 | } | |
3552 | ||
d3ad40de CR |
3553 | /* Quote STRING, returning a new string. This turns "" into QUOTED_NULL, so |
3554 | the W_HASQUOTEDNULL flag needs to be set in any resultant WORD_DESC where | |
3555 | this value is the word. */ | |
ccc6cda3 | 3556 | char * |
726f6388 JA |
3557 | quote_string (string) |
3558 | char *string; | |
3559 | { | |
ccc6cda3 | 3560 | register char *t; |
7117c2d2 JA |
3561 | size_t slen; |
3562 | char *result, *send; | |
726f6388 | 3563 | |
ccc6cda3 | 3564 | if (*string == 0) |
726f6388 | 3565 | { |
f73dda09 | 3566 | result = (char *)xmalloc (2); |
726f6388 JA |
3567 | result[0] = CTLNUL; |
3568 | result[1] = '\0'; | |
3569 | } | |
3570 | else | |
3571 | { | |
7117c2d2 | 3572 | DECLARE_MBSTATE; |
726f6388 | 3573 | |
7117c2d2 JA |
3574 | slen = strlen (string); |
3575 | send = string + slen; | |
3576 | ||
3577 | result = (char *)xmalloc ((slen * 2) + 1); | |
3578 | ||
3579 | for (t = result; string < send; ) | |
726f6388 JA |
3580 | { |
3581 | *t++ = CTLESC; | |
7117c2d2 | 3582 | COPY_CHAR_P (t, string, send); |
726f6388 JA |
3583 | } |
3584 | *t = '\0'; | |
3585 | } | |
3586 | return (result); | |
3587 | } | |
3588 | ||
d3ad40de | 3589 | /* De-quote quoted characters in STRING. */ |
726f6388 JA |
3590 | char * |
3591 | dequote_string (string) | |
3592 | char *string; | |
3593 | { | |
7117c2d2 JA |
3594 | register char *s, *t; |
3595 | size_t slen; | |
3596 | char *result, *send; | |
3597 | DECLARE_MBSTATE; | |
726f6388 | 3598 | |
7117c2d2 JA |
3599 | slen = strlen (string); |
3600 | ||
3601 | t = result = (char *)xmalloc (slen + 1); | |
726f6388 JA |
3602 | |
3603 | if (QUOTED_NULL (string)) | |
3604 | { | |
3605 | result[0] = '\0'; | |
3606 | return (result); | |
3607 | } | |
3608 | ||
3609 | /* If no character in the string can be quoted, don't bother examining | |
3610 | each character. Just return a copy of the string passed to us. */ | |
7117c2d2 JA |
3611 | if (strchr (string, CTLESC) == NULL) |
3612 | return (strcpy (result, string)); | |
726f6388 | 3613 | |
7117c2d2 JA |
3614 | send = string + slen; |
3615 | s = string; | |
3616 | while (*s) | |
726f6388 | 3617 | { |
7117c2d2 | 3618 | if (*s == CTLESC) |
726f6388 | 3619 | { |
7117c2d2 JA |
3620 | s++; |
3621 | if (*s == '\0') | |
726f6388 JA |
3622 | break; |
3623 | } | |
7117c2d2 | 3624 | COPY_CHAR_P (t, s, send); |
726f6388 JA |
3625 | } |
3626 | ||
3627 | *t = '\0'; | |
3628 | return (result); | |
3629 | } | |
3630 | ||
3631 | /* Quote the entire WORD_LIST list. */ | |
ccc6cda3 | 3632 | static WORD_LIST * |
726f6388 JA |
3633 | quote_list (list) |
3634 | WORD_LIST *list; | |
3635 | { | |
3636 | register WORD_LIST *w; | |
ccc6cda3 | 3637 | char *t; |
726f6388 JA |
3638 | |
3639 | for (w = list; w; w = w->next) | |
3640 | { | |
ccc6cda3 | 3641 | t = w->word->word; |
726f6388 | 3642 | w->word->word = quote_string (t); |
e33f2203 CR |
3643 | if (*t == 0) |
3644 | w->word->flags |= W_HASQUOTEDNULL; /* XXX - turn on W_HASQUOTEDNULL here? */ | |
ccc6cda3 | 3645 | w->word->flags |= W_QUOTED; |
e33f2203 | 3646 | free (t); |
726f6388 | 3647 | } |
ccc6cda3 | 3648 | return list; |
726f6388 JA |
3649 | } |
3650 | ||
d3ad40de CR |
3651 | /* De-quote quoted characters in each word in LIST. */ |
3652 | WORD_LIST * | |
7117c2d2 JA |
3653 | dequote_list (list) |
3654 | WORD_LIST *list; | |
3655 | { | |
3656 | register char *s; | |
3657 | register WORD_LIST *tlist; | |
3658 | ||
3659 | for (tlist = list; tlist; tlist = tlist->next) | |
3660 | { | |
3661 | s = dequote_string (tlist->word->word); | |
e33f2203 CR |
3662 | if (QUOTED_NULL (tlist->word->word)) |
3663 | tlist->word->flags &= ~W_HASQUOTEDNULL; | |
7117c2d2 JA |
3664 | free (tlist->word->word); |
3665 | tlist->word->word = s; | |
3666 | } | |
3667 | return list; | |
3668 | } | |
3669 | ||
3670 | /* Remove CTLESC protecting a CTLESC or CTLNUL in place. Return the passed | |
3671 | string. */ | |
b0c16657 | 3672 | char * |
7117c2d2 JA |
3673 | remove_quoted_escapes (string) |
3674 | char *string; | |
3675 | { | |
3676 | char *t; | |
3677 | ||
3678 | if (string) | |
3679 | { | |
3680 | t = dequote_escapes (string); | |
3681 | strcpy (string, t); | |
3682 | free (t); | |
3683 | } | |
3684 | ||
3685 | return (string); | |
3686 | } | |
3687 | ||
cce855bc JA |
3688 | /* Perform quoted null character removal on STRING. We don't allow any |
3689 | quoted null characters in the middle or at the ends of strings because | |
3690 | of how expand_word_internal works. remove_quoted_nulls () turns | |
3691 | STRING into an empty string iff it only consists of a quoted null, | |
3692 | and removes all unquoted CTLNUL characters. */ | |
09767ff0 | 3693 | char * |
cce855bc JA |
3694 | remove_quoted_nulls (string) |
3695 | char *string; | |
3696 | { | |
7117c2d2 JA |
3697 | register size_t slen; |
3698 | register int i, j, prev_i; | |
3699 | DECLARE_MBSTATE; | |
3700 | ||
3701 | if (strchr (string, CTLNUL) == 0) /* XXX */ | |
3702 | return string; /* XXX */ | |
3703 | ||
3704 | slen = strlen (string); | |
3705 | i = j = 0; | |
3706 | ||
3707 | while (i < slen) | |
3708 | { | |
3709 | if (string[i] == CTLESC) | |
704a1a2a | 3710 | { |
d3a24ed2 CR |
3711 | /* Old code had j++, but we cannot assume that i == j at this |
3712 | point -- what if a CTLNUL has already been removed from the | |
3713 | string? We don't want to drop the CTLESC or recopy characters | |
3714 | that we've already copied down. */ | |
3715 | i++; string[j++] = CTLESC; | |
704a1a2a CR |
3716 | if (i == slen) |
3717 | break; | |
3718 | } | |
7117c2d2 | 3719 | else if (string[i] == CTLNUL) |
f4f5e1c2 CR |
3720 | { |
3721 | i++; | |
3722 | continue; | |
3723 | } | |
7117c2d2 JA |
3724 | |
3725 | prev_i = i; | |
3726 | ADVANCE_CHAR (string, slen, i); | |
3727 | if (j < prev_i) | |
cce855bc | 3728 | { |
7117c2d2 | 3729 | do string[j++] = string[prev_i++]; while (prev_i < i); |
cce855bc | 3730 | } |
7117c2d2 JA |
3731 | else |
3732 | j = i; | |
cce855bc | 3733 | } |
7117c2d2 JA |
3734 | string[j] = '\0'; |
3735 | ||
3736 | return (string); | |
cce855bc JA |
3737 | } |
3738 | ||
3739 | /* Perform quoted null character removal on each element of LIST. | |
3740 | This modifies LIST. */ | |
3741 | void | |
3742 | word_list_remove_quoted_nulls (list) | |
3743 | WORD_LIST *list; | |
3744 | { | |
3745 | register WORD_LIST *t; | |
3746 | ||
3747 | for (t = list; t; t = t->next) | |
227f982e CR |
3748 | { |
3749 | remove_quoted_nulls (t->word->word); | |
3750 | t->word->flags &= ~W_HASQUOTEDNULL; | |
3751 | } | |
cce855bc JA |
3752 | } |
3753 | ||
3754 | /* **************************************************************** */ | |
3755 | /* */ | |
3756 | /* Functions for Matching and Removing Patterns */ | |
3757 | /* */ | |
3758 | /* **************************************************************** */ | |
3759 | ||
545f34cf | 3760 | #if defined (HANDLE_MULTIBYTE) |
704a1a2a | 3761 | #if 0 /* Currently unused */ |
545f34cf CR |
3762 | static unsigned char * |
3763 | mb_getcharlens (string, len) | |
3764 | char *string; | |
3765 | int len; | |
3766 | { | |
704a1a2a CR |
3767 | int i, offset, last; |
3768 | unsigned char *ret; | |
545f34cf CR |
3769 | char *p; |
3770 | DECLARE_MBSTATE; | |
3771 | ||
3772 | i = offset = 0; | |
3773 | last = 0; | |
3774 | ret = (unsigned char *)xmalloc (len); | |
3775 | memset (ret, 0, len); | |
3776 | while (string[last]) | |
3777 | { | |
3778 | ADVANCE_CHAR (string, len, offset); | |
3779 | ret[last] = offset - last; | |
3780 | last = offset; | |
3781 | } | |
3782 | return ret; | |
3783 | } | |
3784 | #endif | |
704a1a2a | 3785 | #endif |
545f34cf | 3786 | |
cce855bc JA |
3787 | /* Remove the portion of PARAM matched by PATTERN according to OP, where OP |
3788 | can have one of 4 values: | |
3789 | RP_LONG_LEFT remove longest matching portion at start of PARAM | |
726f6388 JA |
3790 | RP_SHORT_LEFT remove shortest matching portion at start of PARAM |
3791 | RP_LONG_RIGHT remove longest matching portion at end of PARAM | |
3792 | RP_SHORT_RIGHT remove shortest matching portion at end of PARAM | |
3793 | */ | |
3794 | ||
3795 | #define RP_LONG_LEFT 1 | |
3796 | #define RP_SHORT_LEFT 2 | |
3797 | #define RP_LONG_RIGHT 3 | |
3798 | #define RP_SHORT_RIGHT 4 | |
3799 | ||
3d35553a | 3800 | /* Returns its first argument if nothing matched; new memory otherwise */ |
726f6388 | 3801 | static char * |
704a1a2a | 3802 | remove_upattern (param, pattern, op) |
726f6388 JA |
3803 | char *param, *pattern; |
3804 | int op; | |
3805 | { | |
ccc6cda3 JA |
3806 | register int len; |
3807 | register char *end; | |
726f6388 | 3808 | register char *p, *ret, c; |
726f6388 | 3809 | |
ccc6cda3 JA |
3810 | len = STRLEN (param); |
3811 | end = param + len; | |
726f6388 JA |
3812 | |
3813 | switch (op) | |
3814 | { | |
3815 | case RP_LONG_LEFT: /* remove longest match at start */ | |
704a1a2a | 3816 | for (p = end; p >= param; p--) |
726f6388 JA |
3817 | { |
3818 | c = *p; *p = '\0'; | |
f73dda09 | 3819 | if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
726f6388 JA |
3820 | { |
3821 | *p = c; | |
3822 | return (savestring (p)); | |
3823 | } | |
3824 | *p = c; | |
545f34cf | 3825 | |
726f6388 JA |
3826 | } |
3827 | break; | |
3828 | ||
3829 | case RP_SHORT_LEFT: /* remove shortest match at start */ | |
704a1a2a | 3830 | for (p = param; p <= end; p++) |
726f6388 JA |
3831 | { |
3832 | c = *p; *p = '\0'; | |
f73dda09 | 3833 | if (strmatch (pattern, param, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
726f6388 JA |
3834 | { |
3835 | *p = c; | |
3836 | return (savestring (p)); | |
3837 | } | |
3838 | *p = c; | |
3839 | } | |
3840 | break; | |
3841 | ||
ccc6cda3 | 3842 | case RP_LONG_RIGHT: /* remove longest match at end */ |
704a1a2a | 3843 | for (p = param; p <= end; p++) |
ccc6cda3 | 3844 | { |
f73dda09 | 3845 | if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
ccc6cda3 JA |
3846 | { |
3847 | c = *p; *p = '\0'; | |
3848 | ret = savestring (param); | |
3849 | *p = c; | |
3850 | return (ret); | |
3851 | } | |
3852 | } | |
3853 | break; | |
3854 | ||
3855 | case RP_SHORT_RIGHT: /* remove shortest match at end */ | |
704a1a2a | 3856 | for (p = end; p >= param; p--) |
ccc6cda3 | 3857 | { |
f73dda09 | 3858 | if (strmatch (pattern, p, FNMATCH_EXTFLAG) != FNM_NOMATCH) |
ccc6cda3 JA |
3859 | { |
3860 | c = *p; *p = '\0'; | |
3861 | ret = savestring (param); | |
3862 | *p = c; | |
3863 | return (ret); | |
3864 | } | |
704a1a2a CR |
3865 | } |
3866 | break; | |
3867 | } | |
3868 | ||
3d35553a | 3869 | return (param); /* no match, return original string */ |
704a1a2a CR |
3870 | } |
3871 | ||
3872 | #if defined (HANDLE_MULTIBYTE) | |
3d35553a | 3873 | /* Returns its first argument if nothing matched; new memory otherwise */ |
704a1a2a CR |
3874 | static wchar_t * |
3875 | remove_wpattern (wparam, wstrlen, wpattern, op) | |
3876 | wchar_t *wparam; | |
3877 | size_t wstrlen; | |
3878 | wchar_t *wpattern; | |
3879 | int op; | |
3880 | { | |
d3ad40de CR |
3881 | wchar_t wc, *ret; |
3882 | int n; | |
545f34cf | 3883 | |
704a1a2a CR |
3884 | switch (op) |
3885 | { | |
3886 | case RP_LONG_LEFT: /* remove longest match at start */ | |
3887 | for (n = wstrlen; n >= 0; n--) | |
3888 | { | |
3889 | wc = wparam[n]; wparam[n] = L'\0'; | |
3890 | if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
545f34cf | 3891 | { |
704a1a2a CR |
3892 | wparam[n] = wc; |
3893 | return (wcsdup (wparam + n)); | |
3894 | } | |
3895 | wparam[n] = wc; | |
3896 | } | |
3897 | break; | |
3898 | ||
3899 | case RP_SHORT_LEFT: /* remove shortest match at start */ | |
3900 | for (n = 0; n <= wstrlen; n++) | |
3901 | { | |
3902 | wc = wparam[n]; wparam[n] = L'\0'; | |
3903 | if (wcsmatch (wpattern, wparam, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3904 | { | |
3905 | wparam[n] = wc; | |
3906 | return (wcsdup (wparam + n)); | |
3907 | } | |
3908 | wparam[n] = wc; | |
3909 | } | |
3910 | break; | |
3911 | ||
3912 | case RP_LONG_RIGHT: /* remove longest match at end */ | |
3913 | for (n = 0; n <= wstrlen; n++) | |
3914 | { | |
3915 | if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3916 | { | |
3917 | wc = wparam[n]; wparam[n] = L'\0'; | |
3918 | ret = wcsdup (wparam); | |
3919 | wparam[n] = wc; | |
3920 | return (ret); | |
3921 | } | |
3922 | } | |
3923 | break; | |
3924 | ||
3925 | case RP_SHORT_RIGHT: /* remove shortest match at end */ | |
3926 | for (n = wstrlen; n >= 0; n--) | |
3927 | { | |
3928 | if (wcsmatch (wpattern, wparam + n, FNMATCH_EXTFLAG) != FNM_NOMATCH) | |
3929 | { | |
3930 | wc = wparam[n]; wparam[n] = L'\0'; | |
3931 | ret = wcsdup (wparam); | |
3932 | wparam[n] = wc; | |
3933 | return (ret); | |
545f34cf | 3934 | } |
ccc6cda3 JA |
3935 | } |
3936 | break; | |
3937 | } | |
545f34cf | 3938 | |
3d35553a | 3939 | return (wparam); /* no match, return original string */ |
704a1a2a CR |
3940 | } |
3941 | #endif /* HANDLE_MULTIBYTE */ | |
3942 | ||
3943 | static char * | |
3944 | remove_pattern (param, pattern, op) | |
3945 | char *param, *pattern; | |
3946 | int op; | |
3947 | { | |
3d35553a CR |
3948 | char *xret; |
3949 | ||
704a1a2a CR |
3950 | if (param == NULL) |
3951 | return (param); | |
3952 | if (*param == '\0' || pattern == NULL || *pattern == '\0') /* minor optimization */ | |
3953 | return (savestring (param)); | |
3954 | ||
3955 | #if defined (HANDLE_MULTIBYTE) | |
3956 | if (MB_CUR_MAX > 1) | |
3957 | { | |
3958 | wchar_t *ret, *oret; | |
3959 | size_t n; | |
3960 | wchar_t *wparam, *wpattern; | |
3961 | mbstate_t ps; | |
704a1a2a CR |
3962 | |
3963 | n = xdupmbstowcs (&wpattern, NULL, pattern); | |
3964 | if (n == (size_t)-1) | |
3d35553a CR |
3965 | { |
3966 | xret = remove_upattern (param, pattern, op); | |
3967 | return ((xret == param) ? savestring (param) : xret); | |
3968 | } | |
704a1a2a CR |
3969 | n = xdupmbstowcs (&wparam, NULL, param); |
3970 | if (n == (size_t)-1) | |
3971 | { | |
3972 | free (wpattern); | |
3d35553a CR |
3973 | xret = remove_upattern (param, pattern, op); |
3974 | return ((xret == param) ? savestring (param) : xret); | |
704a1a2a CR |
3975 | } |
3976 | oret = ret = remove_wpattern (wparam, n, wpattern, op); | |
3d35553a CR |
3977 | /* Don't bother to convert wparam back to multibyte string if nothing |
3978 | matched; just return copy of original string */ | |
3979 | if (ret == wparam) | |
3980 | { | |
3981 | free (wparam); | |
3982 | free (wpattern); | |
3983 | return (savestring (param)); | |
3984 | } | |
704a1a2a CR |
3985 | |
3986 | free (wparam); | |
3987 | free (wpattern); | |
3988 | ||
3989 | n = strlen (param); | |
dc8fbaf9 | 3990 | xret = (char *)xmalloc (n + 1); |
704a1a2a CR |
3991 | memset (&ps, '\0', sizeof (mbstate_t)); |
3992 | n = wcsrtombs (xret, (const wchar_t **)&ret, n, &ps); | |
3993 | xret[n] = '\0'; /* just to make sure */ | |
3994 | free (oret); | |
3995 | return xret; | |
3996 | } | |
3997 | else | |
3998 | #endif | |
3d35553a CR |
3999 | { |
4000 | xret = remove_upattern (param, pattern, op); | |
4001 | return ((xret == param) ? savestring (param) : xret); | |
4002 | } | |
ccc6cda3 JA |
4003 | } |
4004 | ||
ccc6cda3 JA |
4005 | /* Match PAT anywhere in STRING and return the match boundaries. |
4006 | This returns 1 in case of a successful match, 0 otherwise. SP | |
4007 | and EP are pointers into the string where the match begins and | |
4008 | ends, respectively. MTYPE controls what kind of match is attempted. | |
4009 | MATCH_BEG and MATCH_END anchor the match at the beginning and end | |
4010 | of the string, respectively. The longest match is returned. */ | |
4011 | static int | |
704a1a2a | 4012 | match_upattern (string, pat, mtype, sp, ep) |
ccc6cda3 JA |
4013 | char *string, *pat; |
4014 | int mtype; | |
4015 | char **sp, **ep; | |
4016 | { | |
5f8cde23 | 4017 | int c, len, mlen; |
233564d2 | 4018 | register char *p, *p1, *npat; |
ccc6cda3 | 4019 | char *end; |
06dff54a | 4020 | int n1; |
ccc6cda3 | 4021 | |
233564d2 CR |
4022 | /* If the pattern doesn't match anywhere in the string, go ahead and |
4023 | short-circuit right away. A minor optimization, saves a bunch of | |
4024 | unnecessary calls to strmatch (up to N calls for a string of N | |
4025 | characters) if the match is unsuccessful. To preserve the semantics | |
4026 | of the substring matches below, we make sure that the pattern has | |
4027 | `*' as first and last character, making a new pattern if necessary. */ | |
4028 | /* XXX - check this later if I ever implement `**' with special meaning, | |
4029 | since this will potentially result in `**' at the beginning or end */ | |
4030 | len = STRLEN (pat); | |
176b12ee | 4031 | if (pat[0] != '*' || (pat[0] == '*' && pat[1] == LPAREN && extended_glob) || pat[len - 1] != '*') |
233564d2 | 4032 | { |
dc8fbaf9 | 4033 | p = npat = (char *)xmalloc (len + 3); |
233564d2 | 4034 | p1 = pat; |
176b12ee | 4035 | if (*p1 != '*' || (*p1 == '*' && p1[1] == LPAREN && extended_glob)) |
233564d2 CR |
4036 | *p++ = '*'; |
4037 | while (*p1) | |
4038 | *p++ = *p1++; | |
4039 | if (p1[-1] != '*' || p[-2] == '\\') | |
4040 | *p++ = '*'; | |
4041 | *p = '\0'; | |
4042 | } | |
4043 | else | |
4044 | npat = pat; | |
4045 | c = strmatch (npat, string, FNMATCH_EXTFLAG); | |
4046 | if (npat != pat) | |
4047 | free (npat); | |
4048 | if (c == FNM_NOMATCH) | |
4049 | return (0); | |
4050 | ||
5e13499c CR |
4051 | len = STRLEN (string); |
4052 | end = string + len; | |
ccc6cda3 | 4053 | |
5f8cde23 CR |
4054 | mlen = umatchlen (pat, len); |
4055 | ||
ccc6cda3 JA |
4056 | switch (mtype) |
4057 | { | |
4058 | case MATCH_ANY: | |
704a1a2a | 4059 | for (p = string; p <= end; p++) |
ccc6cda3 JA |
4060 | { |
4061 | if (match_pattern_char (pat, p)) | |
4062 | { | |
5f8cde23 | 4063 | #if 0 |
704a1a2a | 4064 | for (p1 = end; p1 >= p; p1--) |
5f8cde23 CR |
4065 | #else |
4066 | p1 = (mlen == -1) ? end : p + mlen; | |
06dff54a CR |
4067 | /* p1 - p = length of portion of string to be considered |
4068 | p = current position in string | |
4069 | mlen = number of characters consumed by match (-1 for entire string) | |
4070 | end = end of string | |
4071 | we want to break immediately if the potential match len | |
4072 | is greater than the number of characters remaining in the | |
4073 | string | |
4074 | */ | |
4075 | if (p1 > end) | |
5f8cde23 CR |
4076 | break; |
4077 | for ( ; p1 >= p; p1--) | |
4078 | #endif | |
ccc6cda3 JA |
4079 | { |
4080 | c = *p1; *p1 = '\0'; | |
f73dda09 | 4081 | if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0) |
ccc6cda3 JA |
4082 | { |
4083 | *p1 = c; | |
4084 | *sp = p; | |
4085 | *ep = p1; | |
4086 | return 1; | |
4087 | } | |
4088 | *p1 = c; | |
5f8cde23 CR |
4089 | #if 1 |
4090 | /* If MLEN != -1, we have a fixed length pattern. */ | |
4091 | if (mlen != -1) | |
4092 | break; | |
4093 | #endif | |
ccc6cda3 JA |
4094 | } |
4095 | } | |
4096 | } | |
545f34cf | 4097 | |
ccc6cda3 JA |
4098 | return (0); |
4099 | ||
4100 | case MATCH_BEG: | |
4101 | if (match_pattern_char (pat, string) == 0) | |
28ef6c31 | 4102 | return (0); |
545f34cf | 4103 | |
5f8cde23 | 4104 | #if 0 |
704a1a2a | 4105 | for (p = end; p >= string; p--) |
5f8cde23 CR |
4106 | #else |
4107 | for (p = (mlen == -1) ? end : string + mlen; p >= string; p--) | |
4108 | #endif | |
ccc6cda3 JA |
4109 | { |
4110 | c = *p; *p = '\0'; | |
f73dda09 | 4111 | if (strmatch (pat, string, FNMATCH_EXTFLAG) == 0) |
ccc6cda3 JA |
4112 | { |
4113 | *p = c; | |
4114 | *sp = string; | |
4115 | *ep = p; | |
4116 | return 1; | |
4117 | } | |
4118 | *p = c; | |
5f8cde23 CR |
4119 | #if 1 |
4120 | /* If MLEN != -1, we have a fixed length pattern. */ | |
4121 | if (mlen != -1) | |
4122 | break; | |
4123 | #endif | |
ccc6cda3 | 4124 | } |
545f34cf | 4125 | |
ccc6cda3 | 4126 | return (0); |
726f6388 | 4127 | |
ccc6cda3 | 4128 | case MATCH_END: |
5f8cde23 | 4129 | #if 0 |
704a1a2a | 4130 | for (p = string; p <= end; p++) |
5f8cde23 CR |
4131 | #else |
4132 | for (p = end - ((mlen == -1) ? len : mlen); p <= end; p++) | |
4133 | #endif | |
545f34cf CR |
4134 | { |
4135 | if (strmatch (pat, p, FNMATCH_EXTFLAG) == 0) | |
4136 | { | |
4137 | *sp = p; | |
4138 | *ep = end; | |
4139 | return 1; | |
4140 | } | |
5f8cde23 CR |
4141 | #if 1 |
4142 | /* If MLEN != -1, we have a fixed length pattern. */ | |
4143 | if (mlen != -1) | |
4144 | break; | |
4145 | #endif | |
704a1a2a CR |
4146 | } |
4147 | ||
4148 | return (0); | |
4149 | } | |
4150 | ||
4151 | return (0); | |
4152 | } | |
4153 | ||
4154 | #if defined (HANDLE_MULTIBYTE) | |
704a1a2a CR |
4155 | /* Match WPAT anywhere in WSTRING and return the match boundaries. |
4156 | This returns 1 in case of a successful match, 0 otherwise. Wide | |
4157 | character version. */ | |
4158 | static int | |
4159 | match_wpattern (wstring, indices, wstrlen, wpat, mtype, sp, ep) | |
4160 | wchar_t *wstring; | |
4161 | char **indices; | |
4162 | size_t wstrlen; | |
4163 | wchar_t *wpat; | |
4164 | int mtype; | |
4165 | char **sp, **ep; | |
4166 | { | |
233564d2 | 4167 | wchar_t wc, *wp, *nwpat, *wp1; |
5f8cde23 CR |
4168 | size_t len; |
4169 | int mlen; | |
adc6cff5 CR |
4170 | int n, n1, n2, simple; |
4171 | ||
4172 | simple = (wpat[0] != L'\\' && wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'['); | |
4173 | #if defined (EXTENDED_GLOB) | |
4174 | if (extended_glob) | |
e107650c | 4175 | simple &= (wpat[1] != L'(' || (wpat[0] != L'*' && wpat[0] != L'?' && wpat[0] != L'+' && wpat[0] != L'!' && wpat[0] != L'@')); /*)*/ |
adc6cff5 | 4176 | #endif |
704a1a2a | 4177 | |
233564d2 CR |
4178 | /* If the pattern doesn't match anywhere in the string, go ahead and |
4179 | short-circuit right away. A minor optimization, saves a bunch of | |
4180 | unnecessary calls to strmatch (up to N calls for a string of N | |
4181 | characters) if the match is unsuccessful. To preserve the semantics | |
4182 | of the substring matches below, we make sure that the pattern has | |
4183 | `*' as first and last character, making a new pattern if necessary. */ | |
233564d2 | 4184 | len = wcslen (wpat); |
176b12ee | 4185 | if (wpat[0] != L'*' || (wpat[0] == L'*' && wpat[1] == WLPAREN && extended_glob) || wpat[len - 1] != L'*') |
233564d2 | 4186 | { |
dc8fbaf9 | 4187 | wp = nwpat = (wchar_t *)xmalloc ((len + 3) * sizeof (wchar_t)); |
233564d2 | 4188 | wp1 = wpat; |
176b12ee | 4189 | if (*wp1 != L'*' || (*wp1 == '*' && wp1[1] == WLPAREN && extended_glob)) |
233564d2 CR |
4190 | *wp++ = L'*'; |
4191 | while (*wp1 != L'\0') | |
4192 | *wp++ = *wp1++; | |
4193 | if (wp1[-1] != L'*' || wp1[-2] == L'\\') | |
4194 | *wp++ = L'*'; | |
4195 | *wp = '\0'; | |
4196 | } | |
4197 | else | |
4198 | nwpat = wpat; | |
4199 | len = wcsmatch (nwpat, wstring, FNMATCH_EXTFLAG); | |
4200 | if (nwpat != wpat) | |
4201 | free (nwpat); | |
4202 | if (len == FNM_NOMATCH) | |
4203 | return (0); | |
4204 | ||
5f8cde23 | 4205 | mlen = wmatchlen (wpat, wstrlen); |
adc6cff5 | 4206 | |
5f8cde23 | 4207 | /* itrace("wmatchlen (%ls) -> %d", wpat, mlen); */ |
704a1a2a CR |
4208 | switch (mtype) |
4209 | { | |
4210 | case MATCH_ANY: | |
4211 | for (n = 0; n <= wstrlen; n++) | |
4212 | { | |
adc6cff5 CR |
4213 | #if 1 |
4214 | n2 = simple ? (*wpat == wstring[n]) : match_pattern_wchar (wpat, wstring + n); | |
4215 | #else | |
4216 | n2 = match_pattern_wchar (wpat, wstring + n); | |
4217 | #endif | |
4218 | if (n2) | |
704a1a2a | 4219 | { |
5f8cde23 | 4220 | #if 0 |
704a1a2a | 4221 | for (n1 = wstrlen; n1 >= n; n1--) |
5f8cde23 CR |
4222 | #else |
4223 | n1 = (mlen == -1) ? wstrlen : n + mlen; | |
4224 | if (n1 > wstrlen) | |
4225 | break; | |
4226 | ||
4227 | for ( ; n1 >= n; n1--) | |
4228 | #endif | |
704a1a2a CR |
4229 | { |
4230 | wc = wstring[n1]; wstring[n1] = L'\0'; | |
4231 | if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0) | |
4232 | { | |
4233 | wstring[n1] = wc; | |
4234 | *sp = indices[n]; | |
4235 | *ep = indices[n1]; | |
4236 | return 1; | |
4237 | } | |
4238 | wstring[n1] = wc; | |
5f8cde23 CR |
4239 | #if 1 |
4240 | /* If MLEN != -1, we have a fixed length pattern. */ | |
4241 | if (mlen != -1) | |
4242 | break; | |
4243 | #endif | |
704a1a2a CR |
4244 | } |
4245 | } | |
4246 | } | |
4247 | ||
4248 | return (0); | |
4249 | ||
4250 | case MATCH_BEG: | |
4251 | if (match_pattern_wchar (wpat, wstring) == 0) | |
4252 | return (0); | |
4253 | ||
5f8cde23 | 4254 | #if 0 |
704a1a2a | 4255 | for (n = wstrlen; n >= 0; n--) |
5f8cde23 CR |
4256 | #else |
4257 | for (n = (mlen == -1) ? wstrlen : mlen; n >= 0; n--) | |
4258 | #endif | |
704a1a2a CR |
4259 | { |
4260 | wc = wstring[n]; wstring[n] = L'\0'; | |
4261 | if (wcsmatch (wpat, wstring, FNMATCH_EXTFLAG) == 0) | |
4262 | { | |
4263 | wstring[n] = wc; | |
4264 | *sp = indices[0]; | |
4265 | *ep = indices[n]; | |
4266 | return 1; | |
4267 | } | |
4268 | wstring[n] = wc; | |
5f8cde23 CR |
4269 | #if 1 |
4270 | /* If MLEN != -1, we have a fixed length pattern. */ | |
4271 | if (mlen != -1) | |
4272 | break; | |
4273 | #endif | |
704a1a2a CR |
4274 | } |
4275 | ||
4276 | return (0); | |
4277 | ||
4278 | case MATCH_END: | |
5f8cde23 | 4279 | #if 0 |
704a1a2a | 4280 | for (n = 0; n <= wstrlen; n++) |
5f8cde23 CR |
4281 | #else |
4282 | for (n = wstrlen - ((mlen == -1) ? wstrlen : mlen); n <= wstrlen; n++) | |
4283 | #endif | |
704a1a2a CR |
4284 | { |
4285 | if (wcsmatch (wpat, wstring + n, FNMATCH_EXTFLAG) == 0) | |
545f34cf | 4286 | { |
704a1a2a CR |
4287 | *sp = indices[n]; |
4288 | *ep = indices[wstrlen]; | |
4289 | return 1; | |
545f34cf | 4290 | } |
5f8cde23 CR |
4291 | #if 1 |
4292 | /* If MLEN != -1, we have a fixed length pattern. */ | |
4293 | if (mlen != -1) | |
4294 | break; | |
4295 | #endif | |
545f34cf | 4296 | } |
704a1a2a | 4297 | |
ccc6cda3 | 4298 | return (0); |
726f6388 | 4299 | } |
ccc6cda3 JA |
4300 | |
4301 | return (0); | |
726f6388 | 4302 | } |
704a1a2a CR |
4303 | #endif /* HANDLE_MULTIBYTE */ |
4304 | ||
4305 | static int | |
4306 | match_pattern (string, pat, mtype, sp, ep) | |
4307 | char *string, *pat; | |
4308 | int mtype; | |
4309 | char **sp, **ep; | |
4310 | { | |
4311 | #if defined (HANDLE_MULTIBYTE) | |
4312 | int ret; | |
4313 | size_t n; | |
4314 | wchar_t *wstring, *wpat; | |
4315 | char **indices; | |
adc6cff5 | 4316 | size_t slen, plen, mslen, mplen; |
704a1a2a CR |
4317 | #endif |
4318 | ||
4319 | if (string == 0 || *string == 0 || pat == 0 || *pat == 0) | |
4320 | return (0); | |
4321 | ||
4322 | #if defined (HANDLE_MULTIBYTE) | |
4323 | if (MB_CUR_MAX > 1) | |
4324 | { | |
adc6cff5 CR |
4325 | #if 0 |
4326 | slen = STRLEN (string); | |
4327 | mslen = MBSLEN (string); | |
4328 | plen = STRLEN (pat); | |
4329 | mplen = MBSLEN (pat); | |
4330 | if (slen == mslen && plen == mplen) | |
4331 | #else | |
4332 | if (mbsmbchar (string) == 0 && mbsmbchar (pat) == 0) | |
4333 | #endif | |
4334 | return (match_upattern (string, pat, mtype, sp, ep)); | |
4335 | ||
704a1a2a CR |
4336 | n = xdupmbstowcs (&wpat, NULL, pat); |
4337 | if (n == (size_t)-1) | |
4338 | return (match_upattern (string, pat, mtype, sp, ep)); | |
4339 | n = xdupmbstowcs (&wstring, &indices, string); | |
4340 | if (n == (size_t)-1) | |
4341 | { | |
4342 | free (wpat); | |
4343 | return (match_upattern (string, pat, mtype, sp, ep)); | |
4344 | } | |
4345 | ret = match_wpattern (wstring, indices, n, wpat, mtype, sp, ep); | |
4346 | ||
4347 | free (wpat); | |
4348 | free (wstring); | |
4349 | free (indices); | |
4350 | ||
4351 | return (ret); | |
4352 | } | |
4353 | else | |
4354 | #endif | |
4355 | return (match_upattern (string, pat, mtype, sp, ep)); | |
4356 | } | |
726f6388 | 4357 | |
cce855bc JA |
4358 | static int |
4359 | getpatspec (c, value) | |
4360 | int c; | |
4361 | char *value; | |
4362 | { | |
4363 | if (c == '#') | |
4364 | return ((*value == '#') ? RP_LONG_LEFT : RP_SHORT_LEFT); | |
4365 | else /* c == '%' */ | |
4366 | return ((*value == '%') ? RP_LONG_RIGHT : RP_SHORT_RIGHT); | |
4367 | } | |
4368 | ||
4369 | /* Posix.2 says that the WORD should be run through tilde expansion, | |
4370 | parameter expansion, command substitution and arithmetic expansion. | |
4371 | This leaves the result quoted, so quote_string_for_globbing () has | |
f73dda09 | 4372 | to be called to fix it up for strmatch (). If QUOTED is non-zero, |
cce855bc JA |
4373 | it means that the entire expression was enclosed in double quotes. |
4374 | This means that quoting characters in the pattern do not make any | |
4375 | special pattern characters quoted. For example, the `*' in the | |
4376 | following retains its special meaning: "${foo#'*'}". */ | |
4377 | static char * | |
4378 | getpattern (value, quoted, expandpat) | |
4379 | char *value; | |
4380 | int quoted, expandpat; | |
4381 | { | |
4382 | char *pat, *tword; | |
4383 | WORD_LIST *l; | |
d3ad40de | 4384 | #if 0 |
cce855bc | 4385 | int i; |
d3ad40de | 4386 | #endif |
7117c2d2 JA |
4387 | /* There is a problem here: how to handle single or double quotes in the |
4388 | pattern string when the whole expression is between double quotes? | |
4389 | POSIX.2 says that enclosing double quotes do not cause the pattern to | |
4390 | be quoted, but does that leave us a problem with @ and array[@] and their | |
4391 | expansions inside a pattern? */ | |
4392 | #if 0 | |
cce855bc JA |
4393 | if (expandpat && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *tword) |
4394 | { | |
4395 | i = 0; | |
4396 | pat = string_extract_double_quoted (tword, &i, 1); | |
4397 | free (tword); | |
4398 | tword = pat; | |
4399 | } | |
7117c2d2 | 4400 | #endif |
cce855bc | 4401 | |
7117c2d2 JA |
4402 | /* expand_string_for_rhs () leaves WORD quoted and does not perform |
4403 | word splitting. */ | |
22e63b05 | 4404 | l = *value ? expand_string_for_rhs (value, |
7117c2d2 | 4405 | (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) ? Q_PATQUOTE : quoted, |
cce855bc | 4406 | (int *)NULL, (int *)NULL) |
cce855bc | 4407 | : (WORD_LIST *)0; |
cce855bc JA |
4408 | pat = string_list (l); |
4409 | dispose_words (l); | |
4410 | if (pat) | |
4411 | { | |
4412 | tword = quote_string_for_globbing (pat, QGLOB_CVTNULL); | |
4413 | free (pat); | |
4414 | pat = tword; | |
4415 | } | |
4416 | return (pat); | |
4417 | } | |
4418 | ||
7117c2d2 | 4419 | #if 0 |
cce855bc JA |
4420 | /* Handle removing a pattern from a string as a result of ${name%[%]value} |
4421 | or ${name#[#]value}. */ | |
4422 | static char * | |
7117c2d2 JA |
4423 | variable_remove_pattern (value, pattern, patspec, quoted) |
4424 | char *value, *pattern; | |
4425 | int patspec, quoted; | |
cce855bc | 4426 | { |
7117c2d2 | 4427 | char *tword; |
cce855bc | 4428 | |
7117c2d2 | 4429 | tword = remove_pattern (value, pattern, patspec); |
cce855bc | 4430 | |
cce855bc JA |
4431 | return (tword); |
4432 | } | |
7117c2d2 | 4433 | #endif |
cce855bc JA |
4434 | |
4435 | static char * | |
7117c2d2 | 4436 | list_remove_pattern (list, pattern, patspec, itype, quoted) |
cce855bc JA |
4437 | WORD_LIST *list; |
4438 | char *pattern; | |
7117c2d2 | 4439 | int patspec, itype, quoted; |
cce855bc JA |
4440 | { |
4441 | WORD_LIST *new, *l; | |
4442 | WORD_DESC *w; | |
4443 | char *tword; | |
4444 | ||
4445 | for (new = (WORD_LIST *)NULL, l = list; l; l = l->next) | |
4446 | { | |
4447 | tword = remove_pattern (l->word->word, pattern, patspec); | |
227f982e CR |
4448 | w = alloc_word_desc (); |
4449 | w->word = tword ? tword : savestring (""); | |
cce855bc JA |
4450 | new = make_word_list (w, new); |
4451 | } | |
4452 | ||
4453 | l = REVERSE_LIST (new, WORD_LIST *); | |
e33f2203 | 4454 | tword = string_list_pos_params (itype, l, quoted); |
cce855bc | 4455 | dispose_words (l); |
e33f2203 | 4456 | |
cce855bc JA |
4457 | return (tword); |
4458 | } | |
4459 | ||
4460 | static char * | |
7117c2d2 JA |
4461 | parameter_list_remove_pattern (itype, pattern, patspec, quoted) |
4462 | int itype; | |
4463 | char *pattern; | |
4464 | int patspec, quoted; | |
cce855bc | 4465 | { |
7117c2d2 | 4466 | char *ret; |
cce855bc JA |
4467 | WORD_LIST *list; |
4468 | ||
cce855bc | 4469 | list = list_rest_of_args (); |
7117c2d2 JA |
4470 | if (list == 0) |
4471 | return ((char *)NULL); | |
4472 | ret = list_remove_pattern (list, pattern, patspec, itype, quoted); | |
cce855bc | 4473 | dispose_words (list); |
cce855bc JA |
4474 | return (ret); |
4475 | } | |
4476 | ||
4477 | #if defined (ARRAY_VARS) | |
4478 | static char * | |
fdf670ea CR |
4479 | array_remove_pattern (var, pattern, patspec, varname, quoted) |
4480 | SHELL_VAR *var; | |
7117c2d2 JA |
4481 | char *pattern; |
4482 | int patspec; | |
4483 | char *varname; /* so we can figure out how it's indexed */ | |
4484 | int quoted; | |
cce855bc | 4485 | { |
fdf670ea CR |
4486 | ARRAY *a; |
4487 | HASH_TABLE *h; | |
7117c2d2 JA |
4488 | int itype; |
4489 | char *ret; | |
4490 | WORD_LIST *list; | |
4491 | SHELL_VAR *v; | |
cce855bc | 4492 | |
7117c2d2 JA |
4493 | /* compute itype from varname here */ |
4494 | v = array_variable_part (varname, &ret, 0); | |
4495 | itype = ret[0]; | |
4496 | ||
fdf670ea CR |
4497 | a = (v && array_p (v)) ? array_cell (v) : 0; |
4498 | h = (v && assoc_p (v)) ? assoc_cell (v) : 0; | |
4499 | ||
4500 | list = a ? array_to_word_list (a) : (h ? assoc_to_word_list (h) : 0); | |
7117c2d2 JA |
4501 | if (list == 0) |
4502 | return ((char *)NULL); | |
4503 | ret = list_remove_pattern (list, pattern, patspec, itype, quoted); | |
4504 | dispose_words (list); | |
4505 | ||
4506 | return ret; | |
4507 | } | |
4508 | #endif /* ARRAY_VARS */ | |
4509 | ||
4510 | static char * | |
5f8cde23 CR |
4511 | parameter_brace_remove_pattern (varname, value, ind, patstr, rtype, quoted, flags) |
4512 | char *varname, *value; | |
4513 | int ind; | |
4514 | char *patstr; | |
4515 | int rtype, quoted, flags; | |
7117c2d2 | 4516 | { |
d3a24ed2 | 4517 | int vtype, patspec, starsub; |
7117c2d2 JA |
4518 | char *temp1, *val, *pattern; |
4519 | SHELL_VAR *v; | |
4520 | ||
4521 | if (value == 0) | |
4522 | return ((char *)NULL); | |
4523 | ||
4524 | this_command_name = varname; | |
4525 | ||
5f8cde23 | 4526 | vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); |
7117c2d2 | 4527 | if (vtype == -1) |
cce855bc JA |
4528 | return ((char *)NULL); |
4529 | ||
d3a24ed2 CR |
4530 | starsub = vtype & VT_STARSUB; |
4531 | vtype &= ~VT_STARSUB; | |
4532 | ||
7117c2d2 | 4533 | patspec = getpatspec (rtype, patstr); |
cce855bc | 4534 | if (patspec == RP_LONG_LEFT || patspec == RP_LONG_RIGHT) |
7117c2d2 | 4535 | patstr++; |
cce855bc | 4536 | |
c40a57dd CR |
4537 | /* Need to pass getpattern newly-allocated memory in case of expansion -- |
4538 | the expansion code will free the passed string on an error. */ | |
4539 | temp1 = savestring (patstr); | |
4540 | pattern = getpattern (temp1, quoted, 1); | |
4541 | free (temp1); | |
cce855bc | 4542 | |
7117c2d2 JA |
4543 | temp1 = (char *)NULL; /* shut up gcc */ |
4544 | switch (vtype) | |
cce855bc | 4545 | { |
7117c2d2 JA |
4546 | case VT_VARIABLE: |
4547 | case VT_ARRAYMEMBER: | |
4548 | temp1 = remove_pattern (val, pattern, patspec); | |
4549 | if (vtype == VT_VARIABLE) | |
4550 | FREE (val); | |
4551 | if (temp1) | |
28ef6c31 | 4552 | { |
e6598ba4 CR |
4553 | val = (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
4554 | ? quote_string (temp1) | |
4555 | : quote_escapes (temp1); | |
7117c2d2 JA |
4556 | free (temp1); |
4557 | temp1 = val; | |
28ef6c31 | 4558 | } |
7117c2d2 JA |
4559 | break; |
4560 | #if defined (ARRAY_VARS) | |
4561 | case VT_ARRAYVAR: | |
fdf670ea | 4562 | temp1 = array_remove_pattern (v, pattern, patspec, varname, quoted); |
7117c2d2 | 4563 | if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) |
cce855bc | 4564 | { |
7117c2d2 JA |
4565 | val = quote_escapes (temp1); |
4566 | free (temp1); | |
4567 | temp1 = val; | |
cce855bc | 4568 | } |
7117c2d2 JA |
4569 | break; |
4570 | #endif | |
4571 | case VT_POSPARMS: | |
4572 | temp1 = parameter_list_remove_pattern (varname[0], pattern, patspec, quoted); | |
4573 | if (temp1 && ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) == 0)) | |
4574 | { | |
4575 | val = quote_escapes (temp1); | |
4576 | free (temp1); | |
4577 | temp1 = val; | |
4578 | } | |
4579 | break; | |
cce855bc JA |
4580 | } |
4581 | ||
4582 | FREE (pattern); | |
7117c2d2 JA |
4583 | return temp1; |
4584 | } | |
cce855bc | 4585 | |
726f6388 JA |
4586 | /******************************************* |
4587 | * * | |
4588 | * Functions to expand WORD_DESCs * | |
4589 | * * | |
4590 | *******************************************/ | |
4591 | ||
4592 | /* Expand WORD, performing word splitting on the result. This does | |
4593 | parameter expansion, command substitution, arithmetic expansion, | |
4594 | word splitting, and quote removal. */ | |
4595 | ||
4596 | WORD_LIST * | |
4597 | expand_word (word, quoted) | |
4598 | WORD_DESC *word; | |
4599 | int quoted; | |
4600 | { | |
4601 | WORD_LIST *result, *tresult; | |
4602 | ||
b72432fd | 4603 | tresult = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); |
726f6388 JA |
4604 | result = word_list_split (tresult); |
4605 | dispose_words (tresult); | |
ccc6cda3 | 4606 | return (result ? dequote_list (result) : result); |
726f6388 JA |
4607 | } |
4608 | ||
4609 | /* Expand WORD, but do not perform word splitting on the result. This | |
4610 | does parameter expansion, command substitution, arithmetic expansion, | |
4611 | and quote removal. */ | |
4612 | WORD_LIST * | |
28ef6c31 | 4613 | expand_word_unsplit (word, quoted) |
726f6388 JA |
4614 | WORD_DESC *word; |
4615 | int quoted; | |
4616 | { | |
4617 | WORD_LIST *result; | |
4618 | ||
28ef6c31 | 4619 | expand_no_split_dollar_star = 1; |
1231ac47 CR |
4620 | #if defined (HANDLE_MULTIBYTE) |
4621 | if (ifs_firstc[0] == 0) | |
4622 | #else | |
4623 | if (ifs_firstc == 0) | |
4624 | #endif | |
4625 | word->flags |= W_NOSPLIT; | |
40647963 | 4626 | word->flags |= W_NOSPLIT2; |
b72432fd | 4627 | result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); |
28ef6c31 | 4628 | expand_no_split_dollar_star = 0; |
7117c2d2 | 4629 | |
ccc6cda3 | 4630 | return (result ? dequote_list (result) : result); |
726f6388 JA |
4631 | } |
4632 | ||
4633 | /* Perform shell expansions on WORD, but do not perform word splitting or | |
1231ac47 CR |
4634 | quote removal on the result. Virtually identical to expand_word_unsplit; |
4635 | could be combined if implementations don't diverge. */ | |
726f6388 JA |
4636 | WORD_LIST * |
4637 | expand_word_leave_quoted (word, quoted) | |
4638 | WORD_DESC *word; | |
4639 | int quoted; | |
4640 | { | |
1231ac47 CR |
4641 | WORD_LIST *result; |
4642 | ||
4643 | expand_no_split_dollar_star = 1; | |
4644 | #if defined (HANDLE_MULTIBYTE) | |
4645 | if (ifs_firstc[0] == 0) | |
4646 | #else | |
4647 | if (ifs_firstc == 0) | |
4648 | #endif | |
4649 | word->flags |= W_NOSPLIT; | |
e1e48bba | 4650 | word->flags |= W_NOSPLIT2; |
1231ac47 CR |
4651 | result = call_expand_word_internal (word, quoted, 0, (int *)NULL, (int *)NULL); |
4652 | expand_no_split_dollar_star = 0; | |
4653 | ||
4654 | return result; | |
726f6388 JA |
4655 | } |
4656 | ||
726f6388 JA |
4657 | #if defined (PROCESS_SUBSTITUTION) |
4658 | ||
cce855bc JA |
4659 | /*****************************************************************/ |
4660 | /* */ | |
4661 | /* Hacking Process Substitution */ | |
4662 | /* */ | |
4663 | /*****************************************************************/ | |
726f6388 | 4664 | |
726f6388 JA |
4665 | #if !defined (HAVE_DEV_FD) |
4666 | /* Named pipes must be removed explicitly with `unlink'. This keeps a list | |
4667 | of FIFOs the shell has open. unlink_fifo_list will walk the list and | |
4668 | unlink all of them. add_fifo_list adds the name of an open FIFO to the | |
4669 | list. NFIFO is a count of the number of FIFOs in the list. */ | |
4670 | #define FIFO_INCR 20 | |
4671 | ||
f73dda09 JA |
4672 | struct temp_fifo { |
4673 | char *file; | |
4674 | pid_t proc; | |
4675 | }; | |
4676 | ||
4677 | static struct temp_fifo *fifo_list = (struct temp_fifo *)NULL; | |
ccc6cda3 JA |
4678 | static int nfifo; |
4679 | static int fifo_list_size; | |
726f6388 | 4680 | |
112ff2a6 CR |
4681 | char * |
4682 | copy_fifo_list (sizep) | |
4683 | int *sizep; | |
4684 | { | |
4685 | if (sizep) | |
4686 | *sizep = 0; | |
4687 | return (char *)NULL; | |
4688 | } | |
4689 | ||
726f6388 JA |
4690 | static void |
4691 | add_fifo_list (pathname) | |
4692 | char *pathname; | |
4693 | { | |
4694 | if (nfifo >= fifo_list_size - 1) | |
4695 | { | |
4696 | fifo_list_size += FIFO_INCR; | |
f73dda09 JA |
4697 | fifo_list = (struct temp_fifo *)xrealloc (fifo_list, |
4698 | fifo_list_size * sizeof (struct temp_fifo)); | |
726f6388 JA |
4699 | } |
4700 | ||
f73dda09 JA |
4701 | fifo_list[nfifo].file = savestring (pathname); |
4702 | nfifo++; | |
726f6388 JA |
4703 | } |
4704 | ||
112ff2a6 CR |
4705 | void |
4706 | unlink_fifo (i) | |
4707 | int i; | |
4708 | { | |
4709 | if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1)) | |
4710 | { | |
4711 | unlink (fifo_list[i].file); | |
4712 | free (fifo_list[i].file); | |
4713 | fifo_list[i].file = (char *)NULL; | |
4714 | fifo_list[i].proc = -1; | |
4715 | } | |
4716 | } | |
4717 | ||
726f6388 JA |
4718 | void |
4719 | unlink_fifo_list () | |
4720 | { | |
f73dda09 JA |
4721 | int saved, i, j; |
4722 | ||
ccc6cda3 | 4723 | if (nfifo == 0) |
726f6388 JA |
4724 | return; |
4725 | ||
f73dda09 | 4726 | for (i = saved = 0; i < nfifo; i++) |
726f6388 | 4727 | { |
f73dda09 JA |
4728 | if ((fifo_list[i].proc == -1) || (kill(fifo_list[i].proc, 0) == -1)) |
4729 | { | |
7117c2d2 JA |
4730 | unlink (fifo_list[i].file); |
4731 | free (fifo_list[i].file); | |
4732 | fifo_list[i].file = (char *)NULL; | |
4733 | fifo_list[i].proc = -1; | |
f73dda09 JA |
4734 | } |
4735 | else | |
7117c2d2 | 4736 | saved++; |
f73dda09 JA |
4737 | } |
4738 | ||
4739 | /* If we didn't remove some of the FIFOs, compact the list. */ | |
4740 | if (saved) | |
4741 | { | |
4742 | for (i = j = 0; i < nfifo; i++) | |
4743 | if (fifo_list[i].file) | |
4744 | { | |
4745 | fifo_list[j].file = fifo_list[i].file; | |
4746 | fifo_list[j].proc = fifo_list[i].proc; | |
4747 | j++; | |
4748 | } | |
4749 | nfifo = j; | |
726f6388 | 4750 | } |
f73dda09 JA |
4751 | else |
4752 | nfifo = 0; | |
726f6388 JA |
4753 | } |
4754 | ||
112ff2a6 CR |
4755 | /* Take LIST, which is a bitmap denoting active FIFOs in fifo_list |
4756 | from some point in the past, and close all open FIFOs in fifo_list | |
4757 | that are not marked as active in LIST. If LIST is NULL, close | |
4758 | everything in fifo_list. LSIZE is the number of elements in LIST, in | |
4759 | case it's larger than fifo_list_size (size of fifo_list). */ | |
4760 | void | |
4761 | close_new_fifos (list, lsize) | |
4762 | char *list; | |
4763 | int lsize; | |
4764 | { | |
4765 | int i; | |
4766 | ||
4767 | if (list == 0) | |
4768 | { | |
112ff2a6 CR |
4769 | unlink_fifo_list (); |
4770 | return; | |
4771 | } | |
4772 | ||
4773 | for (i = 0; i < lsize; i++) | |
4774 | if (list[i] == 0 && i < fifo_list_size && fifo_list[i].proc != -1) | |
112ff2a6 | 4775 | unlink_fifo (i); |
112ff2a6 CR |
4776 | |
4777 | for (i = lsize; i < fifo_list_size; i++) | |
112ff2a6 CR |
4778 | unlink_fifo (i); |
4779 | } | |
112ff2a6 | 4780 | |
d3ad40de CR |
4781 | int |
4782 | fifos_pending () | |
4783 | { | |
4784 | return nfifo; | |
4785 | } | |
4786 | ||
112ff2a6 CR |
4787 | int |
4788 | num_fifos () | |
4789 | { | |
4790 | return nfifo; | |
4791 | } | |
4792 | ||
726f6388 JA |
4793 | static char * |
4794 | make_named_pipe () | |
4795 | { | |
4796 | char *tname; | |
4797 | ||
d3ad40de | 4798 | tname = sh_mktmpname ("sh-np", MT_USERANDOM|MT_USETMPDIR); |
726f6388 JA |
4799 | if (mkfifo (tname, 0600) < 0) |
4800 | { | |
4801 | free (tname); | |
4802 | return ((char *)NULL); | |
4803 | } | |
4804 | ||
4805 | add_fifo_list (tname); | |
4806 | return (tname); | |
4807 | } | |
4808 | ||
726f6388 JA |
4809 | #else /* HAVE_DEV_FD */ |
4810 | ||
4811 | /* DEV_FD_LIST is a bitmap of file descriptors attached to pipes the shell | |
4812 | has open to children. NFDS is a count of the number of bits currently | |
4813 | set in DEV_FD_LIST. TOTFDS is a count of the highest possible number | |
4814 | of open files. */ | |
4815 | static char *dev_fd_list = (char *)NULL; | |
ccc6cda3 | 4816 | static int nfds; |
726f6388 JA |
4817 | static int totfds; /* The highest possible number of open files. */ |
4818 | ||
112ff2a6 CR |
4819 | char * |
4820 | copy_fifo_list (sizep) | |
4821 | int *sizep; | |
4822 | { | |
4823 | char *ret; | |
4824 | ||
4825 | if (nfds == 0 || totfds == 0) | |
4826 | { | |
4827 | if (sizep) | |
4828 | *sizep = 0; | |
4829 | return (char *)NULL; | |
4830 | } | |
4831 | ||
4832 | if (sizep) | |
4833 | *sizep = totfds; | |
4834 | ret = (char *)xmalloc (totfds); | |
4835 | return (memcpy (ret, dev_fd_list, totfds)); | |
4836 | } | |
4837 | ||
726f6388 JA |
4838 | static void |
4839 | add_fifo_list (fd) | |
4840 | int fd; | |
4841 | { | |
112ff2a6 | 4842 | if (dev_fd_list == 0 || fd >= totfds) |
726f6388 JA |
4843 | { |
4844 | int ofds; | |
4845 | ||
4846 | ofds = totfds; | |
4847 | totfds = getdtablesize (); | |
4848 | if (totfds < 0 || totfds > 256) | |
4849 | totfds = 256; | |
626d0694 | 4850 | if (fd >= totfds) |
726f6388 JA |
4851 | totfds = fd + 2; |
4852 | ||
f73dda09 | 4853 | dev_fd_list = (char *)xrealloc (dev_fd_list, totfds); |
7117c2d2 | 4854 | memset (dev_fd_list + ofds, '\0', totfds - ofds); |
726f6388 JA |
4855 | } |
4856 | ||
4857 | dev_fd_list[fd] = 1; | |
4858 | nfds++; | |
4859 | } | |
4860 | ||
d3ad40de CR |
4861 | int |
4862 | fifos_pending () | |
4863 | { | |
4864 | return 0; /* used for cleanup; not needed with /dev/fd */ | |
4865 | } | |
4866 | ||
112ff2a6 CR |
4867 | int |
4868 | num_fifos () | |
4869 | { | |
4870 | return nfds; | |
4871 | } | |
4872 | ||
4873 | void | |
4874 | unlink_fifo (fd) | |
4875 | int fd; | |
4876 | { | |
4877 | if (dev_fd_list[fd]) | |
4878 | { | |
4879 | close (fd); | |
4880 | dev_fd_list[fd] = 0; | |
4881 | nfds--; | |
4882 | } | |
4883 | } | |
4884 | ||
726f6388 JA |
4885 | void |
4886 | unlink_fifo_list () | |
4887 | { | |
4888 | register int i; | |
4889 | ||
ccc6cda3 | 4890 | if (nfds == 0) |
726f6388 JA |
4891 | return; |
4892 | ||
4893 | for (i = 0; nfds && i < totfds; i++) | |
112ff2a6 | 4894 | unlink_fifo (i); |
726f6388 JA |
4895 | |
4896 | nfds = 0; | |
4897 | } | |
4898 | ||
112ff2a6 CR |
4899 | /* Take LIST, which is a snapshot copy of dev_fd_list from some point in |
4900 | the past, and close all open fds in dev_fd_list that are not marked | |
4901 | as open in LIST. If LIST is NULL, close everything in dev_fd_list. | |
4902 | LSIZE is the number of elements in LIST, in case it's larger than | |
4903 | totfds (size of dev_fd_list). */ | |
4904 | void | |
4905 | close_new_fifos (list, lsize) | |
4906 | char *list; | |
4907 | int lsize; | |
4908 | { | |
4909 | int i; | |
4910 | ||
4911 | if (list == 0) | |
4912 | { | |
4913 | unlink_fifo_list (); | |
4914 | return; | |
4915 | } | |
4916 | ||
4917 | for (i = 0; i < lsize; i++) | |
4918 | if (list[i] == 0 && i < totfds && dev_fd_list[i]) | |
4919 | unlink_fifo (i); | |
4920 | ||
4921 | for (i = lsize; i < totfds; i++) | |
4922 | unlink_fifo (i); | |
4923 | } | |
4924 | ||
726f6388 JA |
4925 | #if defined (NOTDEF) |
4926 | print_dev_fd_list () | |
4927 | { | |
4928 | register int i; | |
4929 | ||
f73dda09 | 4930 | fprintf (stderr, "pid %ld: dev_fd_list:", (long)getpid ()); |
726f6388 JA |
4931 | fflush (stderr); |
4932 | ||
4933 | for (i = 0; i < totfds; i++) | |
4934 | { | |
4935 | if (dev_fd_list[i]) | |
4936 | fprintf (stderr, " %d", i); | |
4937 | } | |
4938 | fprintf (stderr, "\n"); | |
4939 | } | |
4940 | #endif /* NOTDEF */ | |
4941 | ||
4942 | static char * | |
4943 | make_dev_fd_filename (fd) | |
4944 | int fd; | |
4945 | { | |
f73dda09 | 4946 | char *ret, intbuf[INT_STRLEN_BOUND (int) + 1], *p; |
726f6388 | 4947 | |
e141c35a | 4948 | ret = (char *)xmalloc (sizeof (DEV_FD_PREFIX) + 8); |
bb70624e JA |
4949 | |
4950 | strcpy (ret, DEV_FD_PREFIX); | |
4951 | p = inttostr (fd, intbuf, sizeof (intbuf)); | |
4952 | strcpy (ret + sizeof (DEV_FD_PREFIX) - 1, p); | |
4953 | ||
726f6388 JA |
4954 | add_fifo_list (fd); |
4955 | return (ret); | |
4956 | } | |
4957 | ||
4958 | #endif /* HAVE_DEV_FD */ | |
4959 | ||
4960 | /* Return a filename that will open a connection to the process defined by | |
4961 | executing STRING. HAVE_DEV_FD, if defined, means open a pipe and return | |
4962 | a filename in /dev/fd corresponding to a descriptor that is one of the | |
4963 | ends of the pipe. If not defined, we use named pipes on systems that have | |
4964 | them. Systems without /dev/fd and named pipes are out of luck. | |
4965 | ||
4966 | OPEN_FOR_READ_IN_CHILD, if 1, means open the named pipe for reading or | |
4967 | use the read end of the pipe and dup that file descriptor to fd 0 in | |
4968 | the child. If OPEN_FOR_READ_IN_CHILD is 0, we open the named pipe for | |
4969 | writing or use the write end of the pipe in the child, and dup that | |
4970 | file descriptor to fd 1 in the child. The parent does the opposite. */ | |
4971 | ||
4972 | static char * | |
4973 | process_substitute (string, open_for_read_in_child) | |
4974 | char *string; | |
4975 | int open_for_read_in_child; | |
4976 | { | |
4977 | char *pathname; | |
4978 | int fd, result; | |
4979 | pid_t old_pid, pid; | |
4980 | #if defined (HAVE_DEV_FD) | |
4981 | int parent_pipe_fd, child_pipe_fd; | |
4982 | int fildes[2]; | |
4983 | #endif /* HAVE_DEV_FD */ | |
4984 | #if defined (JOB_CONTROL) | |
4985 | pid_t old_pipeline_pgrp; | |
ccc6cda3 | 4986 | #endif |
726f6388 | 4987 | |
cce855bc | 4988 | if (!string || !*string || wordexp_only) |
726f6388 JA |
4989 | return ((char *)NULL); |
4990 | ||
4991 | #if !defined (HAVE_DEV_FD) | |
4992 | pathname = make_named_pipe (); | |
4993 | #else /* HAVE_DEV_FD */ | |
4994 | if (pipe (fildes) < 0) | |
4995 | { | |
5e13499c | 4996 | sys_error (_("cannot make pipe for process substitution")); |
726f6388 JA |
4997 | return ((char *)NULL); |
4998 | } | |
4999 | /* If OPEN_FOR_READ_IN_CHILD == 1, we want to use the write end of | |
5000 | the pipe in the parent, otherwise the read end. */ | |
5001 | parent_pipe_fd = fildes[open_for_read_in_child]; | |
5002 | child_pipe_fd = fildes[1 - open_for_read_in_child]; | |
d166f048 JA |
5003 | /* Move the parent end of the pipe to some high file descriptor, to |
5004 | avoid clashes with FDs used by the script. */ | |
5005 | parent_pipe_fd = move_to_high_fd (parent_pipe_fd, 1, 64); | |
5006 | ||
726f6388 JA |
5007 | pathname = make_dev_fd_filename (parent_pipe_fd); |
5008 | #endif /* HAVE_DEV_FD */ | |
5009 | ||
36211029 | 5010 | if (pathname == 0) |
726f6388 | 5011 | { |
5e13499c | 5012 | sys_error (_("cannot make pipe for process substitution")); |
726f6388 JA |
5013 | return ((char *)NULL); |
5014 | } | |
5015 | ||
5016 | old_pid = last_made_pid; | |
5017 | ||
5018 | #if defined (JOB_CONTROL) | |
5019 | old_pipeline_pgrp = pipeline_pgrp; | |
5020 | pipeline_pgrp = shell_pgrp; | |
ccc6cda3 | 5021 | save_pipeline (1); |
ccc6cda3 JA |
5022 | #endif /* JOB_CONTROL */ |
5023 | ||
726f6388 JA |
5024 | pid = make_child ((char *)NULL, 1); |
5025 | if (pid == 0) | |
5026 | { | |
ccc6cda3 | 5027 | reset_terminating_signals (); /* XXX */ |
d3a24ed2 | 5028 | free_pushed_string_input (); |
726f6388 | 5029 | /* Cancel traps, in trap.c. */ |
d1fab3dc | 5030 | restore_original_signals (); /* XXX - what about special builtins? bash-4.2 */ |
726f6388 | 5031 | setup_async_signals (); |
d3ad40de | 5032 | subshell_environment |= SUBSHELL_COMSUB|SUBSHELL_PROCSUB; |
726f6388 | 5033 | } |
ccc6cda3 JA |
5034 | |
5035 | #if defined (JOB_CONTROL) | |
726f6388 JA |
5036 | set_sigchld_handler (); |
5037 | stop_making_children (); | |
866961ad | 5038 | /* XXX - should we only do this in the parent? (as in command subst) */ |
726f6388 | 5039 | pipeline_pgrp = old_pipeline_pgrp; |
ccc6cda3 | 5040 | #endif /* JOB_CONTROL */ |
726f6388 JA |
5041 | |
5042 | if (pid < 0) | |
5043 | { | |
5e13499c | 5044 | sys_error (_("cannot make child for process substitution")); |
726f6388 JA |
5045 | free (pathname); |
5046 | #if defined (HAVE_DEV_FD) | |
5047 | close (parent_pipe_fd); | |
5048 | close (child_pipe_fd); | |
5049 | #endif /* HAVE_DEV_FD */ | |
5050 | return ((char *)NULL); | |
5051 | } | |
5052 | ||
5053 | if (pid > 0) | |
5054 | { | |
ccc6cda3 JA |
5055 | #if defined (JOB_CONTROL) |
5056 | restore_pipeline (1); | |
5057 | #endif | |
5058 | ||
f73dda09 JA |
5059 | #if !defined (HAVE_DEV_FD) |
5060 | fifo_list[nfifo-1].proc = pid; | |
5061 | #endif | |
5062 | ||
726f6388 JA |
5063 | last_made_pid = old_pid; |
5064 | ||
5065 | #if defined (JOB_CONTROL) && defined (PGRP_PIPE) | |
5066 | close_pgrp_pipe (); | |
5067 | #endif /* JOB_CONTROL && PGRP_PIPE */ | |
5068 | ||
5069 | #if defined (HAVE_DEV_FD) | |
5070 | close (child_pipe_fd); | |
5071 | #endif /* HAVE_DEV_FD */ | |
5072 | ||
5073 | return (pathname); | |
5074 | } | |
5075 | ||
5076 | set_sigint_handler (); | |
5077 | ||
5078 | #if defined (JOB_CONTROL) | |
5079 | set_job_control (0); | |
5080 | #endif /* JOB_CONTROL */ | |
5081 | ||
5082 | #if !defined (HAVE_DEV_FD) | |
5083 | /* Open the named pipe in the child. */ | |
ccc6cda3 | 5084 | fd = open (pathname, open_for_read_in_child ? O_RDONLY|O_NONBLOCK : O_WRONLY); |
726f6388 JA |
5085 | if (fd < 0) |
5086 | { | |
5e13499c CR |
5087 | /* Two separate strings for ease of translation. */ |
5088 | if (open_for_read_in_child) | |
5089 | sys_error (_("cannot open named pipe %s for reading"), pathname); | |
5090 | else | |
5091 | sys_error (_("cannot open named pipe %s for writing"), pathname); | |
5092 | ||
726f6388 JA |
5093 | exit (127); |
5094 | } | |
bb70624e JA |
5095 | if (open_for_read_in_child) |
5096 | { | |
28ef6c31 | 5097 | if (sh_unset_nodelay_mode (fd) < 0) |
bb70624e | 5098 | { |
d3ad40de | 5099 | sys_error (_("cannot reset nodelay mode for fd %d"), fd); |
bb70624e JA |
5100 | exit (127); |
5101 | } | |
5102 | } | |
726f6388 JA |
5103 | #else /* HAVE_DEV_FD */ |
5104 | fd = child_pipe_fd; | |
5105 | #endif /* HAVE_DEV_FD */ | |
5106 | ||
5107 | if (dup2 (fd, open_for_read_in_child ? 0 : 1) < 0) | |
5108 | { | |
5e13499c | 5109 | sys_error (_("cannot duplicate named pipe %s as fd %d"), pathname, |
ccc6cda3 | 5110 | open_for_read_in_child ? 0 : 1); |
726f6388 JA |
5111 | exit (127); |
5112 | } | |
5113 | ||
f73dda09 JA |
5114 | if (fd != (open_for_read_in_child ? 0 : 1)) |
5115 | close (fd); | |
726f6388 JA |
5116 | |
5117 | /* Need to close any files that this process has open to pipes inherited | |
5118 | from its parent. */ | |
5119 | if (current_fds_to_close) | |
5120 | { | |
5121 | close_fd_bitmap (current_fds_to_close); | |
5122 | current_fds_to_close = (struct fd_bitmap *)NULL; | |
5123 | } | |
5124 | ||
5125 | #if defined (HAVE_DEV_FD) | |
5126 | /* Make sure we close the parent's end of the pipe and clear the slot | |
5127 | in the fd list so it is not closed later, if reallocated by, for | |
5128 | instance, pipe(2). */ | |
5129 | close (parent_pipe_fd); | |
5130 | dev_fd_list[parent_pipe_fd] = 0; | |
5131 | #endif /* HAVE_DEV_FD */ | |
5132 | ||
d166f048 | 5133 | result = parse_and_execute (string, "process substitution", (SEVAL_NONINT|SEVAL_NOHIST)); |
726f6388 JA |
5134 | |
5135 | #if !defined (HAVE_DEV_FD) | |
5136 | /* Make sure we close the named pipe in the child before we exit. */ | |
5137 | close (open_for_read_in_child ? 0 : 1); | |
5138 | #endif /* !HAVE_DEV_FD */ | |
5139 | ||
5140 | exit (result); | |
5141 | /*NOTREACHED*/ | |
5142 | } | |
5143 | #endif /* PROCESS_SUBSTITUTION */ | |
5144 | ||
cce855bc JA |
5145 | /***********************************/ |
5146 | /* */ | |
5147 | /* Command Substitution */ | |
5148 | /* */ | |
5149 | /***********************************/ | |
5150 | ||
d166f048 | 5151 | static char * |
d3ad40de | 5152 | read_comsub (fd, quoted, rflag) |
d166f048 | 5153 | int fd, quoted; |
d3ad40de | 5154 | int *rflag; |
d166f048 | 5155 | { |
e6598ba4 CR |
5156 | char *istring, buf[128], *bufp, *s; |
5157 | int istring_index, istring_size, c, tflag, skip_ctlesc, skip_ctlnul; | |
f73dda09 | 5158 | ssize_t bufn; |
d166f048 JA |
5159 | |
5160 | istring = (char *)NULL; | |
d3ad40de | 5161 | istring_index = istring_size = bufn = tflag = 0; |
d166f048 | 5162 | |
e6598ba4 CR |
5163 | for (skip_ctlesc = skip_ctlnul = 0, s = ifs_value; s && *s; s++) |
5164 | skip_ctlesc |= *s == CTLESC, skip_ctlnul |= *s == CTLNUL; | |
5165 | ||
d3ad40de CR |
5166 | /* Read the output of the command through the pipe. This may need to be |
5167 | changed to understand multibyte characters in the future. */ | |
d166f048 JA |
5168 | while (1) |
5169 | { | |
5170 | if (fd < 0) | |
28ef6c31 | 5171 | break; |
d166f048 JA |
5172 | if (--bufn <= 0) |
5173 | { | |
bb70624e | 5174 | bufn = zread (fd, buf, sizeof (buf)); |
d166f048 JA |
5175 | if (bufn <= 0) |
5176 | break; | |
5177 | bufp = buf; | |
5178 | } | |
5179 | c = *bufp++; | |
5180 | ||
28ef6c31 JA |
5181 | if (c == 0) |
5182 | { | |
5183 | #if 0 | |
5184 | internal_warning ("read_comsub: ignored null byte in input"); | |
5185 | #endif | |
5186 | continue; | |
5187 | } | |
5188 | ||
d166f048 JA |
5189 | /* Add the character to ISTRING, possibly after resizing it. */ |
5190 | RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, DEFAULT_ARRAY_SIZE); | |
5191 | ||
d3ad40de CR |
5192 | /* This is essentially quote_string inline */ |
5193 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) /* || c == CTLESC || c == CTLNUL */) | |
5194 | istring[istring_index++] = CTLESC; | |
5195 | /* Escape CTLESC and CTLNUL in the output to protect those characters | |
5196 | from the rest of the word expansions (word splitting and globbing.) | |
5197 | This is essentially quote_escapes inline. */ | |
e6598ba4 | 5198 | else if (skip_ctlesc == 0 && c == CTLESC) |
d3ad40de CR |
5199 | { |
5200 | tflag |= W_HASCTLESC; | |
5201 | istring[istring_index++] = CTLESC; | |
5202 | } | |
e6598ba4 | 5203 | else if ((skip_ctlnul == 0 && c == CTLNUL) || (c == ' ' && (ifs_value && *ifs_value == 0))) |
d166f048 JA |
5204 | istring[istring_index++] = CTLESC; |
5205 | ||
5206 | istring[istring_index++] = c; | |
28ef6c31 JA |
5207 | |
5208 | #if 0 | |
5209 | #if defined (__CYGWIN__) | |
5210 | if (c == '\n' && istring_index > 1 && istring[istring_index - 2] == '\r') | |
5211 | { | |
5212 | istring_index--; | |
5213 | istring[istring_index - 1] = '\n'; | |
5214 | } | |
5215 | #endif | |
5216 | #endif | |
d166f048 JA |
5217 | } |
5218 | ||
5219 | if (istring) | |
5220 | istring[istring_index] = '\0'; | |
5221 | ||
5222 | /* If we read no output, just return now and save ourselves some | |
5223 | trouble. */ | |
5224 | if (istring_index == 0) | |
5225 | { | |
5226 | FREE (istring); | |
d3ad40de CR |
5227 | if (rflag) |
5228 | *rflag = tflag; | |
d166f048 JA |
5229 | return (char *)NULL; |
5230 | } | |
5231 | ||
5232 | /* Strip trailing newlines from the output of the command. */ | |
5233 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
5234 | { | |
5235 | while (istring_index > 0) | |
5236 | { | |
5237 | if (istring[istring_index - 1] == '\n') | |
5238 | { | |
5239 | --istring_index; | |
5240 | ||
5241 | /* If the newline was quoted, remove the quoting char. */ | |
5242 | if (istring[istring_index - 1] == CTLESC) | |
5243 | --istring_index; | |
5244 | } | |
5245 | else | |
5246 | break; | |
5247 | } | |
5248 | istring[istring_index] = '\0'; | |
5249 | } | |
5250 | else | |
5251 | strip_trailing (istring, istring_index - 1, 1); | |
5252 | ||
d3ad40de CR |
5253 | if (rflag) |
5254 | *rflag = tflag; | |
d166f048 JA |
5255 | return istring; |
5256 | } | |
5257 | ||
d3ad40de CR |
5258 | /* Perform command substitution on STRING. This returns a WORD_DESC * with the |
5259 | contained string possibly quoted. */ | |
5260 | WORD_DESC * | |
726f6388 JA |
5261 | command_substitute (string, quoted) |
5262 | char *string; | |
5263 | int quoted; | |
5264 | { | |
cac4cdbf | 5265 | pid_t pid, old_pid, old_pipeline_pgrp, old_async_pid; |
ccc6cda3 | 5266 | char *istring; |
d3ad40de CR |
5267 | int result, fildes[2], function_value, pflags, rc, tflag; |
5268 | WORD_DESC *ret; | |
726f6388 | 5269 | |
ccc6cda3 | 5270 | istring = (char *)NULL; |
726f6388 JA |
5271 | |
5272 | /* Don't fork () if there is no need to. In the case of no command to | |
5273 | run, just return NULL. */ | |
5274 | if (!string || !*string || (string[0] == '\n' && !string[1])) | |
d3ad40de | 5275 | return ((WORD_DESC *)NULL); |
726f6388 | 5276 | |
cce855bc JA |
5277 | if (wordexp_only && read_but_dont_execute) |
5278 | { | |
3eb2d94a | 5279 | last_command_exit_value = EX_WEXPCOMSUB; |
cce855bc JA |
5280 | jump_to_top_level (EXITPROG); |
5281 | } | |
5282 | ||
bb70624e JA |
5283 | /* We're making the assumption here that the command substitution will |
5284 | eventually run a command from the file system. Since we'll run | |
5285 | maybe_make_export_env in this subshell before executing that command, | |
5286 | the parent shell and any other shells it starts will have to remake | |
5287 | the environment. If we make it before we fork, other shells won't | |
5288 | have to. Don't bother if we have any temporary variable assignments, | |
5289 | though, because the export environment will be remade after this | |
5290 | command completes anyway, but do it if all the words to be expanded | |
5291 | are variable assignments. */ | |
5292 | if (subst_assign_varlist == 0 || garglist == 0) | |
5293 | maybe_make_export_env (); /* XXX */ | |
5294 | ||
d3a24ed2 | 5295 | /* Flags to pass to parse_and_execute() */ |
f486d0a1 | 5296 | pflags = (interactive && sourcelevel == 0) ? SEVAL_RESETLINE : 0; |
d3a24ed2 | 5297 | |
726f6388 JA |
5298 | /* Pipe the output of executing STRING into the current shell. */ |
5299 | if (pipe (fildes) < 0) | |
5300 | { | |
5e13499c | 5301 | sys_error (_("cannot make pipe for command substitution")); |
726f6388 JA |
5302 | goto error_exit; |
5303 | } | |
5304 | ||
5305 | old_pid = last_made_pid; | |
5306 | #if defined (JOB_CONTROL) | |
ccc6cda3 | 5307 | old_pipeline_pgrp = pipeline_pgrp; |
28ef6c31 JA |
5308 | /* Don't reset the pipeline pgrp if we're already a subshell in a pipeline. */ |
5309 | if ((subshell_environment & SUBSHELL_PIPE) == 0) | |
5310 | pipeline_pgrp = shell_pgrp; | |
ccc6cda3 | 5311 | cleanup_the_pipeline (); |
da719982 | 5312 | #endif /* JOB_CONTROL */ |
726f6388 | 5313 | |
cac4cdbf | 5314 | old_async_pid = last_asynchronous_pid; |
cac4cdbf | 5315 | pid = make_child ((char *)NULL, subshell_environment&SUBSHELL_ASYNC); |
cac4cdbf CR |
5316 | last_asynchronous_pid = old_async_pid; |
5317 | ||
726f6388 | 5318 | if (pid == 0) |
64419627 CR |
5319 | { |
5320 | /* Reset the signal handlers in the child, but don't free the | |
5321 | trap strings. Set a flag noting that we have to free the | |
5322 | trap strings if we run trap to change a signal disposition. */ | |
5323 | reset_signal_handlers (); | |
5324 | subshell_environment |= SUBSHELL_RESETTRAP; | |
5325 | } | |
ccc6cda3 JA |
5326 | |
5327 | #if defined (JOB_CONTROL) | |
525739ba | 5328 | /* XXX DO THIS ONLY IN PARENT ? XXX */ |
ccc6cda3 JA |
5329 | set_sigchld_handler (); |
5330 | stop_making_children (); | |
525739ba | 5331 | if (pid != 0) |
525739ba | 5332 | pipeline_pgrp = old_pipeline_pgrp; |
f73dda09 JA |
5333 | #else |
5334 | stop_making_children (); | |
ccc6cda3 | 5335 | #endif /* JOB_CONTROL */ |
726f6388 JA |
5336 | |
5337 | if (pid < 0) | |
5338 | { | |
5e13499c | 5339 | sys_error (_("cannot make child for command substitution")); |
726f6388 JA |
5340 | error_exit: |
5341 | ||
5342 | FREE (istring); | |
5343 | close (fildes[0]); | |
5344 | close (fildes[1]); | |
d3ad40de | 5345 | return ((WORD_DESC *)NULL); |
726f6388 JA |
5346 | } |
5347 | ||
5348 | if (pid == 0) | |
5349 | { | |
5350 | set_sigint_handler (); /* XXX */ | |
28ef6c31 | 5351 | |
d3a24ed2 CR |
5352 | free_pushed_string_input (); |
5353 | ||
726f6388 JA |
5354 | if (dup2 (fildes[1], 1) < 0) |
5355 | { | |
5e13499c | 5356 | sys_error (_("command_substitute: cannot duplicate pipe as fd 1")); |
726f6388 JA |
5357 | exit (EXECUTION_FAILURE); |
5358 | } | |
5359 | ||
5360 | /* If standard output is closed in the parent shell | |
5361 | (such as after `exec >&-'), file descriptor 1 will be | |
5362 | the lowest available file descriptor, and end up in | |
5363 | fildes[0]. This can happen for stdin and stderr as well, | |
5364 | but stdout is more important -- it will cause no output | |
5365 | to be generated from this command. */ | |
5366 | if ((fildes[1] != fileno (stdin)) && | |
5367 | (fildes[1] != fileno (stdout)) && | |
5368 | (fildes[1] != fileno (stderr))) | |
5369 | close (fildes[1]); | |
5370 | ||
5371 | if ((fildes[0] != fileno (stdin)) && | |
5372 | (fildes[0] != fileno (stdout)) && | |
5373 | (fildes[0] != fileno (stderr))) | |
5374 | close (fildes[0]); | |
5375 | ||
54a1fa7c CR |
5376 | #ifdef __CYGWIN__ |
5377 | /* Let stdio know the fd may have changed from text to binary mode, and | |
5378 | make sure to preserve stdout line buffering. */ | |
5379 | freopen (NULL, "w", stdout); | |
5380 | sh_setlinebuf (stdout); | |
5381 | #endif /* __CYGWIN__ */ | |
5382 | ||
726f6388 JA |
5383 | /* The currently executing shell is not interactive. */ |
5384 | interactive = 0; | |
5385 | ||
ccc6cda3 | 5386 | /* This is a subshell environment. */ |
28ef6c31 | 5387 | subshell_environment |= SUBSHELL_COMSUB; |
ccc6cda3 | 5388 | |
28ef6c31 | 5389 | /* When not in POSIX mode, command substitution does not inherit |
7117c2d2 | 5390 | the -e flag. */ |
28ef6c31 JA |
5391 | if (posixly_correct == 0) |
5392 | exit_immediately_on_error = 0; | |
726f6388 JA |
5393 | |
5394 | remove_quoted_escapes (string); | |
5395 | ||
ccc6cda3 | 5396 | startup_state = 2; /* see if we can avoid a fork */ |
726f6388 JA |
5397 | /* Give command substitution a place to jump back to on failure, |
5398 | so we don't go back up to main (). */ | |
5399 | result = setjmp (top_level); | |
5400 | ||
bb70624e JA |
5401 | /* If we're running a command substitution inside a shell function, |
5402 | trap `return' so we don't return from the function in the subshell | |
5403 | and go off to never-never land. */ | |
5404 | if (result == 0 && return_catch_flag) | |
5405 | function_value = setjmp (return_catch); | |
5406 | else | |
5407 | function_value = 0; | |
5408 | ||
d3a24ed2 | 5409 | if (result == ERREXIT) |
5e13499c | 5410 | rc = last_command_exit_value; |
d3a24ed2 | 5411 | else if (result == EXITPROG) |
5e13499c | 5412 | rc = last_command_exit_value; |
726f6388 | 5413 | else if (result) |
5e13499c | 5414 | rc = EXECUTION_FAILURE; |
bb70624e | 5415 | else if (function_value) |
5e13499c | 5416 | rc = return_catch_value; |
726f6388 | 5417 | else |
d3a24ed2 CR |
5418 | { |
5419 | subshell_level++; | |
5420 | rc = parse_and_execute (string, "command substitution", pflags|SEVAL_NOHIST); | |
5421 | subshell_level--; | |
d3a24ed2 | 5422 | } |
5e13499c CR |
5423 | |
5424 | last_command_exit_value = rc; | |
5425 | rc = run_exit_trap (); | |
d3ad40de CR |
5426 | #if defined (PROCESS_SUBSTITUTION) |
5427 | unlink_fifo_list (); | |
5428 | #endif | |
5e13499c | 5429 | exit (rc); |
726f6388 JA |
5430 | } |
5431 | else | |
5432 | { | |
726f6388 JA |
5433 | #if defined (JOB_CONTROL) && defined (PGRP_PIPE) |
5434 | close_pgrp_pipe (); | |
5435 | #endif /* JOB_CONTROL && PGRP_PIPE */ | |
5436 | ||
5437 | close (fildes[1]); | |
5438 | ||
d3ad40de CR |
5439 | tflag = 0; |
5440 | istring = read_comsub (fildes[0], quoted, &tflag); | |
ccc6cda3 | 5441 | |
726f6388 JA |
5442 | close (fildes[0]); |
5443 | ||
b72432fd | 5444 | current_command_subst_pid = pid; |
726f6388 JA |
5445 | last_command_exit_value = wait_for (pid); |
5446 | last_command_subst_pid = pid; | |
5447 | last_made_pid = old_pid; | |
5448 | ||
5449 | #if defined (JOB_CONTROL) | |
5450 | /* If last_command_exit_value > 128, then the substituted command | |
5451 | was terminated by a signal. If that signal was SIGINT, then send | |
5452 | SIGINT to ourselves. This will break out of loops, for instance. */ | |
d3a24ed2 | 5453 | if (last_command_exit_value == (128 + SIGINT) && last_command_exit_signal == SIGINT) |
726f6388 JA |
5454 | kill (getpid (), SIGINT); |
5455 | ||
5456 | /* wait_for gives the terminal back to shell_pgrp. If some other | |
cce855bc JA |
5457 | process group should have it, give it away to that group here. |
5458 | pipeline_pgrp is non-zero only while we are constructing a | |
5459 | pipline, so what we are concerned about is whether or not that | |
5460 | pipeline was started in the background. A pipeline started in | |
5461 | the background should never get the tty back here. */ | |
28ef6c31 | 5462 | if (interactive && pipeline_pgrp != (pid_t)0 && (subshell_environment & SUBSHELL_ASYNC) == 0) |
28ef6c31 | 5463 | give_terminal_to (pipeline_pgrp, 0); |
726f6388 JA |
5464 | #endif /* JOB_CONTROL */ |
5465 | ||
d3ad40de CR |
5466 | ret = alloc_word_desc (); |
5467 | ret->word = istring; | |
5468 | ret->flags = tflag; | |
5469 | ||
5470 | return ret; | |
726f6388 JA |
5471 | } |
5472 | } | |
5473 | ||
5474 | /******************************************************** | |
5475 | * * | |
5476 | * Utility functions for parameter expansion * | |
5477 | * * | |
5478 | ********************************************************/ | |
5479 | ||
ccc6cda3 | 5480 | #if defined (ARRAY_VARS) |
ccc6cda3 | 5481 | |
f73dda09 | 5482 | static arrayind_t |
ccc6cda3 JA |
5483 | array_length_reference (s) |
5484 | char *s; | |
5485 | { | |
f73dda09 JA |
5486 | int len; |
5487 | arrayind_t ind; | |
fdf670ea | 5488 | char *akey; |
f73dda09 | 5489 | char *t, c; |
ccc6cda3 | 5490 | ARRAY *array; |
68dfe178 | 5491 | HASH_TABLE *h; |
ccc6cda3 JA |
5492 | SHELL_VAR *var; |
5493 | ||
5494 | var = array_variable_part (s, &t, &len); | |
726f6388 | 5495 | |
ccc6cda3 JA |
5496 | /* If unbound variables should generate an error, report one and return |
5497 | failure. */ | |
fdf670ea | 5498 | if ((var == 0 || (assoc_p (var) == 0 && array_p (var) == 0)) && unbound_vars_is_error) |
726f6388 | 5499 | { |
f73dda09 | 5500 | c = *--t; |
ccc6cda3 | 5501 | *t = '\0'; |
0d8616ff | 5502 | last_command_exit_value = EXECUTION_FAILURE; |
7117c2d2 | 5503 | err_unboundvar (s); |
f73dda09 | 5504 | *t = c; |
ccc6cda3 | 5505 | return (-1); |
726f6388 | 5506 | } |
ccc6cda3 JA |
5507 | else if (var == 0) |
5508 | return 0; | |
726f6388 | 5509 | |
28ef6c31 JA |
5510 | /* We support a couple of expansions for variables that are not arrays. |
5511 | We'll return the length of the value for v[0], and 1 for v[@] or | |
5512 | v[*]. Return 0 for everything else. */ | |
5513 | ||
5514 | array = array_p (var) ? array_cell (var) : (ARRAY *)NULL; | |
68dfe178 | 5515 | h = assoc_p (var) ? assoc_cell (var) : (HASH_TABLE *)NULL; |
726f6388 | 5516 | |
ccc6cda3 | 5517 | if (ALL_ELEMENT_SUB (t[0]) && t[1] == ']') |
ccc6cda3 | 5518 | { |
fdf670ea | 5519 | if (assoc_p (var)) |
68dfe178 | 5520 | return (h ? assoc_num_elements (h) : 0); |
fdf670ea | 5521 | else if (array_p (var)) |
68dfe178 | 5522 | return (array ? array_num_elements (array) : 0); |
fdf670ea | 5523 | else |
68dfe178 | 5524 | return (var_isset (var) ? 1 : 0); |
ccc6cda3 | 5525 | } |
ccc6cda3 | 5526 | |
fdf670ea CR |
5527 | if (assoc_p (var)) |
5528 | { | |
5529 | t[len - 1] = '\0'; | |
09767ff0 | 5530 | akey = expand_assignment_string_to_string (t, 0); /* [ */ |
fdf670ea CR |
5531 | t[len - 1] = ']'; |
5532 | if (akey == 0 || *akey == 0) | |
5533 | { | |
5534 | err_badarraysub (t); | |
631b20c6 | 5535 | FREE (akey); |
fdf670ea CR |
5536 | return (-1); |
5537 | } | |
5538 | t = assoc_reference (assoc_cell (var), akey); | |
631b20c6 | 5539 | free (akey); |
fdf670ea | 5540 | } |
28ef6c31 | 5541 | else |
fdf670ea | 5542 | { |
d9e1f41e | 5543 | ind = array_expand_index (var, t, len); |
fdf670ea CR |
5544 | if (ind < 0) |
5545 | { | |
5546 | err_badarraysub (t); | |
5547 | return (-1); | |
5548 | } | |
5549 | if (array_p (var)) | |
5550 | t = array_reference (array, ind); | |
5551 | else | |
5552 | t = (ind == 0) ? value_cell (var) : (char *)NULL; | |
5553 | } | |
28ef6c31 | 5554 | |
30915f17 | 5555 | len = MB_STRLEN (t); |
ccc6cda3 | 5556 | return (len); |
726f6388 | 5557 | } |
ccc6cda3 | 5558 | #endif /* ARRAY_VARS */ |
726f6388 JA |
5559 | |
5560 | static int | |
5561 | valid_brace_expansion_word (name, var_is_special) | |
5562 | char *name; | |
5563 | int var_is_special; | |
5564 | { | |
f73dda09 | 5565 | if (DIGIT (*name) && all_digits (name)) |
726f6388 JA |
5566 | return 1; |
5567 | else if (var_is_special) | |
5568 | return 1; | |
ccc6cda3 JA |
5569 | #if defined (ARRAY_VARS) |
5570 | else if (valid_array_reference (name)) | |
5571 | return 1; | |
5572 | #endif /* ARRAY_VARS */ | |
726f6388 JA |
5573 | else if (legal_identifier (name)) |
5574 | return 1; | |
5575 | else | |
5576 | return 0; | |
5577 | } | |
ccc6cda3 | 5578 | |
d3a24ed2 CR |
5579 | static int |
5580 | chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at) | |
5581 | char *name; | |
5582 | int quoted; | |
5583 | int *quoted_dollar_atp, *contains_dollar_at; | |
5584 | { | |
5585 | char *temp1; | |
5586 | ||
5587 | if (name == 0) | |
5588 | { | |
5589 | if (quoted_dollar_atp) | |
5590 | *quoted_dollar_atp = 0; | |
5591 | if (contains_dollar_at) | |
5592 | *contains_dollar_at = 0; | |
5593 | return 0; | |
5594 | } | |
5595 | ||
5596 | /* check for $@ and $* */ | |
5597 | if (name[0] == '@' && name[1] == 0) | |
5598 | { | |
5599 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
5600 | *quoted_dollar_atp = 1; | |
5601 | if (contains_dollar_at) | |
5602 | *contains_dollar_at = 1; | |
5603 | return 1; | |
5604 | } | |
5605 | else if (name[0] == '*' && name[1] == '\0' && quoted == 0) | |
5606 | { | |
5607 | if (contains_dollar_at) | |
5608 | *contains_dollar_at = 1; | |
5609 | return 1; | |
5610 | } | |
5611 | ||
5612 | /* Now check for ${array[@]} and ${array[*]} */ | |
5613 | #if defined (ARRAY_VARS) | |
5614 | else if (valid_array_reference (name)) | |
5615 | { | |
d0ca3503 | 5616 | temp1 = mbschr (name, '['); |
d3a24ed2 CR |
5617 | if (temp1 && temp1[1] == '@' && temp1[2] == ']') |
5618 | { | |
5619 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
5620 | *quoted_dollar_atp = 1; | |
5621 | if (contains_dollar_at) | |
5622 | *contains_dollar_at = 1; | |
5623 | return 1; | |
5624 | } /* [ */ | |
5625 | /* ${array[*]}, when unquoted, should be treated like ${array[@]}, | |
5626 | which should result in separate words even when IFS is unset. */ | |
5627 | if (temp1 && temp1[1] == '*' && temp1[2] == ']' && quoted == 0) | |
5628 | { | |
5629 | if (contains_dollar_at) | |
5630 | *contains_dollar_at = 1; | |
5631 | return 1; | |
5632 | } | |
5633 | } | |
5634 | #endif | |
5635 | return 0; | |
5636 | } | |
5637 | ||
726f6388 JA |
5638 | /* Parameter expand NAME, and return a new string which is the expansion, |
5639 | or NULL if there was no expansion. | |
5640 | VAR_IS_SPECIAL is non-zero if NAME is one of the special variables in | |
5641 | the shell, e.g., "@", "$", "*", etc. QUOTED, if non-zero, means that | |
5642 | NAME was found inside of a double-quoted expression. */ | |
227f982e | 5643 | static WORD_DESC * |
5f8cde23 | 5644 | parameter_brace_expand_word (name, var_is_special, quoted, pflags, indp) |
726f6388 | 5645 | char *name; |
1231ac47 | 5646 | int var_is_special, quoted, pflags; |
5f8cde23 | 5647 | arrayind_t *indp; |
726f6388 | 5648 | { |
227f982e | 5649 | WORD_DESC *ret; |
ccc6cda3 | 5650 | char *temp, *tt; |
7117c2d2 | 5651 | intmax_t arg_index; |
ccc6cda3 | 5652 | SHELL_VAR *var; |
b1a26c01 | 5653 | int atype, rflags; |
5f8cde23 | 5654 | arrayind_t ind; |
726f6388 | 5655 | |
227f982e CR |
5656 | ret = 0; |
5657 | temp = 0; | |
b1a26c01 | 5658 | rflags = 0; |
227f982e | 5659 | |
5f8cde23 CR |
5660 | if (indp) |
5661 | *indp = INTMAX_MIN; | |
5662 | ||
227f982e | 5663 | /* Handle multiple digit arguments, as in ${11}. */ |
f73dda09 | 5664 | if (legal_number (name, &arg_index)) |
7117c2d2 JA |
5665 | { |
5666 | tt = get_dollar_var_value (arg_index); | |
762a763b CR |
5667 | if (tt) |
5668 | temp = (*tt && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
5669 | ? quote_string (tt) | |
5670 | : quote_escapes (tt); | |
5671 | else | |
5672 | temp = (char *)NULL; | |
7117c2d2 JA |
5673 | FREE (tt); |
5674 | } | |
726f6388 JA |
5675 | else if (var_is_special) /* ${@} */ |
5676 | { | |
cce855bc | 5677 | int sindex; |
f73dda09 | 5678 | tt = (char *)xmalloc (2 + strlen (name)); |
cce855bc | 5679 | tt[sindex = 0] = '$'; |
726f6388 | 5680 | strcpy (tt + 1, name); |
7117c2d2 | 5681 | |
227f982e | 5682 | ret = param_expand (tt, &sindex, quoted, (int *)NULL, (int *)NULL, |
1231ac47 | 5683 | (int *)NULL, (int *)NULL, pflags); |
cce855bc | 5684 | free (tt); |
726f6388 | 5685 | } |
ccc6cda3 JA |
5686 | #if defined (ARRAY_VARS) |
5687 | else if (valid_array_reference (name)) | |
5688 | { | |
631b20c6 | 5689 | /* XXX - does this leak if name[@] or name[*]? */ |
5f8cde23 | 5690 | temp = array_value (name, quoted, 0, &atype, &ind); |
7117c2d2 | 5691 | if (atype == 0 && temp) |
5f8cde23 CR |
5692 | { |
5693 | temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
5694 | ? quote_string (temp) | |
5695 | : quote_escapes (temp); | |
5696 | rflags |= W_ARRAYIND; | |
5697 | if (indp) | |
5698 | *indp = ind; | |
5699 | } | |
b1a26c01 CR |
5700 | else if (atype == 1 && temp && QUOTED_NULL (temp) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
5701 | rflags |= W_HASQUOTEDNULL; | |
ccc6cda3 JA |
5702 | } |
5703 | #endif | |
5704 | else if (var = find_variable (name)) | |
5705 | { | |
7117c2d2 | 5706 | if (var_isset (var) && invisible_p (var) == 0) |
28ef6c31 | 5707 | { |
ccc6cda3 | 5708 | #if defined (ARRAY_VARS) |
fdf670ea CR |
5709 | if (assoc_p (var)) |
5710 | temp = assoc_reference (assoc_cell (var), "0"); | |
5711 | else if (array_p (var)) | |
5712 | temp = array_reference (array_cell (var), 0); | |
5713 | else | |
5714 | temp = value_cell (var); | |
ccc6cda3 JA |
5715 | #else |
5716 | temp = value_cell (var); | |
5717 | #endif | |
5718 | ||
5719 | if (temp) | |
762a763b CR |
5720 | temp = (*temp && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) |
5721 | ? quote_string (temp) | |
5722 | : quote_escapes (temp); | |
28ef6c31 | 5723 | } |
ccc6cda3 JA |
5724 | else |
5725 | temp = (char *)NULL; | |
5726 | } | |
726f6388 | 5727 | else |
ccc6cda3 | 5728 | temp = (char *)NULL; |
726f6388 | 5729 | |
227f982e CR |
5730 | if (ret == 0) |
5731 | { | |
5732 | ret = alloc_word_desc (); | |
5733 | ret->word = temp; | |
b1a26c01 | 5734 | ret->flags |= rflags; |
227f982e CR |
5735 | } |
5736 | return ret; | |
726f6388 JA |
5737 | } |
5738 | ||
ccc6cda3 JA |
5739 | /* Expand an indirect reference to a variable: ${!NAME} expands to the |
5740 | value of the variable whose name is the value of NAME. */ | |
227f982e | 5741 | static WORD_DESC * |
d3a24ed2 | 5742 | parameter_brace_expand_indir (name, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at) |
ccc6cda3 JA |
5743 | char *name; |
5744 | int var_is_special, quoted; | |
d3a24ed2 | 5745 | int *quoted_dollar_atp, *contains_dollar_at; |
ccc6cda3 JA |
5746 | { |
5747 | char *temp, *t; | |
227f982e | 5748 | WORD_DESC *w; |
ccc6cda3 | 5749 | |
5f8cde23 | 5750 | w = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND, 0); |
227f982e | 5751 | t = w->word; |
762a763b CR |
5752 | /* Have to dequote here if necessary */ |
5753 | if (t) | |
5754 | { | |
5755 | temp = (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
5756 | ? dequote_string (t) | |
5757 | : dequote_escapes (t); | |
5758 | free (t); | |
5759 | t = temp; | |
5760 | } | |
227f982e CR |
5761 | dispose_word_desc (w); |
5762 | ||
d3a24ed2 | 5763 | chk_atstar (t, quoted, quoted_dollar_atp, contains_dollar_at); |
ccc6cda3 | 5764 | if (t == 0) |
227f982e CR |
5765 | return (WORD_DESC *)NULL; |
5766 | ||
5f8cde23 | 5767 | w = parameter_brace_expand_word (t, SPECIAL_VAR(t, 0), quoted, 0, 0); |
ccc6cda3 | 5768 | free (t); |
227f982e CR |
5769 | |
5770 | return w; | |
ccc6cda3 JA |
5771 | } |
5772 | ||
726f6388 JA |
5773 | /* Expand the right side of a parameter expansion of the form ${NAMEcVALUE}, |
5774 | depending on the value of C, the separating character. C can be one of | |
ccc6cda3 JA |
5775 | "-", "+", or "=". QUOTED is true if the entire brace expression occurs |
5776 | between double quotes. */ | |
227f982e | 5777 | static WORD_DESC * |
ccc6cda3 | 5778 | parameter_brace_expand_rhs (name, value, c, quoted, qdollaratp, hasdollarat) |
726f6388 | 5779 | char *name, *value; |
ccc6cda3 | 5780 | int c, quoted, *qdollaratp, *hasdollarat; |
726f6388 | 5781 | { |
227f982e | 5782 | WORD_DESC *w; |
726f6388 JA |
5783 | WORD_LIST *l; |
5784 | char *t, *t1, *temp; | |
ccc6cda3 | 5785 | int hasdol; |
726f6388 | 5786 | |
ccc6cda3 JA |
5787 | /* If the entire expression is between double quotes, we want to treat |
5788 | the value as a double-quoted string, with the exception that we strip | |
d3ad40de | 5789 | embedded unescaped double quotes (for sh backwards compatibility). */ |
22e63b05 | 5790 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && *value) |
726f6388 | 5791 | { |
ccc6cda3 | 5792 | hasdol = 0; |
22e63b05 | 5793 | temp = string_extract_double_quoted (value, &hasdol, 1); |
726f6388 | 5794 | } |
22e63b05 CR |
5795 | else |
5796 | temp = value; | |
ccc6cda3 | 5797 | |
227f982e | 5798 | w = alloc_word_desc (); |
726f6388 | 5799 | hasdol = 0; |
ccc6cda3 JA |
5800 | /* XXX was 0 not quoted */ |
5801 | l = *temp ? expand_string_for_rhs (temp, quoted, &hasdol, (int *)NULL) | |
5802 | : (WORD_LIST *)0; | |
5803 | if (hasdollarat) | |
5804 | *hasdollarat = hasdol || (l && l->next); | |
22e63b05 CR |
5805 | if (temp != value) |
5806 | free (temp); | |
726f6388 JA |
5807 | if (l) |
5808 | { | |
ccc6cda3 | 5809 | /* The expansion of TEMP returned something. We need to treat things |
12d937f9 CR |
5810 | slightly differently if HASDOL is non-zero. If we have "$@", the |
5811 | individual words have already been quoted. We need to turn them | |
5812 | into a string with the words separated by the first character of | |
5813 | $IFS without any additional quoting, so string_list_dollar_at won't | |
5814 | do the right thing. We use string_list_dollar_star instead. */ | |
5815 | temp = (hasdol || l->next) ? string_list_dollar_star (l) : string_list (l); | |
5816 | ||
ccc6cda3 JA |
5817 | /* If l->next is not null, we know that TEMP contained "$@", since that |
5818 | is the only expansion that creates more than one word. */ | |
12d937f9 | 5819 | if (qdollaratp && ((hasdol && quoted) || l->next)) |
ccc6cda3 | 5820 | *qdollaratp = 1; |
ec860d76 CR |
5821 | /* If we have a quoted null result (QUOTED_NULL(temp)) and the word is |
5822 | a quoted null (l->next == 0 && QUOTED_NULL(l->word->word)), the | |
5823 | flags indicate it (l->word->flags & W_HASQUOTEDNULL), and the | |
5824 | expansion is quoted (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
5825 | (which is more paranoia than anything else), we need to return the | |
5826 | quoted null string and set the flags to indicate it. */ | |
5827 | if (l->next == 0 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && QUOTED_NULL (temp) && QUOTED_NULL (l->word->word) && (l->word->flags & W_HASQUOTEDNULL)) | |
5828 | { | |
5829 | w->flags |= W_HASQUOTEDNULL; | |
5830 | } | |
726f6388 JA |
5831 | dispose_words (l); |
5832 | } | |
ccc6cda3 | 5833 | else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && hasdol) |
726f6388 | 5834 | { |
ccc6cda3 JA |
5835 | /* The brace expansion occurred between double quotes and there was |
5836 | a $@ in TEMP. It does not matter if the $@ is quoted, as long as | |
7117c2d2 JA |
5837 | it does not expand to anything. In this case, we want to return |
5838 | a quoted empty string. */ | |
d3ad40de | 5839 | temp = make_quoted_char ('\0'); |
227f982e | 5840 | w->flags |= W_HASQUOTEDNULL; |
726f6388 JA |
5841 | } |
5842 | else | |
5843 | temp = (char *)NULL; | |
5844 | ||
5845 | if (c == '-' || c == '+') | |
227f982e CR |
5846 | { |
5847 | w->word = temp; | |
5848 | return w; | |
5849 | } | |
726f6388 JA |
5850 | |
5851 | /* c == '=' */ | |
ccc6cda3 | 5852 | t = temp ? savestring (temp) : savestring (""); |
726f6388 JA |
5853 | t1 = dequote_string (t); |
5854 | free (t); | |
d3a24ed2 CR |
5855 | #if defined (ARRAY_VARS) |
5856 | if (valid_array_reference (name)) | |
d11b8b46 | 5857 | assign_array_element (name, t1, 0); |
d3a24ed2 CR |
5858 | else |
5859 | #endif /* ARRAY_VARS */ | |
d11b8b46 | 5860 | bind_variable (name, t1, 0); |
227f982e | 5861 | |
9e51a74d | 5862 | /* From Posix group discussion Feb-March 2010. Issue 7 0000221 */ |
64419627 CR |
5863 | free (temp); |
5864 | ||
5865 | w->word = t1; | |
227f982e | 5866 | return w; |
726f6388 JA |
5867 | } |
5868 | ||
5869 | /* Deal with the right hand side of a ${name:?value} expansion in the case | |
5870 | that NAME is null or not set. If VALUE is non-null it is expanded and | |
5871 | used as the error message to print, otherwise a standard message is | |
5872 | printed. */ | |
5873 | static void | |
5874 | parameter_brace_expand_error (name, value) | |
5875 | char *name, *value; | |
5876 | { | |
ccc6cda3 JA |
5877 | WORD_LIST *l; |
5878 | char *temp; | |
5879 | ||
7f947b68 | 5880 | last_command_exit_value = EXECUTION_FAILURE; /* ensure it's non-zero */ |
726f6388 JA |
5881 | if (value && *value) |
5882 | { | |
22e63b05 | 5883 | l = expand_string (value, 0); |
ccc6cda3 JA |
5884 | temp = string_list (l); |
5885 | report_error ("%s: %s", name, temp ? temp : ""); /* XXX was value not "" */ | |
5886 | FREE (temp); | |
726f6388 JA |
5887 | dispose_words (l); |
5888 | } | |
5889 | else | |
5e13499c | 5890 | report_error (_("%s: parameter null or not set"), name); |
726f6388 JA |
5891 | |
5892 | /* Free the data we have allocated during this expansion, since we | |
5893 | are about to longjmp out. */ | |
5894 | free (name); | |
5895 | FREE (value); | |
5896 | } | |
5897 | ||
5898 | /* Return 1 if NAME is something for which parameter_brace_expand_length is | |
5899 | OK to do. */ | |
5900 | static int | |
5901 | valid_length_expression (name) | |
5902 | char *name; | |
5903 | { | |
28ef6c31 | 5904 | return (name[1] == '\0' || /* ${#} */ |
f73dda09 JA |
5905 | ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') || /* special param */ |
5906 | (DIGIT (name[1]) && all_digits (name + 1)) || /* ${#11} */ | |
ccc6cda3 JA |
5907 | #if defined (ARRAY_VARS) |
5908 | valid_array_reference (name + 1) || /* ${#a[7]} */ | |
5909 | #endif | |
726f6388 JA |
5910 | legal_identifier (name + 1)); /* ${#PS1} */ |
5911 | } | |
5912 | ||
5913 | /* Handle the parameter brace expansion that requires us to return the | |
5914 | length of a parameter. */ | |
7117c2d2 | 5915 | static intmax_t |
726f6388 JA |
5916 | parameter_brace_expand_length (name) |
5917 | char *name; | |
5918 | { | |
ccc6cda3 | 5919 | char *t, *newname; |
7117c2d2 | 5920 | intmax_t number, arg_index; |
ccc6cda3 JA |
5921 | WORD_LIST *list; |
5922 | #if defined (ARRAY_VARS) | |
5923 | SHELL_VAR *var; | |
5924 | #endif | |
5925 | ||
5926 | if (name[1] == '\0') /* ${#} */ | |
5927 | number = number_of_args (); | |
cce855bc JA |
5928 | else if ((name[1] == '@' || name[1] == '*') && name[2] == '\0') /* ${#@}, ${#*} */ |
5929 | number = number_of_args (); | |
f73dda09 | 5930 | else if ((sh_syntaxtab[(unsigned char) name[1]] & CSPECVAR) && name[2] == '\0') |
cce855bc JA |
5931 | { |
5932 | /* Take the lengths of some of the shell's special parameters. */ | |
5933 | switch (name[1]) | |
5934 | { | |
5935 | case '-': | |
5936 | t = which_set_flags (); | |
5937 | break; | |
5938 | case '?': | |
5939 | t = itos (last_command_exit_value); | |
5940 | break; | |
5941 | case '$': | |
5942 | t = itos (dollar_dollar_pid); | |
5943 | break; | |
5944 | case '!': | |
5945 | if (last_asynchronous_pid == NO_PID) | |
e05be32d | 5946 | t = (char *)NULL; /* XXX - error if set -u set? */ |
cce855bc | 5947 | else |
f73dda09 | 5948 | t = itos (last_asynchronous_pid); |
cce855bc JA |
5949 | break; |
5950 | case '#': | |
5951 | t = itos (number_of_args ()); | |
5952 | break; | |
5953 | } | |
5954 | number = STRLEN (t); | |
5955 | FREE (t); | |
5956 | } | |
ccc6cda3 JA |
5957 | #if defined (ARRAY_VARS) |
5958 | else if (valid_array_reference (name + 1)) | |
5959 | number = array_length_reference (name + 1); | |
5960 | #endif /* ARRAY_VARS */ | |
cce855bc | 5961 | else |
ccc6cda3 JA |
5962 | { |
5963 | number = 0; | |
5964 | ||
f73dda09 | 5965 | if (legal_number (name + 1, &arg_index)) /* ${#1} */ |
ccc6cda3 | 5966 | { |
f73dda09 | 5967 | t = get_dollar_var_value (arg_index); |
e05be32d CR |
5968 | if (t == 0 && unbound_vars_is_error) |
5969 | return INTMAX_MIN; | |
898cc92e | 5970 | number = MB_STRLEN (t); |
ccc6cda3 JA |
5971 | FREE (t); |
5972 | } | |
5973 | #if defined (ARRAY_VARS) | |
fdf670ea | 5974 | else if ((var = find_variable (name + 1)) && (invisible_p (var) == 0) && (array_p (var) || assoc_p (var))) |
ccc6cda3 | 5975 | { |
fdf670ea CR |
5976 | if (assoc_p (var)) |
5977 | t = assoc_reference (assoc_cell (var), "0"); | |
5978 | else | |
5979 | t = array_reference (array_cell (var), 0); | |
e05be32d CR |
5980 | if (t == 0 && unbound_vars_is_error) |
5981 | return INTMAX_MIN; | |
898cc92e | 5982 | number = MB_STRLEN (t); |
ccc6cda3 JA |
5983 | } |
5984 | #endif | |
5985 | else /* ${#PS1} */ | |
5986 | { | |
5987 | newname = savestring (name); | |
5988 | newname[0] = '$'; | |
5989 | list = expand_string (newname, Q_DOUBLE_QUOTES); | |
5990 | t = list ? string_list (list) : (char *)NULL; | |
5991 | free (newname); | |
5992 | if (list) | |
5993 | dispose_words (list); | |
5994 | ||
adc6cff5 | 5995 | number = t ? MB_STRLEN (t) : 0; |
ccc6cda3 JA |
5996 | FREE (t); |
5997 | } | |
5998 | } | |
ccc6cda3 JA |
5999 | |
6000 | return (number); | |
6001 | } | |
6002 | ||
28ef6c31 JA |
6003 | /* Skip characters in SUBSTR until DELIM. SUBSTR is an arithmetic expression, |
6004 | so we do some ad-hoc parsing of an arithmetic expression to find | |
6005 | the first DELIM, instead of using strchr(3). Two rules: | |
6006 | 1. If the substring contains a `(', read until closing `)'. | |
6007 | 2. If the substring contains a `?', read past one `:' for each `?'. | |
6008 | */ | |
6009 | ||
6010 | static char * | |
6011 | skiparith (substr, delim) | |
6012 | char *substr; | |
6013 | int delim; | |
6014 | { | |
7117c2d2 JA |
6015 | size_t sublen; |
6016 | int skipcol, pcount, i; | |
6017 | DECLARE_MBSTATE; | |
28ef6c31 | 6018 | |
7117c2d2 JA |
6019 | sublen = strlen (substr); |
6020 | i = skipcol = pcount = 0; | |
6021 | while (substr[i]) | |
28ef6c31 JA |
6022 | { |
6023 | /* Balance parens */ | |
7117c2d2 | 6024 | if (substr[i] == LPAREN) |
28ef6c31 JA |
6025 | { |
6026 | pcount++; | |
7117c2d2 | 6027 | i++; |
28ef6c31 JA |
6028 | continue; |
6029 | } | |
7117c2d2 | 6030 | if (substr[i] == RPAREN && pcount) |
28ef6c31 JA |
6031 | { |
6032 | pcount--; | |
7117c2d2 | 6033 | i++; |
28ef6c31 JA |
6034 | continue; |
6035 | } | |
6036 | if (pcount) | |
7117c2d2 JA |
6037 | { |
6038 | ADVANCE_CHAR (substr, sublen, i); | |
6039 | continue; | |
6040 | } | |
28ef6c31 JA |
6041 | |
6042 | /* Skip one `:' for each `?' */ | |
7117c2d2 | 6043 | if (substr[i] == ':' && skipcol) |
28ef6c31 JA |
6044 | { |
6045 | skipcol--; | |
7117c2d2 | 6046 | i++; |
28ef6c31 JA |
6047 | continue; |
6048 | } | |
7117c2d2 | 6049 | if (substr[i] == delim) |
28ef6c31 | 6050 | break; |
7117c2d2 | 6051 | if (substr[i] == '?') |
28ef6c31 JA |
6052 | { |
6053 | skipcol++; | |
7117c2d2 | 6054 | i++; |
28ef6c31 JA |
6055 | continue; |
6056 | } | |
7117c2d2 | 6057 | ADVANCE_CHAR (substr, sublen, i); |
28ef6c31 | 6058 | } |
7117c2d2 JA |
6059 | |
6060 | return (substr + i); | |
28ef6c31 JA |
6061 | } |
6062 | ||
ccc6cda3 JA |
6063 | /* Verify and limit the start and end of the desired substring. If |
6064 | VTYPE == 0, a regular shell variable is being used; if it is 1, | |
cce855bc | 6065 | then the positional parameters are being used; if it is 2, then |
e8ce775d JA |
6066 | VALUE is really a pointer to an array variable that should be used. |
6067 | Return value is 1 if both values were OK, 0 if there was a problem | |
6068 | with an invalid expression, or -1 if the values were out of range. */ | |
ccc6cda3 | 6069 | static int |
09767ff0 CR |
6070 | verify_substring_values (v, value, substr, vtype, e1p, e2p) |
6071 | SHELL_VAR *v; | |
ccc6cda3 | 6072 | char *value, *substr; |
f73dda09 | 6073 | int vtype; |
7117c2d2 | 6074 | intmax_t *e1p, *e2p; |
ccc6cda3 | 6075 | { |
bb70624e | 6076 | char *t, *temp1, *temp2; |
f73dda09 JA |
6077 | arrayind_t len; |
6078 | int expok; | |
ccc6cda3 JA |
6079 | #if defined (ARRAY_VARS) |
6080 | ARRAY *a; | |
09767ff0 | 6081 | HASH_TABLE *h; |
ccc6cda3 JA |
6082 | #endif |
6083 | ||
28ef6c31 JA |
6084 | /* duplicate behavior of strchr(3) */ |
6085 | t = skiparith (substr, ':'); | |
6086 | if (*t && *t == ':') | |
7117c2d2 | 6087 | *t = '\0'; |
28ef6c31 JA |
6088 | else |
6089 | t = (char *)0; | |
f73dda09 | 6090 | |
d3ad40de | 6091 | temp1 = expand_arith_string (substr, Q_DOUBLE_QUOTES); |
d166f048 | 6092 | *e1p = evalexp (temp1, &expok); |
ccc6cda3 | 6093 | free (temp1); |
d166f048 JA |
6094 | if (expok == 0) |
6095 | return (0); | |
ccc6cda3 | 6096 | |
f73dda09 | 6097 | len = -1; /* paranoia */ |
ccc6cda3 JA |
6098 | switch (vtype) |
6099 | { | |
6100 | case VT_VARIABLE: | |
d166f048 | 6101 | case VT_ARRAYMEMBER: |
898cc92e | 6102 | len = MB_STRLEN (value); |
ccc6cda3 JA |
6103 | break; |
6104 | case VT_POSPARMS: | |
6105 | len = number_of_args () + 1; | |
d3ad40de CR |
6106 | if (*e1p == 0) |
6107 | len++; /* add one arg if counting from $0 */ | |
ccc6cda3 JA |
6108 | break; |
6109 | #if defined (ARRAY_VARS) | |
6110 | case VT_ARRAYVAR: | |
301e2142 | 6111 | /* For arrays, the first value deals with array indices. Negative |
09767ff0 CR |
6112 | offsets count from one past the array's maximum index. Associative |
6113 | arrays treat the number of elements as the maximum index. */ | |
6114 | if (assoc_p (v)) | |
6115 | { | |
6116 | h = assoc_cell (v); | |
6117 | len = assoc_num_elements (h) + (*e1p < 0); | |
6118 | } | |
6119 | else | |
6120 | { | |
6121 | a = (ARRAY *)value; | |
6122 | len = array_max_index (a) + (*e1p < 0); /* arrays index from 0 to n - 1 */ | |
6123 | } | |
ccc6cda3 JA |
6124 | break; |
6125 | #endif | |
6126 | } | |
6127 | ||
f73dda09 JA |
6128 | if (len == -1) /* paranoia */ |
6129 | return -1; | |
6130 | ||
ccc6cda3 JA |
6131 | if (*e1p < 0) /* negative offsets count from end */ |
6132 | *e1p += len; | |
6133 | ||
8ed8525c | 6134 | if (*e1p > len || *e1p < 0) |
e8ce775d | 6135 | return (-1); |
d166f048 | 6136 | |
5e13499c CR |
6137 | #if defined (ARRAY_VARS) |
6138 | /* For arrays, the second offset deals with the number of elements. */ | |
6139 | if (vtype == VT_ARRAYVAR) | |
dd4f3dd8 | 6140 | len = assoc_p (v) ? assoc_num_elements (h) : array_num_elements (a); |
5e13499c CR |
6141 | #endif |
6142 | ||
ccc6cda3 JA |
6143 | if (t) |
6144 | { | |
6145 | t++; | |
bb70624e | 6146 | temp2 = savestring (t); |
d3ad40de | 6147 | temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES); |
bb70624e | 6148 | free (temp2); |
ccc6cda3 | 6149 | t[-1] = ':'; |
d166f048 | 6150 | *e2p = evalexp (temp1, &expok); |
ccc6cda3 | 6151 | free (temp1); |
d166f048 | 6152 | if (expok == 0) |
28ef6c31 | 6153 | return (0); |
67362c60 | 6154 | if ((vtype == VT_ARRAYVAR || vtype == VT_POSPARMS) && *e2p < 0) |
28ef6c31 | 6155 | { |
5e13499c | 6156 | internal_error (_("%s: substring expression < 0"), t); |
ccc6cda3 | 6157 | return (0); |
28ef6c31 | 6158 | } |
5e13499c CR |
6159 | #if defined (ARRAY_VARS) |
6160 | /* In order to deal with sparse arrays, push the intelligence about how | |
6161 | to deal with the number of elements desired down to the array- | |
6162 | specific functions. */ | |
6163 | if (vtype != VT_ARRAYVAR) | |
6164 | #endif | |
6165 | { | |
67362c60 CR |
6166 | if (*e2p < 0) |
6167 | { | |
6168 | *e2p += len; | |
6169 | if (*e2p < 0 || *e2p < *e1p) | |
6170 | { | |
6171 | internal_error (_("%s: substring expression < 0"), t); | |
6172 | return (0); | |
6173 | } | |
6174 | } | |
6175 | else | |
6176 | *e2p += *e1p; /* want E2 chars starting at E1 */ | |
5e13499c CR |
6177 | if (*e2p > len) |
6178 | *e2p = len; | |
6179 | } | |
ccc6cda3 JA |
6180 | } |
6181 | else | |
6182 | *e2p = len; | |
6183 | ||
6184 | return (1); | |
6185 | } | |
6186 | ||
ccc6cda3 | 6187 | /* Return the type of variable specified by VARNAME (simple variable, |
cce855bc | 6188 | positional param, or array variable). Also return the value specified |
7117c2d2 | 6189 | by VARNAME (value of a variable or a reference to an array element). |
5f8cde23 CR |
6190 | QUOTED is the standard description of quoting state, using Q_* defines. |
6191 | FLAGS is currently a set of flags to pass to array_value. If IND is | |
6192 | non-null and not INTMAX_MIN, and FLAGS includes AV_USEIND, IND is | |
6193 | passed to array_value so the array index is not computed again. | |
7117c2d2 JA |
6194 | If this returns VT_VARIABLE, the caller assumes that CTLESC and CTLNUL |
6195 | characters in the value are quoted with CTLESC and takes appropriate | |
6196 | steps. For convenience, *VALP is set to the dequoted VALUE. */ | |
ccc6cda3 | 6197 | static int |
5f8cde23 | 6198 | get_var_and_type (varname, value, ind, quoted, flags, varp, valp) |
ccc6cda3 | 6199 | char *varname, *value; |
5f8cde23 CR |
6200 | arrayind_t ind; |
6201 | int quoted, flags; | |
ccc6cda3 JA |
6202 | SHELL_VAR **varp; |
6203 | char **valp; | |
6204 | { | |
6205 | int vtype; | |
6206 | char *temp; | |
6207 | #if defined (ARRAY_VARS) | |
6208 | SHELL_VAR *v; | |
6209 | #endif | |
5f8cde23 | 6210 | arrayind_t lind; |
ccc6cda3 | 6211 | |
7117c2d2 JA |
6212 | /* This sets vtype to VT_VARIABLE or VT_POSPARMS */ |
6213 | vtype = (varname[0] == '@' || varname[0] == '*') && varname[1] == '\0'; | |
d3a24ed2 CR |
6214 | if (vtype == VT_POSPARMS && varname[0] == '*') |
6215 | vtype |= VT_STARSUB; | |
ccc6cda3 JA |
6216 | *varp = (SHELL_VAR *)NULL; |
6217 | ||
6218 | #if defined (ARRAY_VARS) | |
6219 | if (valid_array_reference (varname)) | |
6220 | { | |
6221 | v = array_variable_part (varname, &temp, (int *)0); | |
5f8cde23 CR |
6222 | /* If we want to signal array_value to use an already-computed index, |
6223 | set LIND to that index */ | |
6224 | lind = (ind != INTMAX_MIN && (flags & AV_USEIND)) ? ind : 0; | |
fdf670ea | 6225 | if (v && (array_p (v) || assoc_p (v))) |
f73dda09 JA |
6226 | { /* [ */ |
6227 | if (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']') | |
ccc6cda3 | 6228 | { |
dd4f3dd8 | 6229 | /* Callers have to differentiate betwen indexed and associative */ |
ccc6cda3 | 6230 | vtype = VT_ARRAYVAR; |
d3a24ed2 CR |
6231 | if (temp[0] == '*') |
6232 | vtype |= VT_STARSUB; | |
fdf670ea | 6233 | *valp = array_p (v) ? (char *)array_cell (v) : (char *)assoc_cell (v); |
ccc6cda3 JA |
6234 | } |
6235 | else | |
6236 | { | |
d166f048 | 6237 | vtype = VT_ARRAYMEMBER; |
5f8cde23 | 6238 | *valp = array_value (varname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind); |
ccc6cda3 JA |
6239 | } |
6240 | *varp = v; | |
6241 | } | |
8fed3589 CR |
6242 | else if (v && (ALL_ELEMENT_SUB (temp[0]) && temp[1] == ']')) |
6243 | { | |
6244 | vtype = VT_VARIABLE; | |
6245 | *varp = v; | |
6246 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
6247 | *valp = dequote_string (value); | |
6248 | else | |
6249 | *valp = dequote_escapes (value); | |
6250 | } | |
ccc6cda3 | 6251 | else |
fdf670ea CR |
6252 | { |
6253 | vtype = VT_ARRAYMEMBER; | |
6254 | *varp = v; | |
5f8cde23 | 6255 | *valp = array_value (varname, Q_DOUBLE_QUOTES, flags, (int *)NULL, &lind); |
fdf670ea | 6256 | } |
ccc6cda3 | 6257 | } |
fdf670ea | 6258 | else if ((v = find_variable (varname)) && (invisible_p (v) == 0) && (assoc_p (v) || array_p (v))) |
ccc6cda3 | 6259 | { |
7117c2d2 | 6260 | vtype = VT_ARRAYMEMBER; |
ccc6cda3 | 6261 | *varp = v; |
fdf670ea | 6262 | *valp = assoc_p (v) ? assoc_reference (assoc_cell (v), "0") : array_reference (array_cell (v), 0); |
ccc6cda3 JA |
6263 | } |
6264 | else | |
6265 | #endif | |
762a763b CR |
6266 | { |
6267 | if (value && vtype == VT_VARIABLE) | |
6268 | { | |
6269 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
6270 | *valp = dequote_string (value); | |
6271 | else | |
6272 | *valp = dequote_escapes (value); | |
6273 | } | |
6274 | else | |
6275 | *valp = value; | |
6276 | } | |
ccc6cda3 JA |
6277 | |
6278 | return vtype; | |
6279 | } | |
6280 | ||
cce855bc JA |
6281 | /******************************************************/ |
6282 | /* */ | |
6283 | /* Functions to extract substrings of variable values */ | |
6284 | /* */ | |
6285 | /******************************************************/ | |
6286 | ||
545f34cf CR |
6287 | #if defined (HANDLE_MULTIBYTE) |
6288 | /* Character-oriented rather than strictly byte-oriented substrings. S and | |
6289 | E, rather being strict indices into STRING, indicate character (possibly | |
6290 | multibyte character) positions that require calculation. | |
6291 | Used by the ${param:offset[:length]} expansion. */ | |
6292 | static char * | |
6293 | mb_substring (string, s, e) | |
6294 | char *string; | |
6295 | int s, e; | |
6296 | { | |
6297 | char *tt; | |
6298 | int start, stop, i, slen; | |
6299 | DECLARE_MBSTATE; | |
6300 | ||
6301 | start = 0; | |
da713c23 CR |
6302 | /* Don't need string length in ADVANCE_CHAR unless multibyte chars possible. */ |
6303 | slen = (MB_CUR_MAX > 1) ? STRLEN (string) : 0; | |
545f34cf CR |
6304 | |
6305 | i = s; | |
6306 | while (string[start] && i--) | |
6307 | ADVANCE_CHAR (string, slen, start); | |
6308 | stop = start; | |
6309 | i = e - s; | |
6310 | while (string[stop] && i--) | |
6311 | ADVANCE_CHAR (string, slen, stop); | |
6312 | tt = substring (string, start, stop); | |
6313 | return tt; | |
6314 | } | |
6315 | #endif | |
6316 | ||
ccc6cda3 JA |
6317 | /* Process a variable substring expansion: ${name:e1[:e2]}. If VARNAME |
6318 | is `@', use the positional parameters; otherwise, use the value of | |
6319 | VARNAME. If VARNAME is an array variable, use the array elements. */ | |
6320 | ||
6321 | static char * | |
5f8cde23 CR |
6322 | parameter_brace_substring (varname, value, ind, substr, quoted, flags) |
6323 | char *varname, *value; | |
6324 | int ind; | |
6325 | char *substr; | |
6326 | int quoted, flags; | |
ccc6cda3 | 6327 | { |
7117c2d2 | 6328 | intmax_t e1, e2; |
d3a24ed2 | 6329 | int vtype, r, starsub; |
d3ad40de | 6330 | char *temp, *val, *tt, *oname; |
ccc6cda3 JA |
6331 | SHELL_VAR *v; |
6332 | ||
6333 | if (value == 0) | |
6334 | return ((char *)NULL); | |
6335 | ||
d3ad40de | 6336 | oname = this_command_name; |
ccc6cda3 JA |
6337 | this_command_name = varname; |
6338 | ||
5f8cde23 | 6339 | vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); |
ccc6cda3 | 6340 | if (vtype == -1) |
d3ad40de CR |
6341 | { |
6342 | this_command_name = oname; | |
6343 | return ((char *)NULL); | |
6344 | } | |
ccc6cda3 | 6345 | |
d3a24ed2 CR |
6346 | starsub = vtype & VT_STARSUB; |
6347 | vtype &= ~VT_STARSUB; | |
6348 | ||
09767ff0 | 6349 | r = verify_substring_values (v, val, substr, vtype, &e1, &e2); |
d3ad40de | 6350 | this_command_name = oname; |
e8ce775d | 6351 | if (r <= 0) |
631b20c6 CR |
6352 | { |
6353 | if (vtype == VT_VARIABLE) | |
6354 | FREE (val); | |
6355 | return ((r == 0) ? &expand_param_error : (char *)NULL); | |
6356 | } | |
ccc6cda3 JA |
6357 | |
6358 | switch (vtype) | |
6359 | { | |
6360 | case VT_VARIABLE: | |
d166f048 | 6361 | case VT_ARRAYMEMBER: |
545f34cf CR |
6362 | #if defined (HANDLE_MULTIBYTE) |
6363 | if (MB_CUR_MAX > 1) | |
6364 | tt = mb_substring (val, e1, e2); | |
6365 | else | |
6366 | #endif | |
7117c2d2 | 6367 | tt = substring (val, e1, e2); |
545f34cf | 6368 | |
7117c2d2 JA |
6369 | if (vtype == VT_VARIABLE) |
6370 | FREE (val); | |
6371 | if (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) | |
6372 | temp = quote_string (tt); | |
6373 | else | |
6374 | temp = tt ? quote_escapes (tt) : (char *)NULL; | |
6375 | FREE (tt); | |
ccc6cda3 JA |
6376 | break; |
6377 | case VT_POSPARMS: | |
7117c2d2 JA |
6378 | tt = pos_params (varname, e1, e2, quoted); |
6379 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0) | |
6380 | { | |
6381 | temp = tt ? quote_escapes (tt) : (char *)NULL; | |
6382 | FREE (tt); | |
6383 | } | |
6384 | else | |
6385 | temp = tt; | |
ccc6cda3 JA |
6386 | break; |
6387 | #if defined (ARRAY_VARS) | |
6388 | case VT_ARRAYVAR: | |
fdf670ea | 6389 | if (assoc_p (v)) |
09767ff0 CR |
6390 | /* we convert to list and take first e2 elements starting at e1th |
6391 | element -- officially undefined for now */ | |
6392 | temp = assoc_subrange (assoc_cell (v), e1, e2, starsub, quoted); | |
fdf670ea | 6393 | else |
5e13499c | 6394 | /* We want E2 to be the number of elements desired (arrays can be sparse, |
704a1a2a CR |
6395 | so verify_substring_values just returns the numbers specified and we |
6396 | rely on array_subrange to understand how to deal with them). */ | |
fdf670ea | 6397 | temp = array_subrange (array_cell (v), e1, e2, starsub, quoted); |
d3ad40de CR |
6398 | /* array_subrange now calls array_quote_escapes as appropriate, so the |
6399 | caller no longer needs to. */ | |
ccc6cda3 JA |
6400 | break; |
6401 | #endif | |
f73dda09 JA |
6402 | default: |
6403 | temp = (char *)NULL; | |
ccc6cda3 JA |
6404 | } |
6405 | ||
6406 | return temp; | |
6407 | } | |
6408 | ||
cce855bc JA |
6409 | /****************************************************************/ |
6410 | /* */ | |
6411 | /* Functions to perform pattern substitution on variable values */ | |
6412 | /* */ | |
6413 | /****************************************************************/ | |
6414 | ||
ee56a865 CR |
6415 | static int |
6416 | shouldexp_replacement (s) | |
6417 | char *s; | |
6418 | { | |
6419 | register char *p; | |
6420 | ||
6421 | for (p = s; p && *p; p++) | |
6422 | { | |
6423 | if (*p == '\\') | |
6424 | p++; | |
6425 | else if (*p == '&') | |
6426 | return 1; | |
6427 | } | |
6428 | return 0; | |
6429 | } | |
6430 | ||
ccc6cda3 JA |
6431 | char * |
6432 | pat_subst (string, pat, rep, mflags) | |
6433 | char *string, *pat, *rep; | |
6434 | int mflags; | |
6435 | { | |
ee56a865 CR |
6436 | char *ret, *s, *e, *str, *rstr, *mstr; |
6437 | int rsize, rptr, l, replen, mtype, rxpand, rslen, mlen; | |
ccc6cda3 | 6438 | |
5f8cde23 CR |
6439 | if (string == 0) |
6440 | return (savestring ("")); | |
6441 | ||
b72432fd JA |
6442 | mtype = mflags & MATCH_TYPEMASK; |
6443 | ||
e054386f | 6444 | #if 0 /* bash-4.2 ? */ |
ee56a865 | 6445 | rxpand = (rep && *rep) ? shouldexp_replacement (rep) : 0; |
e054386f CR |
6446 | #else |
6447 | rxpand = 0; | |
6448 | #endif | |
ee56a865 | 6449 | |
b72432fd JA |
6450 | /* Special cases: |
6451 | * 1. A null pattern with mtype == MATCH_BEG means to prefix STRING | |
6452 | * with REP and return the result. | |
6453 | * 2. A null pattern with mtype == MATCH_END means to append REP to | |
6454 | * STRING and return the result. | |
ee56a865 | 6455 | * These don't understand or process `&' in the replacement string. |
b72432fd JA |
6456 | */ |
6457 | if ((pat == 0 || *pat == 0) && (mtype == MATCH_BEG || mtype == MATCH_END)) | |
6458 | { | |
6459 | replen = STRLEN (rep); | |
5f8cde23 | 6460 | l = STRLEN (string); |
f73dda09 | 6461 | ret = (char *)xmalloc (replen + l + 2); |
bb70624e JA |
6462 | if (replen == 0) |
6463 | strcpy (ret, string); | |
6464 | else if (mtype == MATCH_BEG) | |
b72432fd JA |
6465 | { |
6466 | strcpy (ret, rep); | |
6467 | strcpy (ret + replen, string); | |
6468 | } | |
6469 | else | |
6470 | { | |
6471 | strcpy (ret, string); | |
6472 | strcpy (ret + l, rep); | |
6473 | } | |
6474 | return (ret); | |
6475 | } | |
6476 | ||
f73dda09 | 6477 | ret = (char *)xmalloc (rsize = 64); |
ccc6cda3 JA |
6478 | ret[0] = '\0'; |
6479 | ||
ccc6cda3 JA |
6480 | for (replen = STRLEN (rep), rptr = 0, str = string;;) |
6481 | { | |
6482 | if (match_pattern (str, pat, mtype, &s, &e) == 0) | |
6483 | break; | |
6484 | l = s - str; | |
ee56a865 CR |
6485 | |
6486 | if (rxpand) | |
6487 | { | |
6488 | int x; | |
6489 | mlen = e - s; | |
6490 | mstr = xmalloc (mlen + 1); | |
6491 | for (x = 0; x < mlen; x++) | |
6492 | mstr[x] = s[x]; | |
6493 | mstr[mlen] = '\0'; | |
6494 | rstr = strcreplace (rep, '&', mstr, 0); | |
6495 | rslen = strlen (rstr); | |
6496 | } | |
6497 | else | |
6498 | { | |
6499 | rstr = rep; | |
6500 | rslen = replen; | |
6501 | } | |
6502 | ||
6503 | RESIZE_MALLOCED_BUFFER (ret, rptr, (l + rslen), rsize, 64); | |
ccc6cda3 JA |
6504 | |
6505 | /* OK, now copy the leading unmatched portion of the string (from | |
6506 | str to s) to ret starting at rptr (the current offset). Then copy | |
28ef6c31 JA |
6507 | the replacement string at ret + rptr + (s - str). Increment |
6508 | rptr (if necessary) and str and go on. */ | |
ccc6cda3 JA |
6509 | if (l) |
6510 | { | |
6511 | strncpy (ret + rptr, str, l); | |
6512 | rptr += l; | |
6513 | } | |
6514 | if (replen) | |
6515 | { | |
ee56a865 CR |
6516 | strncpy (ret + rptr, rstr, rslen); |
6517 | rptr += rslen; | |
ccc6cda3 JA |
6518 | } |
6519 | str = e; /* e == end of match */ | |
d3a24ed2 | 6520 | |
ee56a865 CR |
6521 | if (rstr != rep) |
6522 | free (rstr); | |
6523 | ||
ccc6cda3 | 6524 | if (((mflags & MATCH_GLOBREP) == 0) || mtype != MATCH_ANY) |
28ef6c31 | 6525 | break; |
d3a24ed2 CR |
6526 | |
6527 | if (s == e) | |
5ccad779 CR |
6528 | { |
6529 | /* On a zero-length match, make sure we copy one character, since | |
6530 | we increment one character to avoid infinite recursion. */ | |
6531 | RESIZE_MALLOCED_BUFFER (ret, rptr, 1, rsize, 64); | |
6532 | ret[rptr++] = *str++; | |
6533 | e++; /* avoid infinite recursion on zero-length match */ | |
6534 | } | |
ccc6cda3 JA |
6535 | } |
6536 | ||
6537 | /* Now copy the unmatched portion of the input string */ | |
5f8cde23 | 6538 | if (str && *str) |
d166f048 JA |
6539 | { |
6540 | RESIZE_MALLOCED_BUFFER (ret, rptr, STRLEN(str) + 1, rsize, 64); | |
6541 | strcpy (ret + rptr, str); | |
6542 | } | |
ccc6cda3 JA |
6543 | else |
6544 | ret[rptr] = '\0'; | |
6545 | ||
6546 | return ret; | |
6547 | } | |
6548 | ||
6549 | /* Do pattern match and replacement on the positional parameters. */ | |
6550 | static char * | |
6551 | pos_params_pat_subst (string, pat, rep, mflags) | |
6552 | char *string, *pat, *rep; | |
6553 | int mflags; | |
6554 | { | |
6555 | WORD_LIST *save, *params; | |
6556 | WORD_DESC *w; | |
d3ad40de | 6557 | char *ret; |
e33f2203 | 6558 | int pchar, qflags; |
ccc6cda3 JA |
6559 | |
6560 | save = params = list_rest_of_args (); | |
6561 | if (save == 0) | |
6562 | return ((char *)NULL); | |
6563 | ||
6564 | for ( ; params; params = params->next) | |
6565 | { | |
6566 | ret = pat_subst (params->word->word, pat, rep, mflags); | |
227f982e CR |
6567 | w = alloc_word_desc (); |
6568 | w->word = ret ? ret : savestring (""); | |
ccc6cda3 JA |
6569 | dispose_word (params->word); |
6570 | params->word = w; | |
ccc6cda3 JA |
6571 | } |
6572 | ||
e33f2203 CR |
6573 | pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@'; |
6574 | qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0; | |
6575 | ||
6576 | #if 0 | |
d3a24ed2 CR |
6577 | if ((mflags & (MATCH_QUOTED|MATCH_STARSUB)) == (MATCH_QUOTED|MATCH_STARSUB)) |
6578 | ret = string_list_dollar_star (quote_list (save)); | |
e33f2203 CR |
6579 | else if ((mflags & MATCH_STARSUB) == MATCH_STARSUB) |
6580 | ret = string_list_dollar_star (save); | |
6581 | else if ((mflags & MATCH_QUOTED) == MATCH_QUOTED) | |
6582 | ret = string_list_dollar_at (save, qflags); | |
d3a24ed2 | 6583 | else |
e33f2203 CR |
6584 | ret = string_list_dollar_star (save); |
6585 | #else | |
6586 | ret = string_list_pos_params (pchar, save, qflags); | |
6587 | #endif | |
6588 | ||
ccc6cda3 JA |
6589 | dispose_words (save); |
6590 | ||
6591 | return (ret); | |
6592 | } | |
6593 | ||
cce855bc JA |
6594 | /* Perform pattern substitution on VALUE, which is the expansion of |
6595 | VARNAME. PATSUB is an expression supplying the pattern to match | |
6596 | and the string to substitute. QUOTED is a flags word containing | |
6597 | the type of quoting currently in effect. */ | |
ccc6cda3 | 6598 | static char * |
5f8cde23 CR |
6599 | parameter_brace_patsub (varname, value, ind, patsub, quoted, flags) |
6600 | char *varname, *value; | |
6601 | int ind; | |
6602 | char *patsub; | |
6603 | int quoted, flags; | |
ccc6cda3 | 6604 | { |
c2a47ea9 | 6605 | int vtype, mflags, starsub, delim; |
7117c2d2 | 6606 | char *val, *temp, *pat, *rep, *p, *lpatsub, *tt; |
ccc6cda3 JA |
6607 | SHELL_VAR *v; |
6608 | ||
6609 | if (value == 0) | |
6610 | return ((char *)NULL); | |
6611 | ||
6612 | this_command_name = varname; | |
6613 | ||
5f8cde23 | 6614 | vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); |
ccc6cda3 JA |
6615 | if (vtype == -1) |
6616 | return ((char *)NULL); | |
6617 | ||
d3a24ed2 CR |
6618 | starsub = vtype & VT_STARSUB; |
6619 | vtype &= ~VT_STARSUB; | |
6620 | ||
ccc6cda3 | 6621 | mflags = 0; |
77638cbf CR |
6622 | /* PATSUB is never NULL when this is called. */ |
6623 | if (*patsub == '/') | |
d3ad40de CR |
6624 | { |
6625 | mflags |= MATCH_GLOBREP; | |
6626 | patsub++; | |
6627 | } | |
7117c2d2 JA |
6628 | |
6629 | /* Malloc this because expand_string_if_necessary or one of the expansion | |
6630 | functions in its call chain may free it on a substitution error. */ | |
bb70624e | 6631 | lpatsub = savestring (patsub); |
ccc6cda3 JA |
6632 | |
6633 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
6634 | mflags |= MATCH_QUOTED; | |
6635 | ||
d3a24ed2 CR |
6636 | if (starsub) |
6637 | mflags |= MATCH_STARSUB; | |
6638 | ||
dc8fbaf9 CR |
6639 | /* If the pattern starts with a `/', make sure we skip over it when looking |
6640 | for the replacement delimiter. */ | |
4d8d005b | 6641 | delim = skip_to_delim (lpatsub, ((*patsub == '/') ? 1 : 0), "/", 0); |
c2a47ea9 CR |
6642 | if (lpatsub[delim] == '/') |
6643 | { | |
6644 | lpatsub[delim] = 0; | |
6645 | rep = lpatsub + delim + 1; | |
6646 | } | |
6647 | else | |
6648 | rep = (char *)NULL; | |
ccc6cda3 JA |
6649 | |
6650 | if (rep && *rep == '\0') | |
6651 | rep = (char *)NULL; | |
6652 | ||
5e13499c CR |
6653 | /* Perform the same expansions on the pattern as performed by the |
6654 | pattern removal expansions. */ | |
6655 | pat = getpattern (lpatsub, quoted, 1); | |
bb70624e | 6656 | |
ccc6cda3 | 6657 | if (rep) |
e73012f1 CR |
6658 | /* We want to perform quote removal on the expanded replacement even if |
6659 | the entire expansion is double-quoted because the parser and string | |
6660 | extraction functions treated quotes in the replacement string as | |
6661 | special. */ | |
6662 | rep = expand_string_if_necessary (rep, quoted & ~(Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT), expand_string_unsplit); | |
ccc6cda3 | 6663 | |
dc8fbaf9 | 6664 | /* ksh93 doesn't allow the match specifier to be a part of the expanded |
d3ad40de CR |
6665 | pattern. This is an extension. Make sure we don't anchor the pattern |
6666 | at the beginning or end of the string if we're doing global replacement, | |
6667 | though. */ | |
ccc6cda3 | 6668 | p = pat; |
d3ad40de CR |
6669 | if (mflags & MATCH_GLOBREP) |
6670 | mflags |= MATCH_ANY; | |
dc8fbaf9 | 6671 | else if (pat && pat[0] == '#') |
ccc6cda3 JA |
6672 | { |
6673 | mflags |= MATCH_BEG; | |
6674 | p++; | |
6675 | } | |
d166f048 | 6676 | else if (pat && pat[0] == '%') |
ccc6cda3 JA |
6677 | { |
6678 | mflags |= MATCH_END; | |
6679 | p++; | |
6680 | } | |
6681 | else | |
6682 | mflags |= MATCH_ANY; | |
6683 | ||
cce855bc JA |
6684 | /* OK, we now want to substitute REP for PAT in VAL. If |
6685 | flags & MATCH_GLOBREP is non-zero, the substitution is done | |
6686 | everywhere, otherwise only the first occurrence of PAT is | |
7117c2d2 JA |
6687 | replaced. The pattern matching code doesn't understand |
6688 | CTLESC quoting CTLESC and CTLNUL so we use the dequoted variable | |
6689 | values passed in (VT_VARIABLE) so the pattern substitution | |
6690 | code works right. We need to requote special chars after | |
6691 | we're done for VT_VARIABLE and VT_ARRAYMEMBER, and for the | |
6692 | other cases if QUOTED == 0, since the posparams and arrays | |
6693 | indexed by * or @ do special things when QUOTED != 0. */ | |
6694 | ||
ccc6cda3 JA |
6695 | switch (vtype) |
6696 | { | |
6697 | case VT_VARIABLE: | |
d166f048 | 6698 | case VT_ARRAYMEMBER: |
ccc6cda3 | 6699 | temp = pat_subst (val, p, rep, mflags); |
7117c2d2 JA |
6700 | if (vtype == VT_VARIABLE) |
6701 | FREE (val); | |
6702 | if (temp) | |
6703 | { | |
e6598ba4 | 6704 | tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp); |
7117c2d2 JA |
6705 | free (temp); |
6706 | temp = tt; | |
6707 | } | |
ccc6cda3 JA |
6708 | break; |
6709 | case VT_POSPARMS: | |
6710 | temp = pos_params_pat_subst (val, p, rep, mflags); | |
7117c2d2 JA |
6711 | if (temp && (mflags & MATCH_QUOTED) == 0) |
6712 | { | |
6713 | tt = quote_escapes (temp); | |
6714 | free (temp); | |
6715 | temp = tt; | |
6716 | } | |
ccc6cda3 JA |
6717 | break; |
6718 | #if defined (ARRAY_VARS) | |
6719 | case VT_ARRAYVAR: | |
fdf670ea CR |
6720 | temp = assoc_p (v) ? assoc_patsub (assoc_cell (v), p, rep, mflags) |
6721 | : array_patsub (array_cell (v), p, rep, mflags); | |
e6598ba4 CR |
6722 | /* Don't call quote_escapes anymore; array_patsub calls |
6723 | array_quote_escapes as appropriate before adding the | |
fdf670ea | 6724 | space separators; ditto for assoc_patsub. */ |
ccc6cda3 JA |
6725 | break; |
6726 | #endif | |
6727 | } | |
6728 | ||
6729 | FREE (pat); | |
6730 | FREE (rep); | |
bb70624e | 6731 | free (lpatsub); |
ccc6cda3 JA |
6732 | |
6733 | return temp; | |
6734 | } | |
6735 | ||
09767ff0 CR |
6736 | /****************************************************************/ |
6737 | /* */ | |
6738 | /* Functions to perform case modification on variable values */ | |
6739 | /* */ | |
6740 | /****************************************************************/ | |
6741 | ||
6742 | /* Do case modification on the positional parameters. */ | |
6743 | ||
6744 | static char * | |
6745 | pos_params_modcase (string, pat, modop, mflags) | |
6746 | char *string, *pat; | |
6747 | int modop; | |
6748 | int mflags; | |
6749 | { | |
6750 | WORD_LIST *save, *params; | |
6751 | WORD_DESC *w; | |
6752 | char *ret; | |
6753 | int pchar, qflags; | |
6754 | ||
6755 | save = params = list_rest_of_args (); | |
6756 | if (save == 0) | |
6757 | return ((char *)NULL); | |
6758 | ||
6759 | for ( ; params; params = params->next) | |
6760 | { | |
6761 | ret = sh_modcase (params->word->word, pat, modop); | |
6762 | w = alloc_word_desc (); | |
6763 | w->word = ret ? ret : savestring (""); | |
6764 | dispose_word (params->word); | |
6765 | params->word = w; | |
6766 | } | |
6767 | ||
6768 | pchar = (mflags & MATCH_STARSUB) == MATCH_STARSUB ? '*' : '@'; | |
6769 | qflags = (mflags & MATCH_QUOTED) == MATCH_QUOTED ? Q_DOUBLE_QUOTES : 0; | |
6770 | ||
6771 | ret = string_list_pos_params (pchar, save, qflags); | |
6772 | dispose_words (save); | |
6773 | ||
6774 | return (ret); | |
6775 | } | |
6776 | ||
6777 | /* Perform case modification on VALUE, which is the expansion of | |
6778 | VARNAME. MODSPEC is an expression supplying the type of modification | |
6779 | to perform. QUOTED is a flags word containing the type of quoting | |
6780 | currently in effect. */ | |
6781 | static char * | |
5f8cde23 | 6782 | parameter_brace_casemod (varname, value, ind, modspec, patspec, quoted, flags) |
09767ff0 | 6783 | char *varname, *value; |
5f8cde23 | 6784 | int ind, modspec; |
09767ff0 | 6785 | char *patspec; |
5f8cde23 | 6786 | int quoted, flags; |
09767ff0 CR |
6787 | { |
6788 | int vtype, starsub, modop, mflags, x; | |
6789 | char *val, *temp, *pat, *p, *lpat, *tt; | |
6790 | SHELL_VAR *v; | |
6791 | ||
6792 | if (value == 0) | |
6793 | return ((char *)NULL); | |
6794 | ||
6795 | this_command_name = varname; | |
6796 | ||
5f8cde23 | 6797 | vtype = get_var_and_type (varname, value, ind, quoted, flags, &v, &val); |
09767ff0 CR |
6798 | if (vtype == -1) |
6799 | return ((char *)NULL); | |
6800 | ||
6801 | starsub = vtype & VT_STARSUB; | |
6802 | vtype &= ~VT_STARSUB; | |
6803 | ||
6804 | modop = 0; | |
6805 | mflags = 0; | |
6806 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
6807 | mflags |= MATCH_QUOTED; | |
6808 | if (starsub) | |
6809 | mflags |= MATCH_STARSUB; | |
6810 | ||
6811 | p = patspec; | |
6812 | if (modspec == '^') | |
6813 | { | |
6814 | x = p && p[0] == modspec; | |
e141c35a | 6815 | modop = x ? CASE_UPPER : CASE_UPFIRST; |
09767ff0 CR |
6816 | p += x; |
6817 | } | |
6818 | else if (modspec == ',') | |
6819 | { | |
6820 | x = p && p[0] == modspec; | |
e141c35a | 6821 | modop = x ? CASE_LOWER : CASE_LOWFIRST; |
09767ff0 CR |
6822 | p += x; |
6823 | } | |
6824 | else if (modspec == '~') | |
6825 | { | |
6826 | x = p && p[0] == modspec; | |
6827 | modop = x ? CASE_TOGGLEALL : CASE_TOGGLE; | |
6828 | p += x; | |
6829 | } | |
6830 | ||
6831 | lpat = p ? savestring (p) : 0; | |
6832 | /* Perform the same expansions on the pattern as performed by the | |
6833 | pattern removal expansions. FOR LATER */ | |
6834 | pat = lpat ? getpattern (lpat, quoted, 1) : 0; | |
6835 | ||
6836 | /* OK, now we do the case modification. */ | |
6837 | switch (vtype) | |
6838 | { | |
6839 | case VT_VARIABLE: | |
6840 | case VT_ARRAYMEMBER: | |
6841 | temp = sh_modcase (val, pat, modop); | |
6842 | if (vtype == VT_VARIABLE) | |
6843 | FREE (val); | |
6844 | if (temp) | |
6845 | { | |
6846 | tt = (mflags & MATCH_QUOTED) ? quote_string (temp) : quote_escapes (temp); | |
6847 | free (temp); | |
6848 | temp = tt; | |
6849 | } | |
6850 | break; | |
6851 | ||
6852 | case VT_POSPARMS: | |
6853 | temp = pos_params_modcase (val, pat, modop, mflags); | |
6854 | if (temp && (mflags & MATCH_QUOTED) == 0) | |
6855 | { | |
6856 | tt = quote_escapes (temp); | |
6857 | free (temp); | |
6858 | temp = tt; | |
6859 | } | |
6860 | break; | |
6861 | ||
6862 | #if defined (ARRAY_VARS) | |
6863 | case VT_ARRAYVAR: | |
6864 | temp = assoc_p (v) ? assoc_modcase (assoc_cell (v), pat, modop, mflags) | |
6865 | : array_modcase (array_cell (v), pat, modop, mflags); | |
6866 | /* Don't call quote_escapes; array_modcase calls array_quote_escapes | |
6867 | as appropriate before adding the space separators; ditto for | |
6868 | assoc_modcase. */ | |
6869 | break; | |
6870 | #endif | |
6871 | } | |
6872 | ||
6873 | FREE (pat); | |
6874 | free (lpat); | |
6875 | ||
6876 | return temp; | |
6877 | } | |
6878 | ||
d3ad40de CR |
6879 | /* Check for unbalanced parens in S, which is the contents of $(( ... )). If |
6880 | any occur, this must be a nested command substitution, so return 0. | |
6881 | Otherwise, return 1. A valid arithmetic expression must always have a | |
6882 | ( before a matching ), so any cases where there are more right parens | |
6883 | means that this must not be an arithmetic expression, though the parser | |
6884 | will not accept it without a balanced total number of parens. */ | |
6885 | static int | |
6886 | chk_arithsub (s, len) | |
6887 | const char *s; | |
6888 | int len; | |
6889 | { | |
6890 | int i, count; | |
6891 | DECLARE_MBSTATE; | |
6892 | ||
6893 | i = count = 0; | |
6894 | while (i < len) | |
6895 | { | |
176b12ee | 6896 | if (s[i] == LPAREN) |
d3ad40de | 6897 | count++; |
176b12ee | 6898 | else if (s[i] == RPAREN) |
d3ad40de CR |
6899 | { |
6900 | count--; | |
6901 | if (count < 0) | |
6902 | return 0; | |
6903 | } | |
6904 | ||
6905 | switch (s[i]) | |
6906 | { | |
6907 | default: | |
6908 | ADVANCE_CHAR (s, len, i); | |
6909 | break; | |
6910 | ||
6911 | case '\\': | |
6912 | i++; | |
6913 | if (s[i]) | |
6914 | ADVANCE_CHAR (s, len, i); | |
6915 | break; | |
6916 | ||
6917 | case '\'': | |
6918 | i = skip_single_quoted (s, len, ++i); | |
6919 | break; | |
6920 | ||
6921 | case '"': | |
6922 | i = skip_double_quoted ((char *)s, len, ++i); | |
6923 | break; | |
6924 | } | |
6925 | } | |
6926 | ||
6927 | return (count == 0); | |
6928 | } | |
6929 | ||
cce855bc JA |
6930 | /****************************************************************/ |
6931 | /* */ | |
6932 | /* Functions to perform parameter expansion on a string */ | |
6933 | /* */ | |
6934 | /****************************************************************/ | |
6935 | ||
09767ff0 | 6936 | /* ${[#][!]name[[:][^[^]][,[,]]#[#]%[%]-=?+[word][:e1[:e2]]]} */ |
227f982e | 6937 | static WORD_DESC * |
e1e48bba | 6938 | parameter_brace_expand (string, indexp, quoted, pflags, quoted_dollar_atp, contains_dollar_at) |
ccc6cda3 | 6939 | char *string; |
e1e48bba | 6940 | int *indexp, quoted, *quoted_dollar_atp, *contains_dollar_at, pflags; |
ccc6cda3 JA |
6941 | { |
6942 | int check_nullness, var_is_set, var_is_null, var_is_special; | |
09767ff0 | 6943 | int want_substring, want_indir, want_patsub, want_casemod; |
ccc6cda3 | 6944 | char *name, *value, *temp, *temp1; |
227f982e | 6945 | WORD_DESC *tdesc, *ret; |
09767ff0 | 6946 | int t_index, sindex, c, tflag, modspec; |
7117c2d2 | 6947 | intmax_t number; |
5f8cde23 | 6948 | arrayind_t ind; |
ccc6cda3 | 6949 | |
c40a57dd | 6950 | temp = temp1 = value = (char *)NULL; |
ccc6cda3 | 6951 | var_is_set = var_is_null = var_is_special = check_nullness = 0; |
09767ff0 | 6952 | want_substring = want_indir = want_patsub = want_casemod = 0; |
ccc6cda3 | 6953 | |
cce855bc JA |
6954 | sindex = *indexp; |
6955 | t_index = ++sindex; | |
11a6f9a9 CR |
6956 | /* ${#var} doesn't have any of the other parameter expansions on it. */ |
6957 | if (string[t_index] == '#' && legal_variable_starter (string[t_index+1])) /* {{ */ | |
e6598ba4 | 6958 | name = string_extract (string, &t_index, "}", SX_VARNAME); |
11a6f9a9 | 6959 | else |
09767ff0 CR |
6960 | #if defined (CASEMOD_EXPANSIONS) |
6961 | /* To enable case-toggling expansions using the `~' operator character | |
6962 | change the 1 to 0. */ | |
6963 | # if defined (CASEMOD_CAPCASE) | |
6964 | name = string_extract (string, &t_index, "#%^,~:-=?+/}", SX_VARNAME); | |
6965 | # else | |
6966 | name = string_extract (string, &t_index, "#%^,:-=?+/}", SX_VARNAME); | |
6967 | # endif /* CASEMOD_CAPCASE */ | |
6968 | #else | |
e6598ba4 | 6969 | name = string_extract (string, &t_index, "#%:-=?+/}", SX_VARNAME); |
09767ff0 | 6970 | #endif /* CASEMOD_EXPANSIONS */ |
cce855bc | 6971 | |
227f982e CR |
6972 | ret = 0; |
6973 | tflag = 0; | |
6974 | ||
5f8cde23 CR |
6975 | ind = INTMAX_MIN; |
6976 | ||
cce855bc JA |
6977 | /* If the name really consists of a special variable, then make sure |
6978 | that we have the entire name. We don't allow indirect references | |
6979 | to special variables except `#', `?', `@' and `*'. */ | |
e05be32d CR |
6980 | if ((sindex == t_index && VALID_SPECIAL_LENGTH_PARAM (string[t_index])) || |
6981 | (sindex == t_index - 1 && string[sindex] == '!' && VALID_INDIR_PARAM (string[t_index]))) | |
ccc6cda3 JA |
6982 | { |
6983 | t_index++; | |
ccc6cda3 | 6984 | temp1 = string_extract (string, &t_index, "#%:-=?+/}", 0); |
631b20c6 | 6985 | name = (char *)xrealloc (name, 3 + (strlen (temp1))); |
ccc6cda3 JA |
6986 | *name = string[sindex]; |
6987 | if (string[sindex] == '!') | |
6988 | { | |
28ef6c31 JA |
6989 | /* indirect reference of $#, $?, $@, or $* */ |
6990 | name[1] = string[sindex + 1]; | |
6991 | strcpy (name + 2, temp1); | |
ccc6cda3 | 6992 | } |
cce855bc | 6993 | else |
ccc6cda3 JA |
6994 | strcpy (name + 1, temp1); |
6995 | free (temp1); | |
6996 | } | |
6997 | sindex = t_index; | |
6998 | ||
6999 | /* Find out what character ended the variable name. Then | |
7000 | do the appropriate thing. */ | |
7001 | if (c = string[sindex]) | |
7002 | sindex++; | |
7003 | ||
7004 | /* If c is followed by one of the valid parameter expansion | |
7005 | characters, move past it as normal. If not, assume that | |
7006 | a substring specification is being given, and do not move | |
7007 | past it. */ | |
28ef6c31 | 7008 | if (c == ':' && VALID_PARAM_EXPAND_CHAR (string[sindex])) |
ccc6cda3 JA |
7009 | { |
7010 | check_nullness++; | |
7011 | if (c = string[sindex]) | |
7012 | sindex++; | |
7013 | } | |
cce855bc | 7014 | else if (c == ':' && string[sindex] != RBRACE) |
ccc6cda3 | 7015 | want_substring = 1; |
77638cbf | 7016 | else if (c == '/' /* && string[sindex] != RBRACE */) /* XXX */ |
ccc6cda3 | 7017 | want_patsub = 1; |
09767ff0 CR |
7018 | #if defined (CASEMOD_EXPANSIONS) |
7019 | else if (c == '^' || c == ',' || c == '~') | |
7020 | { | |
7021 | modspec = c; | |
7022 | want_casemod = 1; | |
7023 | } | |
7024 | #endif | |
ccc6cda3 | 7025 | |
cce855bc JA |
7026 | /* Catch the valid and invalid brace expressions that made it through the |
7027 | tests above. */ | |
7028 | /* ${#-} is a valid expansion and means to take the length of $-. | |
7029 | Similarly for ${#?} and ${##}... */ | |
7030 | if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 && | |
28ef6c31 | 7031 | VALID_SPECIAL_LENGTH_PARAM (c) && string[sindex] == RBRACE) |
cce855bc | 7032 | { |
f73dda09 | 7033 | name = (char *)xrealloc (name, 3); |
cce855bc JA |
7034 | name[1] = c; |
7035 | name[2] = '\0'; | |
7036 | c = string[sindex++]; | |
7037 | } | |
7038 | ||
7039 | /* ...but ${#%}, ${#:}, ${#=}, ${#+}, and ${#/} are errors. */ | |
7040 | if (name[0] == '#' && name[1] == '\0' && check_nullness == 0 && | |
7041 | member (c, "%:=+/") && string[sindex] == RBRACE) | |
7042 | { | |
7043 | temp = (char *)NULL; | |
7044 | goto bad_substitution; | |
7045 | } | |
e33f2203 | 7046 | |
cce855bc JA |
7047 | /* Indirect expansion begins with a `!'. A valid indirect expansion is |
7048 | either a variable name, one of the positional parameters or a special | |
7049 | variable that expands to one of the positional parameters. */ | |
7050 | want_indir = *name == '!' && | |
f73dda09 | 7051 | (legal_variable_starter ((unsigned char)name[1]) || DIGIT (name[1]) |
7117c2d2 | 7052 | || VALID_INDIR_PARAM (name[1])); |
ccc6cda3 JA |
7053 | |
7054 | /* Determine the value of this variable. */ | |
7055 | ||
cce855bc | 7056 | /* Check for special variables, directly referenced. */ |
bb70624e | 7057 | if (SPECIAL_VAR (name, want_indir)) |
ccc6cda3 JA |
7058 | var_is_special++; |
7059 | ||
cce855bc JA |
7060 | /* Check for special expansion things, like the length of a parameter */ |
7061 | if (*name == '#' && name[1]) | |
ccc6cda3 | 7062 | { |
cce855bc | 7063 | /* If we are not pointing at the character just after the |
28ef6c31 JA |
7064 | closing brace, then we haven't gotten all of the name. |
7065 | Since it begins with a special character, this is a bad | |
7066 | substitution. Also check NAME for validity before trying | |
7067 | to go on. */ | |
cce855bc | 7068 | if (string[sindex - 1] != RBRACE || (valid_length_expression (name) == 0)) |
ccc6cda3 JA |
7069 | { |
7070 | temp = (char *)NULL; | |
7071 | goto bad_substitution; | |
7072 | } | |
7073 | ||
7074 | number = parameter_brace_expand_length (name); | |
e05be32d CR |
7075 | if (number == INTMAX_MIN && unbound_vars_is_error) |
7076 | { | |
7077 | last_command_exit_value = EXECUTION_FAILURE; | |
7078 | err_unboundvar (name+1); | |
7079 | free (name); | |
7080 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); | |
7081 | } | |
ccc6cda3 JA |
7082 | free (name); |
7083 | ||
7084 | *indexp = sindex; | |
227f982e CR |
7085 | if (number < 0) |
7086 | return (&expand_wdesc_error); | |
7087 | else | |
7088 | { | |
7089 | ret = alloc_word_desc (); | |
7090 | ret->word = itos (number); | |
7091 | return ret; | |
7092 | } | |
ccc6cda3 JA |
7093 | } |
7094 | ||
7095 | /* ${@} is identical to $@. */ | |
7096 | if (name[0] == '@' && name[1] == '\0') | |
7097 | { | |
7098 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
7099 | *quoted_dollar_atp = 1; | |
7100 | ||
7101 | if (contains_dollar_at) | |
7102 | *contains_dollar_at = 1; | |
7103 | } | |
7104 | ||
d3a24ed2 | 7105 | /* Process ${!PREFIX*} expansion. */ |
bb70624e JA |
7106 | if (want_indir && string[sindex - 1] == RBRACE && |
7107 | (string[sindex - 2] == '*' || string[sindex - 2] == '@') && | |
f73dda09 | 7108 | legal_variable_starter ((unsigned char) name[1])) |
bb70624e JA |
7109 | { |
7110 | char **x; | |
7111 | WORD_LIST *xlist; | |
7112 | ||
7113 | temp1 = savestring (name + 1); | |
7114 | number = strlen (temp1); | |
7115 | temp1[number - 1] = '\0'; | |
7116 | x = all_variables_matching_prefix (temp1); | |
7117c2d2 | 7117 | xlist = strvec_to_word_list (x, 0, 0); |
28ef6c31 JA |
7118 | if (string[sindex - 2] == '*') |
7119 | temp = string_list_dollar_star (xlist); | |
7120 | else | |
7121 | { | |
7122 | temp = string_list_dollar_at (xlist, quoted); | |
7123 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
7124 | *quoted_dollar_atp = 1; | |
7125 | if (contains_dollar_at) | |
7126 | *contains_dollar_at = 1; | |
7127 | } | |
bb70624e | 7128 | free (x); |
af98a2a6 | 7129 | dispose_words (xlist); |
bb70624e JA |
7130 | free (temp1); |
7131 | *indexp = sindex; | |
227f982e | 7132 | |
631b20c6 CR |
7133 | free (name); |
7134 | ||
227f982e CR |
7135 | ret = alloc_word_desc (); |
7136 | ret->word = temp; | |
7137 | return ret; | |
bb70624e | 7138 | } |
d3a24ed2 CR |
7139 | |
7140 | #if defined (ARRAY_VARS) | |
5e13499c CR |
7141 | /* Process ${!ARRAY[@]} and ${!ARRAY[*]} expansion. */ /* [ */ |
7142 | if (want_indir && string[sindex - 1] == RBRACE && | |
7143 | string[sindex - 2] == ']' && valid_array_reference (name+1)) | |
d3a24ed2 CR |
7144 | { |
7145 | char *x, *x1; | |
7146 | ||
7147 | temp1 = savestring (name + 1); | |
7148 | x = array_variable_name (temp1, &x1, (int *)0); /* [ */ | |
7149 | FREE (x); | |
7150 | if (ALL_ELEMENT_SUB (x1[0]) && x1[1] == ']') | |
7151 | { | |
fdf670ea | 7152 | temp = array_keys (temp1, quoted); /* handles assoc vars too */ |
d3a24ed2 CR |
7153 | if (x1[0] == '@') |
7154 | { | |
7155 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
7156 | *quoted_dollar_atp = 1; | |
7157 | if (contains_dollar_at) | |
7158 | *contains_dollar_at = 1; | |
7159 | } | |
7160 | ||
7161 | free (temp1); | |
7162 | *indexp = sindex; | |
227f982e CR |
7163 | |
7164 | ret = alloc_word_desc (); | |
7165 | ret->word = temp; | |
7166 | return ret; | |
d3a24ed2 CR |
7167 | } |
7168 | ||
7169 | free (temp1); | |
7170 | } | |
7171 | #endif /* ARRAY_VARS */ | |
bb70624e | 7172 | |
ccc6cda3 JA |
7173 | /* Make sure that NAME is valid before trying to go on. */ |
7174 | if (valid_brace_expansion_word (want_indir ? name + 1 : name, | |
7175 | var_is_special) == 0) | |
7176 | { | |
7177 | temp = (char *)NULL; | |
7178 | goto bad_substitution; | |
7179 | } | |
7180 | ||
7181 | if (want_indir) | |
227f982e | 7182 | tdesc = parameter_brace_expand_indir (name + 1, var_is_special, quoted, quoted_dollar_atp, contains_dollar_at); |
ccc6cda3 | 7183 | else |
5f8cde23 | 7184 | tdesc = parameter_brace_expand_word (name, var_is_special, quoted, PF_IGNUNBOUND|(pflags&PF_NOSPLIT2), &ind); |
227f982e CR |
7185 | |
7186 | if (tdesc) | |
7187 | { | |
7188 | temp = tdesc->word; | |
7189 | tflag = tdesc->flags; | |
7190 | dispose_word_desc (tdesc); | |
7191 | } | |
7192 | else | |
7193 | temp = (char *)0; | |
ccc6cda3 JA |
7194 | |
7195 | #if defined (ARRAY_VARS) | |
cce855bc | 7196 | if (valid_array_reference (name)) |
d3a24ed2 | 7197 | chk_atstar (name, quoted, quoted_dollar_atp, contains_dollar_at); |
ccc6cda3 JA |
7198 | #endif |
7199 | ||
7200 | var_is_set = temp != (char *)0; | |
7201 | var_is_null = check_nullness && (var_is_set == 0 || *temp == 0); | |
7202 | ||
7203 | /* Get the rest of the stuff inside the braces. */ | |
cce855bc | 7204 | if (c && c != RBRACE) |
ccc6cda3 JA |
7205 | { |
7206 | /* Extract the contents of the ${ ... } expansion | |
28ef6c31 | 7207 | according to the Posix.2 rules. */ |
e192f341 | 7208 | value = extract_dollar_brace_string (string, &sindex, quoted, (c == '%' || c == '#' || c =='/' || c == '^' || c == ',' || c ==':') ? SX_POSIXEXP|SX_WORD : SX_WORD); |
cce855bc | 7209 | if (string[sindex] == RBRACE) |
28ef6c31 | 7210 | sindex++; |
ccc6cda3 JA |
7211 | else |
7212 | goto bad_substitution; | |
7213 | } | |
7214 | else | |
7215 | value = (char *)NULL; | |
726f6388 | 7216 | |
ccc6cda3 JA |
7217 | *indexp = sindex; |
7218 | ||
e05be32d CR |
7219 | /* All the cases where an expansion can possibly generate an unbound |
7220 | variable error. */ | |
7221 | if (want_substring || want_patsub || want_casemod || c == '#' || c == '%' || c == RBRACE) | |
7222 | { | |
7223 | if (var_is_set == 0 && unbound_vars_is_error && ((name[0] != '@' && name[0] != '*') || name[1])) | |
7224 | { | |
7225 | last_command_exit_value = EXECUTION_FAILURE; | |
7226 | err_unboundvar (name); | |
7227 | FREE (value); | |
7228 | FREE (temp); | |
7229 | free (name); | |
7230 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); | |
7231 | } | |
7232 | } | |
7233 | ||
ccc6cda3 JA |
7234 | /* If this is a substring spec, process it and add the result. */ |
7235 | if (want_substring) | |
726f6388 | 7236 | { |
5f8cde23 | 7237 | temp1 = parameter_brace_substring (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); |
ccc6cda3 JA |
7238 | FREE (name); |
7239 | FREE (value); | |
7240 | FREE (temp); | |
227f982e CR |
7241 | |
7242 | if (temp1 == &expand_param_error) | |
7243 | return (&expand_wdesc_error); | |
7244 | else if (temp1 == &expand_param_fatal) | |
7245 | return (&expand_wdesc_fatal); | |
7246 | ||
7247 | ret = alloc_word_desc (); | |
7248 | ret->word = temp1; | |
d3ad40de CR |
7249 | if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
7250 | ret->flags |= W_QUOTED|W_HASQUOTEDNULL; | |
227f982e | 7251 | return ret; |
726f6388 | 7252 | } |
ccc6cda3 | 7253 | else if (want_patsub) |
726f6388 | 7254 | { |
5f8cde23 | 7255 | temp1 = parameter_brace_patsub (name, temp, ind, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); |
ccc6cda3 JA |
7256 | FREE (name); |
7257 | FREE (value); | |
7258 | FREE (temp); | |
227f982e CR |
7259 | |
7260 | if (temp1 == &expand_param_error) | |
7261 | return (&expand_wdesc_error); | |
7262 | else if (temp1 == &expand_param_fatal) | |
7263 | return (&expand_wdesc_fatal); | |
7264 | ||
40c8fbee CR |
7265 | ret = alloc_word_desc (); |
7266 | ret->word = temp1; | |
7267 | if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
7268 | ret->flags |= W_QUOTED|W_HASQUOTEDNULL; | |
227f982e | 7269 | return ret; |
ccc6cda3 | 7270 | } |
09767ff0 CR |
7271 | #if defined (CASEMOD_EXPANSIONS) |
7272 | else if (want_casemod) | |
7273 | { | |
5f8cde23 | 7274 | temp1 = parameter_brace_casemod (name, temp, ind, modspec, value, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); |
09767ff0 CR |
7275 | FREE (name); |
7276 | FREE (value); | |
7277 | FREE (temp); | |
7278 | ||
7279 | if (temp1 == &expand_param_error) | |
7280 | return (&expand_wdesc_error); | |
7281 | else if (temp1 == &expand_param_fatal) | |
7282 | return (&expand_wdesc_fatal); | |
7283 | ||
7284 | ret = alloc_word_desc (); | |
7285 | ret->word = temp1; | |
40c8fbee CR |
7286 | if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
7287 | ret->flags |= W_QUOTED|W_HASQUOTEDNULL; | |
09767ff0 CR |
7288 | return ret; |
7289 | } | |
7290 | #endif | |
726f6388 | 7291 | |
ccc6cda3 JA |
7292 | /* Do the right thing based on which character ended the variable name. */ |
7293 | switch (c) | |
7294 | { | |
7295 | default: | |
7296 | case '\0': | |
7297 | bad_substitution: | |
7f947b68 | 7298 | last_command_exit_value = EXECUTION_FAILURE; |
5e13499c | 7299 | report_error (_("%s: bad substitution"), string ? string : "??"); |
ccc6cda3 JA |
7300 | FREE (value); |
7301 | FREE (temp); | |
7302 | free (name); | |
227f982e | 7303 | return &expand_wdesc_error; |
ccc6cda3 | 7304 | |
cce855bc | 7305 | case RBRACE: |
ccc6cda3 | 7306 | break; |
726f6388 | 7307 | |
ccc6cda3 JA |
7308 | case '#': /* ${param#[#]pattern} */ |
7309 | case '%': /* ${param%[%]pattern} */ | |
7310 | if (value == 0 || *value == '\0' || temp == 0 || *temp == '\0') | |
28ef6c31 JA |
7311 | { |
7312 | FREE (value); | |
ccc6cda3 | 7313 | break; |
28ef6c31 | 7314 | } |
5f8cde23 | 7315 | temp1 = parameter_brace_remove_pattern (name, temp, ind, value, c, quoted, (tflag & W_ARRAYIND) ? AV_USEIND : 0); |
ccc6cda3 JA |
7316 | free (temp); |
7317 | free (value); | |
3d35553a | 7318 | free (name); |
40c8fbee CR |
7319 | |
7320 | ret = alloc_word_desc (); | |
7321 | ret->word = temp1; | |
7322 | if (temp1 && QUOTED_NULL (temp1) && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
7323 | ret->flags |= W_QUOTED|W_HASQUOTEDNULL; | |
7324 | return ret; | |
ccc6cda3 JA |
7325 | |
7326 | case '-': | |
7327 | case '=': | |
7328 | case '?': | |
7329 | case '+': | |
7330 | if (var_is_set && var_is_null == 0) | |
28ef6c31 JA |
7331 | { |
7332 | /* If the operator is `+', we don't want the value of the named | |
7333 | variable for anything, just the value of the right hand side. */ | |
ccc6cda3 JA |
7334 | if (c == '+') |
7335 | { | |
28ef6c31 JA |
7336 | /* XXX -- if we're double-quoted and the named variable is "$@", |
7337 | we want to turn off any special handling of "$@" -- | |
7338 | we're not using it, so whatever is on the rhs applies. */ | |
7339 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
7340 | *quoted_dollar_atp = 0; | |
7341 | if (contains_dollar_at) | |
7342 | *contains_dollar_at = 0; | |
7343 | ||
ccc6cda3 JA |
7344 | FREE (temp); |
7345 | if (value) | |
28ef6c31 | 7346 | { |
6faad625 CR |
7347 | /* From Posix discussion on austin-group list. Issue 221 |
7348 | requires that backslashes escaping `}' inside | |
7349 | double-quoted ${...} be removed. */ | |
64419627 CR |
7350 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
7351 | quoted |= Q_DOLBRACE; | |
227f982e CR |
7352 | ret = parameter_brace_expand_rhs (name, value, c, |
7353 | quoted, | |
7354 | quoted_dollar_atp, | |
7355 | contains_dollar_at); | |
7356 | /* XXX - fix up later, esp. noting presence of | |
7357 | W_HASQUOTEDNULL in ret->flags */ | |
ccc6cda3 JA |
7358 | free (value); |
7359 | } | |
7360 | else | |
28ef6c31 | 7361 | temp = (char *)NULL; |
ccc6cda3 JA |
7362 | } |
7363 | else | |
7364 | { | |
7365 | FREE (value); | |
7366 | } | |
7367 | /* Otherwise do nothing; just use the value in TEMP. */ | |
726f6388 | 7368 | } |
ccc6cda3 | 7369 | else /* VAR not set or VAR is NULL. */ |
28ef6c31 | 7370 | { |
ccc6cda3 JA |
7371 | FREE (temp); |
7372 | temp = (char *)NULL; | |
7373 | if (c == '=' && var_is_special) | |
7374 | { | |
7f947b68 | 7375 | last_command_exit_value = EXECUTION_FAILURE; |
5e13499c | 7376 | report_error (_("$%s: cannot assign in this way"), name); |
ccc6cda3 JA |
7377 | free (name); |
7378 | free (value); | |
227f982e | 7379 | return &expand_wdesc_error; |
ccc6cda3 JA |
7380 | } |
7381 | else if (c == '?') | |
7382 | { | |
7383 | parameter_brace_expand_error (name, value); | |
227f982e | 7384 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
ccc6cda3 JA |
7385 | } |
7386 | else if (c != '+') | |
28ef6c31 JA |
7387 | { |
7388 | /* XXX -- if we're double-quoted and the named variable is "$@", | |
7389 | we want to turn off any special handling of "$@" -- | |
7390 | we're not using it, so whatever is on the rhs applies. */ | |
7391 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && quoted_dollar_atp) | |
7392 | *quoted_dollar_atp = 0; | |
7393 | if (contains_dollar_at) | |
7394 | *contains_dollar_at = 0; | |
7395 | ||
6faad625 CR |
7396 | /* From Posix discussion on austin-group list. Issue 221 requires |
7397 | that backslashes escaping `}' inside double-quoted ${...} be | |
7398 | removed. */ | |
64419627 CR |
7399 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
7400 | quoted |= Q_DOLBRACE; | |
227f982e CR |
7401 | ret = parameter_brace_expand_rhs (name, value, c, quoted, |
7402 | quoted_dollar_atp, | |
7403 | contains_dollar_at); | |
7404 | /* XXX - fix up later, esp. noting presence of | |
7405 | W_HASQUOTEDNULL in tdesc->flags */ | |
28ef6c31 | 7406 | } |
ccc6cda3 | 7407 | free (value); |
726f6388 | 7408 | } |
28ef6c31 | 7409 | |
ccc6cda3 | 7410 | break; |
726f6388 | 7411 | } |
ccc6cda3 | 7412 | free (name); |
227f982e CR |
7413 | |
7414 | if (ret == 0) | |
7415 | { | |
7416 | ret = alloc_word_desc (); | |
7417 | ret->flags = tflag; | |
7418 | ret->word = temp; | |
7419 | } | |
7420 | return (ret); | |
726f6388 JA |
7421 | } |
7422 | ||
cce855bc JA |
7423 | /* Expand a single ${xxx} expansion. The braces are optional. When |
7424 | the braces are used, parameter_brace_expand() does the work, | |
7425 | possibly calling param_expand recursively. */ | |
227f982e | 7426 | static WORD_DESC * |
cce855bc JA |
7427 | param_expand (string, sindex, quoted, expanded_something, |
7428 | contains_dollar_at, quoted_dollar_at_p, had_quoted_null_p, | |
7429 | pflags) | |
7430 | char *string; | |
7431 | int *sindex, quoted, *expanded_something, *contains_dollar_at; | |
7432 | int *quoted_dollar_at_p, *had_quoted_null_p, pflags; | |
7433 | { | |
7117c2d2 | 7434 | char *temp, *temp1, uerror[3]; |
f73dda09 JA |
7435 | int zindex, t_index, expok; |
7436 | unsigned char c; | |
7117c2d2 | 7437 | intmax_t number; |
cce855bc | 7438 | SHELL_VAR *var; |
f73dda09 | 7439 | WORD_LIST *list; |
227f982e CR |
7440 | WORD_DESC *tdesc, *ret; |
7441 | int tflag; | |
cce855bc JA |
7442 | |
7443 | zindex = *sindex; | |
7444 | c = string[++zindex]; | |
7445 | ||
7446 | temp = (char *)NULL; | |
227f982e CR |
7447 | ret = tdesc = (WORD_DESC *)NULL; |
7448 | tflag = 0; | |
cce855bc JA |
7449 | |
7450 | /* Do simple cases first. Switch on what follows '$'. */ | |
7451 | switch (c) | |
7452 | { | |
7453 | /* $0 .. $9? */ | |
7454 | case '0': | |
7455 | case '1': | |
7456 | case '2': | |
7457 | case '3': | |
7458 | case '4': | |
7459 | case '5': | |
7460 | case '6': | |
7461 | case '7': | |
7462 | case '8': | |
7463 | case '9': | |
f73dda09 | 7464 | temp1 = dollar_vars[TODIGIT (c)]; |
cce855bc JA |
7465 | if (unbound_vars_is_error && temp1 == (char *)NULL) |
7466 | { | |
7117c2d2 JA |
7467 | uerror[0] = '$'; |
7468 | uerror[1] = c; | |
7469 | uerror[2] = '\0'; | |
cce855bc | 7470 | last_command_exit_value = EXECUTION_FAILURE; |
0d8616ff | 7471 | err_unboundvar (uerror); |
227f982e | 7472 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
cce855bc | 7473 | } |
5ba8ff6e | 7474 | if (temp1) |
762a763b | 7475 | temp = (*temp1 && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
5ba8ff6e CR |
7476 | ? quote_string (temp1) |
7477 | : quote_escapes (temp1); | |
7478 | else | |
7479 | temp = (char *)NULL; | |
227f982e | 7480 | |
cce855bc JA |
7481 | break; |
7482 | ||
7483 | /* $$ -- pid of the invoking shell. */ | |
7484 | case '$': | |
7485 | temp = itos (dollar_dollar_pid); | |
7486 | break; | |
7487 | ||
7488 | /* $# -- number of positional parameters. */ | |
7489 | case '#': | |
7490 | temp = itos (number_of_args ()); | |
7491 | break; | |
7492 | ||
7493 | /* $? -- return value of the last synchronous command. */ | |
7494 | case '?': | |
7495 | temp = itos (last_command_exit_value); | |
7496 | break; | |
7497 | ||
7498 | /* $- -- flags supplied to the shell on invocation or by `set'. */ | |
7499 | case '-': | |
7500 | temp = which_set_flags (); | |
7501 | break; | |
7502 | ||
7503 | /* $! -- Pid of the last asynchronous command. */ | |
7504 | case '!': | |
7505 | /* If no asynchronous pids have been created, expand to nothing. | |
7506 | If `set -u' has been executed, and no async processes have | |
7507 | been created, this is an expansion error. */ | |
7508 | if (last_asynchronous_pid == NO_PID) | |
7509 | { | |
7510 | if (expanded_something) | |
7511 | *expanded_something = 0; | |
7512 | temp = (char *)NULL; | |
7513 | if (unbound_vars_is_error) | |
7514 | { | |
7117c2d2 JA |
7515 | uerror[0] = '$'; |
7516 | uerror[1] = c; | |
7517 | uerror[2] = '\0'; | |
cce855bc | 7518 | last_command_exit_value = EXECUTION_FAILURE; |
0d8616ff | 7519 | err_unboundvar (uerror); |
227f982e | 7520 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
cce855bc JA |
7521 | } |
7522 | } | |
7523 | else | |
f73dda09 | 7524 | temp = itos (last_asynchronous_pid); |
cce855bc JA |
7525 | break; |
7526 | ||
7527 | /* The only difference between this and $@ is when the arg is quoted. */ | |
7528 | case '*': /* `$*' */ | |
7529 | list = list_rest_of_args (); | |
7530 | ||
12ae1612 CR |
7531 | #if 0 |
7532 | /* According to austin-group posix proposal by Geoff Clare in | |
7533 | <20090505091501.GA10097@squonk.masqnet> of 5 May 2009: | |
7534 | ||
7535 | "The shell shall write a message to standard error and | |
7536 | immediately exit when it tries to expand an unset parameter | |
7537 | other than the '@' and '*' special parameters." | |
7538 | */ | |
7539 | ||
1231ac47 | 7540 | if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0) |
57a3f689 CR |
7541 | { |
7542 | uerror[0] = '$'; | |
7543 | uerror[1] = '*'; | |
7544 | uerror[2] = '\0'; | |
57a3f689 | 7545 | last_command_exit_value = EXECUTION_FAILURE; |
0d8616ff | 7546 | err_unboundvar (uerror); |
57a3f689 CR |
7547 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
7548 | } | |
12ae1612 | 7549 | #endif |
57a3f689 | 7550 | |
cce855bc JA |
7551 | /* If there are no command-line arguments, this should just |
7552 | disappear if there are other characters in the expansion, | |
7553 | even if it's quoted. */ | |
7554 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && list == 0) | |
7555 | temp = (char *)NULL; | |
1231ac47 | 7556 | else if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES|Q_PATQUOTE)) |
cce855bc JA |
7557 | { |
7558 | /* If we have "$*" we want to make a string of the positional | |
7559 | parameters, separated by the first character of $IFS, and | |
7560 | quote the whole string, including the separators. If IFS | |
7561 | is unset, the parameters are separated by ' '; if $IFS is | |
7562 | null, the parameters are concatenated. */ | |
1231ac47 | 7563 | temp = (quoted & (Q_DOUBLE_QUOTES|Q_PATQUOTE)) ? string_list_dollar_star (list) : string_list (list); |
adc6cff5 CR |
7564 | if (temp) |
7565 | { | |
7566 | temp1 = quote_string (temp); | |
7567 | if (*temp == 0) | |
7568 | tflag |= W_HASQUOTEDNULL; | |
7569 | free (temp); | |
7570 | temp = temp1; | |
7571 | } | |
cce855bc JA |
7572 | } |
7573 | else | |
28ef6c31 | 7574 | { |
227f982e CR |
7575 | /* We check whether or not we're eventually going to split $* here, |
7576 | for example when IFS is empty and we are processing the rhs of | |
7577 | an assignment statement. In that case, we don't separate the | |
7578 | arguments at all. Otherwise, if the $* is not quoted it is | |
7579 | identical to $@ */ | |
7580 | #if 1 | |
7581 | # if defined (HANDLE_MULTIBYTE) | |
7582 | if (expand_no_split_dollar_star && ifs_firstc[0] == 0) | |
7583 | # else | |
7584 | if (expand_no_split_dollar_star && ifs_firstc == 0) | |
7585 | # endif | |
7586 | temp = string_list_dollar_star (list); | |
7587 | else | |
7588 | temp = string_list_dollar_at (list, quoted); | |
7589 | #else | |
28ef6c31 | 7590 | temp = string_list_dollar_at (list, quoted); |
227f982e | 7591 | #endif |
28ef6c31 JA |
7592 | if (expand_no_split_dollar_star == 0 && contains_dollar_at) |
7593 | *contains_dollar_at = 1; | |
7594 | } | |
cce855bc JA |
7595 | |
7596 | dispose_words (list); | |
7597 | break; | |
7598 | ||
7599 | /* When we have "$@" what we want is "$1" "$2" "$3" ... This | |
7600 | means that we have to turn quoting off after we split into | |
7601 | the individually quoted arguments so that the final split | |
7602 | on the first character of $IFS is still done. */ | |
7603 | case '@': /* `$@' */ | |
7604 | list = list_rest_of_args (); | |
7605 | ||
12ae1612 CR |
7606 | #if 0 |
7607 | /* According to austin-group posix proposal by Geoff Clare in | |
7608 | <20090505091501.GA10097@squonk.masqnet> of 5 May 2009: | |
7609 | ||
7610 | "The shell shall write a message to standard error and | |
7611 | immediately exit when it tries to expand an unset parameter | |
7612 | other than the '@' and '*' special parameters." | |
7613 | */ | |
7614 | ||
1231ac47 | 7615 | if (list == 0 && unbound_vars_is_error && (pflags & PF_IGNUNBOUND) == 0) |
57a3f689 CR |
7616 | { |
7617 | uerror[0] = '$'; | |
7618 | uerror[1] = '@'; | |
7619 | uerror[2] = '\0'; | |
57a3f689 | 7620 | last_command_exit_value = EXECUTION_FAILURE; |
0d8616ff | 7621 | err_unboundvar (uerror); |
57a3f689 CR |
7622 | return (interactive_shell ? &expand_wdesc_error : &expand_wdesc_fatal); |
7623 | } | |
12ae1612 | 7624 | #endif |
57a3f689 | 7625 | |
cce855bc JA |
7626 | /* We want to flag the fact that we saw this. We can't turn |
7627 | off quoting entirely, because other characters in the | |
7628 | string might need it (consider "\"$@\""), but we need some | |
7629 | way to signal that the final split on the first character | |
7630 | of $IFS should be done, even though QUOTED is 1. */ | |
1231ac47 | 7631 | /* XXX - should this test include Q_PATQUOTE? */ |
cce855bc JA |
7632 | if (quoted_dollar_at_p && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
7633 | *quoted_dollar_at_p = 1; | |
7634 | if (contains_dollar_at) | |
7635 | *contains_dollar_at = 1; | |
7636 | ||
7637 | /* We want to separate the positional parameters with the first | |
7638 | character of $IFS in case $IFS is something other than a space. | |
7639 | We also want to make sure that splitting is done no matter what -- | |
7640 | according to POSIX.2, this expands to a list of the positional | |
7641 | parameters no matter what IFS is set to. */ | |
40647963 | 7642 | #if 0 |
cce855bc | 7643 | temp = string_list_dollar_at (list, quoted); |
40647963 CR |
7644 | #else |
7645 | temp = string_list_dollar_at (list, (pflags & PF_ASSIGNRHS) ? (quoted|Q_DOUBLE_QUOTES) : quoted); | |
7646 | #endif | |
cce855bc JA |
7647 | |
7648 | dispose_words (list); | |
7649 | break; | |
7650 | ||
7651 | case LBRACE: | |
e1e48bba | 7652 | tdesc = parameter_brace_expand (string, &zindex, quoted, pflags, |
227f982e CR |
7653 | quoted_dollar_at_p, |
7654 | contains_dollar_at); | |
7655 | ||
227f982e CR |
7656 | if (tdesc == &expand_wdesc_error || tdesc == &expand_wdesc_fatal) |
7657 | return (tdesc); | |
7658 | temp = tdesc ? tdesc->word : (char *)0; | |
cce855bc JA |
7659 | |
7660 | /* XXX */ | |
bb70624e | 7661 | /* Quoted nulls should be removed if there is anything else |
cce855bc JA |
7662 | in the string. */ |
7663 | /* Note that we saw the quoted null so we can add one back at | |
7664 | the end of this function if there are no other characters | |
28ef6c31 JA |
7665 | in the string, discard TEMP, and go on. The exception to |
7666 | this is when we have "${@}" and $1 is '', since $@ needs | |
7667 | special handling. */ | |
227f982e | 7668 | if (tdesc && tdesc->word && (tdesc->flags & W_HASQUOTEDNULL) && QUOTED_NULL (temp)) |
cce855bc JA |
7669 | { |
7670 | if (had_quoted_null_p) | |
7671 | *had_quoted_null_p = 1; | |
28ef6c31 JA |
7672 | if (*quoted_dollar_at_p == 0) |
7673 | { | |
7674 | free (temp); | |
227f982e | 7675 | tdesc->word = temp = (char *)NULL; |
28ef6c31 JA |
7676 | } |
7677 | ||
cce855bc JA |
7678 | } |
7679 | ||
227f982e | 7680 | ret = tdesc; |
cce855bc JA |
7681 | goto return0; |
7682 | ||
7683 | /* Do command or arithmetic substitution. */ | |
7684 | case LPAREN: | |
7685 | /* We have to extract the contents of this paren substitution. */ | |
7686 | t_index = zindex + 1; | |
e33f2203 | 7687 | temp = extract_command_subst (string, &t_index, 0); |
cce855bc JA |
7688 | zindex = t_index; |
7689 | ||
7690 | /* For Posix.2-style `$(( ))' arithmetic substitution, | |
28ef6c31 | 7691 | extract the expression and pass it to the evaluator. */ |
cce855bc JA |
7692 | if (temp && *temp == LPAREN) |
7693 | { | |
7694 | char *temp2; | |
7695 | temp1 = temp + 1; | |
7696 | temp2 = savestring (temp1); | |
7697 | t_index = strlen (temp2) - 1; | |
7698 | ||
7699 | if (temp2[t_index] != RPAREN) | |
7700 | { | |
7701 | free (temp2); | |
7702 | goto comsub; | |
7703 | } | |
7704 | ||
7705 | /* Cut off ending `)' */ | |
7706 | temp2[t_index] = '\0'; | |
7707 | ||
d3ad40de CR |
7708 | if (chk_arithsub (temp2, t_index) == 0) |
7709 | { | |
7710 | free (temp2); | |
984a1947 | 7711 | #if 0 |
ecf57862 | 7712 | internal_warning (_("future versions of the shell will force evaluation as an arithmetic substitution")); |
984a1947 | 7713 | #endif |
d3ad40de CR |
7714 | goto comsub; |
7715 | } | |
7716 | ||
cce855bc | 7717 | /* Expand variables found inside the expression. */ |
d3ad40de | 7718 | temp1 = expand_arith_string (temp2, Q_DOUBLE_QUOTES); |
cce855bc JA |
7719 | free (temp2); |
7720 | ||
7721 | arithsub: | |
7722 | /* No error messages. */ | |
7723 | this_command_name = (char *)NULL; | |
7724 | number = evalexp (temp1, &expok); | |
7725 | free (temp); | |
7726 | free (temp1); | |
7727 | if (expok == 0) | |
7728 | { | |
7729 | if (interactive_shell == 0 && posixly_correct) | |
7730 | { | |
7731 | last_command_exit_value = EXECUTION_FAILURE; | |
227f982e | 7732 | return (&expand_wdesc_fatal); |
cce855bc JA |
7733 | } |
7734 | else | |
227f982e | 7735 | return (&expand_wdesc_error); |
cce855bc JA |
7736 | } |
7737 | temp = itos (number); | |
7738 | break; | |
7739 | } | |
7740 | ||
7741 | comsub: | |
d3a24ed2 CR |
7742 | if (pflags & PF_NOCOMSUB) |
7743 | /* we need zindex+1 because string[zindex] == RPAREN */ | |
7744 | temp1 = substring (string, *sindex, zindex+1); | |
7745 | else | |
d3ad40de CR |
7746 | { |
7747 | tdesc = command_substitute (temp, quoted); | |
7748 | temp1 = tdesc ? tdesc->word : (char *)NULL; | |
e33f2203 CR |
7749 | if (tdesc) |
7750 | dispose_word_desc (tdesc); | |
d3ad40de | 7751 | } |
cce855bc JA |
7752 | FREE (temp); |
7753 | temp = temp1; | |
7754 | break; | |
7755 | ||
7756 | /* Do POSIX.2d9-style arithmetic substitution. This will probably go | |
7757 | away in a future bash release. */ | |
7758 | case '[': | |
bb70624e | 7759 | /* Extract the contents of this arithmetic substitution. */ |
cce855bc JA |
7760 | t_index = zindex + 1; |
7761 | temp = extract_arithmetic_subst (string, &t_index); | |
7762 | zindex = t_index; | |
35bb237e CR |
7763 | if (temp == 0) |
7764 | { | |
7765 | temp = savestring (string); | |
7766 | if (expanded_something) | |
7767 | *expanded_something = 0; | |
7768 | goto return0; | |
7769 | } | |
cce855bc JA |
7770 | |
7771 | /* Do initial variable expansion. */ | |
d3ad40de | 7772 | temp1 = expand_arith_string (temp, Q_DOUBLE_QUOTES); |
cce855bc JA |
7773 | |
7774 | goto arithsub; | |
7775 | ||
7776 | default: | |
7777 | /* Find the variable in VARIABLE_LIST. */ | |
7778 | temp = (char *)NULL; | |
7779 | ||
7780 | for (t_index = zindex; (c = string[zindex]) && legal_variable_char (c); zindex++) | |
7781 | ; | |
7782 | temp1 = (zindex > t_index) ? substring (string, t_index, zindex) : (char *)NULL; | |
7783 | ||
7784 | /* If this isn't a variable name, then just output the `$'. */ | |
7785 | if (temp1 == 0 || *temp1 == '\0') | |
7786 | { | |
7787 | FREE (temp1); | |
f73dda09 | 7788 | temp = (char *)xmalloc (2); |
cce855bc JA |
7789 | temp[0] = '$'; |
7790 | temp[1] = '\0'; | |
7791 | if (expanded_something) | |
7792 | *expanded_something = 0; | |
7793 | goto return0; | |
7794 | } | |
7795 | ||
7796 | /* If the variable exists, return its value cell. */ | |
7797 | var = find_variable (temp1); | |
7798 | ||
7117c2d2 | 7799 | if (var && invisible_p (var) == 0 && var_isset (var)) |
cce855bc JA |
7800 | { |
7801 | #if defined (ARRAY_VARS) | |
fdf670ea | 7802 | if (assoc_p (var) || array_p (var)) |
cce855bc | 7803 | { |
fdf670ea CR |
7804 | temp = array_p (var) ? array_reference (array_cell (var), 0) |
7805 | : assoc_reference (assoc_cell (var), "0"); | |
cce855bc | 7806 | if (temp) |
762a763b | 7807 | temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) |
5ba8ff6e CR |
7808 | ? quote_string (temp) |
7809 | : quote_escapes (temp); | |
5565fb1a CR |
7810 | else if (unbound_vars_is_error) |
7811 | goto unbound_variable; | |
cce855bc JA |
7812 | } |
7813 | else | |
7814 | #endif | |
762a763b CR |
7815 | { |
7816 | temp = value_cell (var); | |
7817 | ||
7818 | temp = (*temp && (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES))) | |
7819 | ? quote_string (temp) | |
7820 | : quote_escapes (temp); | |
7821 | } | |
5ba8ff6e | 7822 | |
cce855bc | 7823 | free (temp1); |
7117c2d2 | 7824 | |
cce855bc JA |
7825 | goto return0; |
7826 | } | |
7827 | ||
7828 | temp = (char *)NULL; | |
7829 | ||
5565fb1a | 7830 | unbound_variable: |
cce855bc | 7831 | if (unbound_vars_is_error) |
0d8616ff CR |
7832 | { |
7833 | last_command_exit_value = EXECUTION_FAILURE; | |
7834 | err_unboundvar (temp1); | |
7835 | } | |
cce855bc JA |
7836 | else |
7837 | { | |
7838 | free (temp1); | |
7839 | goto return0; | |
7840 | } | |
7841 | ||
7842 | free (temp1); | |
7843 | last_command_exit_value = EXECUTION_FAILURE; | |
7844 | return ((unbound_vars_is_error && interactive_shell == 0) | |
227f982e CR |
7845 | ? &expand_wdesc_fatal |
7846 | : &expand_wdesc_error); | |
cce855bc JA |
7847 | } |
7848 | ||
7849 | if (string[zindex]) | |
7850 | zindex++; | |
7851 | ||
7852 | return0: | |
7853 | *sindex = zindex; | |
227f982e CR |
7854 | |
7855 | if (ret == 0) | |
7856 | { | |
7857 | ret = alloc_word_desc (); | |
7858 | ret->flags = tflag; /* XXX */ | |
7859 | ret->word = temp; | |
7860 | } | |
7861 | return ret; | |
cce855bc JA |
7862 | } |
7863 | ||
7864 | /* Make a word list which is the result of parameter and variable | |
7865 | expansion, command substitution, arithmetic substitution, and | |
7866 | quote removal of WORD. Return a pointer to a WORD_LIST which is | |
7867 | the result of the expansion. If WORD contains a null word, the | |
7868 | word list returned is also null. | |
726f6388 | 7869 | |
ccc6cda3 JA |
7870 | QUOTED contains flag values defined in shell.h. |
7871 | ||
b72432fd JA |
7872 | ISEXP is used to tell expand_word_internal that the word should be |
7873 | treated as the result of an expansion. This has implications for | |
7874 | how IFS characters in the word are treated. | |
7875 | ||
726f6388 JA |
7876 | CONTAINS_DOLLAR_AT and EXPANDED_SOMETHING are return values; when non-null |
7877 | they point to an integer value which receives information about expansion. | |
7878 | CONTAINS_DOLLAR_AT gets non-zero if WORD contained "$@", else zero. | |
7879 | EXPANDED_SOMETHING get non-zero if WORD contained any parameter expansions, | |
7880 | else zero. | |
7881 | ||
7882 | This only does word splitting in the case of $@ expansion. In that | |
7883 | case, we split on ' '. */ | |
7884 | ||
7885 | /* Values for the local variable quoted_state. */ | |
7886 | #define UNQUOTED 0 | |
7887 | #define PARTIALLY_QUOTED 1 | |
7888 | #define WHOLLY_QUOTED 2 | |
7889 | ||
7890 | static WORD_LIST * | |
b72432fd | 7891 | expand_word_internal (word, quoted, isexp, contains_dollar_at, expanded_something) |
726f6388 | 7892 | WORD_DESC *word; |
b72432fd | 7893 | int quoted, isexp; |
726f6388 JA |
7894 | int *contains_dollar_at; |
7895 | int *expanded_something; | |
7896 | { | |
ccc6cda3 JA |
7897 | WORD_LIST *list; |
7898 | WORD_DESC *tword; | |
726f6388 JA |
7899 | |
7900 | /* The intermediate string that we build while expanding. */ | |
ccc6cda3 | 7901 | char *istring; |
726f6388 JA |
7902 | |
7903 | /* The current size of the above object. */ | |
ccc6cda3 | 7904 | int istring_size; |
726f6388 JA |
7905 | |
7906 | /* Index into ISTRING. */ | |
ccc6cda3 | 7907 | int istring_index; |
726f6388 JA |
7908 | |
7909 | /* Temporary string storage. */ | |
ccc6cda3 | 7910 | char *temp, *temp1; |
726f6388 JA |
7911 | |
7912 | /* The text of WORD. */ | |
ccc6cda3 | 7913 | register char *string; |
726f6388 | 7914 | |
7117c2d2 JA |
7915 | /* The size of STRING. */ |
7916 | size_t string_size; | |
7917 | ||
726f6388 | 7918 | /* The index into STRING. */ |
ccc6cda3 | 7919 | int sindex; |
726f6388 JA |
7920 | |
7921 | /* This gets 1 if we see a $@ while quoted. */ | |
ccc6cda3 | 7922 | int quoted_dollar_at; |
726f6388 JA |
7923 | |
7924 | /* One of UNQUOTED, PARTIALLY_QUOTED, or WHOLLY_QUOTED, depending on | |
7925 | whether WORD contains no quoting characters, a partially quoted | |
7926 | string (e.g., "xx"ab), or is fully quoted (e.g., "xxab"). */ | |
ccc6cda3 JA |
7927 | int quoted_state; |
7928 | ||
22e63b05 | 7929 | /* State flags */ |
ccc6cda3 | 7930 | int had_quoted_null; |
cce855bc | 7931 | int has_dollar_at; |
28ef6c31 | 7932 | int tflag; |
e1e48bba | 7933 | int pflags; /* flags passed to param_expand */ |
726f6388 | 7934 | |
22e63b05 CR |
7935 | int assignoff; /* If assignment, offset of `=' */ |
7936 | ||
f73dda09 | 7937 | register unsigned char c; /* Current character. */ |
726f6388 | 7938 | int t_index; /* For calls to string_extract_xxx. */ |
726f6388 | 7939 | |
bb70624e | 7940 | char twochars[2]; |
b72432fd | 7941 | |
7117c2d2 JA |
7942 | DECLARE_MBSTATE; |
7943 | ||
f73dda09 | 7944 | istring = (char *)xmalloc (istring_size = DEFAULT_INITIAL_ARRAY_SIZE); |
ccc6cda3 | 7945 | istring[istring_index = 0] = '\0'; |
cce855bc | 7946 | quoted_dollar_at = had_quoted_null = has_dollar_at = 0; |
ccc6cda3 JA |
7947 | quoted_state = UNQUOTED; |
7948 | ||
7949 | string = word->word; | |
7950 | if (string == 0) | |
7951 | goto finished_with_string; | |
da713c23 CR |
7952 | /* Don't need the string length for the SADD... and COPY_ macros unless |
7953 | multibyte characters are possible. */ | |
7954 | string_size = (MB_CUR_MAX > 1) ? strlen (string) : 1; | |
726f6388 JA |
7955 | |
7956 | if (contains_dollar_at) | |
7957 | *contains_dollar_at = 0; | |
7958 | ||
22e63b05 CR |
7959 | assignoff = -1; |
7960 | ||
726f6388 JA |
7961 | /* Begin the expansion. */ |
7962 | ||
ccc6cda3 | 7963 | for (sindex = 0; ;) |
726f6388 JA |
7964 | { |
7965 | c = string[sindex]; | |
7966 | ||
7967 | /* Case on toplevel character. */ | |
7968 | switch (c) | |
7969 | { | |
7970 | case '\0': | |
7971 | goto finished_with_string; | |
7972 | ||
7973 | case CTLESC: | |
7117c2d2 JA |
7974 | sindex++; |
7975 | #if HANDLE_MULTIBYTE | |
7976 | if (MB_CUR_MAX > 1 && string[sindex]) | |
7977 | { | |
545f34cf | 7978 | SADD_MBQCHAR_BODY(temp, string, sindex, string_size); |
7117c2d2 JA |
7979 | } |
7980 | else | |
7981 | #endif | |
7982 | { | |
7983 | temp = (char *)xmalloc (3); | |
7984 | temp[0] = CTLESC; | |
7985 | temp[1] = c = string[sindex]; | |
7986 | temp[2] = '\0'; | |
7987 | } | |
726f6388 | 7988 | |
cce855bc | 7989 | dollar_add_string: |
726f6388 JA |
7990 | if (string[sindex]) |
7991 | sindex++; | |
7992 | ||
cce855bc JA |
7993 | add_string: |
7994 | if (temp) | |
7995 | { | |
7996 | istring = sub_append_string (temp, istring, &istring_index, &istring_size); | |
7997 | temp = (char *)0; | |
7998 | } | |
7999 | ||
8000 | break; | |
726f6388 JA |
8001 | |
8002 | #if defined (PROCESS_SUBSTITUTION) | |
8003 | /* Process substitution. */ | |
8004 | case '<': | |
8005 | case '>': | |
8006 | { | |
dc8fbaf9 | 8007 | if (string[++sindex] != LPAREN || (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (word->flags & (W_DQUOTE|W_NOPROCSUB)) || posixly_correct) |
726f6388 | 8008 | { |
bb70624e | 8009 | sindex--; /* add_character: label increments sindex */ |
726f6388 JA |
8010 | goto add_character; |
8011 | } | |
8012 | else | |
cce855bc | 8013 | t_index = sindex + 1; /* skip past both '<' and LPAREN */ |
726f6388 | 8014 | |
cce855bc | 8015 | temp1 = extract_process_subst (string, (c == '<') ? "<(" : ">(", &t_index); /*))*/ |
ccc6cda3 | 8016 | sindex = t_index; |
726f6388 JA |
8017 | |
8018 | /* If the process substitution specification is `<()', we want to | |
8019 | open the pipe for writing in the child and produce output; if | |
8020 | it is `>()', we want to open the pipe for reading in the child | |
8021 | and consume input. */ | |
ccc6cda3 | 8022 | temp = temp1 ? process_substitute (temp1, (c == '>')) : (char *)0; |
726f6388 JA |
8023 | |
8024 | FREE (temp1); | |
8025 | ||
8026 | goto dollar_add_string; | |
8027 | } | |
8028 | #endif /* PROCESS_SUBSTITUTION */ | |
8029 | ||
22e63b05 CR |
8030 | case '=': |
8031 | /* Posix.2 section 3.6.1 says that tildes following `=' in words | |
8032 | which are not assignment statements are not expanded. If the | |
8033 | shell isn't in posix mode, though, we perform tilde expansion | |
8034 | on `likely candidate' unquoted assignment statements (flags | |
8035 | include W_ASSIGNMENT but not W_QUOTED). A likely candidate | |
8036 | contains an unquoted :~ or =~. Something to think about: we | |
8037 | now have a flag that says to perform tilde expansion on arguments | |
8038 | to `assignment builtins' like declare and export that look like | |
8039 | assignment statements. We now do tilde expansion on such words | |
8040 | even in POSIX mode. */ | |
8041 | if (word->flags & (W_ASSIGNRHS|W_NOTILDE)) | |
c1d39fb8 | 8042 | { |
e1e48bba | 8043 | if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) |
c1d39fb8 CR |
8044 | goto add_ifs_character; |
8045 | else | |
8046 | goto add_character; | |
8047 | } | |
22e63b05 CR |
8048 | /* If we're not in posix mode or forcing assignment-statement tilde |
8049 | expansion, note where the `=' appears in the word and prepare to | |
8050 | do tilde expansion following the first `='. */ | |
8051 | if ((word->flags & W_ASSIGNMENT) && | |
8052 | (posixly_correct == 0 || (word->flags & W_TILDEEXP)) && | |
8053 | assignoff == -1 && sindex > 0) | |
8054 | assignoff = sindex; | |
8055 | if (sindex == assignoff && string[sindex+1] == '~') /* XXX */ | |
8056 | word->flags |= W_ITILDE; | |
d90269dd | 8057 | #if 0 |
22e63b05 CR |
8058 | else if ((word->flags & W_ASSIGNMENT) && |
8059 | (posixly_correct == 0 || (word->flags & W_TILDEEXP)) && | |
8060 | string[sindex+1] == '~') | |
8061 | word->flags |= W_ITILDE; | |
d90269dd | 8062 | #endif |
e1e48bba | 8063 | if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) |
c1d39fb8 CR |
8064 | goto add_ifs_character; |
8065 | else | |
8066 | goto add_character; | |
22e63b05 CR |
8067 | |
8068 | case ':': | |
8069 | if (word->flags & W_NOTILDE) | |
c1d39fb8 | 8070 | { |
e1e48bba | 8071 | if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) |
c1d39fb8 CR |
8072 | goto add_ifs_character; |
8073 | else | |
8074 | goto add_character; | |
8075 | } | |
cdb32d45 CR |
8076 | |
8077 | if ((word->flags & (W_ASSIGNMENT|W_ASSIGNRHS|W_TILDEEXP)) && | |
22e63b05 CR |
8078 | string[sindex+1] == '~') |
8079 | word->flags |= W_ITILDE; | |
c1d39fb8 | 8080 | |
e1e48bba | 8081 | if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c)) |
c1d39fb8 CR |
8082 | goto add_ifs_character; |
8083 | else | |
8084 | goto add_character; | |
22e63b05 CR |
8085 | |
8086 | case '~': | |
8087 | /* If the word isn't supposed to be tilde expanded, or we're not | |
8088 | at the start of a word or after an unquoted : or = in an | |
8089 | assignment statement, we don't do tilde expansion. */ | |
da719982 | 8090 | if ((word->flags & (W_NOTILDE|W_DQUOTE)) || |
22e63b05 CR |
8091 | (sindex > 0 && ((word->flags & W_ITILDE) == 0)) || |
8092 | (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
8093 | { | |
8094 | word->flags &= ~W_ITILDE; | |
e1e48bba | 8095 | if (isexp == 0 && (word->flags & (W_NOSPLIT|W_NOSPLIT2)) == 0 && isifs (c) && (quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) == 0) |
c1d39fb8 CR |
8096 | goto add_ifs_character; |
8097 | else | |
8098 | goto add_character; | |
22e63b05 CR |
8099 | } |
8100 | ||
8101 | if (word->flags & W_ASSIGNRHS) | |
8102 | tflag = 2; | |
cdb32d45 | 8103 | else if (word->flags & (W_ASSIGNMENT|W_TILDEEXP)) |
22e63b05 CR |
8104 | tflag = 1; |
8105 | else | |
8106 | tflag = 0; | |
8107 | ||
8108 | temp = bash_tilde_find_word (string + sindex, tflag, &t_index); | |
8109 | ||
8110 | word->flags &= ~W_ITILDE; | |
8111 | ||
8112 | if (temp && *temp && t_index > 0) | |
8113 | { | |
8114 | temp1 = bash_tilde_expand (temp, tflag); | |
dc8fbaf9 CR |
8115 | if (temp1 && *temp1 == '~' && STREQ (temp, temp1)) |
8116 | { | |
8117 | FREE (temp); | |
8118 | FREE (temp1); | |
8119 | goto add_character; /* tilde expansion failed */ | |
8120 | } | |
22e63b05 CR |
8121 | free (temp); |
8122 | temp = temp1; | |
8123 | sindex += t_index; | |
866961ad | 8124 | goto add_quoted_string; /* XXX was add_string */ |
22e63b05 CR |
8125 | } |
8126 | else | |
8127 | { | |
8128 | FREE (temp); | |
8129 | goto add_character; | |
8130 | } | |
8131 | ||
726f6388 | 8132 | case '$': |
726f6388 JA |
8133 | if (expanded_something) |
8134 | *expanded_something = 1; | |
8135 | ||
cce855bc | 8136 | has_dollar_at = 0; |
e1e48bba CR |
8137 | pflags = (word->flags & W_NOCOMSUB) ? PF_NOCOMSUB : 0; |
8138 | if (word->flags & W_NOSPLIT2) | |
8139 | pflags |= PF_NOSPLIT2; | |
40647963 CR |
8140 | if (word->flags & W_ASSIGNRHS) |
8141 | pflags |= PF_ASSIGNRHS; | |
227f982e | 8142 | tword = param_expand (string, &sindex, quoted, expanded_something, |
cce855bc | 8143 | &has_dollar_at, "ed_dollar_at, |
e1e48bba | 8144 | &had_quoted_null, pflags); |
726f6388 | 8145 | |
227f982e | 8146 | if (tword == &expand_wdesc_error || tword == &expand_wdesc_fatal) |
726f6388 | 8147 | { |
cce855bc JA |
8148 | free (string); |
8149 | free (istring); | |
227f982e CR |
8150 | return ((tword == &expand_wdesc_error) ? &expand_word_error |
8151 | : &expand_word_fatal); | |
cce855bc JA |
8152 | } |
8153 | if (contains_dollar_at && has_dollar_at) | |
8154 | *contains_dollar_at = 1; | |
227f982e CR |
8155 | |
8156 | if (tword && (tword->flags & W_HASQUOTEDNULL)) | |
8157 | had_quoted_null = 1; | |
8158 | ||
77638cbf | 8159 | temp = tword ? tword->word : (char *)NULL; |
227f982e CR |
8160 | dispose_word_desc (tword); |
8161 | ||
cce855bc JA |
8162 | goto add_string; |
8163 | break; | |
726f6388 | 8164 | |
cce855bc JA |
8165 | case '`': /* Backquoted command substitution. */ |
8166 | { | |
d3a24ed2 | 8167 | t_index = sindex++; |
726f6388 | 8168 | |
e6598ba4 | 8169 | temp = string_extract (string, &sindex, "`", SX_REQMATCH); |
bc7bed50 CR |
8170 | /* The test of sindex against t_index is to allow bare instances of |
8171 | ` to pass through, for backwards compatibility. */ | |
7027abcb CR |
8172 | if (temp == &extract_string_error || temp == &extract_string_fatal) |
8173 | { | |
bc7bed50 CR |
8174 | if (sindex - 1 == t_index) |
8175 | { | |
8176 | sindex = t_index; | |
8177 | goto add_character; | |
8178 | } | |
7f947b68 | 8179 | last_command_exit_value = EXECUTION_FAILURE; |
d3ad40de | 8180 | report_error (_("bad substitution: no closing \"`\" in %s") , string+t_index); |
7027abcb CR |
8181 | free (string); |
8182 | free (istring); | |
8183 | return ((temp == &extract_string_error) ? &expand_word_error | |
8184 | : &expand_word_fatal); | |
8185 | } | |
8186 | ||
bc7bed50 CR |
8187 | if (expanded_something) |
8188 | *expanded_something = 1; | |
8189 | ||
d3a24ed2 CR |
8190 | if (word->flags & W_NOCOMSUB) |
8191 | /* sindex + 1 because string[sindex] == '`' */ | |
8192 | temp1 = substring (string, t_index, sindex + 1); | |
8193 | else | |
8194 | { | |
8195 | de_backslash (temp); | |
d3ad40de CR |
8196 | tword = command_substitute (temp, quoted); |
8197 | temp1 = tword ? tword->word : (char *)NULL; | |
e33f2203 CR |
8198 | if (tword) |
8199 | dispose_word_desc (tword); | |
d3a24ed2 | 8200 | } |
cce855bc JA |
8201 | FREE (temp); |
8202 | temp = temp1; | |
8203 | goto dollar_add_string; | |
8204 | } | |
ccc6cda3 | 8205 | |
cce855bc JA |
8206 | case '\\': |
8207 | if (string[sindex + 1] == '\n') | |
8208 | { | |
8209 | sindex += 2; | |
8210 | continue; | |
8211 | } | |
726f6388 | 8212 | |
cce855bc | 8213 | c = string[++sindex]; |
726f6388 | 8214 | |
cce855bc | 8215 | if (quoted & Q_HERE_DOCUMENT) |
28ef6c31 | 8216 | tflag = CBSHDOC; |
cce855bc | 8217 | else if (quoted & Q_DOUBLE_QUOTES) |
28ef6c31 | 8218 | tflag = CBSDQUOTE; |
cce855bc | 8219 | else |
28ef6c31 JA |
8220 | tflag = 0; |
8221 | ||
64419627 | 8222 | /* From Posix discussion on austin-group list: Backslash escaping |
9e51a74d CR |
8223 | a } in ${...} is removed. Issue 0000221 */ |
8224 | if ((quoted & Q_DOLBRACE) && c == RBRACE) | |
64419627 CR |
8225 | { |
8226 | SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size); | |
8227 | } | |
8228 | else if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) && ((sh_syntaxtab[c] & tflag) == 0)) | |
cce855bc | 8229 | { |
7117c2d2 | 8230 | SCOPY_CHAR_I (twochars, '\\', c, string, sindex, string_size); |
bb70624e JA |
8231 | } |
8232 | else if (c == 0) | |
8233 | { | |
8234 | c = CTLNUL; | |
8235 | sindex--; /* add_character: label increments sindex */ | |
8236 | goto add_character; | |
cce855bc JA |
8237 | } |
8238 | else | |
bb70624e | 8239 | { |
7117c2d2 | 8240 | SCOPY_CHAR_I (twochars, CTLESC, c, string, sindex, string_size); |
bb70624e | 8241 | } |
726f6388 | 8242 | |
bb70624e JA |
8243 | sindex++; |
8244 | add_twochars: | |
8245 | /* BEFORE jumping here, we need to increment sindex if appropriate */ | |
8246 | RESIZE_MALLOCED_BUFFER (istring, istring_index, 2, istring_size, | |
8247 | DEFAULT_ARRAY_SIZE); | |
8248 | istring[istring_index++] = twochars[0]; | |
8249 | istring[istring_index++] = twochars[1]; | |
8250 | istring[istring_index] = '\0'; | |
8251 | ||
8252 | break; | |
726f6388 | 8253 | |
cce855bc | 8254 | case '"': |
da719982 CR |
8255 | #if 0 |
8256 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE)) | |
8257 | #else | |
8258 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
8259 | #endif | |
cce855bc | 8260 | goto add_character; |
ccc6cda3 JA |
8261 | |
8262 | t_index = ++sindex; | |
8263 | temp = string_extract_double_quoted (string, &sindex, 0); | |
8264 | ||
8265 | /* If the quotes surrounded the entire string, then the | |
8266 | whole word was quoted. */ | |
8267 | quoted_state = (t_index == 1 && string[sindex] == '\0') | |
8268 | ? WHOLLY_QUOTED | |
7117c2d2 | 8269 | : PARTIALLY_QUOTED; |
ccc6cda3 JA |
8270 | |
8271 | if (temp && *temp) | |
726f6388 | 8272 | { |
227f982e CR |
8273 | tword = alloc_word_desc (); |
8274 | tword->word = temp; | |
8275 | ||
ccc6cda3 JA |
8276 | temp = (char *)NULL; |
8277 | ||
cce855bc | 8278 | has_dollar_at = 0; |
227f982e | 8279 | /* Need to get W_HASQUOTEDNULL flag through this function. */ |
b72432fd | 8280 | list = expand_word_internal (tword, Q_DOUBLE_QUOTES, 0, &has_dollar_at, (int *)NULL); |
726f6388 | 8281 | |
ccc6cda3 JA |
8282 | if (list == &expand_word_error || list == &expand_word_fatal) |
8283 | { | |
8284 | free (istring); | |
8285 | free (string); | |
8286 | /* expand_word_internal has already freed temp_word->word | |
8287 | for us because of the way it prints error messages. */ | |
8288 | tword->word = (char *)NULL; | |
8289 | dispose_word (tword); | |
8290 | return list; | |
8291 | } | |
726f6388 | 8292 | |
ccc6cda3 | 8293 | dispose_word (tword); |
726f6388 | 8294 | |
ccc6cda3 JA |
8295 | /* "$@" (a double-quoted dollar-at) expands into nothing, |
8296 | not even a NULL word, when there are no positional | |
8297 | parameters. */ | |
cce855bc | 8298 | if (list == 0 && has_dollar_at) |
726f6388 | 8299 | { |
ccc6cda3 JA |
8300 | quoted_dollar_at++; |
8301 | break; | |
8302 | } | |
8303 | ||
8304 | /* If we get "$@", we know we have expanded something, so we | |
8305 | need to remember it for the final split on $IFS. This is | |
8306 | a special case; it's the only case where a quoted string | |
8307 | can expand into more than one word. It's going to come back | |
8308 | from the above call to expand_word_internal as a list with | |
8309 | a single word, in which all characters are quoted and | |
8310 | separated by blanks. What we want to do is to turn it back | |
8311 | into a list for the next piece of code. */ | |
8312 | if (list) | |
8313 | dequote_list (list); | |
8314 | ||
227f982e CR |
8315 | if (list && list->word && (list->word->flags & W_HASQUOTEDNULL)) |
8316 | had_quoted_null = 1; | |
8317 | ||
cce855bc | 8318 | if (has_dollar_at) |
ccc6cda3 JA |
8319 | { |
8320 | quoted_dollar_at++; | |
8321 | if (contains_dollar_at) | |
8322 | *contains_dollar_at = 1; | |
8323 | if (expanded_something) | |
8324 | *expanded_something = 1; | |
8325 | } | |
8326 | } | |
8327 | else | |
8328 | { | |
8329 | /* What we have is "". This is a minor optimization. */ | |
f73dda09 | 8330 | FREE (temp); |
ccc6cda3 JA |
8331 | list = (WORD_LIST *)NULL; |
8332 | } | |
8333 | ||
8334 | /* The code above *might* return a list (consider the case of "$@", | |
8335 | where it returns "$1", "$2", etc.). We can't throw away the | |
8336 | rest of the list, and we have to make sure each word gets added | |
8337 | as quoted. We test on tresult->next: if it is non-NULL, we | |
8338 | quote the whole list, save it to a string with string_list, and | |
8339 | add that string. We don't need to quote the results of this | |
8340 | (and it would be wrong, since that would quote the separators | |
8341 | as well), so we go directly to add_string. */ | |
8342 | if (list) | |
8343 | { | |
8344 | if (list->next) | |
8345 | { | |
e1e48bba | 8346 | #if 0 |
adc6cff5 | 8347 | if (quoted_dollar_at && (word->flags & W_NOSPLIT2)) |
e1e48bba CR |
8348 | temp = string_list_internal (quote_list (list), " "); |
8349 | else | |
8350 | #endif | |
bc4cd23c JA |
8351 | /* Testing quoted_dollar_at makes sure that "$@" is |
8352 | split correctly when $IFS does not contain a space. */ | |
8353 | temp = quoted_dollar_at | |
8354 | ? string_list_dollar_at (list, Q_DOUBLE_QUOTES) | |
8355 | : string_list (quote_list (list)); | |
ccc6cda3 | 8356 | dispose_words (list); |
726f6388 JA |
8357 | goto add_string; |
8358 | } | |
8359 | else | |
8360 | { | |
ccc6cda3 | 8361 | temp = savestring (list->word->word); |
227f982e | 8362 | tflag = list->word->flags; |
ccc6cda3 | 8363 | dispose_words (list); |
227f982e | 8364 | |
cce855bc JA |
8365 | /* If the string is not a quoted null string, we want |
8366 | to remove any embedded unquoted CTLNUL characters. | |
8367 | We do not want to turn quoted null strings back into | |
8368 | the empty string, though. We do this because we | |
8369 | want to remove any quoted nulls from expansions that | |
8370 | contain other characters. For example, if we have | |
8371 | x"$*"y or "x$*y" and there are no positional parameters, | |
7117c2d2 | 8372 | the $* should expand into nothing. */ |
227f982e CR |
8373 | /* We use the W_HASQUOTEDNULL flag to differentiate the |
8374 | cases: a quoted null character as above and when | |
8375 | CTLNUL is contained in the (non-null) expansion | |
8376 | of some variable. We use the had_quoted_null flag to | |
10590446 | 8377 | pass the value through this function to its caller. */ |
227f982e | 8378 | if ((tflag & W_HASQUOTEDNULL) && QUOTED_NULL (temp) == 0) |
cce855bc | 8379 | remove_quoted_nulls (temp); /* XXX */ |
726f6388 JA |
8380 | } |
8381 | } | |
ccc6cda3 JA |
8382 | else |
8383 | temp = (char *)NULL; | |
726f6388 | 8384 | |
ccc6cda3 | 8385 | /* We do not want to add quoted nulls to strings that are only |
f4f5e1c2 CR |
8386 | partially quoted; we can throw them away. The execption to |
8387 | this is when we are going to be performing word splitting, | |
8388 | since we have to preserve a null argument if the next character | |
8389 | will cause word splitting. */ | |
adc6cff5 | 8390 | if (temp == 0 && quoted_state == PARTIALLY_QUOTED && (word->flags & (W_NOSPLIT|W_NOSPLIT2))) |
cce855bc | 8391 | continue; |
726f6388 | 8392 | |
ccc6cda3 | 8393 | add_quoted_string: |
726f6388 | 8394 | |
ccc6cda3 JA |
8395 | if (temp) |
8396 | { | |
8397 | temp1 = temp; | |
8398 | temp = quote_string (temp); | |
8399 | free (temp1); | |
bb70624e | 8400 | goto add_string; |
ccc6cda3 JA |
8401 | } |
8402 | else | |
8403 | { | |
8404 | /* Add NULL arg. */ | |
bb70624e JA |
8405 | c = CTLNUL; |
8406 | sindex--; /* add_character: label increments sindex */ | |
8407 | goto add_character; | |
ccc6cda3 | 8408 | } |
bb70624e | 8409 | |
ccc6cda3 | 8410 | /* break; */ |
726f6388 | 8411 | |
ccc6cda3 | 8412 | case '\'': |
da719982 CR |
8413 | #if 0 |
8414 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (word->flags & W_DQUOTE)) | |
8415 | #else | |
8416 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT))) | |
8417 | #endif | |
ccc6cda3 | 8418 | goto add_character; |
726f6388 | 8419 | |
ccc6cda3 JA |
8420 | t_index = ++sindex; |
8421 | temp = string_extract_single_quoted (string, &sindex); | |
726f6388 | 8422 | |
ccc6cda3 JA |
8423 | /* If the entire STRING was surrounded by single quotes, |
8424 | then the string is wholly quoted. */ | |
8425 | quoted_state = (t_index == 1 && string[sindex] == '\0') | |
8426 | ? WHOLLY_QUOTED | |
7117c2d2 | 8427 | : PARTIALLY_QUOTED; |
726f6388 | 8428 | |
ccc6cda3 JA |
8429 | /* If all we had was '', it is a null expansion. */ |
8430 | if (*temp == '\0') | |
8431 | { | |
8432 | free (temp); | |
8433 | temp = (char *)NULL; | |
8434 | } | |
8435 | else | |
7117c2d2 | 8436 | remove_quoted_escapes (temp); /* ??? */ |
726f6388 | 8437 | |
ccc6cda3 JA |
8438 | /* We do not want to add quoted nulls to strings that are only |
8439 | partially quoted; such nulls are discarded. */ | |
8440 | if (temp == 0 && (quoted_state == PARTIALLY_QUOTED)) | |
8441 | continue; | |
726f6388 | 8442 | |
bb70624e JA |
8443 | /* If we have a quoted null expansion, add a quoted NULL to istring. */ |
8444 | if (temp == 0) | |
8445 | { | |
8446 | c = CTLNUL; | |
8447 | sindex--; /* add_character: label increments sindex */ | |
8448 | goto add_character; | |
8449 | } | |
8450 | else | |
8451 | goto add_quoted_string; | |
8452 | ||
ccc6cda3 | 8453 | /* break; */ |
726f6388 JA |
8454 | |
8455 | default: | |
726f6388 | 8456 | /* This is the fix for " $@ " */ |
c1d39fb8 | 8457 | add_ifs_character: |
7117c2d2 | 8458 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || (isexp == 0 && isifs (c))) |
726f6388 | 8459 | { |
bb70624e JA |
8460 | if (string[sindex]) /* from old goto dollar_add_string */ |
8461 | sindex++; | |
8462 | if (c == 0) | |
8463 | { | |
8464 | c = CTLNUL; | |
8465 | goto add_character; | |
8466 | } | |
8467 | else | |
8468 | { | |
7117c2d2 | 8469 | #if HANDLE_MULTIBYTE |
545f34cf CR |
8470 | if (MB_CUR_MAX > 1) |
8471 | sindex--; | |
8472 | ||
7117c2d2 JA |
8473 | if (MB_CUR_MAX > 1) |
8474 | { | |
545f34cf | 8475 | SADD_MBQCHAR_BODY(temp, string, sindex, string_size); |
7117c2d2 JA |
8476 | } |
8477 | else | |
8478 | #endif | |
8479 | { | |
8480 | twochars[0] = CTLESC; | |
8481 | twochars[1] = c; | |
8482 | goto add_twochars; | |
8483 | } | |
bb70624e | 8484 | } |
726f6388 JA |
8485 | } |
8486 | ||
7117c2d2 JA |
8487 | SADD_MBCHAR (temp, string, sindex, string_size); |
8488 | ||
726f6388 | 8489 | add_character: |
ccc6cda3 JA |
8490 | RESIZE_MALLOCED_BUFFER (istring, istring_index, 1, istring_size, |
8491 | DEFAULT_ARRAY_SIZE); | |
726f6388 JA |
8492 | istring[istring_index++] = c; |
8493 | istring[istring_index] = '\0'; | |
8494 | ||
8495 | /* Next character. */ | |
8496 | sindex++; | |
8497 | } | |
8498 | } | |
8499 | ||
8500 | finished_with_string: | |
726f6388 JA |
8501 | /* OK, we're ready to return. If we have a quoted string, and |
8502 | quoted_dollar_at is not set, we do no splitting at all; otherwise | |
8503 | we split on ' '. The routines that call this will handle what to | |
8504 | do if nothing has been expanded. */ | |
ccc6cda3 JA |
8505 | |
8506 | /* Partially and wholly quoted strings which expand to the empty | |
8507 | string are retained as an empty arguments. Unquoted strings | |
8508 | which expand to the empty string are discarded. The single | |
8509 | exception is the case of expanding "$@" when there are no | |
8510 | positional parameters. In that case, we discard the expansion. */ | |
8511 | ||
8512 | /* Because of how the code that handles "" and '' in partially | |
8513 | quoted strings works, we need to make ISTRING into a QUOTED_NULL | |
8514 | if we saw quoting characters, but the expansion was empty. | |
8515 | "" and '' are tossed away before we get to this point when | |
8516 | processing partially quoted strings. This makes "" and $xxx"" | |
8517 | equivalent when xxx is unset. We also look to see whether we | |
8518 | saw a quoted null from a ${} expansion and add one back if we | |
8519 | need to. */ | |
8520 | ||
8521 | /* If we expand to nothing and there were no single or double quotes | |
8522 | in the word, we throw it away. Otherwise, we return a NULL word. | |
8523 | The single exception is for $@ surrounded by double quotes when | |
8524 | there are no positional parameters. In that case, we also throw | |
8525 | the word away. */ | |
8526 | ||
8527 | if (*istring == '\0') | |
8528 | { | |
8529 | if (quoted_dollar_at == 0 && (had_quoted_null || quoted_state == PARTIALLY_QUOTED)) | |
726f6388 | 8530 | { |
726f6388 JA |
8531 | istring[0] = CTLNUL; |
8532 | istring[1] = '\0'; | |
ccc6cda3 | 8533 | tword = make_bare_word (istring); |
227f982e | 8534 | tword->flags |= W_HASQUOTEDNULL; /* XXX */ |
ccc6cda3 JA |
8535 | list = make_word_list (tword, (WORD_LIST *)NULL); |
8536 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) | |
8537 | tword->flags |= W_QUOTED; | |
726f6388 | 8538 | } |
ccc6cda3 JA |
8539 | /* According to sh, ksh, and Posix.2, if a word expands into nothing |
8540 | and a double-quoted "$@" appears anywhere in it, then the entire | |
8541 | word is removed. */ | |
8542 | else if (quoted_state == UNQUOTED || quoted_dollar_at) | |
8543 | list = (WORD_LIST *)NULL; | |
8544 | #if 0 | |
8545 | else | |
726f6388 | 8546 | { |
ccc6cda3 | 8547 | tword = make_bare_word (istring); |
ccc6cda3 JA |
8548 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
8549 | tword->flags |= W_QUOTED; | |
227f982e | 8550 | list = make_word_list (tword, (WORD_LIST *)NULL); |
726f6388 | 8551 | } |
f73dda09 JA |
8552 | #else |
8553 | else | |
8554 | list = (WORD_LIST *)NULL; | |
ccc6cda3 JA |
8555 | #endif |
8556 | } | |
8557 | else if (word->flags & W_NOSPLIT) | |
8558 | { | |
8559 | tword = make_bare_word (istring); | |
ccc6cda3 JA |
8560 | if (word->flags & W_ASSIGNMENT) |
8561 | tword->flags |= W_ASSIGNMENT; /* XXX */ | |
43df7bbb CR |
8562 | if (word->flags & W_COMPASSIGN) |
8563 | tword->flags |= W_COMPASSIGN; /* XXX */ | |
b72432fd JA |
8564 | if (word->flags & W_NOGLOB) |
8565 | tword->flags |= W_NOGLOB; /* XXX */ | |
43df7bbb CR |
8566 | if (word->flags & W_NOEXPAND) |
8567 | tword->flags |= W_NOEXPAND; /* XXX */ | |
ccc6cda3 | 8568 | if (quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) |
28ef6c31 | 8569 | tword->flags |= W_QUOTED; |
227f982e CR |
8570 | if (had_quoted_null) |
8571 | tword->flags |= W_HASQUOTEDNULL; | |
8572 | list = make_word_list (tword, (WORD_LIST *)NULL); | |
ccc6cda3 JA |
8573 | } |
8574 | else | |
8575 | { | |
8576 | char *ifs_chars; | |
8577 | ||
7117c2d2 | 8578 | ifs_chars = (quoted_dollar_at || has_dollar_at) ? ifs_value : (char *)NULL; |
726f6388 | 8579 | |
cce855bc JA |
8580 | /* If we have $@, we need to split the results no matter what. If |
8581 | IFS is unset or NULL, string_list_dollar_at has separated the | |
8582 | positional parameters with a space, so we split on space (we have | |
8583 | set ifs_chars to " \t\n" above if ifs is unset). If IFS is set, | |
8584 | string_list_dollar_at has separated the positional parameters | |
8585 | with the first character of $IFS, so we split on $IFS. */ | |
8586 | if (has_dollar_at && ifs_chars) | |
8587 | list = list_string (istring, *ifs_chars ? ifs_chars : " ", 1); | |
ccc6cda3 JA |
8588 | else |
8589 | { | |
8590 | tword = make_bare_word (istring); | |
ccc6cda3 JA |
8591 | if ((quoted & (Q_DOUBLE_QUOTES|Q_HERE_DOCUMENT)) || (quoted_state == WHOLLY_QUOTED)) |
8592 | tword->flags |= W_QUOTED; | |
8593 | if (word->flags & W_ASSIGNMENT) | |
8594 | tword->flags |= W_ASSIGNMENT; | |
43df7bbb CR |
8595 | if (word->flags & W_COMPASSIGN) |
8596 | tword->flags |= W_COMPASSIGN; | |
b72432fd JA |
8597 | if (word->flags & W_NOGLOB) |
8598 | tword->flags |= W_NOGLOB; | |
43df7bbb CR |
8599 | if (word->flags & W_NOEXPAND) |
8600 | tword->flags |= W_NOEXPAND; | |
227f982e CR |
8601 | if (had_quoted_null) |
8602 | tword->flags |= W_HASQUOTEDNULL; /* XXX */ | |
8603 | list = make_word_list (tword, (WORD_LIST *)NULL); | |
726f6388 | 8604 | } |
726f6388 | 8605 | } |
726f6388 | 8606 | |
ccc6cda3 JA |
8607 | free (istring); |
8608 | return (list); | |
726f6388 JA |
8609 | } |
8610 | ||
8611 | /* **************************************************************** */ | |
8612 | /* */ | |
8613 | /* Functions for Quote Removal */ | |
8614 | /* */ | |
8615 | /* **************************************************************** */ | |
8616 | ||
8617 | /* Perform quote removal on STRING. If QUOTED > 0, assume we are obeying the | |
7117c2d2 | 8618 | backslash quoting rules for within double quotes or a here document. */ |
726f6388 JA |
8619 | char * |
8620 | string_quote_removal (string, quoted) | |
8621 | char *string; | |
8622 | int quoted; | |
8623 | { | |
7117c2d2 JA |
8624 | size_t slen; |
8625 | char *r, *result_string, *temp, *send; | |
f73dda09 JA |
8626 | int sindex, tindex, dquote; |
8627 | unsigned char c; | |
7117c2d2 | 8628 | DECLARE_MBSTATE; |
726f6388 JA |
8629 | |
8630 | /* The result can be no longer than the original string. */ | |
7117c2d2 JA |
8631 | slen = strlen (string); |
8632 | send = string + slen; | |
8633 | ||
8634 | r = result_string = (char *)xmalloc (slen + 1); | |
726f6388 | 8635 | |
ccc6cda3 | 8636 | for (dquote = sindex = 0; c = string[sindex];) |
726f6388 JA |
8637 | { |
8638 | switch (c) | |
8639 | { | |
8640 | case '\\': | |
8641 | c = string[++sindex]; | |
33fe8777 CR |
8642 | if (c == 0) |
8643 | { | |
8644 | *r++ = '\\'; | |
8645 | break; | |
8646 | } | |
28ef6c31 | 8647 | if (((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) && (sh_syntaxtab[c] & CBSDQUOTE) == 0) |
726f6388 | 8648 | *r++ = '\\'; |
ccc6cda3 | 8649 | /* FALLTHROUGH */ |
726f6388 JA |
8650 | |
8651 | default: | |
7117c2d2 | 8652 | SCOPY_CHAR_M (r, string, send, sindex); |
726f6388 JA |
8653 | break; |
8654 | ||
8655 | case '\'': | |
ccc6cda3 | 8656 | if ((quoted & (Q_HERE_DOCUMENT|Q_DOUBLE_QUOTES)) || dquote) |
726f6388 JA |
8657 | { |
8658 | *r++ = c; | |
8659 | sindex++; | |
ccc6cda3 | 8660 | break; |
726f6388 | 8661 | } |
ccc6cda3 JA |
8662 | tindex = sindex + 1; |
8663 | temp = string_extract_single_quoted (string, &tindex); | |
8664 | if (temp) | |
726f6388 | 8665 | { |
ccc6cda3 JA |
8666 | strcpy (r, temp); |
8667 | r += strlen (r); | |
8668 | free (temp); | |
726f6388 | 8669 | } |
ccc6cda3 | 8670 | sindex = tindex; |
726f6388 JA |
8671 | break; |
8672 | ||
8673 | case '"': | |
8674 | dquote = 1 - dquote; | |
8675 | sindex++; | |
8676 | break; | |
8677 | } | |
8678 | } | |
8679 | *r = '\0'; | |
8680 | return (result_string); | |
8681 | } | |
8682 | ||
ccc6cda3 JA |
8683 | #if 0 |
8684 | /* UNUSED */ | |
726f6388 JA |
8685 | /* Perform quote removal on word WORD. This allocates and returns a new |
8686 | WORD_DESC *. */ | |
8687 | WORD_DESC * | |
8688 | word_quote_removal (word, quoted) | |
8689 | WORD_DESC *word; | |
8690 | int quoted; | |
8691 | { | |
8692 | WORD_DESC *w; | |
8693 | char *t; | |
8694 | ||
8695 | t = string_quote_removal (word->word, quoted); | |
227f982e CR |
8696 | w = alloc_word_desc (); |
8697 | w->word = t ? t : savestring (""); | |
726f6388 JA |
8698 | return (w); |
8699 | } | |
8700 | ||
8701 | /* Perform quote removal on all words in LIST. If QUOTED is non-zero, | |
8702 | the members of the list are treated as if they are surrounded by | |
8703 | double quotes. Return a new list, or NULL if LIST is NULL. */ | |
8704 | WORD_LIST * | |
8705 | word_list_quote_removal (list, quoted) | |
8706 | WORD_LIST *list; | |
8707 | int quoted; | |
8708 | { | |
227f982e | 8709 | WORD_LIST *result, *t, *tresult, *e; |
726f6388 | 8710 | |
ccc6cda3 | 8711 | for (t = list, result = (WORD_LIST *)NULL; t; t = t->next) |
726f6388 | 8712 | { |
7117c2d2 | 8713 | tresult = make_word_list (word_quote_removal (t->word, quoted), (WORD_LIST *)NULL); |
227f982e | 8714 | #if 0 |
726f6388 | 8715 | result = (WORD_LIST *) list_append (result, tresult); |
227f982e CR |
8716 | #else |
8717 | if (result == 0) | |
8718 | result = e = tresult; | |
8719 | else | |
8720 | { | |
8721 | e->next = tresult; | |
8722 | while (e->next) | |
8723 | e = e->next; | |
8724 | } | |
8725 | #endif | |
726f6388 JA |
8726 | } |
8727 | return (result); | |
8728 | } | |
ccc6cda3 | 8729 | #endif |
726f6388 | 8730 | |
726f6388 JA |
8731 | /******************************************* |
8732 | * * | |
8733 | * Functions to perform word splitting * | |
8734 | * * | |
8735 | *******************************************/ | |
8736 | ||
7117c2d2 JA |
8737 | void |
8738 | setifs (v) | |
8739 | SHELL_VAR *v; | |
b72432fd | 8740 | { |
7117c2d2 JA |
8741 | char *t; |
8742 | unsigned char uc; | |
8743 | ||
8744 | ifs_var = v; | |
cdb32d45 | 8745 | ifs_value = (v && value_cell (v)) ? value_cell (v) : " \t\n"; |
b72432fd | 8746 | |
633e5c6d CR |
8747 | /* Should really merge ifs_cmap with sh_syntaxtab. XXX - doesn't yet |
8748 | handle multibyte chars in IFS */ | |
7117c2d2 JA |
8749 | memset (ifs_cmap, '\0', sizeof (ifs_cmap)); |
8750 | for (t = ifs_value ; t && *t; t++) | |
8751 | { | |
8752 | uc = *t; | |
8753 | ifs_cmap[uc] = 1; | |
8754 | } | |
8755 | ||
633e5c6d CR |
8756 | #if defined (HANDLE_MULTIBYTE) |
8757 | if (ifs_value == 0) | |
8758 | { | |
8759 | ifs_firstc[0] = '\0'; | |
8760 | ifs_firstc_len = 1; | |
8761 | } | |
8762 | else | |
8763 | { | |
8764 | size_t ifs_len; | |
8765 | ifs_len = strnlen (ifs_value, MB_CUR_MAX); | |
8766 | ifs_firstc_len = MBLEN (ifs_value, ifs_len); | |
22e63b05 | 8767 | if (ifs_firstc_len == 1 || ifs_firstc_len == 0 || MB_INVALIDCH (ifs_firstc_len)) |
633e5c6d CR |
8768 | { |
8769 | ifs_firstc[0] = ifs_value[0]; | |
8770 | ifs_firstc[1] = '\0'; | |
8771 | ifs_firstc_len = 1; | |
8772 | } | |
8773 | else | |
8774 | memcpy (ifs_firstc, ifs_value, ifs_firstc_len); | |
8775 | } | |
8776 | #else | |
7117c2d2 | 8777 | ifs_firstc = ifs_value ? *ifs_value : 0; |
633e5c6d | 8778 | #endif |
7117c2d2 JA |
8779 | } |
8780 | ||
8781 | char * | |
8782 | getifs () | |
8783 | { | |
8784 | return ifs_value; | |
b72432fd JA |
8785 | } |
8786 | ||
726f6388 JA |
8787 | /* This splits a single word into a WORD LIST on $IFS, but only if the word |
8788 | is not quoted. list_string () performs quote removal for us, even if we | |
8789 | don't do any splitting. */ | |
8790 | WORD_LIST * | |
7117c2d2 | 8791 | word_split (w, ifs_chars) |
726f6388 | 8792 | WORD_DESC *w; |
7117c2d2 | 8793 | char *ifs_chars; |
726f6388 JA |
8794 | { |
8795 | WORD_LIST *result; | |
8796 | ||
8797 | if (w) | |
8798 | { | |
7117c2d2 | 8799 | char *xifs; |
726f6388 | 8800 | |
7117c2d2 JA |
8801 | xifs = ((w->flags & W_QUOTED) || ifs_chars == 0) ? "" : ifs_chars; |
8802 | result = list_string (w->word, xifs, w->flags & W_QUOTED); | |
726f6388 JA |
8803 | } |
8804 | else | |
8805 | result = (WORD_LIST *)NULL; | |
ccc6cda3 | 8806 | |
726f6388 JA |
8807 | return (result); |
8808 | } | |
8809 | ||
8810 | /* Perform word splitting on LIST and return the RESULT. It is possible | |
8811 | to return (WORD_LIST *)NULL. */ | |
8812 | static WORD_LIST * | |
8813 | word_list_split (list) | |
8814 | WORD_LIST *list; | |
8815 | { | |
37c41ab1 | 8816 | WORD_LIST *result, *t, *tresult, *e; |
726f6388 | 8817 | |
ccc6cda3 | 8818 | for (t = list, result = (WORD_LIST *)NULL; t; t = t->next) |
726f6388 | 8819 | { |
7117c2d2 | 8820 | tresult = word_split (t->word, ifs_value); |
37c41ab1 CR |
8821 | if (result == 0) |
8822 | result = e = tresult; | |
8823 | else | |
8824 | { | |
8825 | e->next = tresult; | |
8826 | while (e->next) | |
8827 | e = e->next; | |
8828 | } | |
726f6388 JA |
8829 | } |
8830 | return (result); | |
8831 | } | |
8832 | ||
8833 | /************************************************** | |
8834 | * * | |
cce855bc | 8835 | * Functions to expand an entire WORD_LIST * |
726f6388 JA |
8836 | * * |
8837 | **************************************************/ | |
8838 | ||
d3a24ed2 CR |
8839 | /* Do any word-expansion-specific cleanup and jump to top_level */ |
8840 | static void | |
8841 | exp_jump_to_top_level (v) | |
8842 | int v; | |
8843 | { | |
f11997f8 CR |
8844 | set_pipestatus_from_exit (last_command_exit_value); |
8845 | ||
d3a24ed2 CR |
8846 | /* Cleanup code goes here. */ |
8847 | expand_no_split_dollar_star = 0; /* XXX */ | |
8848 | expanding_redir = 0; | |
d7f49990 | 8849 | assigning_in_environment = 0; |
d3a24ed2 | 8850 | |
2e4498b3 CR |
8851 | if (parse_and_execute_level == 0) |
8852 | top_level_cleanup (); /* from sig.c */ | |
8853 | ||
d3a24ed2 CR |
8854 | jump_to_top_level (v); |
8855 | } | |
8856 | ||
cce855bc JA |
8857 | /* Put NLIST (which is a WORD_LIST * of only one element) at the front of |
8858 | ELIST, and set ELIST to the new list. */ | |
8859 | #define PREPEND_LIST(nlist, elist) \ | |
8860 | do { nlist->next = elist; elist = nlist; } while (0) | |
8861 | ||
726f6388 JA |
8862 | /* Separate out any initial variable assignments from TLIST. If set -k has |
8863 | been executed, remove all assignment statements from TLIST. Initial | |
8864 | variable assignments and other environment assignments are placed | |
bb70624e | 8865 | on SUBST_ASSIGN_VARLIST. */ |
726f6388 JA |
8866 | static WORD_LIST * |
8867 | separate_out_assignments (tlist) | |
8868 | WORD_LIST *tlist; | |
8869 | { | |
8870 | register WORD_LIST *vp, *lp; | |
8871 | ||
0527c903 | 8872 | if (tlist == 0) |
726f6388 JA |
8873 | return ((WORD_LIST *)NULL); |
8874 | ||
bb70624e JA |
8875 | if (subst_assign_varlist) |
8876 | dispose_words (subst_assign_varlist); /* Clean up after previous error */ | |
b72432fd | 8877 | |
bb70624e | 8878 | subst_assign_varlist = (WORD_LIST *)NULL; |
726f6388 JA |
8879 | vp = lp = tlist; |
8880 | ||
8881 | /* Separate out variable assignments at the start of the command. | |
8882 | Loop invariant: vp->next == lp | |
8883 | Loop postcondition: | |
7117c2d2 JA |
8884 | lp = list of words left after assignment statements skipped |
8885 | tlist = original list of words | |
726f6388 | 8886 | */ |
ccc6cda3 | 8887 | while (lp && (lp->word->flags & W_ASSIGNMENT)) |
726f6388 JA |
8888 | { |
8889 | vp = lp; | |
8890 | lp = lp->next; | |
8891 | } | |
8892 | ||
bb70624e JA |
8893 | /* If lp != tlist, we have some initial assignment statements. |
8894 | We make SUBST_ASSIGN_VARLIST point to the list of assignment | |
8895 | words and TLIST point to the remaining words. */ | |
726f6388 JA |
8896 | if (lp != tlist) |
8897 | { | |
bb70624e | 8898 | subst_assign_varlist = tlist; |
726f6388 JA |
8899 | /* ASSERT(vp->next == lp); */ |
8900 | vp->next = (WORD_LIST *)NULL; /* terminate variable list */ | |
8901 | tlist = lp; /* remainder of word list */ | |
8902 | } | |
8903 | ||
8904 | /* vp == end of variable list */ | |
8905 | /* tlist == remainder of original word list without variable assignments */ | |
8906 | if (!tlist) | |
8907 | /* All the words in tlist were assignment statements */ | |
8908 | return ((WORD_LIST *)NULL); | |
8909 | ||
8910 | /* ASSERT(tlist != NULL); */ | |
ccc6cda3 | 8911 | /* ASSERT((tlist->word->flags & W_ASSIGNMENT) == 0); */ |
726f6388 JA |
8912 | |
8913 | /* If the -k option is in effect, we need to go through the remaining | |
bb70624e JA |
8914 | words, separate out the assignment words, and place them on |
8915 | SUBST_ASSIGN_VARLIST. */ | |
726f6388 JA |
8916 | if (place_keywords_in_env) |
8917 | { | |
8918 | WORD_LIST *tp; /* tp == running pointer into tlist */ | |
8919 | ||
8920 | tp = tlist; | |
8921 | lp = tlist->next; | |
8922 | ||
8923 | /* Loop Invariant: tp->next == lp */ | |
8924 | /* Loop postcondition: tlist == word list without assignment statements */ | |
8925 | while (lp) | |
8926 | { | |
ccc6cda3 | 8927 | if (lp->word->flags & W_ASSIGNMENT) |
726f6388 JA |
8928 | { |
8929 | /* Found an assignment statement, add this word to end of | |
bb70624e JA |
8930 | subst_assign_varlist (vp). */ |
8931 | if (!subst_assign_varlist) | |
8932 | subst_assign_varlist = vp = lp; | |
726f6388 JA |
8933 | else |
8934 | { | |
8935 | vp->next = lp; | |
8936 | vp = lp; | |
8937 | } | |
8938 | ||
8939 | /* Remove the word pointed to by LP from TLIST. */ | |
8940 | tp->next = lp->next; | |
8941 | /* ASSERT(vp == lp); */ | |
8942 | lp->next = (WORD_LIST *)NULL; | |
8943 | lp = tp->next; | |
8944 | } | |
8945 | else | |
8946 | { | |
8947 | tp = lp; | |
8948 | lp = lp->next; | |
8949 | } | |
8950 | } | |
8951 | } | |
8952 | return (tlist); | |
8953 | } | |
8954 | ||
cce855bc JA |
8955 | #define WEXP_VARASSIGN 0x001 |
8956 | #define WEXP_BRACEEXP 0x002 | |
8957 | #define WEXP_TILDEEXP 0x004 | |
8958 | #define WEXP_PARAMEXP 0x008 | |
8959 | #define WEXP_PATHEXP 0x010 | |
8960 | ||
8961 | /* All of the expansions, including variable assignments at the start of | |
8962 | the list. */ | |
8963 | #define WEXP_ALL (WEXP_VARASSIGN|WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP) | |
8964 | ||
8965 | /* All of the expansions except variable assignments at the start of | |
8966 | the list. */ | |
8967 | #define WEXP_NOVARS (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP|WEXP_PATHEXP) | |
8968 | ||
8969 | /* All of the `shell expansions': brace expansion, tilde expansion, parameter | |
8970 | expansion, command substitution, arithmetic expansion, word splitting, and | |
8971 | quote removal. */ | |
8972 | #define WEXP_SHELLEXP (WEXP_BRACEEXP|WEXP_TILDEEXP|WEXP_PARAMEXP) | |
8973 | ||
726f6388 JA |
8974 | /* Take the list of words in LIST and do the various substitutions. Return |
8975 | a new list of words which is the expanded list, and without things like | |
8976 | variable assignments. */ | |
8977 | ||
8978 | WORD_LIST * | |
8979 | expand_words (list) | |
8980 | WORD_LIST *list; | |
8981 | { | |
cce855bc | 8982 | return (expand_word_list_internal (list, WEXP_ALL)); |
726f6388 JA |
8983 | } |
8984 | ||
8985 | /* Same as expand_words (), but doesn't hack variable or environment | |
8986 | variables. */ | |
8987 | WORD_LIST * | |
8988 | expand_words_no_vars (list) | |
8989 | WORD_LIST *list; | |
8990 | { | |
cce855bc | 8991 | return (expand_word_list_internal (list, WEXP_NOVARS)); |
726f6388 JA |
8992 | } |
8993 | ||
cce855bc JA |
8994 | WORD_LIST * |
8995 | expand_words_shellexp (list) | |
726f6388 | 8996 | WORD_LIST *list; |
726f6388 | 8997 | { |
cce855bc JA |
8998 | return (expand_word_list_internal (list, WEXP_SHELLEXP)); |
8999 | } | |
726f6388 | 9000 | |
cce855bc JA |
9001 | static WORD_LIST * |
9002 | glob_expand_word_list (tlist, eflags) | |
9003 | WORD_LIST *tlist; | |
9004 | int eflags; | |
9005 | { | |
9006 | char **glob_array, *temp_string; | |
9007 | register int glob_index; | |
9008 | WORD_LIST *glob_list, *output_list, *disposables, *next; | |
9009 | WORD_DESC *tword; | |
726f6388 | 9010 | |
cce855bc JA |
9011 | output_list = disposables = (WORD_LIST *)NULL; |
9012 | glob_array = (char **)NULL; | |
9013 | while (tlist) | |
9014 | { | |
9015 | /* For each word, either globbing is attempted or the word is | |
9016 | added to orig_list. If globbing succeeds, the results are | |
9017 | added to orig_list and the word (tlist) is added to the list | |
9018 | of disposable words. If globbing fails and failed glob | |
9019 | expansions are left unchanged (the shell default), the | |
9020 | original word is added to orig_list. If globbing fails and | |
9021 | failed glob expansions are removed, the original word is | |
9022 | added to the list of disposable words. orig_list ends up | |
7117c2d2 | 9023 | in reverse order and requires a call to REVERSE_LIST to |
cce855bc JA |
9024 | be set right. After all words are examined, the disposable |
9025 | words are freed. */ | |
9026 | next = tlist->next; | |
726f6388 | 9027 | |
cce855bc | 9028 | /* If the word isn't an assignment and contains an unquoted |
28ef6c31 | 9029 | pattern matching character, then glob it. */ |
b72432fd | 9030 | if ((tlist->word->flags & W_NOGLOB) == 0 && |
cce855bc | 9031 | unquoted_glob_pattern_p (tlist->word->word)) |
726f6388 | 9032 | { |
cce855bc JA |
9033 | glob_array = shell_glob_filename (tlist->word->word); |
9034 | ||
9035 | /* Handle error cases. | |
9036 | I don't think we should report errors like "No such file | |
9037 | or directory". However, I would like to report errors | |
9038 | like "Read failed". */ | |
9039 | ||
d3a24ed2 | 9040 | if (glob_array == 0 || GLOB_FAILED (glob_array)) |
726f6388 | 9041 | { |
bb70624e | 9042 | glob_array = (char **)xmalloc (sizeof (char *)); |
cce855bc JA |
9043 | glob_array[0] = (char *)NULL; |
9044 | } | |
9045 | ||
9046 | /* Dequote the current word in case we have to use it. */ | |
9047 | if (glob_array[0] == NULL) | |
9048 | { | |
9049 | temp_string = dequote_string (tlist->word->word); | |
9050 | free (tlist->word->word); | |
9051 | tlist->word->word = temp_string; | |
9052 | } | |
9053 | ||
9054 | /* Make the array into a word list. */ | |
9055 | glob_list = (WORD_LIST *)NULL; | |
9056 | for (glob_index = 0; glob_array[glob_index]; glob_index++) | |
9057 | { | |
9058 | tword = make_bare_word (glob_array[glob_index]); | |
9059 | tword->flags |= W_GLOBEXP; /* XXX */ | |
9060 | glob_list = make_word_list (tword, glob_list); | |
9061 | } | |
9062 | ||
9063 | if (glob_list) | |
9064 | { | |
9065 | output_list = (WORD_LIST *)list_append (glob_list, output_list); | |
9066 | PREPEND_LIST (tlist, disposables); | |
9067 | } | |
d3a24ed2 CR |
9068 | else if (fail_glob_expansion != 0) |
9069 | { | |
7f947b68 | 9070 | last_command_exit_value = EXECUTION_FAILURE; |
5e13499c | 9071 | report_error (_("no match: %s"), tlist->word->word); |
c184f645 | 9072 | exp_jump_to_top_level (DISCARD); |
d3a24ed2 | 9073 | } |
cce855bc JA |
9074 | else if (allow_null_glob_expansion == 0) |
9075 | { | |
9076 | /* Failed glob expressions are left unchanged. */ | |
9077 | PREPEND_LIST (tlist, output_list); | |
9078 | } | |
9079 | else | |
9080 | { | |
9081 | /* Failed glob expressions are removed. */ | |
9082 | PREPEND_LIST (tlist, disposables); | |
726f6388 | 9083 | } |
726f6388 | 9084 | } |
cce855bc JA |
9085 | else |
9086 | { | |
9087 | /* Dequote the string. */ | |
9088 | temp_string = dequote_string (tlist->word->word); | |
9089 | free (tlist->word->word); | |
9090 | tlist->word->word = temp_string; | |
9091 | PREPEND_LIST (tlist, output_list); | |
9092 | } | |
9093 | ||
7117c2d2 | 9094 | strvec_dispose (glob_array); |
cce855bc JA |
9095 | glob_array = (char **)NULL; |
9096 | ||
9097 | tlist = next; | |
726f6388 JA |
9098 | } |
9099 | ||
cce855bc JA |
9100 | if (disposables) |
9101 | dispose_words (disposables); | |
9102 | ||
9103 | if (output_list) | |
9104 | output_list = REVERSE_LIST (output_list, WORD_LIST *); | |
9105 | ||
9106 | return (output_list); | |
9107 | } | |
726f6388 JA |
9108 | |
9109 | #if defined (BRACE_EXPANSION) | |
cce855bc JA |
9110 | static WORD_LIST * |
9111 | brace_expand_word_list (tlist, eflags) | |
9112 | WORD_LIST *tlist; | |
9113 | int eflags; | |
9114 | { | |
9115 | register char **expansions; | |
9116 | char *temp_string; | |
9117 | WORD_LIST *disposables, *output_list, *next; | |
9118 | WORD_DESC *w; | |
9119 | int eindex; | |
9120 | ||
9121 | for (disposables = output_list = (WORD_LIST *)NULL; tlist; tlist = next) | |
726f6388 | 9122 | { |
cce855bc | 9123 | next = tlist->next; |
726f6388 | 9124 | |
6669f0e5 CR |
9125 | if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG)) |
9126 | { | |
9127 | /*itrace("brace_expand_word_list: %s: W_COMPASSIGN|W_ASSIGNARG", tlist->word->word);*/ | |
9128 | PREPEND_LIST (tlist, output_list); | |
9129 | continue; | |
9130 | } | |
9131 | ||
cce855bc JA |
9132 | /* Only do brace expansion if the word has a brace character. If |
9133 | not, just add the word list element to BRACES and continue. In | |
9134 | the common case, at least when running shell scripts, this will | |
d0ca3503 | 9135 | degenerate to a bunch of calls to `mbschr', and then what is |
cce855bc | 9136 | basically a reversal of TLIST into BRACES, which is corrected |
7117c2d2 | 9137 | by a call to REVERSE_LIST () on BRACES when the end of TLIST |
cce855bc | 9138 | is reached. */ |
d0ca3503 | 9139 | if (mbschr (tlist->word->word, LBRACE)) |
726f6388 | 9140 | { |
cce855bc | 9141 | expansions = brace_expand (tlist->word->word); |
726f6388 | 9142 | |
cce855bc | 9143 | for (eindex = 0; temp_string = expansions[eindex]; eindex++) |
726f6388 | 9144 | { |
51f7ea36 CR |
9145 | w = alloc_word_desc (); |
9146 | w->word = temp_string; | |
9147 | ||
cce855bc JA |
9148 | /* If brace expansion didn't change the word, preserve |
9149 | the flags. We may want to preserve the flags | |
9150 | unconditionally someday -- XXX */ | |
9151 | if (STREQ (temp_string, tlist->word->word)) | |
9152 | w->flags = tlist->word->flags; | |
51f7ea36 CR |
9153 | else |
9154 | w = make_word_flags (w, temp_string); | |
9155 | ||
cce855bc | 9156 | output_list = make_word_list (w, output_list); |
726f6388 | 9157 | } |
cce855bc | 9158 | free (expansions); |
726f6388 | 9159 | |
cce855bc JA |
9160 | /* Add TLIST to the list of words to be freed after brace |
9161 | expansion has been performed. */ | |
9162 | PREPEND_LIST (tlist, disposables); | |
9163 | } | |
9164 | else | |
9165 | PREPEND_LIST (tlist, output_list); | |
726f6388 | 9166 | } |
cce855bc JA |
9167 | |
9168 | if (disposables) | |
9169 | dispose_words (disposables); | |
9170 | ||
9171 | if (output_list) | |
9172 | output_list = REVERSE_LIST (output_list, WORD_LIST *); | |
9173 | ||
9174 | return (output_list); | |
9175 | } | |
9176 | #endif | |
9177 | ||
fdf670ea CR |
9178 | #if defined (ARRAY_VARS) |
9179 | /* Take WORD, a compound associative array assignment, and internally run | |
9180 | 'declare -A w', where W is the variable name portion of WORD. */ | |
9181 | static int | |
9182 | make_internal_declare (word, option) | |
9183 | char *word; | |
9184 | char *option; | |
9185 | { | |
9186 | int t; | |
9187 | WORD_LIST *wl; | |
9188 | WORD_DESC *w; | |
9189 | ||
9190 | w = make_word (word); | |
9191 | ||
9192 | t = assignment (w->word, 0); | |
9193 | w->word[t] = '\0'; | |
9194 | ||
9195 | wl = make_word_list (w, (WORD_LIST *)NULL); | |
9196 | wl = make_word_list (make_word (option), wl); | |
9197 | ||
9198 | return (declare_builtin (wl)); | |
9199 | } | |
9200 | #endif | |
9201 | ||
cce855bc JA |
9202 | static WORD_LIST * |
9203 | shell_expand_word_list (tlist, eflags) | |
9204 | WORD_LIST *tlist; | |
9205 | int eflags; | |
9206 | { | |
9207 | WORD_LIST *expanded, *orig_list, *new_list, *next, *temp_list; | |
9208 | int expanded_something, has_dollar_at; | |
9209 | char *temp_string; | |
726f6388 | 9210 | |
726f6388 | 9211 | /* We do tilde expansion all the time. This is what 1003.2 says. */ |
cce855bc JA |
9212 | new_list = (WORD_LIST *)NULL; |
9213 | for (orig_list = tlist; tlist; tlist = next) | |
726f6388 | 9214 | { |
ccc6cda3 | 9215 | temp_string = tlist->word->word; |
726f6388 JA |
9216 | |
9217 | next = tlist->next; | |
9218 | ||
43df7bbb CR |
9219 | #if defined (ARRAY_VARS) |
9220 | /* If this is a compound array assignment to a builtin that accepts | |
9221 | such assignments (e.g., `declare'), take the assignment and perform | |
9222 | it separately, handling the semantics of declarations inside shell | |
9223 | functions. This avoids the double-evaluation of such arguments, | |
9224 | because `declare' does some evaluation of compound assignments on | |
9225 | its own. */ | |
9226 | if ((tlist->word->flags & (W_COMPASSIGN|W_ASSIGNARG)) == (W_COMPASSIGN|W_ASSIGNARG)) | |
9227 | { | |
9228 | int t; | |
9229 | ||
fdf670ea CR |
9230 | if (tlist->word->flags & W_ASSIGNASSOC) |
9231 | make_internal_declare (tlist->word->word, "-A"); | |
9232 | ||
adc6cff5 | 9233 | t = do_word_assignment (tlist->word, 0); |
43df7bbb CR |
9234 | if (t == 0) |
9235 | { | |
9236 | last_command_exit_value = EXECUTION_FAILURE; | |
9237 | exp_jump_to_top_level (DISCARD); | |
9238 | } | |
9239 | ||
9240 | /* Now transform the word as ksh93 appears to do and go on */ | |
9241 | t = assignment (tlist->word->word, 0); | |
9242 | tlist->word->word[t] = '\0'; | |
fdf670ea | 9243 | tlist->word->flags &= ~(W_ASSIGNMENT|W_NOSPLIT|W_COMPASSIGN|W_ASSIGNARG|W_ASSIGNASSOC); |
43df7bbb CR |
9244 | } |
9245 | #endif | |
9246 | ||
ccc6cda3 | 9247 | expanded_something = 0; |
726f6388 | 9248 | expanded = expand_word_internal |
b72432fd | 9249 | (tlist->word, 0, 0, &has_dollar_at, &expanded_something); |
726f6388 JA |
9250 | |
9251 | if (expanded == &expand_word_error || expanded == &expand_word_fatal) | |
9252 | { | |
9253 | /* By convention, each time this error is returned, | |
9254 | tlist->word->word has already been freed. */ | |
9255 | tlist->word->word = (char *)NULL; | |
ccc6cda3 | 9256 | |
726f6388 JA |
9257 | /* Dispose our copy of the original list. */ |
9258 | dispose_words (orig_list); | |
d166f048 | 9259 | /* Dispose the new list we're building. */ |
726f6388 JA |
9260 | dispose_words (new_list); |
9261 | ||
28ef6c31 | 9262 | last_command_exit_value = EXECUTION_FAILURE; |
726f6388 | 9263 | if (expanded == &expand_word_error) |
d3a24ed2 | 9264 | exp_jump_to_top_level (DISCARD); |
726f6388 | 9265 | else |
d3a24ed2 | 9266 | exp_jump_to_top_level (FORCE_EOF); |
726f6388 JA |
9267 | } |
9268 | ||
ccc6cda3 JA |
9269 | /* Don't split words marked W_NOSPLIT. */ |
9270 | if (expanded_something && (tlist->word->flags & W_NOSPLIT) == 0) | |
726f6388 | 9271 | { |
ccc6cda3 | 9272 | temp_list = word_list_split (expanded); |
726f6388 JA |
9273 | dispose_words (expanded); |
9274 | } | |
9275 | else | |
9276 | { | |
9277 | /* If no parameter expansion, command substitution, process | |
9278 | substitution, or arithmetic substitution took place, then | |
9279 | do not do word splitting. We still have to remove quoted | |
9280 | null characters from the result. */ | |
9281 | word_list_remove_quoted_nulls (expanded); | |
ccc6cda3 | 9282 | temp_list = expanded; |
726f6388 JA |
9283 | } |
9284 | ||
ccc6cda3 JA |
9285 | expanded = REVERSE_LIST (temp_list, WORD_LIST *); |
9286 | new_list = (WORD_LIST *)list_append (expanded, new_list); | |
726f6388 JA |
9287 | } |
9288 | ||
cce855bc JA |
9289 | if (orig_list) |
9290 | dispose_words (orig_list); | |
726f6388 | 9291 | |
726f6388 | 9292 | if (new_list) |
cce855bc | 9293 | new_list = REVERSE_LIST (new_list, WORD_LIST *); |
726f6388 | 9294 | |
cce855bc JA |
9295 | return (new_list); |
9296 | } | |
726f6388 | 9297 | |
cce855bc JA |
9298 | /* The workhorse for expand_words () and expand_words_no_vars (). |
9299 | First arg is LIST, a WORD_LIST of words. | |
b72432fd JA |
9300 | Second arg EFLAGS is a flags word controlling which expansions are |
9301 | performed. | |
726f6388 | 9302 | |
cce855bc JA |
9303 | This does all of the substitutions: brace expansion, tilde expansion, |
9304 | parameter expansion, command substitution, arithmetic expansion, | |
9305 | process substitution, word splitting, and pathname expansion, according | |
9306 | to the bits set in EFLAGS. Words with the W_QUOTED or W_NOSPLIT bits | |
9307 | set, or for which no expansion is done, do not undergo word splitting. | |
b72432fd | 9308 | Words with the W_NOGLOB bit set do not undergo pathname expansion. */ |
cce855bc JA |
9309 | static WORD_LIST * |
9310 | expand_word_list_internal (list, eflags) | |
9311 | WORD_LIST *list; | |
9312 | int eflags; | |
9313 | { | |
9314 | WORD_LIST *new_list, *temp_list; | |
9315 | int tint; | |
726f6388 | 9316 | |
cce855bc JA |
9317 | if (list == 0) |
9318 | return ((WORD_LIST *)NULL); | |
726f6388 | 9319 | |
bb70624e | 9320 | garglist = new_list = copy_word_list (list); |
cce855bc JA |
9321 | if (eflags & WEXP_VARASSIGN) |
9322 | { | |
bb70624e | 9323 | garglist = new_list = separate_out_assignments (new_list); |
cce855bc JA |
9324 | if (new_list == 0) |
9325 | { | |
bb70624e | 9326 | if (subst_assign_varlist) |
cce855bc JA |
9327 | { |
9328 | /* All the words were variable assignments, so they are placed | |
9329 | into the shell's environment. */ | |
bb70624e | 9330 | for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next) |
cce855bc JA |
9331 | { |
9332 | this_command_name = (char *)NULL; /* no arithmetic errors */ | |
adc6cff5 | 9333 | tint = do_word_assignment (temp_list->word, 0); |
cce855bc JA |
9334 | /* Variable assignment errors in non-interactive shells |
9335 | running in Posix.2 mode cause the shell to exit. */ | |
28ef6c31 | 9336 | if (tint == 0) |
ccc6cda3 | 9337 | { |
cce855bc | 9338 | last_command_exit_value = EXECUTION_FAILURE; |
28ef6c31 | 9339 | if (interactive_shell == 0 && posixly_correct) |
d3a24ed2 | 9340 | exp_jump_to_top_level (FORCE_EOF); |
28ef6c31 | 9341 | else |
d3a24ed2 | 9342 | exp_jump_to_top_level (DISCARD); |
ccc6cda3 | 9343 | } |
726f6388 | 9344 | } |
bb70624e JA |
9345 | dispose_words (subst_assign_varlist); |
9346 | subst_assign_varlist = (WORD_LIST *)NULL; | |
cce855bc JA |
9347 | } |
9348 | return ((WORD_LIST *)NULL); | |
9349 | } | |
9350 | } | |
726f6388 | 9351 | |
cce855bc JA |
9352 | /* Begin expanding the words that remain. The expansions take place on |
9353 | things that aren't really variable assignments. */ | |
726f6388 | 9354 | |
cce855bc JA |
9355 | #if defined (BRACE_EXPANSION) |
9356 | /* Do brace expansion on this word if there are any brace characters | |
9357 | in the string. */ | |
9358 | if ((eflags & WEXP_BRACEEXP) && brace_expansion && new_list) | |
9359 | new_list = brace_expand_word_list (new_list, eflags); | |
9360 | #endif /* BRACE_EXPANSION */ | |
726f6388 | 9361 | |
cce855bc JA |
9362 | /* Perform the `normal' shell expansions: tilde expansion, parameter and |
9363 | variable substitution, command substitution, arithmetic expansion, | |
9364 | and word splitting. */ | |
9365 | new_list = shell_expand_word_list (new_list, eflags); | |
726f6388 | 9366 | |
cce855bc JA |
9367 | /* Okay, we're almost done. Now let's just do some filename |
9368 | globbing. */ | |
9369 | if (new_list) | |
9370 | { | |
9371 | if ((eflags & WEXP_PATHEXP) && disallow_filename_globbing == 0) | |
9372 | /* Glob expand the word list unless globbing has been disabled. */ | |
9373 | new_list = glob_expand_word_list (new_list, eflags); | |
726f6388 | 9374 | else |
cce855bc JA |
9375 | /* Dequote the words, because we're not performing globbing. */ |
9376 | new_list = dequote_list (new_list); | |
726f6388 JA |
9377 | } |
9378 | ||
bb70624e | 9379 | if ((eflags & WEXP_VARASSIGN) && subst_assign_varlist) |
726f6388 | 9380 | { |
43df7bbb | 9381 | sh_wassign_func_t *assign_func; |
adc6cff5 | 9382 | int is_special_builtin, is_builtin_or_func; |
726f6388 JA |
9383 | |
9384 | /* If the remainder of the words expand to nothing, Posix.2 requires | |
9385 | that the variable and environment assignments affect the shell's | |
9386 | environment. */ | |
43df7bbb | 9387 | assign_func = new_list ? assign_in_env : do_word_assignment; |
56299fa5 | 9388 | tempenv_assign_error = 0; |
726f6388 | 9389 | |
adc6cff5 | 9390 | is_builtin_or_func = (new_list && new_list->word && (find_shell_builtin (new_list->word->word) || find_function (new_list->word->word))); |
e05be32d CR |
9391 | /* Posix says that special builtins exit if a variable assignment error |
9392 | occurs in an assignment preceding it. */ | |
adc6cff5 | 9393 | is_special_builtin = (posixly_correct && new_list && new_list->word && find_special_builtin (new_list->word->word)); |
e05be32d | 9394 | |
bb70624e | 9395 | for (temp_list = subst_assign_varlist; temp_list; temp_list = temp_list->next) |
726f6388 | 9396 | { |
ccc6cda3 | 9397 | this_command_name = (char *)NULL; |
d7f49990 | 9398 | assigning_in_environment = (assign_func == assign_in_env); |
adc6cff5 | 9399 | tint = (*assign_func) (temp_list->word, is_builtin_or_func); |
d7f49990 | 9400 | assigning_in_environment = 0; |
ccc6cda3 JA |
9401 | /* Variable assignment errors in non-interactive shells running |
9402 | in Posix.2 mode cause the shell to exit. */ | |
56299fa5 | 9403 | if (tint == 0) |
ccc6cda3 | 9404 | { |
43df7bbb | 9405 | if (assign_func == do_word_assignment) |
56299fa5 CR |
9406 | { |
9407 | last_command_exit_value = EXECUTION_FAILURE; | |
e05be32d | 9408 | if (interactive_shell == 0 && posixly_correct && is_special_builtin) |
56299fa5 CR |
9409 | exp_jump_to_top_level (FORCE_EOF); |
9410 | else | |
9411 | exp_jump_to_top_level (DISCARD); | |
9412 | } | |
28ef6c31 | 9413 | else |
56299fa5 | 9414 | tempenv_assign_error++; |
ccc6cda3 | 9415 | } |
726f6388 | 9416 | } |
726f6388 | 9417 | |
bb70624e JA |
9418 | dispose_words (subst_assign_varlist); |
9419 | subst_assign_varlist = (WORD_LIST *)NULL; | |
726f6388 JA |
9420 | } |
9421 | ||
cce855bc | 9422 | #if 0 |
ccc6cda3 JA |
9423 | tint = list_length (new_list) + 1; |
9424 | RESIZE_MALLOCED_BUFFER (glob_argv_flags, 0, tint, glob_argv_flags_size, 16); | |
cce855bc JA |
9425 | for (tint = 0, temp_list = new_list; temp_list; temp_list = temp_list->next) |
9426 | glob_argv_flags[tint++] = (temp_list->word->flags & W_GLOBEXP) ? '1' : '0'; | |
ccc6cda3 | 9427 | glob_argv_flags[tint] = '\0'; |
ccc6cda3 | 9428 | #endif |
726f6388 | 9429 | |
cce855bc | 9430 | return (new_list); |
ccc6cda3 | 9431 | } |