1 |
adcroft |
1.1 |
/* Extended regular expression matching and search library, |
2 |
|
|
version 0.12. |
3 |
|
|
(Implements POSIX draft P10003.2/D11.2, except for |
4 |
|
|
internationalization features.) |
5 |
|
|
|
6 |
|
|
Copyright (C) 1993 Free Software Foundation, Inc. |
7 |
|
|
|
8 |
|
|
This program is free software; you can redistribute it and/or modify |
9 |
|
|
it under the terms of the GNU General Public License as published by |
10 |
|
|
the Free Software Foundation; either version 2, or (at your option) |
11 |
|
|
any later version. |
12 |
|
|
|
13 |
|
|
This program is distributed in the hope that it will be useful, |
14 |
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of |
15 |
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
16 |
|
|
GNU General Public License for more details. |
17 |
|
|
|
18 |
|
|
You should have received a copy of the GNU General Public License |
19 |
|
|
along with this program; if not, write to the Free Software |
20 |
|
|
Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. */ |
21 |
|
|
|
22 |
|
|
/* AIX requires this to be the first thing in the file. */ |
23 |
|
|
#if defined (_AIX) && !defined (REGEX_MALLOC) |
24 |
|
|
#pragma alloca |
25 |
|
|
#endif |
26 |
|
|
|
27 |
|
|
#define _GNU_SOURCE |
28 |
|
|
|
29 |
|
|
#ifdef _WIN32 |
30 |
|
|
#define HAVE_STRING_H 1 /* Win32 */ |
31 |
|
|
#define REGEX_MALLOC 1 /* Win32 */ |
32 |
|
|
#endif |
33 |
|
|
|
34 |
|
|
|
35 |
|
|
/* We need this for `regex.h', and perhaps for the Emacs include files. */ |
36 |
|
|
#include <sys/types.h> |
37 |
|
|
|
38 |
|
|
#ifdef HAVE_CONFIG_H |
39 |
|
|
#include "config.h" |
40 |
|
|
#endif |
41 |
|
|
|
42 |
|
|
/* The `emacs' switch turns on certain matching commands |
43 |
|
|
that make sense only in Emacs. */ |
44 |
|
|
#ifdef emacs |
45 |
|
|
|
46 |
|
|
#include "lisp.h" |
47 |
|
|
#include "buffer.h" |
48 |
|
|
#include "syntax.h" |
49 |
|
|
|
50 |
|
|
/* Emacs uses `NULL' as a predicate. */ |
51 |
|
|
#undef NULL |
52 |
|
|
|
53 |
|
|
#else /* not emacs */ |
54 |
|
|
|
55 |
|
|
/* We used to test for `BSTRING' here, but only GCC and Emacs define |
56 |
|
|
`BSTRING', as far as I know, and neither of them use this code. */ |
57 |
|
|
#if HAVE_STRING_H || STDC_HEADERS |
58 |
|
|
#include <string.h> |
59 |
|
|
#ifndef bcmp |
60 |
|
|
#define bcmp(s1, s2, n) memcmp ((s1), (s2), (n)) |
61 |
|
|
#endif |
62 |
|
|
#ifndef bcopy |
63 |
|
|
#define bcopy(s, d, n) memcpy ((d), (s), (n)) |
64 |
|
|
#endif |
65 |
|
|
#ifndef bzero |
66 |
|
|
#define bzero(s, n) memset ((s), 0, (n)) |
67 |
|
|
#endif |
68 |
|
|
#else |
69 |
|
|
#include <strings.h> |
70 |
|
|
#endif |
71 |
|
|
|
72 |
|
|
#ifdef STDC_HEADERS |
73 |
|
|
#include <stdlib.h> |
74 |
|
|
#else |
75 |
|
|
char *malloc (); |
76 |
|
|
char *realloc (); |
77 |
|
|
#endif |
78 |
|
|
|
79 |
|
|
|
80 |
|
|
/* Define the syntax stuff for \<, \>, etc. */ |
81 |
|
|
|
82 |
|
|
/* This must be nonzero for the wordchar and notwordchar pattern |
83 |
|
|
commands in re_match_2. */ |
84 |
|
|
#ifndef Sword |
85 |
|
|
#define Sword 1 |
86 |
|
|
#endif |
87 |
|
|
|
88 |
|
|
#ifdef SYNTAX_TABLE |
89 |
|
|
|
90 |
|
|
extern char *re_syntax_table; |
91 |
|
|
|
92 |
|
|
#else /* not SYNTAX_TABLE */ |
93 |
|
|
|
94 |
|
|
/* How many characters in the character set. */ |
95 |
|
|
#define CHAR_SET_SIZE 256 |
96 |
|
|
|
97 |
|
|
static char re_syntax_table[CHAR_SET_SIZE]; |
98 |
|
|
|
99 |
|
|
static void |
100 |
|
|
init_syntax_once () |
101 |
|
|
{ |
102 |
|
|
register int c; |
103 |
|
|
static int done = 0; |
104 |
|
|
|
105 |
|
|
if (done) |
106 |
|
|
return; |
107 |
|
|
|
108 |
|
|
bzero (re_syntax_table, sizeof re_syntax_table); |
109 |
|
|
|
110 |
|
|
for (c = 'a'; c <= 'z'; c++) |
111 |
|
|
re_syntax_table[c] = Sword; |
112 |
|
|
|
113 |
|
|
for (c = 'A'; c <= 'Z'; c++) |
114 |
|
|
re_syntax_table[c] = Sword; |
115 |
|
|
|
116 |
|
|
for (c = '0'; c <= '9'; c++) |
117 |
|
|
re_syntax_table[c] = Sword; |
118 |
|
|
|
119 |
|
|
re_syntax_table['_'] = Sword; |
120 |
|
|
|
121 |
|
|
done = 1; |
122 |
|
|
} |
123 |
|
|
|
124 |
|
|
#endif /* not SYNTAX_TABLE */ |
125 |
|
|
|
126 |
|
|
#define SYNTAX(c) re_syntax_table[c] |
127 |
|
|
|
128 |
|
|
#endif /* not emacs */ |
129 |
|
|
|
130 |
|
|
/* Get the interface, including the syntax bits. */ |
131 |
|
|
#include "regex.h" |
132 |
|
|
|
133 |
|
|
/* isalpha etc. are used for the character classes. */ |
134 |
|
|
#include <ctype.h> |
135 |
|
|
|
136 |
|
|
#ifndef isascii |
137 |
|
|
#define isascii(c) 1 |
138 |
|
|
#endif |
139 |
|
|
|
140 |
|
|
#ifdef isblank |
141 |
|
|
#define ISBLANK(c) (isascii (c) && isblank (c)) |
142 |
|
|
#else |
143 |
|
|
#define ISBLANK(c) ((c) == ' ' || (c) == '\t') |
144 |
|
|
#endif |
145 |
|
|
#ifdef isgraph |
146 |
|
|
#define ISGRAPH(c) (isascii (c) && isgraph (c)) |
147 |
|
|
#else |
148 |
|
|
#define ISGRAPH(c) (isascii (c) && isprint (c) && !isspace (c)) |
149 |
|
|
#endif |
150 |
|
|
|
151 |
|
|
#define ISPRINT(c) (isascii (c) && isprint (c)) |
152 |
|
|
#define ISDIGIT(c) (isascii (c) && isdigit (c)) |
153 |
|
|
#define ISALNUM(c) (isascii (c) && isalnum (c)) |
154 |
|
|
#define ISALPHA(c) (isascii (c) && isalpha (c)) |
155 |
|
|
#define ISCNTRL(c) (isascii (c) && iscntrl (c)) |
156 |
|
|
#define ISLOWER(c) (isascii (c) && islower (c)) |
157 |
|
|
#define ISPUNCT(c) (isascii (c) && ispunct (c)) |
158 |
|
|
#define ISSPACE(c) (isascii (c) && isspace (c)) |
159 |
|
|
#define ISUPPER(c) (isascii (c) && isupper (c)) |
160 |
|
|
#define ISXDIGIT(c) (isascii (c) && isxdigit (c)) |
161 |
|
|
|
162 |
|
|
#ifndef NULL |
163 |
|
|
#define NULL 0 |
164 |
|
|
#endif |
165 |
|
|
|
166 |
|
|
/* We remove any previous definition of `SIGN_EXTEND_CHAR', |
167 |
|
|
since ours (we hope) works properly with all combinations of |
168 |
|
|
machines, compilers, `char' and `unsigned char' argument types. |
169 |
|
|
(Per Bothner suggested the basic approach.) */ |
170 |
|
|
#undef SIGN_EXTEND_CHAR |
171 |
|
|
#if __STDC__ |
172 |
|
|
#define SIGN_EXTEND_CHAR(c) ((signed char) (c)) |
173 |
|
|
#else /* not __STDC__ */ |
174 |
|
|
/* As in Harbison and Steele. */ |
175 |
|
|
#define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128) |
176 |
|
|
#endif |
177 |
|
|
|
178 |
|
|
/* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we |
179 |
|
|
use `alloca' instead of `malloc'. This is because using malloc in |
180 |
|
|
re_search* or re_match* could cause memory leaks when C-g is used in |
181 |
|
|
Emacs; also, malloc is slower and causes storage fragmentation. On |
182 |
|
|
the other hand, malloc is more portable, and easier to debug. |
183 |
|
|
|
184 |
|
|
Because we sometimes use alloca, some routines have to be macros, |
185 |
|
|
not functions -- `alloca'-allocated space disappears at the end of the |
186 |
|
|
function it is called in. */ |
187 |
|
|
|
188 |
|
|
#ifdef REGEX_MALLOC |
189 |
|
|
|
190 |
|
|
#define REGEX_ALLOCATE malloc |
191 |
|
|
#define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize) |
192 |
|
|
|
193 |
|
|
#else /* not REGEX_MALLOC */ |
194 |
|
|
|
195 |
|
|
/* Emacs already defines alloca, sometimes. */ |
196 |
|
|
#ifndef alloca |
197 |
|
|
|
198 |
|
|
/* Make alloca work the best possible way. */ |
199 |
|
|
#ifdef __GNUC__ |
200 |
|
|
#define alloca __builtin_alloca |
201 |
|
|
#else /* not __GNUC__ */ |
202 |
|
|
#if HAVE_ALLOCA_H |
203 |
|
|
#include <alloca.h> |
204 |
|
|
#else /* not __GNUC__ or HAVE_ALLOCA_H */ |
205 |
|
|
#ifndef _AIX /* Already did AIX, up at the top. */ |
206 |
|
|
char *alloca (); |
207 |
|
|
#endif /* not _AIX */ |
208 |
|
|
#endif /* not HAVE_ALLOCA_H */ |
209 |
|
|
#endif /* not __GNUC__ */ |
210 |
|
|
|
211 |
|
|
#endif /* not alloca */ |
212 |
|
|
|
213 |
|
|
#define REGEX_ALLOCATE alloca |
214 |
|
|
|
215 |
|
|
/* Assumes a `char *destination' variable. */ |
216 |
|
|
#define REGEX_REALLOCATE(source, osize, nsize) \ |
217 |
|
|
(destination = (char *) alloca (nsize), \ |
218 |
|
|
bcopy (source, destination, osize), \ |
219 |
|
|
destination) |
220 |
|
|
|
221 |
|
|
#endif /* not REGEX_MALLOC */ |
222 |
|
|
|
223 |
|
|
|
224 |
|
|
/* True if `size1' is non-NULL and PTR is pointing anywhere inside |
225 |
|
|
`string1' or just past its end. This works if PTR is NULL, which is |
226 |
|
|
a good thing. */ |
227 |
|
|
#define FIRST_STRING_P(ptr) \ |
228 |
|
|
(size1 && string1 <= (ptr) && (ptr) <= string1 + size1) |
229 |
|
|
|
230 |
|
|
/* (Re)Allocate N items of type T using malloc, or fail. */ |
231 |
|
|
#define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t))) |
232 |
|
|
#define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t))) |
233 |
|
|
#define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t))) |
234 |
|
|
|
235 |
|
|
#define BYTEWIDTH 8 /* In bits. */ |
236 |
|
|
|
237 |
|
|
#define STREQ(s1, s2) ((strcmp (s1, s2) == 0)) |
238 |
|
|
|
239 |
|
|
#define MAX(a, b) ((a) > (b) ? (a) : (b)) |
240 |
|
|
#define MIN(a, b) ((a) < (b) ? (a) : (b)) |
241 |
|
|
|
242 |
|
|
typedef char boolean; |
243 |
|
|
#define false 0 |
244 |
|
|
#define true 1 |
245 |
|
|
|
246 |
|
|
/* These are the command codes that appear in compiled regular |
247 |
|
|
expressions. Some opcodes are followed by argument bytes. A |
248 |
|
|
command code can specify any interpretation whatsoever for its |
249 |
|
|
arguments. Zero bytes may appear in the compiled regular expression. |
250 |
|
|
|
251 |
|
|
The value of `exactn' is needed in search.c (search_buffer) in Emacs. |
252 |
|
|
So regex.h defines a symbol `RE_EXACTN_VALUE' to be 1; the value of |
253 |
|
|
`exactn' we use here must also be 1. */ |
254 |
|
|
|
255 |
|
|
typedef enum |
256 |
|
|
{ |
257 |
|
|
no_op = 0, |
258 |
|
|
|
259 |
|
|
/* Followed by one byte giving n, then by n literal bytes. */ |
260 |
|
|
exactn = 1, |
261 |
|
|
|
262 |
|
|
/* Matches any (more or less) character. */ |
263 |
|
|
anychar, |
264 |
|
|
|
265 |
|
|
/* Matches any one char belonging to specified set. First |
266 |
|
|
following byte is number of bitmap bytes. Then come bytes |
267 |
|
|
for a bitmap saying which chars are in. Bits in each byte |
268 |
|
|
are ordered low-bit-first. A character is in the set if its |
269 |
|
|
bit is 1. A character too large to have a bit in the map is |
270 |
|
|
automatically not in the set. */ |
271 |
|
|
charset, |
272 |
|
|
|
273 |
|
|
/* Same parameters as charset, but match any character that is |
274 |
|
|
not one of those specified. */ |
275 |
|
|
charset_not, |
276 |
|
|
|
277 |
|
|
/* Start remembering the text that is matched, for storing in a |
278 |
|
|
register. Followed by one byte with the register number, in |
279 |
|
|
the range 0 to one less than the pattern buffer's re_nsub |
280 |
|
|
field. Then followed by one byte with the number of groups |
281 |
|
|
inner to this one. (This last has to be part of the |
282 |
|
|
start_memory only because we need it in the on_failure_jump |
283 |
|
|
of re_match_2.) */ |
284 |
|
|
start_memory, |
285 |
|
|
|
286 |
|
|
/* Stop remembering the text that is matched and store it in a |
287 |
|
|
memory register. Followed by one byte with the register |
288 |
|
|
number, in the range 0 to one less than `re_nsub' in the |
289 |
|
|
pattern buffer, and one byte with the number of inner groups, |
290 |
|
|
just like `start_memory'. (We need the number of inner |
291 |
|
|
groups here because we don't have any easy way of finding the |
292 |
|
|
corresponding start_memory when we're at a stop_memory.) */ |
293 |
|
|
stop_memory, |
294 |
|
|
|
295 |
|
|
/* Match a duplicate of something remembered. Followed by one |
296 |
|
|
byte containing the register number. */ |
297 |
|
|
duplicate, |
298 |
|
|
|
299 |
|
|
/* Fail unless at beginning of line. */ |
300 |
|
|
begline, |
301 |
|
|
|
302 |
|
|
/* Fail unless at end of line. */ |
303 |
|
|
endline, |
304 |
|
|
|
305 |
|
|
/* Succeeds if at beginning of buffer (if emacs) or at beginning |
306 |
|
|
of string to be matched (if not). */ |
307 |
|
|
begbuf, |
308 |
|
|
|
309 |
|
|
/* Analogously, for end of buffer/string. */ |
310 |
|
|
endbuf, |
311 |
|
|
|
312 |
|
|
/* Followed by two byte relative address to which to jump. */ |
313 |
|
|
jump, |
314 |
|
|
|
315 |
|
|
/* Same as jump, but marks the end of an alternative. */ |
316 |
|
|
jump_past_alt, |
317 |
|
|
|
318 |
|
|
/* Followed by two-byte relative address of place to resume at |
319 |
|
|
in case of failure. */ |
320 |
|
|
on_failure_jump, |
321 |
|
|
|
322 |
|
|
/* Like on_failure_jump, but pushes a placeholder instead of the |
323 |
|
|
current string position when executed. */ |
324 |
|
|
on_failure_keep_string_jump, |
325 |
|
|
|
326 |
|
|
/* Throw away latest failure point and then jump to following |
327 |
|
|
two-byte relative address. */ |
328 |
|
|
pop_failure_jump, |
329 |
|
|
|
330 |
|
|
/* Change to pop_failure_jump if know won't have to backtrack to |
331 |
|
|
match; otherwise change to jump. This is used to jump |
332 |
|
|
back to the beginning of a repeat. If what follows this jump |
333 |
|
|
clearly won't match what the repeat does, such that we can be |
334 |
|
|
sure that there is no use backtracking out of repetitions |
335 |
|
|
already matched, then we change it to a pop_failure_jump. |
336 |
|
|
Followed by two-byte address. */ |
337 |
|
|
maybe_pop_jump, |
338 |
|
|
|
339 |
|
|
/* Jump to following two-byte address, and push a dummy failure |
340 |
|
|
point. This failure point will be thrown away if an attempt |
341 |
|
|
is made to use it for a failure. A `+' construct makes this |
342 |
|
|
before the first repeat. Also used as an intermediary kind |
343 |
|
|
of jump when compiling an alternative. */ |
344 |
|
|
dummy_failure_jump, |
345 |
|
|
|
346 |
|
|
/* Push a dummy failure point and continue. Used at the end of |
347 |
|
|
alternatives. */ |
348 |
|
|
push_dummy_failure, |
349 |
|
|
|
350 |
|
|
/* Followed by two-byte relative address and two-byte number n. |
351 |
|
|
After matching N times, jump to the address upon failure. */ |
352 |
|
|
succeed_n, |
353 |
|
|
|
354 |
|
|
/* Followed by two-byte relative address, and two-byte number n. |
355 |
|
|
Jump to the address N times, then fail. */ |
356 |
|
|
jump_n, |
357 |
|
|
|
358 |
|
|
/* Set the following two-byte relative address to the |
359 |
|
|
subsequent two-byte number. The address *includes* the two |
360 |
|
|
bytes of number. */ |
361 |
|
|
set_number_at, |
362 |
|
|
|
363 |
|
|
wordchar, /* Matches any word-constituent character. */ |
364 |
|
|
notwordchar, /* Matches any char that is not a word-constituent. */ |
365 |
|
|
|
366 |
|
|
wordbeg, /* Succeeds if at word beginning. */ |
367 |
|
|
wordend, /* Succeeds if at word end. */ |
368 |
|
|
|
369 |
|
|
wordbound, /* Succeeds if at a word boundary. */ |
370 |
|
|
notwordbound /* Succeeds if not at a word boundary. */ |
371 |
|
|
|
372 |
|
|
#ifdef emacs |
373 |
|
|
,before_dot, /* Succeeds if before point. */ |
374 |
|
|
at_dot, /* Succeeds if at point. */ |
375 |
|
|
after_dot, /* Succeeds if after point. */ |
376 |
|
|
|
377 |
|
|
/* Matches any character whose syntax is specified. Followed by |
378 |
|
|
a byte which contains a syntax code, e.g., Sword. */ |
379 |
|
|
syntaxspec, |
380 |
|
|
|
381 |
|
|
/* Matches any character whose syntax is not that specified. */ |
382 |
|
|
notsyntaxspec |
383 |
|
|
#endif /* emacs */ |
384 |
|
|
} re_opcode_t; |
385 |
|
|
|
386 |
|
|
/* Common operations on the compiled pattern. */ |
387 |
|
|
|
388 |
|
|
/* Store NUMBER in two contiguous bytes starting at DESTINATION. */ |
389 |
|
|
|
390 |
|
|
#define STORE_NUMBER(destination, number) \ |
391 |
|
|
do { \ |
392 |
|
|
(destination)[0] = (number) & 0377; \ |
393 |
|
|
(destination)[1] = (number) >> 8; \ |
394 |
|
|
} while (0) |
395 |
|
|
|
396 |
|
|
/* Same as STORE_NUMBER, except increment DESTINATION to |
397 |
|
|
the byte after where the number is stored. Therefore, DESTINATION |
398 |
|
|
must be an lvalue. */ |
399 |
|
|
|
400 |
|
|
#define STORE_NUMBER_AND_INCR(destination, number) \ |
401 |
|
|
do { \ |
402 |
|
|
STORE_NUMBER (destination, number); \ |
403 |
|
|
(destination) += 2; \ |
404 |
|
|
} while (0) |
405 |
|
|
|
406 |
|
|
/* Put into DESTINATION a number stored in two contiguous bytes starting |
407 |
|
|
at SOURCE. */ |
408 |
|
|
|
409 |
|
|
#define EXTRACT_NUMBER(destination, source) \ |
410 |
|
|
do { \ |
411 |
|
|
(destination) = *(source) & 0377; \ |
412 |
|
|
(destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \ |
413 |
|
|
} while (0) |
414 |
|
|
|
415 |
|
|
#ifdef DEBUG |
416 |
|
|
static void |
417 |
|
|
extract_number (dest, source) |
418 |
|
|
int *dest; |
419 |
|
|
unsigned char *source; |
420 |
|
|
{ |
421 |
|
|
int temp = SIGN_EXTEND_CHAR (*(source + 1)); |
422 |
|
|
*dest = *source & 0377; |
423 |
|
|
*dest += temp << 8; |
424 |
|
|
} |
425 |
|
|
|
426 |
|
|
#ifndef EXTRACT_MACROS /* To debug the macros. */ |
427 |
|
|
#undef EXTRACT_NUMBER |
428 |
|
|
#define EXTRACT_NUMBER(dest, src) extract_number (&dest, src) |
429 |
|
|
#endif /* not EXTRACT_MACROS */ |
430 |
|
|
|
431 |
|
|
#endif /* DEBUG */ |
432 |
|
|
|
433 |
|
|
/* Same as EXTRACT_NUMBER, except increment SOURCE to after the number. |
434 |
|
|
SOURCE must be an lvalue. */ |
435 |
|
|
|
436 |
|
|
#define EXTRACT_NUMBER_AND_INCR(destination, source) \ |
437 |
|
|
do { \ |
438 |
|
|
EXTRACT_NUMBER (destination, source); \ |
439 |
|
|
(source) += 2; \ |
440 |
|
|
} while (0) |
441 |
|
|
|
442 |
|
|
#ifdef DEBUG |
443 |
|
|
static void |
444 |
|
|
extract_number_and_incr (destination, source) |
445 |
|
|
int *destination; |
446 |
|
|
unsigned char **source; |
447 |
|
|
{ |
448 |
|
|
extract_number (destination, *source); |
449 |
|
|
*source += 2; |
450 |
|
|
} |
451 |
|
|
|
452 |
|
|
#ifndef EXTRACT_MACROS |
453 |
|
|
#undef EXTRACT_NUMBER_AND_INCR |
454 |
|
|
#define EXTRACT_NUMBER_AND_INCR(dest, src) \ |
455 |
|
|
extract_number_and_incr (&dest, &src) |
456 |
|
|
#endif /* not EXTRACT_MACROS */ |
457 |
|
|
|
458 |
|
|
#endif /* DEBUG */ |
459 |
|
|
|
460 |
|
|
/* If DEBUG is defined, Regex prints many voluminous messages about what |
461 |
|
|
it is doing (if the variable `debug' is nonzero). If linked with the |
462 |
|
|
main program in `iregex.c', you can enter patterns and strings |
463 |
|
|
interactively. And if linked with the main program in `main.c' and |
464 |
|
|
the other test files, you can run the already-written tests. */ |
465 |
|
|
|
466 |
|
|
#ifdef DEBUG |
467 |
|
|
|
468 |
|
|
/* We use standard I/O for debugging. */ |
469 |
|
|
#include <stdio.h> |
470 |
|
|
|
471 |
|
|
/* It is useful to test things that ``must'' be true when debugging. */ |
472 |
|
|
#include <assert.h> |
473 |
|
|
|
474 |
|
|
static int debug = 0; |
475 |
|
|
|
476 |
|
|
#define DEBUG_STATEMENT(e) e |
477 |
|
|
#define DEBUG_PRINT1(x) if (debug) printf (x) |
478 |
|
|
#define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2) |
479 |
|
|
#define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3) |
480 |
|
|
#define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4) |
481 |
|
|
#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \ |
482 |
|
|
if (debug) print_partial_compiled_pattern (s, e) |
483 |
|
|
#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \ |
484 |
|
|
if (debug) print_double_string (w, s1, sz1, s2, sz2) |
485 |
|
|
|
486 |
|
|
|
487 |
|
|
extern void printchar (); |
488 |
|
|
|
489 |
|
|
/* Print the fastmap in human-readable form. */ |
490 |
|
|
|
491 |
|
|
void |
492 |
|
|
print_fastmap (fastmap) |
493 |
|
|
char *fastmap; |
494 |
|
|
{ |
495 |
|
|
unsigned was_a_range = 0; |
496 |
|
|
unsigned i = 0; |
497 |
|
|
|
498 |
|
|
while (i < (1 << BYTEWIDTH)) |
499 |
|
|
{ |
500 |
|
|
if (fastmap[i++]) |
501 |
|
|
{ |
502 |
|
|
was_a_range = 0; |
503 |
|
|
printchar (i - 1); |
504 |
|
|
while (i < (1 << BYTEWIDTH) && fastmap[i]) |
505 |
|
|
{ |
506 |
|
|
was_a_range = 1; |
507 |
|
|
i++; |
508 |
|
|
} |
509 |
|
|
if (was_a_range) |
510 |
|
|
{ |
511 |
|
|
printf ("-"); |
512 |
|
|
printchar (i - 1); |
513 |
|
|
} |
514 |
|
|
} |
515 |
|
|
} |
516 |
|
|
putchar ('\n'); |
517 |
|
|
} |
518 |
|
|
|
519 |
|
|
|
520 |
|
|
/* Print a compiled pattern string in human-readable form, starting at |
521 |
|
|
the START pointer into it and ending just before the pointer END. */ |
522 |
|
|
|
523 |
|
|
void |
524 |
|
|
print_partial_compiled_pattern (start, end) |
525 |
|
|
unsigned char *start; |
526 |
|
|
unsigned char *end; |
527 |
|
|
{ |
528 |
|
|
int mcnt, mcnt2; |
529 |
|
|
unsigned char *p = start; |
530 |
|
|
unsigned char *pend = end; |
531 |
|
|
|
532 |
|
|
if (start == NULL) |
533 |
|
|
{ |
534 |
|
|
printf ("(null)\n"); |
535 |
|
|
return; |
536 |
|
|
} |
537 |
|
|
|
538 |
|
|
/* Loop over pattern commands. */ |
539 |
|
|
while (p < pend) |
540 |
|
|
{ |
541 |
|
|
switch ((re_opcode_t) *p++) |
542 |
|
|
{ |
543 |
|
|
case no_op: |
544 |
|
|
printf ("/no_op"); |
545 |
|
|
break; |
546 |
|
|
|
547 |
|
|
case exactn: |
548 |
|
|
mcnt = *p++; |
549 |
|
|
printf ("/exactn/%d", mcnt); |
550 |
|
|
do |
551 |
|
|
{ |
552 |
|
|
putchar ('/'); |
553 |
|
|
printchar (*p++); |
554 |
|
|
} |
555 |
|
|
while (--mcnt); |
556 |
|
|
break; |
557 |
|
|
|
558 |
|
|
case start_memory: |
559 |
|
|
mcnt = *p++; |
560 |
|
|
printf ("/start_memory/%d/%d", mcnt, *p++); |
561 |
|
|
break; |
562 |
|
|
|
563 |
|
|
case stop_memory: |
564 |
|
|
mcnt = *p++; |
565 |
|
|
printf ("/stop_memory/%d/%d", mcnt, *p++); |
566 |
|
|
break; |
567 |
|
|
|
568 |
|
|
case duplicate: |
569 |
|
|
printf ("/duplicate/%d", *p++); |
570 |
|
|
break; |
571 |
|
|
|
572 |
|
|
case anychar: |
573 |
|
|
printf ("/anychar"); |
574 |
|
|
break; |
575 |
|
|
|
576 |
|
|
case charset: |
577 |
|
|
case charset_not: |
578 |
|
|
{ |
579 |
|
|
register int c; |
580 |
|
|
|
581 |
|
|
printf ("/charset%s", |
582 |
|
|
(re_opcode_t) *(p - 1) == charset_not ? "_not" : ""); |
583 |
|
|
|
584 |
|
|
assert (p + *p < pend); |
585 |
|
|
|
586 |
|
|
for (c = 0; c < *p; c++) |
587 |
|
|
{ |
588 |
|
|
unsigned bit; |
589 |
|
|
unsigned char map_byte = p[1 + c]; |
590 |
|
|
|
591 |
|
|
putchar ('/'); |
592 |
|
|
|
593 |
|
|
for (bit = 0; bit < BYTEWIDTH; bit++) |
594 |
|
|
if (map_byte & (1 << bit)) |
595 |
|
|
printchar (c * BYTEWIDTH + bit); |
596 |
|
|
} |
597 |
|
|
p += 1 + *p; |
598 |
|
|
break; |
599 |
|
|
} |
600 |
|
|
|
601 |
|
|
case begline: |
602 |
|
|
printf ("/begline"); |
603 |
|
|
break; |
604 |
|
|
|
605 |
|
|
case endline: |
606 |
|
|
printf ("/endline"); |
607 |
|
|
break; |
608 |
|
|
|
609 |
|
|
case on_failure_jump: |
610 |
|
|
extract_number_and_incr (&mcnt, &p); |
611 |
|
|
printf ("/on_failure_jump/0/%d", mcnt); |
612 |
|
|
break; |
613 |
|
|
|
614 |
|
|
case on_failure_keep_string_jump: |
615 |
|
|
extract_number_and_incr (&mcnt, &p); |
616 |
|
|
printf ("/on_failure_keep_string_jump/0/%d", mcnt); |
617 |
|
|
break; |
618 |
|
|
|
619 |
|
|
case dummy_failure_jump: |
620 |
|
|
extract_number_and_incr (&mcnt, &p); |
621 |
|
|
printf ("/dummy_failure_jump/0/%d", mcnt); |
622 |
|
|
break; |
623 |
|
|
|
624 |
|
|
case push_dummy_failure: |
625 |
|
|
printf ("/push_dummy_failure"); |
626 |
|
|
break; |
627 |
|
|
|
628 |
|
|
case maybe_pop_jump: |
629 |
|
|
extract_number_and_incr (&mcnt, &p); |
630 |
|
|
printf ("/maybe_pop_jump/0/%d", mcnt); |
631 |
|
|
break; |
632 |
|
|
|
633 |
|
|
case pop_failure_jump: |
634 |
|
|
extract_number_and_incr (&mcnt, &p); |
635 |
|
|
printf ("/pop_failure_jump/0/%d", mcnt); |
636 |
|
|
break; |
637 |
|
|
|
638 |
|
|
case jump_past_alt: |
639 |
|
|
extract_number_and_incr (&mcnt, &p); |
640 |
|
|
printf ("/jump_past_alt/0/%d", mcnt); |
641 |
|
|
break; |
642 |
|
|
|
643 |
|
|
case jump: |
644 |
|
|
extract_number_and_incr (&mcnt, &p); |
645 |
|
|
printf ("/jump/0/%d", mcnt); |
646 |
|
|
break; |
647 |
|
|
|
648 |
|
|
case succeed_n: |
649 |
|
|
extract_number_and_incr (&mcnt, &p); |
650 |
|
|
extract_number_and_incr (&mcnt2, &p); |
651 |
|
|
printf ("/succeed_n/0/%d/0/%d", mcnt, mcnt2); |
652 |
|
|
break; |
653 |
|
|
|
654 |
|
|
case jump_n: |
655 |
|
|
extract_number_and_incr (&mcnt, &p); |
656 |
|
|
extract_number_and_incr (&mcnt2, &p); |
657 |
|
|
printf ("/jump_n/0/%d/0/%d", mcnt, mcnt2); |
658 |
|
|
break; |
659 |
|
|
|
660 |
|
|
case set_number_at: |
661 |
|
|
extract_number_and_incr (&mcnt, &p); |
662 |
|
|
extract_number_and_incr (&mcnt2, &p); |
663 |
|
|
printf ("/set_number_at/0/%d/0/%d", mcnt, mcnt2); |
664 |
|
|
break; |
665 |
|
|
|
666 |
|
|
case wordbound: |
667 |
|
|
printf ("/wordbound"); |
668 |
|
|
break; |
669 |
|
|
|
670 |
|
|
case notwordbound: |
671 |
|
|
printf ("/notwordbound"); |
672 |
|
|
break; |
673 |
|
|
|
674 |
|
|
case wordbeg: |
675 |
|
|
printf ("/wordbeg"); |
676 |
|
|
break; |
677 |
|
|
|
678 |
|
|
case wordend: |
679 |
|
|
printf ("/wordend"); |
680 |
|
|
|
681 |
|
|
#ifdef emacs |
682 |
|
|
case before_dot: |
683 |
|
|
printf ("/before_dot"); |
684 |
|
|
break; |
685 |
|
|
|
686 |
|
|
case at_dot: |
687 |
|
|
printf ("/at_dot"); |
688 |
|
|
break; |
689 |
|
|
|
690 |
|
|
case after_dot: |
691 |
|
|
printf ("/after_dot"); |
692 |
|
|
break; |
693 |
|
|
|
694 |
|
|
case syntaxspec: |
695 |
|
|
printf ("/syntaxspec"); |
696 |
|
|
mcnt = *p++; |
697 |
|
|
printf ("/%d", mcnt); |
698 |
|
|
break; |
699 |
|
|
|
700 |
|
|
case notsyntaxspec: |
701 |
|
|
printf ("/notsyntaxspec"); |
702 |
|
|
mcnt = *p++; |
703 |
|
|
printf ("/%d", mcnt); |
704 |
|
|
break; |
705 |
|
|
#endif /* emacs */ |
706 |
|
|
|
707 |
|
|
case wordchar: |
708 |
|
|
printf ("/wordchar"); |
709 |
|
|
break; |
710 |
|
|
|
711 |
|
|
case notwordchar: |
712 |
|
|
printf ("/notwordchar"); |
713 |
|
|
break; |
714 |
|
|
|
715 |
|
|
case begbuf: |
716 |
|
|
printf ("/begbuf"); |
717 |
|
|
break; |
718 |
|
|
|
719 |
|
|
case endbuf: |
720 |
|
|
printf ("/endbuf"); |
721 |
|
|
break; |
722 |
|
|
|
723 |
|
|
default: |
724 |
|
|
printf ("?%d", *(p-1)); |
725 |
|
|
} |
726 |
|
|
} |
727 |
|
|
printf ("/\n"); |
728 |
|
|
} |
729 |
|
|
|
730 |
|
|
|
731 |
|
|
void |
732 |
|
|
print_compiled_pattern (bufp) |
733 |
|
|
struct re_pattern_buffer *bufp; |
734 |
|
|
{ |
735 |
|
|
unsigned char *buffer = bufp->buffer; |
736 |
|
|
|
737 |
|
|
print_partial_compiled_pattern (buffer, buffer + bufp->used); |
738 |
|
|
printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated); |
739 |
|
|
|
740 |
|
|
if (bufp->fastmap_accurate && bufp->fastmap) |
741 |
|
|
{ |
742 |
|
|
printf ("fastmap: "); |
743 |
|
|
print_fastmap (bufp->fastmap); |
744 |
|
|
} |
745 |
|
|
|
746 |
|
|
printf ("re_nsub: %d\t", bufp->re_nsub); |
747 |
|
|
printf ("regs_alloc: %d\t", bufp->regs_allocated); |
748 |
|
|
printf ("can_be_null: %d\t", bufp->can_be_null); |
749 |
|
|
printf ("newline_anchor: %d\n", bufp->newline_anchor); |
750 |
|
|
printf ("no_sub: %d\t", bufp->no_sub); |
751 |
|
|
printf ("not_bol: %d\t", bufp->not_bol); |
752 |
|
|
printf ("not_eol: %d\t", bufp->not_eol); |
753 |
|
|
printf ("syntax: %d\n", bufp->syntax); |
754 |
|
|
/* Perhaps we should print the translate table? */ |
755 |
|
|
} |
756 |
|
|
|
757 |
|
|
|
758 |
|
|
void |
759 |
|
|
print_double_string (where, string1, size1, string2, size2) |
760 |
|
|
const char *where; |
761 |
|
|
const char *string1; |
762 |
|
|
const char *string2; |
763 |
|
|
int size1; |
764 |
|
|
int size2; |
765 |
|
|
{ |
766 |
|
|
unsigned this_char; |
767 |
|
|
|
768 |
|
|
if (where == NULL) |
769 |
|
|
printf ("(null)"); |
770 |
|
|
else |
771 |
|
|
{ |
772 |
|
|
if (FIRST_STRING_P (where)) |
773 |
|
|
{ |
774 |
|
|
for (this_char = where - string1; this_char < size1; this_char++) |
775 |
|
|
printchar (string1[this_char]); |
776 |
|
|
|
777 |
|
|
where = string2; |
778 |
|
|
} |
779 |
|
|
|
780 |
|
|
for (this_char = where - string2; this_char < size2; this_char++) |
781 |
|
|
printchar (string2[this_char]); |
782 |
|
|
} |
783 |
|
|
} |
784 |
|
|
|
785 |
|
|
#else /* not DEBUG */ |
786 |
|
|
|
787 |
|
|
#undef assert |
788 |
|
|
#define assert(e) |
789 |
|
|
|
790 |
|
|
#define DEBUG_STATEMENT(e) |
791 |
|
|
#define DEBUG_PRINT1(x) |
792 |
|
|
#define DEBUG_PRINT2(x1, x2) |
793 |
|
|
#define DEBUG_PRINT3(x1, x2, x3) |
794 |
|
|
#define DEBUG_PRINT4(x1, x2, x3, x4) |
795 |
|
|
#define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) |
796 |
|
|
#define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) |
797 |
|
|
|
798 |
|
|
#endif /* not DEBUG */ |
799 |
|
|
|
800 |
|
|
/* Set by `re_set_syntax' to the current regexp syntax to recognize. Can |
801 |
|
|
also be assigned to arbitrarily: each pattern buffer stores its own |
802 |
|
|
syntax, so it can be changed between regex compilations. */ |
803 |
|
|
reg_syntax_t re_syntax_options = RE_SYNTAX_EMACS; |
804 |
|
|
|
805 |
|
|
|
806 |
|
|
/* Specify the precise syntax of regexps for compilation. This provides |
807 |
|
|
for compatibility for various utilities which historically have |
808 |
|
|
different, incompatible syntaxes. |
809 |
|
|
|
810 |
|
|
The argument SYNTAX is a bit mask comprised of the various bits |
811 |
|
|
defined in regex.h. We return the old syntax. */ |
812 |
|
|
|
813 |
|
|
reg_syntax_t |
814 |
|
|
re_set_syntax (syntax) |
815 |
|
|
reg_syntax_t syntax; |
816 |
|
|
{ |
817 |
|
|
reg_syntax_t ret = re_syntax_options; |
818 |
|
|
|
819 |
|
|
re_syntax_options = syntax; |
820 |
|
|
return ret; |
821 |
|
|
} |
822 |
|
|
|
823 |
|
|
/* This table gives an error message for each of the error codes listed |
824 |
|
|
in regex.h. Obviously the order here has to be same as there. */ |
825 |
|
|
|
826 |
|
|
static const char *re_error_msg[] = |
827 |
|
|
{ NULL, /* REG_NOERROR */ |
828 |
|
|
"No match", /* REG_NOMATCH */ |
829 |
|
|
"Invalid regular expression", /* REG_BADPAT */ |
830 |
|
|
"Invalid collation character", /* REG_ECOLLATE */ |
831 |
|
|
"Invalid character class name", /* REG_ECTYPE */ |
832 |
|
|
"Trailing backslash", /* REG_EESCAPE */ |
833 |
|
|
"Invalid back reference", /* REG_ESUBREG */ |
834 |
|
|
"Unmatched [ or [^", /* REG_EBRACK */ |
835 |
|
|
"Unmatched ( or \\(", /* REG_EPAREN */ |
836 |
|
|
"Unmatched \\{", /* REG_EBRACE */ |
837 |
|
|
"Invalid content of \\{\\}", /* REG_BADBR */ |
838 |
|
|
"Invalid range end", /* REG_ERANGE */ |
839 |
|
|
"Memory exhausted", /* REG_ESPACE */ |
840 |
|
|
"Invalid preceding regular expression", /* REG_BADRPT */ |
841 |
|
|
"Premature end of regular expression", /* REG_EEND */ |
842 |
|
|
"Regular expression too big", /* REG_ESIZE */ |
843 |
|
|
"Unmatched ) or \\)", /* REG_ERPAREN */ |
844 |
|
|
}; |
845 |
|
|
|
846 |
|
|
/* Subroutine declarations and macros for regex_compile. */ |
847 |
|
|
|
848 |
|
|
static void store_op1 (), store_op2 (); |
849 |
|
|
static void insert_op1 (), insert_op2 (); |
850 |
|
|
static boolean at_begline_loc_p (), at_endline_loc_p (); |
851 |
|
|
static boolean group_in_compile_stack (); |
852 |
|
|
static reg_errcode_t compile_range (); |
853 |
|
|
|
854 |
|
|
/* Fetch the next character in the uncompiled pattern---translating it |
855 |
|
|
if necessary. Also cast from a signed character in the constant |
856 |
|
|
string passed to us by the user to an unsigned char that we can use |
857 |
|
|
as an array index (in, e.g., `translate'). */ |
858 |
|
|
#define PATFETCH(c) \ |
859 |
|
|
do {if (p == pend) return REG_EEND; \ |
860 |
|
|
c = (unsigned char) *p++; \ |
861 |
|
|
if (translate) c = translate[c]; \ |
862 |
|
|
} while (0) |
863 |
|
|
|
864 |
|
|
/* Fetch the next character in the uncompiled pattern, with no |
865 |
|
|
translation. */ |
866 |
|
|
#define PATFETCH_RAW(c) \ |
867 |
|
|
do {if (p == pend) return REG_EEND; \ |
868 |
|
|
c = (unsigned char) *p++; \ |
869 |
|
|
} while (0) |
870 |
|
|
|
871 |
|
|
/* Go backwards one character in the pattern. */ |
872 |
|
|
#define PATUNFETCH p-- |
873 |
|
|
|
874 |
|
|
|
875 |
|
|
/* If `translate' is non-null, return translate[D], else just D. We |
876 |
|
|
cast the subscript to translate because some data is declared as |
877 |
|
|
`char *', to avoid warnings when a string constant is passed. But |
878 |
|
|
when we use a character as a subscript we must make it unsigned. */ |
879 |
|
|
#define TRANSLATE(d) (translate ? translate[(unsigned char) (d)] : (d)) |
880 |
|
|
|
881 |
|
|
|
882 |
|
|
/* Macros for outputting the compiled pattern into `buffer'. */ |
883 |
|
|
|
884 |
|
|
/* If the buffer isn't allocated when it comes in, use this. */ |
885 |
|
|
#define INIT_BUF_SIZE 32 |
886 |
|
|
|
887 |
|
|
/* Make sure we have at least N more bytes of space in buffer. */ |
888 |
|
|
#define GET_BUFFER_SPACE(n) \ |
889 |
|
|
while (b - bufp->buffer + (n) > bufp->allocated) \ |
890 |
|
|
EXTEND_BUFFER () |
891 |
|
|
|
892 |
|
|
/* Make sure we have one more byte of buffer space and then add C to it. */ |
893 |
|
|
#define BUF_PUSH(c) \ |
894 |
|
|
do { \ |
895 |
|
|
GET_BUFFER_SPACE (1); \ |
896 |
|
|
*b++ = (unsigned char) (c); \ |
897 |
|
|
} while (0) |
898 |
|
|
|
899 |
|
|
|
900 |
|
|
/* Ensure we have two more bytes of buffer space and then append C1 and C2. */ |
901 |
|
|
#define BUF_PUSH_2(c1, c2) \ |
902 |
|
|
do { \ |
903 |
|
|
GET_BUFFER_SPACE (2); \ |
904 |
|
|
*b++ = (unsigned char) (c1); \ |
905 |
|
|
*b++ = (unsigned char) (c2); \ |
906 |
|
|
} while (0) |
907 |
|
|
|
908 |
|
|
|
909 |
|
|
/* As with BUF_PUSH_2, except for three bytes. */ |
910 |
|
|
#define BUF_PUSH_3(c1, c2, c3) \ |
911 |
|
|
do { \ |
912 |
|
|
GET_BUFFER_SPACE (3); \ |
913 |
|
|
*b++ = (unsigned char) (c1); \ |
914 |
|
|
*b++ = (unsigned char) (c2); \ |
915 |
|
|
*b++ = (unsigned char) (c3); \ |
916 |
|
|
} while (0) |
917 |
|
|
|
918 |
|
|
|
919 |
|
|
/* Store a jump with opcode OP at LOC to location TO. We store a |
920 |
|
|
relative address offset by the three bytes the jump itself occupies. */ |
921 |
|
|
#define STORE_JUMP(op, loc, to) \ |
922 |
|
|
store_op1 (op, loc, (to) - (loc) - 3) |
923 |
|
|
|
924 |
|
|
/* Likewise, for a two-argument jump. */ |
925 |
|
|
#define STORE_JUMP2(op, loc, to, arg) \ |
926 |
|
|
store_op2 (op, loc, (to) - (loc) - 3, arg) |
927 |
|
|
|
928 |
|
|
/* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */ |
929 |
|
|
#define INSERT_JUMP(op, loc, to) \ |
930 |
|
|
insert_op1 (op, loc, (to) - (loc) - 3, b) |
931 |
|
|
|
932 |
|
|
/* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */ |
933 |
|
|
#define INSERT_JUMP2(op, loc, to, arg) \ |
934 |
|
|
insert_op2 (op, loc, (to) - (loc) - 3, arg, b) |
935 |
|
|
|
936 |
|
|
|
937 |
|
|
/* This is not an arbitrary limit: the arguments which represent offsets |
938 |
|
|
into the pattern are two bytes long. So if 2^16 bytes turns out to |
939 |
|
|
be too small, many things would have to change. */ |
940 |
|
|
#define MAX_BUF_SIZE (1L << 16) |
941 |
|
|
|
942 |
|
|
|
943 |
|
|
/* Extend the buffer by twice its current size via realloc and |
944 |
|
|
reset the pointers that pointed into the old block to point to the |
945 |
|
|
correct places in the new one. If extending the buffer results in it |
946 |
|
|
being larger than MAX_BUF_SIZE, then flag memory exhausted. */ |
947 |
|
|
#define EXTEND_BUFFER() \ |
948 |
|
|
do { \ |
949 |
|
|
unsigned char *old_buffer = bufp->buffer; \ |
950 |
|
|
if (bufp->allocated == MAX_BUF_SIZE) \ |
951 |
|
|
return REG_ESIZE; \ |
952 |
|
|
bufp->allocated <<= 1; \ |
953 |
|
|
if (bufp->allocated > MAX_BUF_SIZE) \ |
954 |
|
|
bufp->allocated = MAX_BUF_SIZE; \ |
955 |
|
|
bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\ |
956 |
|
|
if (bufp->buffer == NULL) \ |
957 |
|
|
return REG_ESPACE; \ |
958 |
|
|
/* If the buffer moved, move all the pointers into it. */ \ |
959 |
|
|
if (old_buffer != bufp->buffer) \ |
960 |
|
|
{ \ |
961 |
|
|
b = (b - old_buffer) + bufp->buffer; \ |
962 |
|
|
begalt = (begalt - old_buffer) + bufp->buffer; \ |
963 |
|
|
if (fixup_alt_jump) \ |
964 |
|
|
fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\ |
965 |
|
|
if (laststart) \ |
966 |
|
|
laststart = (laststart - old_buffer) + bufp->buffer; \ |
967 |
|
|
if (pending_exact) \ |
968 |
|
|
pending_exact = (pending_exact - old_buffer) + bufp->buffer; \ |
969 |
|
|
} \ |
970 |
|
|
} while (0) |
971 |
|
|
|
972 |
|
|
|
973 |
|
|
/* Since we have one byte reserved for the register number argument to |
974 |
|
|
{start,stop}_memory, the maximum number of groups we can report |
975 |
|
|
things about is what fits in that byte. */ |
976 |
|
|
#define MAX_REGNUM 255 |
977 |
|
|
|
978 |
|
|
/* But patterns can have more than `MAX_REGNUM' registers. We just |
979 |
|
|
ignore the excess. */ |
980 |
|
|
typedef unsigned regnum_t; |
981 |
|
|
|
982 |
|
|
|
983 |
|
|
/* Macros for the compile stack. */ |
984 |
|
|
|
985 |
|
|
/* Since offsets can go either forwards or backwards, this type needs to |
986 |
|
|
be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */ |
987 |
|
|
typedef int pattern_offset_t; |
988 |
|
|
|
989 |
|
|
typedef struct |
990 |
|
|
{ |
991 |
|
|
pattern_offset_t begalt_offset; |
992 |
|
|
pattern_offset_t fixup_alt_jump; |
993 |
|
|
pattern_offset_t inner_group_offset; |
994 |
|
|
pattern_offset_t laststart_offset; |
995 |
|
|
regnum_t regnum; |
996 |
|
|
} compile_stack_elt_t; |
997 |
|
|
|
998 |
|
|
|
999 |
|
|
typedef struct |
1000 |
|
|
{ |
1001 |
|
|
compile_stack_elt_t *stack; |
1002 |
|
|
unsigned size; |
1003 |
|
|
unsigned avail; /* Offset of next open position. */ |
1004 |
|
|
} compile_stack_type; |
1005 |
|
|
|
1006 |
|
|
|
1007 |
|
|
#define INIT_COMPILE_STACK_SIZE 32 |
1008 |
|
|
|
1009 |
|
|
#define COMPILE_STACK_EMPTY (compile_stack.avail == 0) |
1010 |
|
|
#define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size) |
1011 |
|
|
|
1012 |
|
|
/* The next available element. */ |
1013 |
|
|
#define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail]) |
1014 |
|
|
|
1015 |
|
|
|
1016 |
|
|
/* Set the bit for character C in a list. */ |
1017 |
|
|
#define SET_LIST_BIT(c) \ |
1018 |
|
|
(b[((unsigned char) (c)) / BYTEWIDTH] \ |
1019 |
|
|
|= 1 << (((unsigned char) c) % BYTEWIDTH)) |
1020 |
|
|
|
1021 |
|
|
|
1022 |
|
|
/* Get the next unsigned number in the uncompiled pattern. */ |
1023 |
|
|
#define GET_UNSIGNED_NUMBER(num) \ |
1024 |
|
|
{ if (p != pend) \ |
1025 |
|
|
{ \ |
1026 |
|
|
PATFETCH (c); \ |
1027 |
|
|
while (ISDIGIT (c)) \ |
1028 |
|
|
{ \ |
1029 |
|
|
if (num < 0) \ |
1030 |
|
|
num = 0; \ |
1031 |
|
|
num = num * 10 + c - '0'; \ |
1032 |
|
|
if (p == pend) \ |
1033 |
|
|
break; \ |
1034 |
|
|
PATFETCH (c); \ |
1035 |
|
|
} \ |
1036 |
|
|
} \ |
1037 |
|
|
} |
1038 |
|
|
|
1039 |
|
|
#define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */ |
1040 |
|
|
|
1041 |
|
|
#define IS_CHAR_CLASS(string) \ |
1042 |
|
|
(STREQ (string, "alpha") || STREQ (string, "upper") \ |
1043 |
|
|
|| STREQ (string, "lower") || STREQ (string, "digit") \ |
1044 |
|
|
|| STREQ (string, "alnum") || STREQ (string, "xdigit") \ |
1045 |
|
|
|| STREQ (string, "space") || STREQ (string, "print") \ |
1046 |
|
|
|| STREQ (string, "punct") || STREQ (string, "graph") \ |
1047 |
|
|
|| STREQ (string, "cntrl") || STREQ (string, "blank")) |
1048 |
|
|
|
1049 |
|
|
/* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX. |
1050 |
|
|
Returns one of error codes defined in `regex.h', or zero for success. |
1051 |
|
|
|
1052 |
|
|
Assumes the `allocated' (and perhaps `buffer') and `translate' |
1053 |
|
|
fields are set in BUFP on entry. |
1054 |
|
|
|
1055 |
|
|
If it succeeds, results are put in BUFP (if it returns an error, the |
1056 |
|
|
contents of BUFP are undefined): |
1057 |
|
|
`buffer' is the compiled pattern; |
1058 |
|
|
`syntax' is set to SYNTAX; |
1059 |
|
|
`used' is set to the length of the compiled pattern; |
1060 |
|
|
`fastmap_accurate' is zero; |
1061 |
|
|
`re_nsub' is the number of subexpressions in PATTERN; |
1062 |
|
|
`not_bol' and `not_eol' are zero; |
1063 |
|
|
|
1064 |
|
|
The `fastmap' and `newline_anchor' fields are neither |
1065 |
|
|
examined nor set. */ |
1066 |
|
|
|
1067 |
|
|
static reg_errcode_t |
1068 |
|
|
regex_compile (pattern, size, syntax, bufp) |
1069 |
|
|
const char *pattern; |
1070 |
|
|
int size; |
1071 |
|
|
reg_syntax_t syntax; |
1072 |
|
|
struct re_pattern_buffer *bufp; |
1073 |
|
|
{ |
1074 |
|
|
/* We fetch characters from PATTERN here. Even though PATTERN is |
1075 |
|
|
`char *' (i.e., signed), we declare these variables as unsigned, so |
1076 |
|
|
they can be reliably used as array indices. */ |
1077 |
|
|
register unsigned char c, c1; |
1078 |
|
|
|
1079 |
|
|
/* A random tempory spot in PATTERN. */ |
1080 |
|
|
const char *p1; |
1081 |
|
|
|
1082 |
|
|
/* Points to the end of the buffer, where we should append. */ |
1083 |
|
|
register unsigned char *b; |
1084 |
|
|
|
1085 |
|
|
/* Keeps track of unclosed groups. */ |
1086 |
|
|
compile_stack_type compile_stack; |
1087 |
|
|
|
1088 |
|
|
/* Points to the current (ending) position in the pattern. */ |
1089 |
|
|
const char *p = pattern; |
1090 |
|
|
const char *pend = pattern + size; |
1091 |
|
|
|
1092 |
|
|
/* How to translate the characters in the pattern. */ |
1093 |
|
|
char *translate = bufp->translate; |
1094 |
|
|
|
1095 |
|
|
/* Address of the count-byte of the most recently inserted `exactn' |
1096 |
|
|
command. This makes it possible to tell if a new exact-match |
1097 |
|
|
character can be added to that command or if the character requires |
1098 |
|
|
a new `exactn' command. */ |
1099 |
|
|
unsigned char *pending_exact = 0; |
1100 |
|
|
|
1101 |
|
|
/* Address of start of the most recently finished expression. |
1102 |
|
|
This tells, e.g., postfix * where to find the start of its |
1103 |
|
|
operand. Reset at the beginning of groups and alternatives. */ |
1104 |
|
|
unsigned char *laststart = 0; |
1105 |
|
|
|
1106 |
|
|
/* Address of beginning of regexp, or inside of last group. */ |
1107 |
|
|
unsigned char *begalt; |
1108 |
|
|
|
1109 |
|
|
/* Place in the uncompiled pattern (i.e., the {) to |
1110 |
|
|
which to go back if the interval is invalid. */ |
1111 |
|
|
const char *beg_interval; |
1112 |
|
|
|
1113 |
|
|
/* Address of the place where a forward jump should go to the end of |
1114 |
|
|
the containing expression. Each alternative of an `or' -- except the |
1115 |
|
|
last -- ends with a forward jump of this sort. */ |
1116 |
|
|
unsigned char *fixup_alt_jump = 0; |
1117 |
|
|
|
1118 |
|
|
/* Counts open-groups as they are encountered. Remembered for the |
1119 |
|
|
matching close-group on the compile stack, so the same register |
1120 |
|
|
number is put in the stop_memory as the start_memory. */ |
1121 |
|
|
regnum_t regnum = 0; |
1122 |
|
|
|
1123 |
|
|
#ifdef DEBUG |
1124 |
|
|
DEBUG_PRINT1 ("\nCompiling pattern: "); |
1125 |
|
|
if (debug) |
1126 |
|
|
{ |
1127 |
|
|
unsigned debug_count; |
1128 |
|
|
|
1129 |
|
|
for (debug_count = 0; debug_count < size; debug_count++) |
1130 |
|
|
printchar (pattern[debug_count]); |
1131 |
|
|
putchar ('\n'); |
1132 |
|
|
} |
1133 |
|
|
#endif /* DEBUG */ |
1134 |
|
|
|
1135 |
|
|
/* Initialize the compile stack. */ |
1136 |
|
|
compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t); |
1137 |
|
|
if (compile_stack.stack == NULL) |
1138 |
|
|
return REG_ESPACE; |
1139 |
|
|
|
1140 |
|
|
compile_stack.size = INIT_COMPILE_STACK_SIZE; |
1141 |
|
|
compile_stack.avail = 0; |
1142 |
|
|
|
1143 |
|
|
/* Initialize the pattern buffer. */ |
1144 |
|
|
bufp->syntax = syntax; |
1145 |
|
|
bufp->fastmap_accurate = 0; |
1146 |
|
|
bufp->not_bol = bufp->not_eol = 0; |
1147 |
|
|
|
1148 |
|
|
/* Set `used' to zero, so that if we return an error, the pattern |
1149 |
|
|
printer (for debugging) will think there's no pattern. We reset it |
1150 |
|
|
at the end. */ |
1151 |
|
|
bufp->used = 0; |
1152 |
|
|
|
1153 |
|
|
/* Always count groups, whether or not bufp->no_sub is set. */ |
1154 |
|
|
bufp->re_nsub = 0; |
1155 |
|
|
|
1156 |
|
|
#if !defined (emacs) && !defined (SYNTAX_TABLE) |
1157 |
|
|
/* Initialize the syntax table. */ |
1158 |
|
|
init_syntax_once (); |
1159 |
|
|
#endif |
1160 |
|
|
|
1161 |
|
|
if (bufp->allocated == 0) |
1162 |
|
|
{ |
1163 |
|
|
if (bufp->buffer) |
1164 |
|
|
{ /* If zero allocated, but buffer is non-null, try to realloc |
1165 |
|
|
enough space. This loses if buffer's address is bogus, but |
1166 |
|
|
that is the user's responsibility. */ |
1167 |
|
|
RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char); |
1168 |
|
|
} |
1169 |
|
|
else |
1170 |
|
|
{ /* Caller did not allocate a buffer. Do it for them. */ |
1171 |
|
|
bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char); |
1172 |
|
|
} |
1173 |
|
|
if (!bufp->buffer) return REG_ESPACE; |
1174 |
|
|
|
1175 |
|
|
bufp->allocated = INIT_BUF_SIZE; |
1176 |
|
|
} |
1177 |
|
|
|
1178 |
|
|
begalt = b = bufp->buffer; |
1179 |
|
|
|
1180 |
|
|
/* Loop through the uncompiled pattern until we're at the end. */ |
1181 |
|
|
while (p != pend) |
1182 |
|
|
{ |
1183 |
|
|
PATFETCH (c); |
1184 |
|
|
|
1185 |
|
|
switch (c) |
1186 |
|
|
{ |
1187 |
|
|
case '^': |
1188 |
|
|
{ |
1189 |
|
|
if ( /* If at start of pattern, it's an operator. */ |
1190 |
|
|
p == pattern + 1 |
1191 |
|
|
/* If context independent, it's an operator. */ |
1192 |
|
|
|| syntax & RE_CONTEXT_INDEP_ANCHORS |
1193 |
|
|
/* Otherwise, depends on what's come before. */ |
1194 |
|
|
|| at_begline_loc_p (pattern, p, syntax)) |
1195 |
|
|
BUF_PUSH (begline); |
1196 |
|
|
else |
1197 |
|
|
goto normal_char; |
1198 |
|
|
} |
1199 |
|
|
break; |
1200 |
|
|
|
1201 |
|
|
|
1202 |
|
|
case '$': |
1203 |
|
|
{ |
1204 |
|
|
if ( /* If at end of pattern, it's an operator. */ |
1205 |
|
|
p == pend |
1206 |
|
|
/* If context independent, it's an operator. */ |
1207 |
|
|
|| syntax & RE_CONTEXT_INDEP_ANCHORS |
1208 |
|
|
/* Otherwise, depends on what's next. */ |
1209 |
|
|
|| at_endline_loc_p (p, pend, syntax)) |
1210 |
|
|
BUF_PUSH (endline); |
1211 |
|
|
else |
1212 |
|
|
goto normal_char; |
1213 |
|
|
} |
1214 |
|
|
break; |
1215 |
|
|
|
1216 |
|
|
|
1217 |
|
|
case '+': |
1218 |
|
|
case '?': |
1219 |
|
|
if ((syntax & RE_BK_PLUS_QM) |
1220 |
|
|
|| (syntax & RE_LIMITED_OPS)) |
1221 |
|
|
goto normal_char; |
1222 |
|
|
handle_plus: |
1223 |
|
|
case '*': |
1224 |
|
|
/* If there is no previous pattern... */ |
1225 |
|
|
if (!laststart) |
1226 |
|
|
{ |
1227 |
|
|
if (syntax & RE_CONTEXT_INVALID_OPS) |
1228 |
|
|
return REG_BADRPT; |
1229 |
|
|
else if (!(syntax & RE_CONTEXT_INDEP_OPS)) |
1230 |
|
|
goto normal_char; |
1231 |
|
|
} |
1232 |
|
|
|
1233 |
|
|
{ |
1234 |
|
|
/* Are we optimizing this jump? */ |
1235 |
|
|
boolean keep_string_p = false; |
1236 |
|
|
|
1237 |
|
|
/* 1 means zero (many) matches is allowed. */ |
1238 |
|
|
char zero_times_ok = 0, many_times_ok = 0; |
1239 |
|
|
|
1240 |
|
|
/* If there is a sequence of repetition chars, collapse it |
1241 |
|
|
down to just one (the right one). We can't combine |
1242 |
|
|
interval operators with these because of, e.g., `a{2}*', |
1243 |
|
|
which should only match an even number of `a's. */ |
1244 |
|
|
|
1245 |
|
|
for (;;) |
1246 |
|
|
{ |
1247 |
|
|
zero_times_ok |= c != '+'; |
1248 |
|
|
many_times_ok |= c != '?'; |
1249 |
|
|
|
1250 |
|
|
if (p == pend) |
1251 |
|
|
break; |
1252 |
|
|
|
1253 |
|
|
PATFETCH (c); |
1254 |
|
|
|
1255 |
|
|
if (c == '*' |
1256 |
|
|
|| (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?'))) |
1257 |
|
|
; |
1258 |
|
|
|
1259 |
|
|
else if (syntax & RE_BK_PLUS_QM && c == '\\') |
1260 |
|
|
{ |
1261 |
|
|
if (p == pend) return REG_EESCAPE; |
1262 |
|
|
|
1263 |
|
|
PATFETCH (c1); |
1264 |
|
|
if (!(c1 == '+' || c1 == '?')) |
1265 |
|
|
{ |
1266 |
|
|
PATUNFETCH; |
1267 |
|
|
PATUNFETCH; |
1268 |
|
|
break; |
1269 |
|
|
} |
1270 |
|
|
|
1271 |
|
|
c = c1; |
1272 |
|
|
} |
1273 |
|
|
else |
1274 |
|
|
{ |
1275 |
|
|
PATUNFETCH; |
1276 |
|
|
break; |
1277 |
|
|
} |
1278 |
|
|
|
1279 |
|
|
/* If we get here, we found another repeat character. */ |
1280 |
|
|
} |
1281 |
|
|
|
1282 |
|
|
/* Star, etc. applied to an empty pattern is equivalent |
1283 |
|
|
to an empty pattern. */ |
1284 |
|
|
if (!laststart) |
1285 |
|
|
break; |
1286 |
|
|
|
1287 |
|
|
/* Now we know whether or not zero matches is allowed |
1288 |
|
|
and also whether or not two or more matches is allowed. */ |
1289 |
|
|
if (many_times_ok) |
1290 |
|
|
{ /* More than one repetition is allowed, so put in at the |
1291 |
|
|
end a backward relative jump from `b' to before the next |
1292 |
|
|
jump we're going to put in below (which jumps from |
1293 |
|
|
laststart to after this jump). |
1294 |
|
|
|
1295 |
|
|
But if we are at the `*' in the exact sequence `.*\n', |
1296 |
|
|
insert an unconditional jump backwards to the ., |
1297 |
|
|
instead of the beginning of the loop. This way we only |
1298 |
|
|
push a failure point once, instead of every time |
1299 |
|
|
through the loop. */ |
1300 |
|
|
assert (p - 1 > pattern); |
1301 |
|
|
|
1302 |
|
|
/* Allocate the space for the jump. */ |
1303 |
|
|
GET_BUFFER_SPACE (3); |
1304 |
|
|
|
1305 |
|
|
/* We know we are not at the first character of the pattern, |
1306 |
|
|
because laststart was nonzero. And we've already |
1307 |
|
|
incremented `p', by the way, to be the character after |
1308 |
|
|
the `*'. Do we have to do something analogous here |
1309 |
|
|
for null bytes, because of RE_DOT_NOT_NULL? */ |
1310 |
|
|
if (TRANSLATE (*(p - 2)) == TRANSLATE ('.') |
1311 |
|
|
&& zero_times_ok |
1312 |
|
|
&& p < pend && TRANSLATE (*p) == TRANSLATE ('\n') |
1313 |
|
|
&& !(syntax & RE_DOT_NEWLINE)) |
1314 |
|
|
{ /* We have .*\n. */ |
1315 |
|
|
STORE_JUMP (jump, b, laststart); |
1316 |
|
|
keep_string_p = true; |
1317 |
|
|
} |
1318 |
|
|
else |
1319 |
|
|
/* Anything else. */ |
1320 |
|
|
STORE_JUMP (maybe_pop_jump, b, laststart - 3); |
1321 |
|
|
|
1322 |
|
|
/* We've added more stuff to the buffer. */ |
1323 |
|
|
b += 3; |
1324 |
|
|
} |
1325 |
|
|
|
1326 |
|
|
/* On failure, jump from laststart to b + 3, which will be the |
1327 |
|
|
end of the buffer after this jump is inserted. */ |
1328 |
|
|
GET_BUFFER_SPACE (3); |
1329 |
|
|
INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump |
1330 |
|
|
: on_failure_jump, |
1331 |
|
|
laststart, b + 3); |
1332 |
|
|
pending_exact = 0; |
1333 |
|
|
b += 3; |
1334 |
|
|
|
1335 |
|
|
if (!zero_times_ok) |
1336 |
|
|
{ |
1337 |
|
|
/* At least one repetition is required, so insert a |
1338 |
|
|
`dummy_failure_jump' before the initial |
1339 |
|
|
`on_failure_jump' instruction of the loop. This |
1340 |
|
|
effects a skip over that instruction the first time |
1341 |
|
|
we hit that loop. */ |
1342 |
|
|
GET_BUFFER_SPACE (3); |
1343 |
|
|
INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6); |
1344 |
|
|
b += 3; |
1345 |
|
|
} |
1346 |
|
|
} |
1347 |
|
|
break; |
1348 |
|
|
|
1349 |
|
|
|
1350 |
|
|
case '.': |
1351 |
|
|
laststart = b; |
1352 |
|
|
BUF_PUSH (anychar); |
1353 |
|
|
break; |
1354 |
|
|
|
1355 |
|
|
|
1356 |
|
|
case '[': |
1357 |
|
|
{ |
1358 |
|
|
boolean had_char_class = false; |
1359 |
|
|
|
1360 |
|
|
if (p == pend) return REG_EBRACK; |
1361 |
|
|
|
1362 |
|
|
/* Ensure that we have enough space to push a charset: the |
1363 |
|
|
opcode, the length count, and the bitset; 34 bytes in all. */ |
1364 |
|
|
GET_BUFFER_SPACE (34); |
1365 |
|
|
|
1366 |
|
|
laststart = b; |
1367 |
|
|
|
1368 |
|
|
/* We test `*p == '^' twice, instead of using an if |
1369 |
|
|
statement, so we only need one BUF_PUSH. */ |
1370 |
|
|
BUF_PUSH (*p == '^' ? charset_not : charset); |
1371 |
|
|
if (*p == '^') |
1372 |
|
|
p++; |
1373 |
|
|
|
1374 |
|
|
/* Remember the first position in the bracket expression. */ |
1375 |
|
|
p1 = p; |
1376 |
|
|
|
1377 |
|
|
/* Push the number of bytes in the bitmap. */ |
1378 |
|
|
BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH); |
1379 |
|
|
|
1380 |
|
|
/* Clear the whole map. */ |
1381 |
|
|
bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH); |
1382 |
|
|
|
1383 |
|
|
/* charset_not matches newline according to a syntax bit. */ |
1384 |
|
|
if ((re_opcode_t) b[-2] == charset_not |
1385 |
|
|
&& (syntax & RE_HAT_LISTS_NOT_NEWLINE)) |
1386 |
|
|
SET_LIST_BIT ('\n'); |
1387 |
|
|
|
1388 |
|
|
/* Read in characters and ranges, setting map bits. */ |
1389 |
|
|
for (;;) |
1390 |
|
|
{ |
1391 |
|
|
if (p == pend) return REG_EBRACK; |
1392 |
|
|
|
1393 |
|
|
PATFETCH (c); |
1394 |
|
|
|
1395 |
|
|
/* \ might escape characters inside [...] and [^...]. */ |
1396 |
|
|
if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\') |
1397 |
|
|
{ |
1398 |
|
|
if (p == pend) return REG_EESCAPE; |
1399 |
|
|
|
1400 |
|
|
PATFETCH (c1); |
1401 |
|
|
SET_LIST_BIT (c1); |
1402 |
|
|
continue; |
1403 |
|
|
} |
1404 |
|
|
|
1405 |
|
|
/* Could be the end of the bracket expression. If it's |
1406 |
|
|
not (i.e., when the bracket expression is `[]' so |
1407 |
|
|
far), the ']' character bit gets set way below. */ |
1408 |
|
|
if (c == ']' && p != p1 + 1) |
1409 |
|
|
break; |
1410 |
|
|
|
1411 |
|
|
/* Look ahead to see if it's a range when the last thing |
1412 |
|
|
was a character class. */ |
1413 |
|
|
if (had_char_class && c == '-' && *p != ']') |
1414 |
|
|
return REG_ERANGE; |
1415 |
|
|
|
1416 |
|
|
/* Look ahead to see if it's a range when the last thing |
1417 |
|
|
was a character: if this is a hyphen not at the |
1418 |
|
|
beginning or the end of a list, then it's the range |
1419 |
|
|
operator. */ |
1420 |
|
|
if (c == '-' |
1421 |
|
|
&& !(p - 2 >= pattern && p[-2] == '[') |
1422 |
|
|
&& !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^') |
1423 |
|
|
&& *p != ']') |
1424 |
|
|
{ |
1425 |
|
|
reg_errcode_t ret |
1426 |
|
|
= compile_range (&p, pend, translate, syntax, b); |
1427 |
|
|
if (ret != REG_NOERROR) return ret; |
1428 |
|
|
} |
1429 |
|
|
|
1430 |
|
|
else if (p[0] == '-' && p[1] != ']') |
1431 |
|
|
{ /* This handles ranges made up of characters only. */ |
1432 |
|
|
reg_errcode_t ret; |
1433 |
|
|
|
1434 |
|
|
/* Move past the `-'. */ |
1435 |
|
|
PATFETCH (c1); |
1436 |
|
|
|
1437 |
|
|
ret = compile_range (&p, pend, translate, syntax, b); |
1438 |
|
|
if (ret != REG_NOERROR) return ret; |
1439 |
|
|
} |
1440 |
|
|
|
1441 |
|
|
/* See if we're at the beginning of a possible character |
1442 |
|
|
class. */ |
1443 |
|
|
|
1444 |
|
|
else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':') |
1445 |
|
|
{ /* Leave room for the null. */ |
1446 |
|
|
char str[CHAR_CLASS_MAX_LENGTH + 1]; |
1447 |
|
|
|
1448 |
|
|
PATFETCH (c); |
1449 |
|
|
c1 = 0; |
1450 |
|
|
|
1451 |
|
|
/* If pattern is `[[:'. */ |
1452 |
|
|
if (p == pend) return REG_EBRACK; |
1453 |
|
|
|
1454 |
|
|
for (;;) |
1455 |
|
|
{ |
1456 |
|
|
PATFETCH (c); |
1457 |
|
|
if (c == ':' || c == ']' || p == pend |
1458 |
|
|
|| c1 == CHAR_CLASS_MAX_LENGTH) |
1459 |
|
|
break; |
1460 |
|
|
str[c1++] = c; |
1461 |
|
|
} |
1462 |
|
|
str[c1] = '\0'; |
1463 |
|
|
|
1464 |
|
|
/* If isn't a word bracketed by `[:' and:`]': |
1465 |
|
|
undo the ending character, the letters, and leave |
1466 |
|
|
the leading `:' and `[' (but set bits for them). */ |
1467 |
|
|
if (c == ':' && *p == ']') |
1468 |
|
|
{ |
1469 |
|
|
int ch; |
1470 |
|
|
boolean is_alnum = STREQ (str, "alnum"); |
1471 |
|
|
boolean is_alpha = STREQ (str, "alpha"); |
1472 |
|
|
boolean is_blank = STREQ (str, "blank"); |
1473 |
|
|
boolean is_cntrl = STREQ (str, "cntrl"); |
1474 |
|
|
boolean is_digit = STREQ (str, "digit"); |
1475 |
|
|
boolean is_graph = STREQ (str, "graph"); |
1476 |
|
|
boolean is_lower = STREQ (str, "lower"); |
1477 |
|
|
boolean is_print = STREQ (str, "print"); |
1478 |
|
|
boolean is_punct = STREQ (str, "punct"); |
1479 |
|
|
boolean is_space = STREQ (str, "space"); |
1480 |
|
|
boolean is_upper = STREQ (str, "upper"); |
1481 |
|
|
boolean is_xdigit = STREQ (str, "xdigit"); |
1482 |
|
|
|
1483 |
|
|
if (!IS_CHAR_CLASS (str)) return REG_ECTYPE; |
1484 |
|
|
|
1485 |
|
|
/* Throw away the ] at the end of the character |
1486 |
|
|
class. */ |
1487 |
|
|
PATFETCH (c); |
1488 |
|
|
|
1489 |
|
|
if (p == pend) return REG_EBRACK; |
1490 |
|
|
|
1491 |
|
|
for (ch = 0; ch < 1 << BYTEWIDTH; ch++) |
1492 |
|
|
{ |
1493 |
|
|
if ( (is_alnum && ISALNUM (ch)) |
1494 |
|
|
|| (is_alpha && ISALPHA (ch)) |
1495 |
|
|
|| (is_blank && ISBLANK (ch)) |
1496 |
|
|
|| (is_cntrl && ISCNTRL (ch)) |
1497 |
|
|
|| (is_digit && ISDIGIT (ch)) |
1498 |
|
|
|| (is_graph && ISGRAPH (ch)) |
1499 |
|
|
|| (is_lower && ISLOWER (ch)) |
1500 |
|
|
|| (is_print && ISPRINT (ch)) |
1501 |
|
|
|| (is_punct && ISPUNCT (ch)) |
1502 |
|
|
|| (is_space && ISSPACE (ch)) |
1503 |
|
|
|| (is_upper && ISUPPER (ch)) |
1504 |
|
|
|| (is_xdigit && ISXDIGIT (ch))) |
1505 |
|
|
SET_LIST_BIT (ch); |
1506 |
|
|
} |
1507 |
|
|
had_char_class = true; |
1508 |
|
|
} |
1509 |
|
|
else |
1510 |
|
|
{ |
1511 |
|
|
c1++; |
1512 |
|
|
while (c1--) |
1513 |
|
|
PATUNFETCH; |
1514 |
|
|
SET_LIST_BIT ('['); |
1515 |
|
|
SET_LIST_BIT (':'); |
1516 |
|
|
had_char_class = false; |
1517 |
|
|
} |
1518 |
|
|
} |
1519 |
|
|
else |
1520 |
|
|
{ |
1521 |
|
|
had_char_class = false; |
1522 |
|
|
SET_LIST_BIT (c); |
1523 |
|
|
} |
1524 |
|
|
} |
1525 |
|
|
|
1526 |
|
|
/* Discard any (non)matching list bytes that are all 0 at the |
1527 |
|
|
end of the map. Decrease the map-length byte too. */ |
1528 |
|
|
while ((int) b[-1] > 0 && b[b[-1] - 1] == 0) |
1529 |
|
|
b[-1]--; |
1530 |
|
|
b += b[-1]; |
1531 |
|
|
} |
1532 |
|
|
break; |
1533 |
|
|
|
1534 |
|
|
|
1535 |
|
|
case '(': |
1536 |
|
|
if (syntax & RE_NO_BK_PARENS) |
1537 |
|
|
goto handle_open; |
1538 |
|
|
else |
1539 |
|
|
goto normal_char; |
1540 |
|
|
|
1541 |
|
|
|
1542 |
|
|
case ')': |
1543 |
|
|
if (syntax & RE_NO_BK_PARENS) |
1544 |
|
|
goto handle_close; |
1545 |
|
|
else |
1546 |
|
|
goto normal_char; |
1547 |
|
|
|
1548 |
|
|
|
1549 |
|
|
case '\n': |
1550 |
|
|
if (syntax & RE_NEWLINE_ALT) |
1551 |
|
|
goto handle_alt; |
1552 |
|
|
else |
1553 |
|
|
goto normal_char; |
1554 |
|
|
|
1555 |
|
|
|
1556 |
|
|
case '|': |
1557 |
|
|
if (syntax & RE_NO_BK_VBAR) |
1558 |
|
|
goto handle_alt; |
1559 |
|
|
else |
1560 |
|
|
goto normal_char; |
1561 |
|
|
|
1562 |
|
|
|
1563 |
|
|
case '{': |
1564 |
|
|
if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES) |
1565 |
|
|
goto handle_interval; |
1566 |
|
|
else |
1567 |
|
|
goto normal_char; |
1568 |
|
|
|
1569 |
|
|
|
1570 |
|
|
case '\\': |
1571 |
|
|
if (p == pend) return REG_EESCAPE; |
1572 |
|
|
|
1573 |
|
|
/* Do not translate the character after the \, so that we can |
1574 |
|
|
distinguish, e.g., \B from \b, even if we normally would |
1575 |
|
|
translate, e.g., B to b. */ |
1576 |
|
|
PATFETCH_RAW (c); |
1577 |
|
|
|
1578 |
|
|
switch (c) |
1579 |
|
|
{ |
1580 |
|
|
case '(': |
1581 |
|
|
if (syntax & RE_NO_BK_PARENS) |
1582 |
|
|
goto normal_backslash; |
1583 |
|
|
|
1584 |
|
|
handle_open: |
1585 |
|
|
bufp->re_nsub++; |
1586 |
|
|
regnum++; |
1587 |
|
|
|
1588 |
|
|
if (COMPILE_STACK_FULL) |
1589 |
|
|
{ |
1590 |
|
|
RETALLOC (compile_stack.stack, compile_stack.size << 1, |
1591 |
|
|
compile_stack_elt_t); |
1592 |
|
|
if (compile_stack.stack == NULL) return REG_ESPACE; |
1593 |
|
|
|
1594 |
|
|
compile_stack.size <<= 1; |
1595 |
|
|
} |
1596 |
|
|
|
1597 |
|
|
/* These are the values to restore when we hit end of this |
1598 |
|
|
group. They are all relative offsets, so that if the |
1599 |
|
|
whole pattern moves because of realloc, they will still |
1600 |
|
|
be valid. */ |
1601 |
|
|
COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer; |
1602 |
|
|
COMPILE_STACK_TOP.fixup_alt_jump |
1603 |
|
|
= fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0; |
1604 |
|
|
COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer; |
1605 |
|
|
COMPILE_STACK_TOP.regnum = regnum; |
1606 |
|
|
|
1607 |
|
|
/* We will eventually replace the 0 with the number of |
1608 |
|
|
groups inner to this one. But do not push a |
1609 |
|
|
start_memory for groups beyond the last one we can |
1610 |
|
|
represent in the compiled pattern. */ |
1611 |
|
|
if (regnum <= MAX_REGNUM) |
1612 |
|
|
{ |
1613 |
|
|
COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2; |
1614 |
|
|
BUF_PUSH_3 (start_memory, regnum, 0); |
1615 |
|
|
} |
1616 |
|
|
|
1617 |
|
|
compile_stack.avail++; |
1618 |
|
|
|
1619 |
|
|
fixup_alt_jump = 0; |
1620 |
|
|
laststart = 0; |
1621 |
|
|
begalt = b; |
1622 |
|
|
/* If we've reached MAX_REGNUM groups, then this open |
1623 |
|
|
won't actually generate any code, so we'll have to |
1624 |
|
|
clear pending_exact explicitly. */ |
1625 |
|
|
pending_exact = 0; |
1626 |
|
|
break; |
1627 |
|
|
|
1628 |
|
|
|
1629 |
|
|
case ')': |
1630 |
|
|
if (syntax & RE_NO_BK_PARENS) goto normal_backslash; |
1631 |
|
|
|
1632 |
|
|
if (COMPILE_STACK_EMPTY) |
1633 |
|
|
if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) |
1634 |
|
|
goto normal_backslash; |
1635 |
|
|
else |
1636 |
|
|
return REG_ERPAREN; |
1637 |
|
|
|
1638 |
|
|
handle_close: |
1639 |
|
|
if (fixup_alt_jump) |
1640 |
|
|
{ /* Push a dummy failure point at the end of the |
1641 |
|
|
alternative for a possible future |
1642 |
|
|
`pop_failure_jump' to pop. See comments at |
1643 |
|
|
`push_dummy_failure' in `re_match_2'. */ |
1644 |
|
|
BUF_PUSH (push_dummy_failure); |
1645 |
|
|
|
1646 |
|
|
/* We allocated space for this jump when we assigned |
1647 |
|
|
to `fixup_alt_jump', in the `handle_alt' case below. */ |
1648 |
|
|
STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1); |
1649 |
|
|
} |
1650 |
|
|
|
1651 |
|
|
/* See similar code for backslashed left paren above. */ |
1652 |
|
|
if (COMPILE_STACK_EMPTY) |
1653 |
|
|
if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD) |
1654 |
|
|
goto normal_char; |
1655 |
|
|
else |
1656 |
|
|
return REG_ERPAREN; |
1657 |
|
|
|
1658 |
|
|
/* Since we just checked for an empty stack above, this |
1659 |
|
|
``can't happen''. */ |
1660 |
|
|
assert (compile_stack.avail != 0); |
1661 |
|
|
{ |
1662 |
|
|
/* We don't just want to restore into `regnum', because |
1663 |
|
|
later groups should continue to be numbered higher, |
1664 |
|
|
as in `(ab)c(de)' -- the second group is #2. */ |
1665 |
|
|
regnum_t this_group_regnum; |
1666 |
|
|
|
1667 |
|
|
compile_stack.avail--; |
1668 |
|
|
begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset; |
1669 |
|
|
fixup_alt_jump |
1670 |
|
|
= COMPILE_STACK_TOP.fixup_alt_jump |
1671 |
|
|
? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1 |
1672 |
|
|
: 0; |
1673 |
|
|
laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset; |
1674 |
|
|
this_group_regnum = COMPILE_STACK_TOP.regnum; |
1675 |
|
|
/* If we've reached MAX_REGNUM groups, then this open |
1676 |
|
|
won't actually generate any code, so we'll have to |
1677 |
|
|
clear pending_exact explicitly. */ |
1678 |
|
|
pending_exact = 0; |
1679 |
|
|
|
1680 |
|
|
/* We're at the end of the group, so now we know how many |
1681 |
|
|
groups were inside this one. */ |
1682 |
|
|
if (this_group_regnum <= MAX_REGNUM) |
1683 |
|
|
{ |
1684 |
|
|
unsigned char *inner_group_loc |
1685 |
|
|
= bufp->buffer + COMPILE_STACK_TOP.inner_group_offset; |
1686 |
|
|
|
1687 |
|
|
*inner_group_loc = regnum - this_group_regnum; |
1688 |
|
|
BUF_PUSH_3 (stop_memory, this_group_regnum, |
1689 |
|
|
regnum - this_group_regnum); |
1690 |
|
|
} |
1691 |
|
|
} |
1692 |
|
|
break; |
1693 |
|
|
|
1694 |
|
|
|
1695 |
|
|
case '|': /* `\|'. */ |
1696 |
|
|
if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR) |
1697 |
|
|
goto normal_backslash; |
1698 |
|
|
handle_alt: |
1699 |
|
|
if (syntax & RE_LIMITED_OPS) |
1700 |
|
|
goto normal_char; |
1701 |
|
|
|
1702 |
|
|
/* Insert before the previous alternative a jump which |
1703 |
|
|
jumps to this alternative if the former fails. */ |
1704 |
|
|
GET_BUFFER_SPACE (3); |
1705 |
|
|
INSERT_JUMP (on_failure_jump, begalt, b + 6); |
1706 |
|
|
pending_exact = 0; |
1707 |
|
|
b += 3; |
1708 |
|
|
|
1709 |
|
|
/* The alternative before this one has a jump after it |
1710 |
|
|
which gets executed if it gets matched. Adjust that |
1711 |
|
|
jump so it will jump to this alternative's analogous |
1712 |
|
|
jump (put in below, which in turn will jump to the next |
1713 |
|
|
(if any) alternative's such jump, etc.). The last such |
1714 |
|
|
jump jumps to the correct final destination. A picture: |
1715 |
|
|
_____ _____ |
1716 |
|
|
| | | | |
1717 |
|
|
| v | v |
1718 |
|
|
a | b | c |
1719 |
|
|
|
1720 |
|
|
If we are at `b', then fixup_alt_jump right now points to a |
1721 |
|
|
three-byte space after `a'. We'll put in the jump, set |
1722 |
|
|
fixup_alt_jump to right after `b', and leave behind three |
1723 |
|
|
bytes which we'll fill in when we get to after `c'. */ |
1724 |
|
|
|
1725 |
|
|
if (fixup_alt_jump) |
1726 |
|
|
STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
1727 |
|
|
|
1728 |
|
|
/* Mark and leave space for a jump after this alternative, |
1729 |
|
|
to be filled in later either by next alternative or |
1730 |
|
|
when know we're at the end of a series of alternatives. */ |
1731 |
|
|
fixup_alt_jump = b; |
1732 |
|
|
GET_BUFFER_SPACE (3); |
1733 |
|
|
b += 3; |
1734 |
|
|
|
1735 |
|
|
laststart = 0; |
1736 |
|
|
begalt = b; |
1737 |
|
|
break; |
1738 |
|
|
|
1739 |
|
|
|
1740 |
|
|
case '{': |
1741 |
|
|
/* If \{ is a literal. */ |
1742 |
|
|
if (!(syntax & RE_INTERVALS) |
1743 |
|
|
/* If we're at `\{' and it's not the open-interval |
1744 |
|
|
operator. */ |
1745 |
|
|
|| ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES)) |
1746 |
|
|
|| (p - 2 == pattern && p == pend)) |
1747 |
|
|
goto normal_backslash; |
1748 |
|
|
|
1749 |
|
|
handle_interval: |
1750 |
|
|
{ |
1751 |
|
|
/* If got here, then the syntax allows intervals. */ |
1752 |
|
|
|
1753 |
|
|
/* At least (most) this many matches must be made. */ |
1754 |
|
|
int lower_bound = -1, upper_bound = -1; |
1755 |
|
|
|
1756 |
|
|
beg_interval = p - 1; |
1757 |
|
|
|
1758 |
|
|
if (p == pend) |
1759 |
|
|
{ |
1760 |
|
|
if (syntax & RE_NO_BK_BRACES) |
1761 |
|
|
goto unfetch_interval; |
1762 |
|
|
else |
1763 |
|
|
return REG_EBRACE; |
1764 |
|
|
} |
1765 |
|
|
|
1766 |
|
|
GET_UNSIGNED_NUMBER (lower_bound); |
1767 |
|
|
|
1768 |
|
|
if (c == ',') |
1769 |
|
|
{ |
1770 |
|
|
GET_UNSIGNED_NUMBER (upper_bound); |
1771 |
|
|
if (upper_bound < 0) upper_bound = RE_DUP_MAX; |
1772 |
|
|
} |
1773 |
|
|
else |
1774 |
|
|
/* Interval such as `{1}' => match exactly once. */ |
1775 |
|
|
upper_bound = lower_bound; |
1776 |
|
|
|
1777 |
|
|
if (lower_bound < 0 || upper_bound > RE_DUP_MAX |
1778 |
|
|
|| lower_bound > upper_bound) |
1779 |
|
|
{ |
1780 |
|
|
if (syntax & RE_NO_BK_BRACES) |
1781 |
|
|
goto unfetch_interval; |
1782 |
|
|
else |
1783 |
|
|
return REG_BADBR; |
1784 |
|
|
} |
1785 |
|
|
|
1786 |
|
|
if (!(syntax & RE_NO_BK_BRACES)) |
1787 |
|
|
{ |
1788 |
|
|
if (c != '\\') return REG_EBRACE; |
1789 |
|
|
|
1790 |
|
|
PATFETCH (c); |
1791 |
|
|
} |
1792 |
|
|
|
1793 |
|
|
if (c != '}') |
1794 |
|
|
{ |
1795 |
|
|
if (syntax & RE_NO_BK_BRACES) |
1796 |
|
|
goto unfetch_interval; |
1797 |
|
|
else |
1798 |
|
|
return REG_BADBR; |
1799 |
|
|
} |
1800 |
|
|
|
1801 |
|
|
/* We just parsed a valid interval. */ |
1802 |
|
|
|
1803 |
|
|
/* If it's invalid to have no preceding re. */ |
1804 |
|
|
if (!laststart) |
1805 |
|
|
{ |
1806 |
|
|
if (syntax & RE_CONTEXT_INVALID_OPS) |
1807 |
|
|
return REG_BADRPT; |
1808 |
|
|
else if (syntax & RE_CONTEXT_INDEP_OPS) |
1809 |
|
|
laststart = b; |
1810 |
|
|
else |
1811 |
|
|
goto unfetch_interval; |
1812 |
|
|
} |
1813 |
|
|
|
1814 |
|
|
/* If the upper bound is zero, don't want to succeed at |
1815 |
|
|
all; jump from `laststart' to `b + 3', which will be |
1816 |
|
|
the end of the buffer after we insert the jump. */ |
1817 |
|
|
if (upper_bound == 0) |
1818 |
|
|
{ |
1819 |
|
|
GET_BUFFER_SPACE (3); |
1820 |
|
|
INSERT_JUMP (jump, laststart, b + 3); |
1821 |
|
|
b += 3; |
1822 |
|
|
} |
1823 |
|
|
|
1824 |
|
|
/* Otherwise, we have a nontrivial interval. When |
1825 |
|
|
we're all done, the pattern will look like: |
1826 |
|
|
set_number_at <jump count> <upper bound> |
1827 |
|
|
set_number_at <succeed_n count> <lower bound> |
1828 |
|
|
succeed_n <after jump addr> <succed_n count> |
1829 |
|
|
<body of loop> |
1830 |
|
|
jump_n <succeed_n addr> <jump count> |
1831 |
|
|
(The upper bound and `jump_n' are omitted if |
1832 |
|
|
`upper_bound' is 1, though.) */ |
1833 |
|
|
else |
1834 |
|
|
{ /* If the upper bound is > 1, we need to insert |
1835 |
|
|
more at the end of the loop. */ |
1836 |
|
|
unsigned nbytes = 10 + (upper_bound > 1) * 10; |
1837 |
|
|
|
1838 |
|
|
GET_BUFFER_SPACE (nbytes); |
1839 |
|
|
|
1840 |
|
|
/* Initialize lower bound of the `succeed_n', even |
1841 |
|
|
though it will be set during matching by its |
1842 |
|
|
attendant `set_number_at' (inserted next), |
1843 |
|
|
because `re_compile_fastmap' needs to know. |
1844 |
|
|
Jump to the `jump_n' we might insert below. */ |
1845 |
|
|
INSERT_JUMP2 (succeed_n, laststart, |
1846 |
|
|
b + 5 + (upper_bound > 1) * 5, |
1847 |
|
|
lower_bound); |
1848 |
|
|
b += 5; |
1849 |
|
|
|
1850 |
|
|
/* Code to initialize the lower bound. Insert |
1851 |
|
|
before the `succeed_n'. The `5' is the last two |
1852 |
|
|
bytes of this `set_number_at', plus 3 bytes of |
1853 |
|
|
the following `succeed_n'. */ |
1854 |
|
|
insert_op2 (set_number_at, laststart, 5, lower_bound, b); |
1855 |
|
|
b += 5; |
1856 |
|
|
|
1857 |
|
|
if (upper_bound > 1) |
1858 |
|
|
{ /* More than one repetition is allowed, so |
1859 |
|
|
append a backward jump to the `succeed_n' |
1860 |
|
|
that starts this interval. |
1861 |
|
|
|
1862 |
|
|
When we've reached this during matching, |
1863 |
|
|
we'll have matched the interval once, so |
1864 |
|
|
jump back only `upper_bound - 1' times. */ |
1865 |
|
|
STORE_JUMP2 (jump_n, b, laststart + 5, |
1866 |
|
|
upper_bound - 1); |
1867 |
|
|
b += 5; |
1868 |
|
|
|
1869 |
|
|
/* The location we want to set is the second |
1870 |
|
|
parameter of the `jump_n'; that is `b-2' as |
1871 |
|
|
an absolute address. `laststart' will be |
1872 |
|
|
the `set_number_at' we're about to insert; |
1873 |
|
|
`laststart+3' the number to set, the source |
1874 |
|
|
for the relative address. But we are |
1875 |
|
|
inserting into the middle of the pattern -- |
1876 |
|
|
so everything is getting moved up by 5. |
1877 |
|
|
Conclusion: (b - 2) - (laststart + 3) + 5, |
1878 |
|
|
i.e., b - laststart. |
1879 |
|
|
|
1880 |
|
|
We insert this at the beginning of the loop |
1881 |
|
|
so that if we fail during matching, we'll |
1882 |
|
|
reinitialize the bounds. */ |
1883 |
|
|
insert_op2 (set_number_at, laststart, b - laststart, |
1884 |
|
|
upper_bound - 1, b); |
1885 |
|
|
b += 5; |
1886 |
|
|
} |
1887 |
|
|
} |
1888 |
|
|
pending_exact = 0; |
1889 |
|
|
beg_interval = NULL; |
1890 |
|
|
} |
1891 |
|
|
break; |
1892 |
|
|
|
1893 |
|
|
unfetch_interval: |
1894 |
|
|
/* If an invalid interval, match the characters as literals. */ |
1895 |
|
|
assert (beg_interval); |
1896 |
|
|
p = beg_interval; |
1897 |
|
|
beg_interval = NULL; |
1898 |
|
|
|
1899 |
|
|
/* normal_char and normal_backslash need `c'. */ |
1900 |
|
|
PATFETCH (c); |
1901 |
|
|
|
1902 |
|
|
if (!(syntax & RE_NO_BK_BRACES)) |
1903 |
|
|
{ |
1904 |
|
|
if (p > pattern && p[-1] == '\\') |
1905 |
|
|
goto normal_backslash; |
1906 |
|
|
} |
1907 |
|
|
goto normal_char; |
1908 |
|
|
|
1909 |
|
|
#ifdef emacs |
1910 |
|
|
/* There is no way to specify the before_dot and after_dot |
1911 |
|
|
operators. rms says this is ok. --karl */ |
1912 |
|
|
case '=': |
1913 |
|
|
BUF_PUSH (at_dot); |
1914 |
|
|
break; |
1915 |
|
|
|
1916 |
|
|
case 's': |
1917 |
|
|
laststart = b; |
1918 |
|
|
PATFETCH (c); |
1919 |
|
|
BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]); |
1920 |
|
|
break; |
1921 |
|
|
|
1922 |
|
|
case 'S': |
1923 |
|
|
laststart = b; |
1924 |
|
|
PATFETCH (c); |
1925 |
|
|
BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]); |
1926 |
|
|
break; |
1927 |
|
|
#endif /* emacs */ |
1928 |
|
|
|
1929 |
|
|
|
1930 |
|
|
case 'w': |
1931 |
|
|
laststart = b; |
1932 |
|
|
BUF_PUSH (wordchar); |
1933 |
|
|
break; |
1934 |
|
|
|
1935 |
|
|
|
1936 |
|
|
case 'W': |
1937 |
|
|
laststart = b; |
1938 |
|
|
BUF_PUSH (notwordchar); |
1939 |
|
|
break; |
1940 |
|
|
|
1941 |
|
|
|
1942 |
|
|
case '<': |
1943 |
|
|
BUF_PUSH (wordbeg); |
1944 |
|
|
break; |
1945 |
|
|
|
1946 |
|
|
case '>': |
1947 |
|
|
BUF_PUSH (wordend); |
1948 |
|
|
break; |
1949 |
|
|
|
1950 |
|
|
case 'b': |
1951 |
|
|
BUF_PUSH (wordbound); |
1952 |
|
|
break; |
1953 |
|
|
|
1954 |
|
|
case 'B': |
1955 |
|
|
BUF_PUSH (notwordbound); |
1956 |
|
|
break; |
1957 |
|
|
|
1958 |
|
|
case '`': |
1959 |
|
|
BUF_PUSH (begbuf); |
1960 |
|
|
break; |
1961 |
|
|
|
1962 |
|
|
case '\'': |
1963 |
|
|
BUF_PUSH (endbuf); |
1964 |
|
|
break; |
1965 |
|
|
|
1966 |
|
|
case '1': case '2': case '3': case '4': case '5': |
1967 |
|
|
case '6': case '7': case '8': case '9': |
1968 |
|
|
if (syntax & RE_NO_BK_REFS) |
1969 |
|
|
goto normal_char; |
1970 |
|
|
|
1971 |
|
|
c1 = c - '0'; |
1972 |
|
|
|
1973 |
|
|
if (c1 > regnum) |
1974 |
|
|
return REG_ESUBREG; |
1975 |
|
|
|
1976 |
|
|
/* Can't back reference to a subexpression if inside of it. */ |
1977 |
|
|
if (group_in_compile_stack (compile_stack, c1)) |
1978 |
|
|
goto normal_char; |
1979 |
|
|
|
1980 |
|
|
laststart = b; |
1981 |
|
|
BUF_PUSH_2 (duplicate, c1); |
1982 |
|
|
break; |
1983 |
|
|
|
1984 |
|
|
|
1985 |
|
|
case '+': |
1986 |
|
|
case '?': |
1987 |
|
|
if (syntax & RE_BK_PLUS_QM) |
1988 |
|
|
goto handle_plus; |
1989 |
|
|
else |
1990 |
|
|
goto normal_backslash; |
1991 |
|
|
|
1992 |
|
|
default: |
1993 |
|
|
normal_backslash: |
1994 |
|
|
/* You might think it would be useful for \ to mean |
1995 |
|
|
not to translate; but if we don't translate it |
1996 |
|
|
it will never match anything. */ |
1997 |
|
|
c = TRANSLATE (c); |
1998 |
|
|
goto normal_char; |
1999 |
|
|
} |
2000 |
|
|
break; |
2001 |
|
|
|
2002 |
|
|
|
2003 |
|
|
default: |
2004 |
|
|
/* Expects the character in `c'. */ |
2005 |
|
|
normal_char: |
2006 |
|
|
/* If no exactn currently being built. */ |
2007 |
|
|
if (!pending_exact |
2008 |
|
|
|
2009 |
|
|
/* If last exactn not at current position. */ |
2010 |
|
|
|| pending_exact + *pending_exact + 1 != b |
2011 |
|
|
|
2012 |
|
|
/* We have only one byte following the exactn for the count. */ |
2013 |
|
|
|| *pending_exact == (1 << BYTEWIDTH) - 1 |
2014 |
|
|
|
2015 |
|
|
/* If followed by a repetition operator. */ |
2016 |
|
|
|| *p == '*' || *p == '^' |
2017 |
|
|
|| ((syntax & RE_BK_PLUS_QM) |
2018 |
|
|
? *p == '\\' && (p[1] == '+' || p[1] == '?') |
2019 |
|
|
: (*p == '+' || *p == '?')) |
2020 |
|
|
|| ((syntax & RE_INTERVALS) |
2021 |
|
|
&& ((syntax & RE_NO_BK_BRACES) |
2022 |
|
|
? *p == '{' |
2023 |
|
|
: (p[0] == '\\' && p[1] == '{')))) |
2024 |
|
|
{ |
2025 |
|
|
/* Start building a new exactn. */ |
2026 |
|
|
|
2027 |
|
|
laststart = b; |
2028 |
|
|
|
2029 |
|
|
BUF_PUSH_2 (exactn, 0); |
2030 |
|
|
pending_exact = b - 1; |
2031 |
|
|
} |
2032 |
|
|
|
2033 |
|
|
BUF_PUSH (c); |
2034 |
|
|
(*pending_exact)++; |
2035 |
|
|
break; |
2036 |
|
|
} /* switch (c) */ |
2037 |
|
|
} /* while p != pend */ |
2038 |
|
|
|
2039 |
|
|
|
2040 |
|
|
/* Through the pattern now. */ |
2041 |
|
|
|
2042 |
|
|
if (fixup_alt_jump) |
2043 |
|
|
STORE_JUMP (jump_past_alt, fixup_alt_jump, b); |
2044 |
|
|
|
2045 |
|
|
if (!COMPILE_STACK_EMPTY) |
2046 |
|
|
return REG_EPAREN; |
2047 |
|
|
|
2048 |
|
|
free (compile_stack.stack); |
2049 |
|
|
|
2050 |
|
|
/* We have succeeded; set the length of the buffer. */ |
2051 |
|
|
bufp->used = b - bufp->buffer; |
2052 |
|
|
|
2053 |
|
|
#ifdef DEBUG |
2054 |
|
|
if (debug) |
2055 |
|
|
{ |
2056 |
|
|
DEBUG_PRINT1 ("\nCompiled pattern: "); |
2057 |
|
|
print_compiled_pattern (bufp); |
2058 |
|
|
} |
2059 |
|
|
#endif /* DEBUG */ |
2060 |
|
|
|
2061 |
|
|
return REG_NOERROR; |
2062 |
|
|
} /* regex_compile */ |
2063 |
|
|
|
2064 |
|
|
/* Subroutines for `regex_compile'. */ |
2065 |
|
|
|
2066 |
|
|
/* Store OP at LOC followed by two-byte integer parameter ARG. */ |
2067 |
|
|
|
2068 |
|
|
static void |
2069 |
|
|
store_op1 (op, loc, arg) |
2070 |
|
|
re_opcode_t op; |
2071 |
|
|
unsigned char *loc; |
2072 |
|
|
int arg; |
2073 |
|
|
{ |
2074 |
|
|
*loc = (unsigned char) op; |
2075 |
|
|
STORE_NUMBER (loc + 1, arg); |
2076 |
|
|
} |
2077 |
|
|
|
2078 |
|
|
|
2079 |
|
|
/* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */ |
2080 |
|
|
|
2081 |
|
|
static void |
2082 |
|
|
store_op2 (op, loc, arg1, arg2) |
2083 |
|
|
re_opcode_t op; |
2084 |
|
|
unsigned char *loc; |
2085 |
|
|
int arg1, arg2; |
2086 |
|
|
{ |
2087 |
|
|
*loc = (unsigned char) op; |
2088 |
|
|
STORE_NUMBER (loc + 1, arg1); |
2089 |
|
|
STORE_NUMBER (loc + 3, arg2); |
2090 |
|
|
} |
2091 |
|
|
|
2092 |
|
|
|
2093 |
|
|
/* Copy the bytes from LOC to END to open up three bytes of space at LOC |
2094 |
|
|
for OP followed by two-byte integer parameter ARG. */ |
2095 |
|
|
|
2096 |
|
|
static void |
2097 |
|
|
insert_op1 (op, loc, arg, end) |
2098 |
|
|
re_opcode_t op; |
2099 |
|
|
unsigned char *loc; |
2100 |
|
|
int arg; |
2101 |
|
|
unsigned char *end; |
2102 |
|
|
{ |
2103 |
|
|
register unsigned char *pfrom = end; |
2104 |
|
|
register unsigned char *pto = end + 3; |
2105 |
|
|
|
2106 |
|
|
while (pfrom != loc) |
2107 |
|
|
*--pto = *--pfrom; |
2108 |
|
|
|
2109 |
|
|
store_op1 (op, loc, arg); |
2110 |
|
|
} |
2111 |
|
|
|
2112 |
|
|
|
2113 |
|
|
/* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */ |
2114 |
|
|
|
2115 |
|
|
static void |
2116 |
|
|
insert_op2 (op, loc, arg1, arg2, end) |
2117 |
|
|
re_opcode_t op; |
2118 |
|
|
unsigned char *loc; |
2119 |
|
|
int arg1, arg2; |
2120 |
|
|
unsigned char *end; |
2121 |
|
|
{ |
2122 |
|
|
register unsigned char *pfrom = end; |
2123 |
|
|
register unsigned char *pto = end + 5; |
2124 |
|
|
|
2125 |
|
|
while (pfrom != loc) |
2126 |
|
|
*--pto = *--pfrom; |
2127 |
|
|
|
2128 |
|
|
store_op2 (op, loc, arg1, arg2); |
2129 |
|
|
} |
2130 |
|
|
|
2131 |
|
|
|
2132 |
|
|
/* P points to just after a ^ in PATTERN. Return true if that ^ comes |
2133 |
|
|
after an alternative or a begin-subexpression. We assume there is at |
2134 |
|
|
least one character before the ^. */ |
2135 |
|
|
|
2136 |
|
|
static boolean |
2137 |
|
|
at_begline_loc_p (pattern, p, syntax) |
2138 |
|
|
const char *pattern, *p; |
2139 |
|
|
reg_syntax_t syntax; |
2140 |
|
|
{ |
2141 |
|
|
const char *prev = p - 2; |
2142 |
|
|
boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\'; |
2143 |
|
|
|
2144 |
|
|
return |
2145 |
|
|
/* After a subexpression? */ |
2146 |
|
|
(*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash)) |
2147 |
|
|
/* After an alternative? */ |
2148 |
|
|
|| (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash)); |
2149 |
|
|
} |
2150 |
|
|
|
2151 |
|
|
|
2152 |
|
|
/* The dual of at_begline_loc_p. This one is for $. We assume there is |
2153 |
|
|
at least one character after the $, i.e., `P < PEND'. */ |
2154 |
|
|
|
2155 |
|
|
static boolean |
2156 |
|
|
at_endline_loc_p (p, pend, syntax) |
2157 |
|
|
const char *p, *pend; |
2158 |
|
|
int syntax; |
2159 |
|
|
{ |
2160 |
|
|
const char *next = p; |
2161 |
|
|
boolean next_backslash = *next == '\\'; |
2162 |
|
|
const char *next_next = p + 1 < pend ? p + 1 : NULL; |
2163 |
|
|
|
2164 |
|
|
return |
2165 |
|
|
/* Before a subexpression? */ |
2166 |
|
|
(syntax & RE_NO_BK_PARENS ? *next == ')' |
2167 |
|
|
: next_backslash && next_next && *next_next == ')') |
2168 |
|
|
/* Before an alternative? */ |
2169 |
|
|
|| (syntax & RE_NO_BK_VBAR ? *next == '|' |
2170 |
|
|
: next_backslash && next_next && *next_next == '|'); |
2171 |
|
|
} |
2172 |
|
|
|
2173 |
|
|
|
2174 |
|
|
/* Returns true if REGNUM is in one of COMPILE_STACK's elements and |
2175 |
|
|
false if it's not. */ |
2176 |
|
|
|
2177 |
|
|
static boolean |
2178 |
|
|
group_in_compile_stack (compile_stack, regnum) |
2179 |
|
|
compile_stack_type compile_stack; |
2180 |
|
|
regnum_t regnum; |
2181 |
|
|
{ |
2182 |
|
|
int this_element; |
2183 |
|
|
|
2184 |
|
|
for (this_element = compile_stack.avail - 1; |
2185 |
|
|
this_element >= 0; |
2186 |
|
|
this_element--) |
2187 |
|
|
if (compile_stack.stack[this_element].regnum == regnum) |
2188 |
|
|
return true; |
2189 |
|
|
|
2190 |
|
|
return false; |
2191 |
|
|
} |
2192 |
|
|
|
2193 |
|
|
|
2194 |
|
|
/* Read the ending character of a range (in a bracket expression) from the |
2195 |
|
|
uncompiled pattern *P_PTR (which ends at PEND). We assume the |
2196 |
|
|
starting character is in `P[-2]'. (`P[-1]' is the character `-'.) |
2197 |
|
|
Then we set the translation of all bits between the starting and |
2198 |
|
|
ending characters (inclusive) in the compiled pattern B. |
2199 |
|
|
|
2200 |
|
|
Return an error code. |
2201 |
|
|
|
2202 |
|
|
We use these short variable names so we can use the same macros as |
2203 |
|
|
`regex_compile' itself. */ |
2204 |
|
|
|
2205 |
|
|
static reg_errcode_t |
2206 |
|
|
compile_range (p_ptr, pend, translate, syntax, b) |
2207 |
|
|
const char **p_ptr, *pend; |
2208 |
|
|
char *translate; |
2209 |
|
|
reg_syntax_t syntax; |
2210 |
|
|
unsigned char *b; |
2211 |
|
|
{ |
2212 |
|
|
unsigned this_char; |
2213 |
|
|
|
2214 |
|
|
const char *p = *p_ptr; |
2215 |
|
|
int range_start, range_end; |
2216 |
|
|
|
2217 |
|
|
if (p == pend) |
2218 |
|
|
return REG_ERANGE; |
2219 |
|
|
|
2220 |
|
|
/* Even though the pattern is a signed `char *', we need to fetch |
2221 |
|
|
with unsigned char *'s; if the high bit of the pattern character |
2222 |
|
|
is set, the range endpoints will be negative if we fetch using a |
2223 |
|
|
signed char *. |
2224 |
|
|
|
2225 |
|
|
We also want to fetch the endpoints without translating them; the |
2226 |
|
|
appropriate translation is done in the bit-setting loop below. */ |
2227 |
|
|
range_start = ((unsigned char *) p)[-2]; |
2228 |
|
|
range_end = ((unsigned char *) p)[0]; |
2229 |
|
|
|
2230 |
|
|
/* Have to increment the pointer into the pattern string, so the |
2231 |
|
|
caller isn't still at the ending character. */ |
2232 |
|
|
(*p_ptr)++; |
2233 |
|
|
|
2234 |
|
|
/* If the start is after the end, the range is empty. */ |
2235 |
|
|
if (range_start > range_end) |
2236 |
|
|
return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR; |
2237 |
|
|
|
2238 |
|
|
/* Here we see why `this_char' has to be larger than an `unsigned |
2239 |
|
|
char' -- the range is inclusive, so if `range_end' == 0xff |
2240 |
|
|
(assuming 8-bit characters), we would otherwise go into an infinite |
2241 |
|
|
loop, since all characters <= 0xff. */ |
2242 |
|
|
for (this_char = range_start; this_char <= range_end; this_char++) |
2243 |
|
|
{ |
2244 |
|
|
SET_LIST_BIT (TRANSLATE (this_char)); |
2245 |
|
|
} |
2246 |
|
|
|
2247 |
|
|
return REG_NOERROR; |
2248 |
|
|
} |
2249 |
|
|
|
2250 |
|
|
/* Failure stack declarations and macros; both re_compile_fastmap and |
2251 |
|
|
re_match_2 use a failure stack. These have to be macros because of |
2252 |
|
|
REGEX_ALLOCATE. */ |
2253 |
|
|
|
2254 |
|
|
|
2255 |
|
|
/* Number of failure points for which to initially allocate space |
2256 |
|
|
when matching. If this number is exceeded, we allocate more |
2257 |
|
|
space, so it is not a hard limit. */ |
2258 |
|
|
#ifndef INIT_FAILURE_ALLOC |
2259 |
|
|
#define INIT_FAILURE_ALLOC 5 |
2260 |
|
|
#endif |
2261 |
|
|
|
2262 |
|
|
/* Roughly the maximum number of failure points on the stack. Would be |
2263 |
|
|
exactly that if always used MAX_FAILURE_SPACE each time we failed. |
2264 |
|
|
This is a variable only so users of regex can assign to it; we never |
2265 |
|
|
change it ourselves. */ |
2266 |
|
|
int re_max_failures = 2000; |
2267 |
|
|
|
2268 |
|
|
typedef const unsigned char *fail_stack_elt_t; |
2269 |
|
|
|
2270 |
|
|
typedef struct |
2271 |
|
|
{ |
2272 |
|
|
fail_stack_elt_t *stack; |
2273 |
|
|
unsigned size; |
2274 |
|
|
unsigned avail; /* Offset of next open position. */ |
2275 |
|
|
} fail_stack_type; |
2276 |
|
|
|
2277 |
|
|
#define FAIL_STACK_EMPTY() (fail_stack.avail == 0) |
2278 |
|
|
#define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0) |
2279 |
|
|
#define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size) |
2280 |
|
|
#define FAIL_STACK_TOP() (fail_stack.stack[fail_stack.avail]) |
2281 |
|
|
|
2282 |
|
|
|
2283 |
|
|
/* Initialize `fail_stack'. Do `return -2' if the alloc fails. */ |
2284 |
|
|
|
2285 |
|
|
#define INIT_FAIL_STACK() \ |
2286 |
|
|
do { \ |
2287 |
|
|
fail_stack.stack = (fail_stack_elt_t *) \ |
2288 |
|
|
REGEX_ALLOCATE (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \ |
2289 |
|
|
\ |
2290 |
|
|
if (fail_stack.stack == NULL) \ |
2291 |
|
|
return -2; \ |
2292 |
|
|
\ |
2293 |
|
|
fail_stack.size = INIT_FAILURE_ALLOC; \ |
2294 |
|
|
fail_stack.avail = 0; \ |
2295 |
|
|
} while (0) |
2296 |
|
|
|
2297 |
|
|
|
2298 |
|
|
/* Double the size of FAIL_STACK, up to approximately `re_max_failures' items. |
2299 |
|
|
|
2300 |
|
|
Return 1 if succeeds, and 0 if either ran out of memory |
2301 |
|
|
allocating space for it or it was already too large. |
2302 |
|
|
|
2303 |
|
|
REGEX_REALLOCATE requires `destination' be declared. */ |
2304 |
|
|
|
2305 |
|
|
#define DOUBLE_FAIL_STACK(fail_stack) \ |
2306 |
|
|
((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \ |
2307 |
|
|
? 0 \ |
2308 |
|
|
: ((fail_stack).stack = (fail_stack_elt_t *) \ |
2309 |
|
|
REGEX_REALLOCATE ((fail_stack).stack, \ |
2310 |
|
|
(fail_stack).size * sizeof (fail_stack_elt_t), \ |
2311 |
|
|
((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \ |
2312 |
|
|
\ |
2313 |
|
|
(fail_stack).stack == NULL \ |
2314 |
|
|
? 0 \ |
2315 |
|
|
: ((fail_stack).size <<= 1, \ |
2316 |
|
|
1))) |
2317 |
|
|
|
2318 |
|
|
|
2319 |
|
|
/* Push PATTERN_OP on FAIL_STACK. |
2320 |
|
|
|
2321 |
|
|
Return 1 if was able to do so and 0 if ran out of memory allocating |
2322 |
|
|
space to do so. */ |
2323 |
|
|
#define PUSH_PATTERN_OP(pattern_op, fail_stack) \ |
2324 |
|
|
((FAIL_STACK_FULL () \ |
2325 |
|
|
&& !DOUBLE_FAIL_STACK (fail_stack)) \ |
2326 |
|
|
? 0 \ |
2327 |
|
|
: ((fail_stack).stack[(fail_stack).avail++] = pattern_op, \ |
2328 |
|
|
1)) |
2329 |
|
|
|
2330 |
|
|
/* This pushes an item onto the failure stack. Must be a four-byte |
2331 |
|
|
value. Assumes the variable `fail_stack'. Probably should only |
2332 |
|
|
be called from within `PUSH_FAILURE_POINT'. */ |
2333 |
|
|
#define PUSH_FAILURE_ITEM(item) \ |
2334 |
|
|
fail_stack.stack[fail_stack.avail++] = (fail_stack_elt_t) item |
2335 |
|
|
|
2336 |
|
|
/* The complement operation. Assumes `fail_stack' is nonempty. */ |
2337 |
|
|
#define POP_FAILURE_ITEM() fail_stack.stack[--fail_stack.avail] |
2338 |
|
|
|
2339 |
|
|
/* Used to omit pushing failure point id's when we're not debugging. */ |
2340 |
|
|
#ifdef DEBUG |
2341 |
|
|
#define DEBUG_PUSH PUSH_FAILURE_ITEM |
2342 |
|
|
#define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_ITEM () |
2343 |
|
|
#else |
2344 |
|
|
#define DEBUG_PUSH(item) |
2345 |
|
|
#define DEBUG_POP(item_addr) |
2346 |
|
|
#endif |
2347 |
|
|
|
2348 |
|
|
|
2349 |
|
|
/* Push the information about the state we will need |
2350 |
|
|
if we ever fail back to it. |
2351 |
|
|
|
2352 |
|
|
Requires variables fail_stack, regstart, regend, reg_info, and |
2353 |
|
|
num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be |
2354 |
|
|
declared. |
2355 |
|
|
|
2356 |
|
|
Does `return FAILURE_CODE' if runs out of memory. */ |
2357 |
|
|
|
2358 |
|
|
#define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \ |
2359 |
|
|
do { \ |
2360 |
|
|
char *destination; \ |
2361 |
|
|
/* Must be int, so when we don't save any registers, the arithmetic \ |
2362 |
|
|
of 0 + -1 isn't done as unsigned. */ \ |
2363 |
|
|
int this_reg; \ |
2364 |
|
|
\ |
2365 |
|
|
DEBUG_STATEMENT (failure_id++); \ |
2366 |
|
|
DEBUG_STATEMENT (nfailure_points_pushed++); \ |
2367 |
|
|
DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \ |
2368 |
|
|
DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\ |
2369 |
|
|
DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\ |
2370 |
|
|
\ |
2371 |
|
|
DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \ |
2372 |
|
|
DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \ |
2373 |
|
|
\ |
2374 |
|
|
/* Ensure we have enough space allocated for what we will push. */ \ |
2375 |
|
|
while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \ |
2376 |
|
|
{ \ |
2377 |
|
|
if (!DOUBLE_FAIL_STACK (fail_stack)) \ |
2378 |
|
|
return failure_code; \ |
2379 |
|
|
\ |
2380 |
|
|
DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \ |
2381 |
|
|
(fail_stack).size); \ |
2382 |
|
|
DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\ |
2383 |
|
|
} \ |
2384 |
|
|
\ |
2385 |
|
|
/* Push the info, starting with the registers. */ \ |
2386 |
|
|
DEBUG_PRINT1 ("\n"); \ |
2387 |
|
|
\ |
2388 |
|
|
for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \ |
2389 |
|
|
this_reg++) \ |
2390 |
|
|
{ \ |
2391 |
|
|
DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \ |
2392 |
|
|
DEBUG_STATEMENT (num_regs_pushed++); \ |
2393 |
|
|
\ |
2394 |
|
|
DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ |
2395 |
|
|
PUSH_FAILURE_ITEM (regstart[this_reg]); \ |
2396 |
|
|
\ |
2397 |
|
|
DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ |
2398 |
|
|
PUSH_FAILURE_ITEM (regend[this_reg]); \ |
2399 |
|
|
\ |
2400 |
|
|
DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \ |
2401 |
|
|
DEBUG_PRINT2 (" match_null=%d", \ |
2402 |
|
|
REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \ |
2403 |
|
|
DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \ |
2404 |
|
|
DEBUG_PRINT2 (" matched_something=%d", \ |
2405 |
|
|
MATCHED_SOMETHING (reg_info[this_reg])); \ |
2406 |
|
|
DEBUG_PRINT2 (" ever_matched=%d", \ |
2407 |
|
|
EVER_MATCHED_SOMETHING (reg_info[this_reg])); \ |
2408 |
|
|
DEBUG_PRINT1 ("\n"); \ |
2409 |
|
|
PUSH_FAILURE_ITEM (reg_info[this_reg].word); \ |
2410 |
|
|
} \ |
2411 |
|
|
\ |
2412 |
|
|
DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\ |
2413 |
|
|
PUSH_FAILURE_ITEM (lowest_active_reg); \ |
2414 |
|
|
\ |
2415 |
|
|
DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\ |
2416 |
|
|
PUSH_FAILURE_ITEM (highest_active_reg); \ |
2417 |
|
|
\ |
2418 |
|
|
DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \ |
2419 |
|
|
DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \ |
2420 |
|
|
PUSH_FAILURE_ITEM (pattern_place); \ |
2421 |
|
|
\ |
2422 |
|
|
DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \ |
2423 |
|
|
DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \ |
2424 |
|
|
size2); \ |
2425 |
|
|
DEBUG_PRINT1 ("'\n"); \ |
2426 |
|
|
PUSH_FAILURE_ITEM (string_place); \ |
2427 |
|
|
\ |
2428 |
|
|
DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \ |
2429 |
|
|
DEBUG_PUSH (failure_id); \ |
2430 |
|
|
} while (0) |
2431 |
|
|
|
2432 |
|
|
/* This is the number of items that are pushed and popped on the stack |
2433 |
|
|
for each register. */ |
2434 |
|
|
#define NUM_REG_ITEMS 3 |
2435 |
|
|
|
2436 |
|
|
/* Individual items aside from the registers. */ |
2437 |
|
|
#ifdef DEBUG |
2438 |
|
|
#define NUM_NONREG_ITEMS 5 /* Includes failure point id. */ |
2439 |
|
|
#else |
2440 |
|
|
#define NUM_NONREG_ITEMS 4 |
2441 |
|
|
#endif |
2442 |
|
|
|
2443 |
|
|
/* We push at most this many items on the stack. */ |
2444 |
|
|
#define MAX_FAILURE_ITEMS ((num_regs - 1) * NUM_REG_ITEMS + NUM_NONREG_ITEMS) |
2445 |
|
|
|
2446 |
|
|
/* We actually push this many items. */ |
2447 |
|
|
#define NUM_FAILURE_ITEMS \ |
2448 |
|
|
((highest_active_reg - lowest_active_reg + 1) * NUM_REG_ITEMS \ |
2449 |
|
|
+ NUM_NONREG_ITEMS) |
2450 |
|
|
|
2451 |
|
|
/* How many items can still be added to the stack without overflowing it. */ |
2452 |
|
|
#define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail) |
2453 |
|
|
|
2454 |
|
|
|
2455 |
|
|
/* Pops what PUSH_FAIL_STACK pushes. |
2456 |
|
|
|
2457 |
|
|
We restore into the parameters, all of which should be lvalues: |
2458 |
|
|
STR -- the saved data position. |
2459 |
|
|
PAT -- the saved pattern position. |
2460 |
|
|
LOW_REG, HIGH_REG -- the highest and lowest active registers. |
2461 |
|
|
REGSTART, REGEND -- arrays of string positions. |
2462 |
|
|
REG_INFO -- array of information about each subexpression. |
2463 |
|
|
|
2464 |
|
|
Also assumes the variables `fail_stack' and (if debugging), `bufp', |
2465 |
|
|
`pend', `string1', `size1', `string2', and `size2'. */ |
2466 |
|
|
|
2467 |
|
|
#define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\ |
2468 |
|
|
{ \ |
2469 |
|
|
DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \ |
2470 |
|
|
int this_reg; \ |
2471 |
|
|
const unsigned char *string_temp; \ |
2472 |
|
|
\ |
2473 |
|
|
assert (!FAIL_STACK_EMPTY ()); \ |
2474 |
|
|
\ |
2475 |
|
|
/* Remove failure points and point to how many regs pushed. */ \ |
2476 |
|
|
DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \ |
2477 |
|
|
DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \ |
2478 |
|
|
DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \ |
2479 |
|
|
\ |
2480 |
|
|
assert (fail_stack.avail >= NUM_NONREG_ITEMS); \ |
2481 |
|
|
\ |
2482 |
|
|
DEBUG_POP (&failure_id); \ |
2483 |
|
|
DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \ |
2484 |
|
|
\ |
2485 |
|
|
/* If the saved string location is NULL, it came from an \ |
2486 |
|
|
on_failure_keep_string_jump opcode, and we want to throw away the \ |
2487 |
|
|
saved NULL, thus retaining our current position in the string. */ \ |
2488 |
|
|
string_temp = POP_FAILURE_ITEM (); \ |
2489 |
|
|
if (string_temp != NULL) \ |
2490 |
|
|
str = (const char *) string_temp; \ |
2491 |
|
|
\ |
2492 |
|
|
DEBUG_PRINT2 (" Popping string 0x%x: `", str); \ |
2493 |
|
|
DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \ |
2494 |
|
|
DEBUG_PRINT1 ("'\n"); \ |
2495 |
|
|
\ |
2496 |
|
|
pat = (unsigned char *) POP_FAILURE_ITEM (); \ |
2497 |
|
|
DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \ |
2498 |
|
|
DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \ |
2499 |
|
|
\ |
2500 |
|
|
/* Restore register info. */ \ |
2501 |
|
|
high_reg = (unsigned) POP_FAILURE_ITEM (); \ |
2502 |
|
|
DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \ |
2503 |
|
|
\ |
2504 |
|
|
low_reg = (unsigned) POP_FAILURE_ITEM (); \ |
2505 |
|
|
DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \ |
2506 |
|
|
\ |
2507 |
|
|
for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \ |
2508 |
|
|
{ \ |
2509 |
|
|
DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \ |
2510 |
|
|
\ |
2511 |
|
|
reg_info[this_reg].word = POP_FAILURE_ITEM (); \ |
2512 |
|
|
DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \ |
2513 |
|
|
\ |
2514 |
|
|
regend[this_reg] = (const char *) POP_FAILURE_ITEM (); \ |
2515 |
|
|
DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \ |
2516 |
|
|
\ |
2517 |
|
|
regstart[this_reg] = (const char *) POP_FAILURE_ITEM (); \ |
2518 |
|
|
DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \ |
2519 |
|
|
} \ |
2520 |
|
|
\ |
2521 |
|
|
DEBUG_STATEMENT (nfailure_points_popped++); \ |
2522 |
|
|
} /* POP_FAILURE_POINT */ |
2523 |
|
|
|
2524 |
|
|
/* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in |
2525 |
|
|
BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible |
2526 |
|
|
characters can start a string that matches the pattern. This fastmap |
2527 |
|
|
is used by re_search to skip quickly over impossible starting points. |
2528 |
|
|
|
2529 |
|
|
The caller must supply the address of a (1 << BYTEWIDTH)-byte data |
2530 |
|
|
area as BUFP->fastmap. |
2531 |
|
|
|
2532 |
|
|
We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in |
2533 |
|
|
the pattern buffer. |
2534 |
|
|
|
2535 |
|
|
Returns 0 if we succeed, -2 if an internal error. */ |
2536 |
|
|
|
2537 |
|
|
int |
2538 |
|
|
re_compile_fastmap (bufp) |
2539 |
|
|
struct re_pattern_buffer *bufp; |
2540 |
|
|
{ |
2541 |
|
|
int j, k; |
2542 |
|
|
fail_stack_type fail_stack; |
2543 |
|
|
#ifndef REGEX_MALLOC |
2544 |
|
|
char *destination; |
2545 |
|
|
#endif |
2546 |
|
|
/* We don't push any register information onto the failure stack. */ |
2547 |
|
|
unsigned num_regs = 0; |
2548 |
|
|
|
2549 |
|
|
register char *fastmap = bufp->fastmap; |
2550 |
|
|
unsigned char *pattern = bufp->buffer; |
2551 |
|
|
unsigned long size = bufp->used; |
2552 |
|
|
const unsigned char *p = pattern; |
2553 |
|
|
register unsigned char *pend = pattern + size; |
2554 |
|
|
|
2555 |
|
|
/* Assume that each path through the pattern can be null until |
2556 |
|
|
proven otherwise. We set this false at the bottom of switch |
2557 |
|
|
statement, to which we get only if a particular path doesn't |
2558 |
|
|
match the empty string. */ |
2559 |
|
|
boolean path_can_be_null = true; |
2560 |
|
|
|
2561 |
|
|
/* We aren't doing a `succeed_n' to begin with. */ |
2562 |
|
|
boolean succeed_n_p = false; |
2563 |
|
|
|
2564 |
|
|
assert (fastmap != NULL && p != NULL); |
2565 |
|
|
|
2566 |
|
|
INIT_FAIL_STACK (); |
2567 |
|
|
bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */ |
2568 |
|
|
bufp->fastmap_accurate = 1; /* It will be when we're done. */ |
2569 |
|
|
bufp->can_be_null = 0; |
2570 |
|
|
|
2571 |
|
|
while (p != pend || !FAIL_STACK_EMPTY ()) |
2572 |
|
|
{ |
2573 |
|
|
if (p == pend) |
2574 |
|
|
{ |
2575 |
|
|
bufp->can_be_null |= path_can_be_null; |
2576 |
|
|
|
2577 |
|
|
/* Reset for next path. */ |
2578 |
|
|
path_can_be_null = true; |
2579 |
|
|
|
2580 |
|
|
p = fail_stack.stack[--fail_stack.avail]; |
2581 |
|
|
} |
2582 |
|
|
|
2583 |
|
|
/* We should never be about to go beyond the end of the pattern. */ |
2584 |
|
|
assert (p < pend); |
2585 |
|
|
|
2586 |
|
|
#ifdef SWITCH_ENUM_BUG |
2587 |
|
|
switch ((int) ((re_opcode_t) *p++)) |
2588 |
|
|
#else |
2589 |
|
|
switch ((re_opcode_t) *p++) |
2590 |
|
|
#endif |
2591 |
|
|
{ |
2592 |
|
|
|
2593 |
|
|
/* I guess the idea here is to simply not bother with a fastmap |
2594 |
|
|
if a backreference is used, since it's too hard to figure out |
2595 |
|
|
the fastmap for the corresponding group. Setting |
2596 |
|
|
`can_be_null' stops `re_search_2' from using the fastmap, so |
2597 |
|
|
that is all we do. */ |
2598 |
|
|
case duplicate: |
2599 |
|
|
bufp->can_be_null = 1; |
2600 |
|
|
return 0; |
2601 |
|
|
|
2602 |
|
|
|
2603 |
|
|
/* Following are the cases which match a character. These end |
2604 |
|
|
with `break'. */ |
2605 |
|
|
|
2606 |
|
|
case exactn: |
2607 |
|
|
fastmap[p[1]] = 1; |
2608 |
|
|
break; |
2609 |
|
|
|
2610 |
|
|
|
2611 |
|
|
case charset: |
2612 |
|
|
for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) |
2613 |
|
|
if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))) |
2614 |
|
|
fastmap[j] = 1; |
2615 |
|
|
break; |
2616 |
|
|
|
2617 |
|
|
|
2618 |
|
|
case charset_not: |
2619 |
|
|
/* Chars beyond end of map must be allowed. */ |
2620 |
|
|
for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++) |
2621 |
|
|
fastmap[j] = 1; |
2622 |
|
|
|
2623 |
|
|
for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--) |
2624 |
|
|
if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))) |
2625 |
|
|
fastmap[j] = 1; |
2626 |
|
|
break; |
2627 |
|
|
|
2628 |
|
|
|
2629 |
|
|
case wordchar: |
2630 |
|
|
for (j = 0; j < (1 << BYTEWIDTH); j++) |
2631 |
|
|
if (SYNTAX (j) == Sword) |
2632 |
|
|
fastmap[j] = 1; |
2633 |
|
|
break; |
2634 |
|
|
|
2635 |
|
|
|
2636 |
|
|
case notwordchar: |
2637 |
|
|
for (j = 0; j < (1 << BYTEWIDTH); j++) |
2638 |
|
|
if (SYNTAX (j) != Sword) |
2639 |
|
|
fastmap[j] = 1; |
2640 |
|
|
break; |
2641 |
|
|
|
2642 |
|
|
|
2643 |
|
|
case anychar: |
2644 |
|
|
/* `.' matches anything ... */ |
2645 |
|
|
for (j = 0; j < (1 << BYTEWIDTH); j++) |
2646 |
|
|
fastmap[j] = 1; |
2647 |
|
|
|
2648 |
|
|
/* ... except perhaps newline. */ |
2649 |
|
|
if (!(bufp->syntax & RE_DOT_NEWLINE)) |
2650 |
|
|
fastmap['\n'] = 0; |
2651 |
|
|
|
2652 |
|
|
/* Return if we have already set `can_be_null'; if we have, |
2653 |
|
|
then the fastmap is irrelevant. Something's wrong here. */ |
2654 |
|
|
else if (bufp->can_be_null) |
2655 |
|
|
return 0; |
2656 |
|
|
|
2657 |
|
|
/* Otherwise, have to check alternative paths. */ |
2658 |
|
|
break; |
2659 |
|
|
|
2660 |
|
|
|
2661 |
|
|
#ifdef emacs |
2662 |
|
|
case syntaxspec: |
2663 |
|
|
k = *p++; |
2664 |
|
|
for (j = 0; j < (1 << BYTEWIDTH); j++) |
2665 |
|
|
if (SYNTAX (j) == (enum syntaxcode) k) |
2666 |
|
|
fastmap[j] = 1; |
2667 |
|
|
break; |
2668 |
|
|
|
2669 |
|
|
|
2670 |
|
|
case notsyntaxspec: |
2671 |
|
|
k = *p++; |
2672 |
|
|
for (j = 0; j < (1 << BYTEWIDTH); j++) |
2673 |
|
|
if (SYNTAX (j) != (enum syntaxcode) k) |
2674 |
|
|
fastmap[j] = 1; |
2675 |
|
|
break; |
2676 |
|
|
|
2677 |
|
|
|
2678 |
|
|
/* All cases after this match the empty string. These end with |
2679 |
|
|
`continue'. */ |
2680 |
|
|
|
2681 |
|
|
|
2682 |
|
|
case before_dot: |
2683 |
|
|
case at_dot: |
2684 |
|
|
case after_dot: |
2685 |
|
|
continue; |
2686 |
|
|
#endif /* not emacs */ |
2687 |
|
|
|
2688 |
|
|
|
2689 |
|
|
case no_op: |
2690 |
|
|
case begline: |
2691 |
|
|
case endline: |
2692 |
|
|
case begbuf: |
2693 |
|
|
case endbuf: |
2694 |
|
|
case wordbound: |
2695 |
|
|
case notwordbound: |
2696 |
|
|
case wordbeg: |
2697 |
|
|
case wordend: |
2698 |
|
|
case push_dummy_failure: |
2699 |
|
|
continue; |
2700 |
|
|
|
2701 |
|
|
|
2702 |
|
|
case jump_n: |
2703 |
|
|
case pop_failure_jump: |
2704 |
|
|
case maybe_pop_jump: |
2705 |
|
|
case jump: |
2706 |
|
|
case jump_past_alt: |
2707 |
|
|
case dummy_failure_jump: |
2708 |
|
|
EXTRACT_NUMBER_AND_INCR (j, p); |
2709 |
|
|
p += j; |
2710 |
|
|
if (j > 0) |
2711 |
|
|
continue; |
2712 |
|
|
|
2713 |
|
|
/* Jump backward implies we just went through the body of a |
2714 |
|
|
loop and matched nothing. Opcode jumped to should be |
2715 |
|
|
`on_failure_jump' or `succeed_n'. Just treat it like an |
2716 |
|
|
ordinary jump. For a * loop, it has pushed its failure |
2717 |
|
|
point already; if so, discard that as redundant. */ |
2718 |
|
|
if ((re_opcode_t) *p != on_failure_jump |
2719 |
|
|
&& (re_opcode_t) *p != succeed_n) |
2720 |
|
|
continue; |
2721 |
|
|
|
2722 |
|
|
p++; |
2723 |
|
|
EXTRACT_NUMBER_AND_INCR (j, p); |
2724 |
|
|
p += j; |
2725 |
|
|
|
2726 |
|
|
/* If what's on the stack is where we are now, pop it. */ |
2727 |
|
|
if (!FAIL_STACK_EMPTY () |
2728 |
|
|
&& fail_stack.stack[fail_stack.avail - 1] == p) |
2729 |
|
|
fail_stack.avail--; |
2730 |
|
|
|
2731 |
|
|
continue; |
2732 |
|
|
|
2733 |
|
|
|
2734 |
|
|
case on_failure_jump: |
2735 |
|
|
case on_failure_keep_string_jump: |
2736 |
|
|
handle_on_failure_jump: |
2737 |
|
|
EXTRACT_NUMBER_AND_INCR (j, p); |
2738 |
|
|
|
2739 |
|
|
/* For some patterns, e.g., `(a?)?', `p+j' here points to the |
2740 |
|
|
end of the pattern. We don't want to push such a point, |
2741 |
|
|
since when we restore it above, entering the switch will |
2742 |
|
|
increment `p' past the end of the pattern. We don't need |
2743 |
|
|
to push such a point since we obviously won't find any more |
2744 |
|
|
fastmap entries beyond `pend'. Such a pattern can match |
2745 |
|
|
the null string, though. */ |
2746 |
|
|
if (p + j < pend) |
2747 |
|
|
{ |
2748 |
|
|
if (!PUSH_PATTERN_OP (p + j, fail_stack)) |
2749 |
|
|
return -2; |
2750 |
|
|
} |
2751 |
|
|
else |
2752 |
|
|
bufp->can_be_null = 1; |
2753 |
|
|
|
2754 |
|
|
if (succeed_n_p) |
2755 |
|
|
{ |
2756 |
|
|
EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */ |
2757 |
|
|
succeed_n_p = false; |
2758 |
|
|
} |
2759 |
|
|
|
2760 |
|
|
continue; |
2761 |
|
|
|
2762 |
|
|
|
2763 |
|
|
case succeed_n: |
2764 |
|
|
/* Get to the number of times to succeed. */ |
2765 |
|
|
p += 2; |
2766 |
|
|
|
2767 |
|
|
/* Increment p past the n for when k != 0. */ |
2768 |
|
|
EXTRACT_NUMBER_AND_INCR (k, p); |
2769 |
|
|
if (k == 0) |
2770 |
|
|
{ |
2771 |
|
|
p -= 4; |
2772 |
|
|
succeed_n_p = true; /* Spaghetti code alert. */ |
2773 |
|
|
goto handle_on_failure_jump; |
2774 |
|
|
} |
2775 |
|
|
continue; |
2776 |
|
|
|
2777 |
|
|
|
2778 |
|
|
case set_number_at: |
2779 |
|
|
p += 4; |
2780 |
|
|
continue; |
2781 |
|
|
|
2782 |
|
|
|
2783 |
|
|
case start_memory: |
2784 |
|
|
case stop_memory: |
2785 |
|
|
p += 2; |
2786 |
|
|
continue; |
2787 |
|
|
|
2788 |
|
|
|
2789 |
|
|
default: |
2790 |
|
|
abort (); /* We have listed all the cases. */ |
2791 |
|
|
} /* switch *p++ */ |
2792 |
|
|
|
2793 |
|
|
/* Getting here means we have found the possible starting |
2794 |
|
|
characters for one path of the pattern -- and that the empty |
2795 |
|
|
string does not match. We need not follow this path further. |
2796 |
|
|
Instead, look at the next alternative (remembered on the |
2797 |
|
|
stack), or quit if no more. The test at the top of the loop |
2798 |
|
|
does these things. */ |
2799 |
|
|
path_can_be_null = false; |
2800 |
|
|
p = pend; |
2801 |
|
|
} /* while p */ |
2802 |
|
|
|
2803 |
|
|
/* Set `can_be_null' for the last path (also the first path, if the |
2804 |
|
|
pattern is empty). */ |
2805 |
|
|
bufp->can_be_null |= path_can_be_null; |
2806 |
|
|
return 0; |
2807 |
|
|
} /* re_compile_fastmap */ |
2808 |
|
|
|
2809 |
|
|
/* Set REGS to hold NUM_REGS registers, storing them in STARTS and |
2810 |
|
|
ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use |
2811 |
|
|
this memory for recording register information. STARTS and ENDS |
2812 |
|
|
must be allocated using the malloc library routine, and must each |
2813 |
|
|
be at least NUM_REGS * sizeof (regoff_t) bytes long. |
2814 |
|
|
|
2815 |
|
|
If NUM_REGS == 0, then subsequent matches should allocate their own |
2816 |
|
|
register data. |
2817 |
|
|
|
2818 |
|
|
Unless this function is called, the first search or match using |
2819 |
|
|
PATTERN_BUFFER will allocate its own register data, without |
2820 |
|
|
freeing the old data. */ |
2821 |
|
|
|
2822 |
|
|
void |
2823 |
|
|
re_set_registers (bufp, regs, num_regs, starts, ends) |
2824 |
|
|
struct re_pattern_buffer *bufp; |
2825 |
|
|
struct re_registers *regs; |
2826 |
|
|
unsigned num_regs; |
2827 |
|
|
regoff_t *starts, *ends; |
2828 |
|
|
{ |
2829 |
|
|
if (num_regs) |
2830 |
|
|
{ |
2831 |
|
|
bufp->regs_allocated = REGS_REALLOCATE; |
2832 |
|
|
regs->num_regs = num_regs; |
2833 |
|
|
regs->start = starts; |
2834 |
|
|
regs->end = ends; |
2835 |
|
|
} |
2836 |
|
|
else |
2837 |
|
|
{ |
2838 |
|
|
bufp->regs_allocated = REGS_UNALLOCATED; |
2839 |
|
|
regs->num_regs = 0; |
2840 |
|
|
regs->start = regs->end = (regoff_t) 0; |
2841 |
|
|
} |
2842 |
|
|
} |
2843 |
|
|
|
2844 |
|
|
/* Searching routines. */ |
2845 |
|
|
|
2846 |
|
|
/* Like re_search_2, below, but only one string is specified, and |
2847 |
|
|
doesn't let you say where to stop matching. */ |
2848 |
|
|
|
2849 |
|
|
int |
2850 |
|
|
re_search (bufp, string, size, startpos, range, regs) |
2851 |
|
|
struct re_pattern_buffer *bufp; |
2852 |
|
|
const char *string; |
2853 |
|
|
int size, startpos, range; |
2854 |
|
|
struct re_registers *regs; |
2855 |
|
|
{ |
2856 |
|
|
return re_search_2 (bufp, NULL, 0, string, size, startpos, range, |
2857 |
|
|
regs, size); |
2858 |
|
|
} |
2859 |
|
|
|
2860 |
|
|
|
2861 |
|
|
/* Using the compiled pattern in BUFP->buffer, first tries to match the |
2862 |
|
|
virtual concatenation of STRING1 and STRING2, starting first at index |
2863 |
|
|
STARTPOS, then at STARTPOS + 1, and so on. |
2864 |
|
|
|
2865 |
|
|
STRING1 and STRING2 have length SIZE1 and SIZE2, respectively. |
2866 |
|
|
|
2867 |
|
|
RANGE is how far to scan while trying to match. RANGE = 0 means try |
2868 |
|
|
only at STARTPOS; in general, the last start tried is STARTPOS + |
2869 |
|
|
RANGE. |
2870 |
|
|
|
2871 |
|
|
In REGS, return the indices of the virtual concatenation of STRING1 |
2872 |
|
|
and STRING2 that matched the entire BUFP->buffer and its contained |
2873 |
|
|
subexpressions. |
2874 |
|
|
|
2875 |
|
|
Do not consider matching one past the index STOP in the virtual |
2876 |
|
|
concatenation of STRING1 and STRING2. |
2877 |
|
|
|
2878 |
|
|
We return either the position in the strings at which the match was |
2879 |
|
|
found, -1 if no match, or -2 if error (such as failure |
2880 |
|
|
stack overflow). */ |
2881 |
|
|
|
2882 |
|
|
int |
2883 |
|
|
re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop) |
2884 |
|
|
struct re_pattern_buffer *bufp; |
2885 |
|
|
const char *string1, *string2; |
2886 |
|
|
int size1, size2; |
2887 |
|
|
int startpos; |
2888 |
|
|
int range; |
2889 |
|
|
struct re_registers *regs; |
2890 |
|
|
int stop; |
2891 |
|
|
{ |
2892 |
|
|
int val; |
2893 |
|
|
register char *fastmap = bufp->fastmap; |
2894 |
|
|
register char *translate = bufp->translate; |
2895 |
|
|
int total_size = size1 + size2; |
2896 |
|
|
int endpos = startpos + range; |
2897 |
|
|
|
2898 |
|
|
/* Check for out-of-range STARTPOS. */ |
2899 |
|
|
if (startpos < 0 || startpos > total_size) |
2900 |
|
|
return -1; |
2901 |
|
|
|
2902 |
|
|
/* Fix up RANGE if it might eventually take us outside |
2903 |
|
|
the virtual concatenation of STRING1 and STRING2. */ |
2904 |
|
|
if (endpos < -1) |
2905 |
|
|
range = -1 - startpos; |
2906 |
|
|
else if (endpos > total_size) |
2907 |
|
|
range = total_size - startpos; |
2908 |
|
|
|
2909 |
|
|
/* If the search isn't to be a backwards one, don't waste time in a |
2910 |
|
|
search for a pattern that must be anchored. */ |
2911 |
|
|
if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0) |
2912 |
|
|
{ |
2913 |
|
|
if (startpos > 0) |
2914 |
|
|
return -1; |
2915 |
|
|
else |
2916 |
|
|
range = 1; |
2917 |
|
|
} |
2918 |
|
|
|
2919 |
|
|
/* Update the fastmap now if not correct already. */ |
2920 |
|
|
if (fastmap && !bufp->fastmap_accurate) |
2921 |
|
|
if (re_compile_fastmap (bufp) == -2) |
2922 |
|
|
return -2; |
2923 |
|
|
|
2924 |
|
|
/* Loop through the string, looking for a place to start matching. */ |
2925 |
|
|
for (;;) |
2926 |
|
|
{ |
2927 |
|
|
/* If a fastmap is supplied, skip quickly over characters that |
2928 |
|
|
cannot be the start of a match. If the pattern can match the |
2929 |
|
|
null string, however, we don't need to skip characters; we want |
2930 |
|
|
the first null string. */ |
2931 |
|
|
if (fastmap && startpos < total_size && !bufp->can_be_null) |
2932 |
|
|
{ |
2933 |
|
|
if (range > 0) /* Searching forwards. */ |
2934 |
|
|
{ |
2935 |
|
|
register const char *d; |
2936 |
|
|
register int lim = 0; |
2937 |
|
|
int irange = range; |
2938 |
|
|
|
2939 |
|
|
if (startpos < size1 && startpos + range >= size1) |
2940 |
|
|
lim = range - (size1 - startpos); |
2941 |
|
|
|
2942 |
|
|
d = (startpos >= size1 ? string2 - size1 : string1) + startpos; |
2943 |
|
|
|
2944 |
|
|
/* Written out as an if-else to avoid testing `translate' |
2945 |
|
|
inside the loop. */ |
2946 |
|
|
if (translate) |
2947 |
|
|
while (range > lim |
2948 |
|
|
&& !fastmap[(unsigned char) |
2949 |
|
|
translate[(unsigned char) *d++]]) |
2950 |
|
|
range--; |
2951 |
|
|
else |
2952 |
|
|
while (range > lim && !fastmap[(unsigned char) *d++]) |
2953 |
|
|
range--; |
2954 |
|
|
|
2955 |
|
|
startpos += irange - range; |
2956 |
|
|
} |
2957 |
|
|
else /* Searching backwards. */ |
2958 |
|
|
{ |
2959 |
|
|
register char c = (size1 == 0 || startpos >= size1 |
2960 |
|
|
? string2[startpos - size1] |
2961 |
|
|
: string1[startpos]); |
2962 |
|
|
|
2963 |
|
|
if (!fastmap[(unsigned char) TRANSLATE (c)]) |
2964 |
|
|
goto advance; |
2965 |
|
|
} |
2966 |
|
|
} |
2967 |
|
|
|
2968 |
|
|
/* If can't match the null string, and that's all we have left, fail. */ |
2969 |
|
|
if (range >= 0 && startpos == total_size && fastmap |
2970 |
|
|
&& !bufp->can_be_null) |
2971 |
|
|
return -1; |
2972 |
|
|
|
2973 |
|
|
val = re_match_2 (bufp, string1, size1, string2, size2, |
2974 |
|
|
startpos, regs, stop); |
2975 |
|
|
if (val >= 0) |
2976 |
|
|
return startpos; |
2977 |
|
|
|
2978 |
|
|
if (val == -2) |
2979 |
|
|
return -2; |
2980 |
|
|
|
2981 |
|
|
advance: |
2982 |
|
|
if (!range) |
2983 |
|
|
break; |
2984 |
|
|
else if (range > 0) |
2985 |
|
|
{ |
2986 |
|
|
range--; |
2987 |
|
|
startpos++; |
2988 |
|
|
} |
2989 |
|
|
else |
2990 |
|
|
{ |
2991 |
|
|
range++; |
2992 |
|
|
startpos--; |
2993 |
|
|
} |
2994 |
|
|
} |
2995 |
|
|
return -1; |
2996 |
|
|
} /* re_search_2 */ |
2997 |
|
|
|
2998 |
|
|
/* Declarations and macros for re_match_2. */ |
2999 |
|
|
|
3000 |
|
|
static int bcmp_translate (); |
3001 |
|
|
static boolean alt_match_null_string_p (), |
3002 |
|
|
common_op_match_null_string_p (), |
3003 |
|
|
group_match_null_string_p (); |
3004 |
|
|
|
3005 |
|
|
/* Structure for per-register (a.k.a. per-group) information. |
3006 |
|
|
This must not be longer than one word, because we push this value |
3007 |
|
|
onto the failure stack. Other register information, such as the |
3008 |
|
|
starting and ending positions (which are addresses), and the list of |
3009 |
|
|
inner groups (which is a bits list) are maintained in separate |
3010 |
|
|
variables. |
3011 |
|
|
|
3012 |
|
|
We are making a (strictly speaking) nonportable assumption here: that |
3013 |
|
|
the compiler will pack our bit fields into something that fits into |
3014 |
|
|
the type of `word', i.e., is something that fits into one item on the |
3015 |
|
|
failure stack. */ |
3016 |
|
|
typedef union |
3017 |
|
|
{ |
3018 |
|
|
fail_stack_elt_t word; |
3019 |
|
|
struct |
3020 |
|
|
{ |
3021 |
|
|
/* This field is one if this group can match the empty string, |
3022 |
|
|
zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */ |
3023 |
|
|
#define MATCH_NULL_UNSET_VALUE 3 |
3024 |
|
|
unsigned match_null_string_p : 2; |
3025 |
|
|
unsigned is_active : 1; |
3026 |
|
|
unsigned matched_something : 1; |
3027 |
|
|
unsigned ever_matched_something : 1; |
3028 |
|
|
} bits; |
3029 |
|
|
} register_info_type; |
3030 |
|
|
|
3031 |
|
|
#define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p) |
3032 |
|
|
#define IS_ACTIVE(R) ((R).bits.is_active) |
3033 |
|
|
#define MATCHED_SOMETHING(R) ((R).bits.matched_something) |
3034 |
|
|
#define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something) |
3035 |
|
|
|
3036 |
|
|
|
3037 |
|
|
/* Call this when have matched a real character; it sets `matched' flags |
3038 |
|
|
for the subexpressions which we are currently inside. Also records |
3039 |
|
|
that those subexprs have matched. */ |
3040 |
|
|
#define SET_REGS_MATCHED() \ |
3041 |
|
|
do \ |
3042 |
|
|
{ \ |
3043 |
|
|
unsigned r; \ |
3044 |
|
|
for (r = lowest_active_reg; r <= highest_active_reg; r++) \ |
3045 |
|
|
{ \ |
3046 |
|
|
MATCHED_SOMETHING (reg_info[r]) \ |
3047 |
|
|
= EVER_MATCHED_SOMETHING (reg_info[r]) \ |
3048 |
|
|
= 1; \ |
3049 |
|
|
} \ |
3050 |
|
|
} \ |
3051 |
|
|
while (0) |
3052 |
|
|
|
3053 |
|
|
|
3054 |
|
|
/* This converts PTR, a pointer into one of the search strings `string1' |
3055 |
|
|
and `string2' into an offset from the beginning of that string. */ |
3056 |
|
|
#define POINTER_TO_OFFSET(ptr) \ |
3057 |
|
|
(FIRST_STRING_P (ptr) ? (ptr) - string1 : (ptr) - string2 + size1) |
3058 |
|
|
|
3059 |
|
|
/* Registers are set to a sentinel when they haven't yet matched. */ |
3060 |
|
|
#define REG_UNSET_VALUE ((char *) -1) |
3061 |
|
|
#define REG_UNSET(e) ((e) == REG_UNSET_VALUE) |
3062 |
|
|
|
3063 |
|
|
|
3064 |
|
|
/* Macros for dealing with the split strings in re_match_2. */ |
3065 |
|
|
|
3066 |
|
|
#define MATCHING_IN_FIRST_STRING (dend == end_match_1) |
3067 |
|
|
|
3068 |
|
|
/* Call before fetching a character with *d. This switches over to |
3069 |
|
|
string2 if necessary. */ |
3070 |
|
|
#define PREFETCH() \ |
3071 |
|
|
while (d == dend) \ |
3072 |
|
|
{ \ |
3073 |
|
|
/* End of string2 => fail. */ \ |
3074 |
|
|
if (dend == end_match_2) \ |
3075 |
|
|
goto fail; \ |
3076 |
|
|
/* End of string1 => advance to string2. */ \ |
3077 |
|
|
d = string2; \ |
3078 |
|
|
dend = end_match_2; \ |
3079 |
|
|
} |
3080 |
|
|
|
3081 |
|
|
|
3082 |
|
|
/* Test if at very beginning or at very end of the virtual concatenation |
3083 |
|
|
of `string1' and `string2'. If only one string, it's `string2'. */ |
3084 |
|
|
#define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2) |
3085 |
|
|
#define AT_STRINGS_END(d) ((d) == end2) |
3086 |
|
|
|
3087 |
|
|
|
3088 |
|
|
/* Test if D points to a character which is word-constituent. We have |
3089 |
|
|
two special cases to check for: if past the end of string1, look at |
3090 |
|
|
the first character in string2; and if before the beginning of |
3091 |
|
|
string2, look at the last character in string1. */ |
3092 |
|
|
#define WORDCHAR_P(d) \ |
3093 |
|
|
(SYNTAX ((d) == end1 ? *string2 \ |
3094 |
|
|
: (d) == string2 - 1 ? *(end1 - 1) : *(d)) \ |
3095 |
|
|
== Sword) |
3096 |
|
|
|
3097 |
|
|
/* Test if the character before D and the one at D differ with respect |
3098 |
|
|
to being word-constituent. */ |
3099 |
|
|
#define AT_WORD_BOUNDARY(d) \ |
3100 |
|
|
(AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \ |
3101 |
|
|
|| WORDCHAR_P (d - 1) != WORDCHAR_P (d)) |
3102 |
|
|
|
3103 |
|
|
|
3104 |
|
|
/* Free everything we malloc. */ |
3105 |
|
|
#ifdef REGEX_MALLOC |
3106 |
|
|
#define FREE_VAR(var) if (var) free (var); var = NULL |
3107 |
|
|
#define FREE_VARIABLES() \ |
3108 |
|
|
do { \ |
3109 |
|
|
FREE_VAR (fail_stack.stack); \ |
3110 |
|
|
FREE_VAR (regstart); \ |
3111 |
|
|
FREE_VAR (regend); \ |
3112 |
|
|
FREE_VAR (old_regstart); \ |
3113 |
|
|
FREE_VAR (old_regend); \ |
3114 |
|
|
FREE_VAR (best_regstart); \ |
3115 |
|
|
FREE_VAR (best_regend); \ |
3116 |
|
|
FREE_VAR (reg_info); \ |
3117 |
|
|
FREE_VAR (reg_dummy); \ |
3118 |
|
|
FREE_VAR (reg_info_dummy); \ |
3119 |
|
|
} while (0) |
3120 |
|
|
#else /* not REGEX_MALLOC */ |
3121 |
|
|
/* Some MIPS systems (at least) want this to free alloca'd storage. */ |
3122 |
|
|
#define FREE_VARIABLES() alloca (0) |
3123 |
|
|
#endif /* not REGEX_MALLOC */ |
3124 |
|
|
|
3125 |
|
|
|
3126 |
|
|
/* These values must meet several constraints. They must not be valid |
3127 |
|
|
register values; since we have a limit of 255 registers (because |
3128 |
|
|
we use only one byte in the pattern for the register number), we can |
3129 |
|
|
use numbers larger than 255. They must differ by 1, because of |
3130 |
|
|
NUM_FAILURE_ITEMS above. And the value for the lowest register must |
3131 |
|
|
be larger than the value for the highest register, so we do not try |
3132 |
|
|
to actually save any registers when none are active. */ |
3133 |
|
|
#define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH) |
3134 |
|
|
#define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1) |
3135 |
|
|
|
3136 |
|
|
/* Matching routines. */ |
3137 |
|
|
|
3138 |
|
|
#ifndef emacs /* Emacs never uses this. */ |
3139 |
|
|
/* re_match is like re_match_2 except it takes only a single string. */ |
3140 |
|
|
|
3141 |
|
|
int |
3142 |
|
|
re_match (bufp, string, size, pos, regs) |
3143 |
|
|
struct re_pattern_buffer *bufp; |
3144 |
|
|
const char *string; |
3145 |
|
|
int size, pos; |
3146 |
|
|
struct re_registers *regs; |
3147 |
|
|
{ |
3148 |
|
|
return re_match_2 (bufp, NULL, 0, string, size, pos, regs, size); |
3149 |
|
|
} |
3150 |
|
|
#endif /* not emacs */ |
3151 |
|
|
|
3152 |
|
|
|
3153 |
|
|
/* re_match_2 matches the compiled pattern in BUFP against the |
3154 |
|
|
the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1 |
3155 |
|
|
and SIZE2, respectively). We start matching at POS, and stop |
3156 |
|
|
matching at STOP. |
3157 |
|
|
|
3158 |
|
|
If REGS is non-null and the `no_sub' field of BUFP is nonzero, we |
3159 |
|
|
store offsets for the substring each group matched in REGS. See the |
3160 |
|
|
documentation for exactly how many groups we fill. |
3161 |
|
|
|
3162 |
|
|
We return -1 if no match, -2 if an internal error (such as the |
3163 |
|
|
failure stack overflowing). Otherwise, we return the length of the |
3164 |
|
|
matched substring. */ |
3165 |
|
|
|
3166 |
|
|
int |
3167 |
|
|
re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop) |
3168 |
|
|
struct re_pattern_buffer *bufp; |
3169 |
|
|
const char *string1, *string2; |
3170 |
|
|
int size1, size2; |
3171 |
|
|
int pos; |
3172 |
|
|
struct re_registers *regs; |
3173 |
|
|
int stop; |
3174 |
|
|
{ |
3175 |
|
|
/* General temporaries. */ |
3176 |
|
|
int mcnt; |
3177 |
|
|
unsigned char *p1; |
3178 |
|
|
|
3179 |
|
|
/* Just past the end of the corresponding string. */ |
3180 |
|
|
const char *end1, *end2; |
3181 |
|
|
|
3182 |
|
|
/* Pointers into string1 and string2, just past the last characters in |
3183 |
|
|
each to consider matching. */ |
3184 |
|
|
const char *end_match_1, *end_match_2; |
3185 |
|
|
|
3186 |
|
|
/* Where we are in the data, and the end of the current string. */ |
3187 |
|
|
const char *d, *dend; |
3188 |
|
|
|
3189 |
|
|
/* Where we are in the pattern, and the end of the pattern. */ |
3190 |
|
|
unsigned char *p = bufp->buffer; |
3191 |
|
|
register unsigned char *pend = p + bufp->used; |
3192 |
|
|
|
3193 |
|
|
/* We use this to map every character in the string. */ |
3194 |
|
|
char *translate = bufp->translate; |
3195 |
|
|
|
3196 |
|
|
/* Failure point stack. Each place that can handle a failure further |
3197 |
|
|
down the line pushes a failure point on this stack. It consists of |
3198 |
|
|
restart, regend, and reg_info for all registers corresponding to |
3199 |
|
|
the subexpressions we're currently inside, plus the number of such |
3200 |
|
|
registers, and, finally, two char *'s. The first char * is where |
3201 |
|
|
to resume scanning the pattern; the second one is where to resume |
3202 |
|
|
scanning the strings. If the latter is zero, the failure point is |
3203 |
|
|
a ``dummy''; if a failure happens and the failure point is a dummy, |
3204 |
|
|
it gets discarded and the next next one is tried. */ |
3205 |
|
|
fail_stack_type fail_stack; |
3206 |
|
|
#ifdef DEBUG |
3207 |
|
|
static unsigned failure_id = 0; |
3208 |
|
|
unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0; |
3209 |
|
|
#endif |
3210 |
|
|
|
3211 |
|
|
/* We fill all the registers internally, independent of what we |
3212 |
|
|
return, for use in backreferences. The number here includes |
3213 |
|
|
an element for register zero. */ |
3214 |
|
|
unsigned num_regs = bufp->re_nsub + 1; |
3215 |
|
|
|
3216 |
|
|
/* The currently active registers. */ |
3217 |
|
|
unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
3218 |
|
|
unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
3219 |
|
|
|
3220 |
|
|
/* Information on the contents of registers. These are pointers into |
3221 |
|
|
the input strings; they record just what was matched (on this |
3222 |
|
|
attempt) by a subexpression part of the pattern, that is, the |
3223 |
|
|
regnum-th regstart pointer points to where in the pattern we began |
3224 |
|
|
matching and the regnum-th regend points to right after where we |
3225 |
|
|
stopped matching the regnum-th subexpression. (The zeroth register |
3226 |
|
|
keeps track of what the whole pattern matches.) */ |
3227 |
|
|
const char **regstart, **regend; |
3228 |
|
|
|
3229 |
|
|
/* If a group that's operated upon by a repetition operator fails to |
3230 |
|
|
match anything, then the register for its start will need to be |
3231 |
|
|
restored because it will have been set to wherever in the string we |
3232 |
|
|
are when we last see its open-group operator. Similarly for a |
3233 |
|
|
register's end. */ |
3234 |
|
|
const char **old_regstart, **old_regend; |
3235 |
|
|
|
3236 |
|
|
/* The is_active field of reg_info helps us keep track of which (possibly |
3237 |
|
|
nested) subexpressions we are currently in. The matched_something |
3238 |
|
|
field of reg_info[reg_num] helps us tell whether or not we have |
3239 |
|
|
matched any of the pattern so far this time through the reg_num-th |
3240 |
|
|
subexpression. These two fields get reset each time through any |
3241 |
|
|
loop their register is in. */ |
3242 |
|
|
register_info_type *reg_info; |
3243 |
|
|
|
3244 |
|
|
/* The following record the register info as found in the above |
3245 |
|
|
variables when we find a match better than any we've seen before. |
3246 |
|
|
This happens as we backtrack through the failure points, which in |
3247 |
|
|
turn happens only if we have not yet matched the entire string. */ |
3248 |
|
|
unsigned best_regs_set = false; |
3249 |
|
|
const char **best_regstart, **best_regend; |
3250 |
|
|
|
3251 |
|
|
/* Logically, this is `best_regend[0]'. But we don't want to have to |
3252 |
|
|
allocate space for that if we're not allocating space for anything |
3253 |
|
|
else (see below). Also, we never need info about register 0 for |
3254 |
|
|
any of the other register vectors, and it seems rather a kludge to |
3255 |
|
|
treat `best_regend' differently than the rest. So we keep track of |
3256 |
|
|
the end of the best match so far in a separate variable. We |
3257 |
|
|
initialize this to NULL so that when we backtrack the first time |
3258 |
|
|
and need to test it, it's not garbage. */ |
3259 |
|
|
const char *match_end = NULL; |
3260 |
|
|
|
3261 |
|
|
/* Used when we pop values we don't care about. */ |
3262 |
|
|
const char **reg_dummy; |
3263 |
|
|
register_info_type *reg_info_dummy; |
3264 |
|
|
|
3265 |
|
|
#ifdef DEBUG |
3266 |
|
|
/* Counts the total number of registers pushed. */ |
3267 |
|
|
unsigned num_regs_pushed = 0; |
3268 |
|
|
#endif |
3269 |
|
|
|
3270 |
|
|
DEBUG_PRINT1 ("\n\nEntering re_match_2.\n"); |
3271 |
|
|
|
3272 |
|
|
INIT_FAIL_STACK (); |
3273 |
|
|
|
3274 |
|
|
/* Do not bother to initialize all the register variables if there are |
3275 |
|
|
no groups in the pattern, as it takes a fair amount of time. If |
3276 |
|
|
there are groups, we include space for register 0 (the whole |
3277 |
|
|
pattern), even though we never use it, since it simplifies the |
3278 |
|
|
array indexing. We should fix this. */ |
3279 |
|
|
if (bufp->re_nsub) |
3280 |
|
|
{ |
3281 |
|
|
regstart = REGEX_TALLOC (num_regs, const char *); |
3282 |
|
|
regend = REGEX_TALLOC (num_regs, const char *); |
3283 |
|
|
old_regstart = REGEX_TALLOC (num_regs, const char *); |
3284 |
|
|
old_regend = REGEX_TALLOC (num_regs, const char *); |
3285 |
|
|
best_regstart = REGEX_TALLOC (num_regs, const char *); |
3286 |
|
|
best_regend = REGEX_TALLOC (num_regs, const char *); |
3287 |
|
|
reg_info = REGEX_TALLOC (num_regs, register_info_type); |
3288 |
|
|
reg_dummy = REGEX_TALLOC (num_regs, const char *); |
3289 |
|
|
reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type); |
3290 |
|
|
|
3291 |
|
|
if (!(regstart && regend && old_regstart && old_regend && reg_info |
3292 |
|
|
&& best_regstart && best_regend && reg_dummy && reg_info_dummy)) |
3293 |
|
|
{ |
3294 |
|
|
FREE_VARIABLES (); |
3295 |
|
|
return -2; |
3296 |
|
|
} |
3297 |
|
|
} |
3298 |
|
|
#ifdef REGEX_MALLOC |
3299 |
|
|
else |
3300 |
|
|
{ |
3301 |
|
|
/* We must initialize all our variables to NULL, so that |
3302 |
|
|
`FREE_VARIABLES' doesn't try to free them. */ |
3303 |
|
|
regstart = regend = old_regstart = old_regend = best_regstart |
3304 |
|
|
= best_regend = reg_dummy = NULL; |
3305 |
|
|
reg_info = reg_info_dummy = (register_info_type *) NULL; |
3306 |
|
|
} |
3307 |
|
|
#endif /* REGEX_MALLOC */ |
3308 |
|
|
|
3309 |
|
|
/* The starting position is bogus. */ |
3310 |
|
|
if (pos < 0 || pos > size1 + size2) |
3311 |
|
|
{ |
3312 |
|
|
FREE_VARIABLES (); |
3313 |
|
|
return -1; |
3314 |
|
|
} |
3315 |
|
|
|
3316 |
|
|
/* Initialize subexpression text positions to -1 to mark ones that no |
3317 |
|
|
start_memory/stop_memory has been seen for. Also initialize the |
3318 |
|
|
register information struct. */ |
3319 |
|
|
for (mcnt = 1; mcnt < num_regs; mcnt++) |
3320 |
|
|
{ |
3321 |
|
|
regstart[mcnt] = regend[mcnt] |
3322 |
|
|
= old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE; |
3323 |
|
|
|
3324 |
|
|
REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE; |
3325 |
|
|
IS_ACTIVE (reg_info[mcnt]) = 0; |
3326 |
|
|
MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
3327 |
|
|
EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0; |
3328 |
|
|
} |
3329 |
|
|
|
3330 |
|
|
/* We move `string1' into `string2' if the latter's empty -- but not if |
3331 |
|
|
`string1' is null. */ |
3332 |
|
|
if (size2 == 0 && string1 != NULL) |
3333 |
|
|
{ |
3334 |
|
|
string2 = string1; |
3335 |
|
|
size2 = size1; |
3336 |
|
|
string1 = 0; |
3337 |
|
|
size1 = 0; |
3338 |
|
|
} |
3339 |
|
|
end1 = string1 + size1; |
3340 |
|
|
end2 = string2 + size2; |
3341 |
|
|
|
3342 |
|
|
/* Compute where to stop matching, within the two strings. */ |
3343 |
|
|
if (stop <= size1) |
3344 |
|
|
{ |
3345 |
|
|
end_match_1 = string1 + stop; |
3346 |
|
|
end_match_2 = string2; |
3347 |
|
|
} |
3348 |
|
|
else |
3349 |
|
|
{ |
3350 |
|
|
end_match_1 = end1; |
3351 |
|
|
end_match_2 = string2 + stop - size1; |
3352 |
|
|
} |
3353 |
|
|
|
3354 |
|
|
/* `p' scans through the pattern as `d' scans through the data. |
3355 |
|
|
`dend' is the end of the input string that `d' points within. `d' |
3356 |
|
|
is advanced into the following input string whenever necessary, but |
3357 |
|
|
this happens before fetching; therefore, at the beginning of the |
3358 |
|
|
loop, `d' can be pointing at the end of a string, but it cannot |
3359 |
|
|
equal `string2'. */ |
3360 |
|
|
if (size1 > 0 && pos <= size1) |
3361 |
|
|
{ |
3362 |
|
|
d = string1 + pos; |
3363 |
|
|
dend = end_match_1; |
3364 |
|
|
} |
3365 |
|
|
else |
3366 |
|
|
{ |
3367 |
|
|
d = string2 + pos - size1; |
3368 |
|
|
dend = end_match_2; |
3369 |
|
|
} |
3370 |
|
|
|
3371 |
|
|
DEBUG_PRINT1 ("The compiled pattern is: "); |
3372 |
|
|
DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend); |
3373 |
|
|
DEBUG_PRINT1 ("The string to match is: `"); |
3374 |
|
|
DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2); |
3375 |
|
|
DEBUG_PRINT1 ("'\n"); |
3376 |
|
|
|
3377 |
|
|
/* This loops over pattern commands. It exits by returning from the |
3378 |
|
|
function if the match is complete, or it drops through if the match |
3379 |
|
|
fails at this starting point in the input data. */ |
3380 |
|
|
for (;;) |
3381 |
|
|
{ |
3382 |
|
|
DEBUG_PRINT2 ("\n0x%x: ", p); |
3383 |
|
|
|
3384 |
|
|
if (p == pend) |
3385 |
|
|
{ /* End of pattern means we might have succeeded. */ |
3386 |
|
|
DEBUG_PRINT1 ("end of pattern ... "); |
3387 |
|
|
|
3388 |
|
|
/* If we haven't matched the entire string, and we want the |
3389 |
|
|
longest match, try backtracking. */ |
3390 |
|
|
if (d != end_match_2) |
3391 |
|
|
{ |
3392 |
|
|
DEBUG_PRINT1 ("backtracking.\n"); |
3393 |
|
|
|
3394 |
|
|
if (!FAIL_STACK_EMPTY ()) |
3395 |
|
|
{ /* More failure points to try. */ |
3396 |
|
|
boolean same_str_p = (FIRST_STRING_P (match_end) |
3397 |
|
|
== MATCHING_IN_FIRST_STRING); |
3398 |
|
|
|
3399 |
|
|
/* If exceeds best match so far, save it. */ |
3400 |
|
|
if (!best_regs_set |
3401 |
|
|
|| (same_str_p && d > match_end) |
3402 |
|
|
|| (!same_str_p && !MATCHING_IN_FIRST_STRING)) |
3403 |
|
|
{ |
3404 |
|
|
best_regs_set = true; |
3405 |
|
|
match_end = d; |
3406 |
|
|
|
3407 |
|
|
DEBUG_PRINT1 ("\nSAVING match as best so far.\n"); |
3408 |
|
|
|
3409 |
|
|
for (mcnt = 1; mcnt < num_regs; mcnt++) |
3410 |
|
|
{ |
3411 |
|
|
best_regstart[mcnt] = regstart[mcnt]; |
3412 |
|
|
best_regend[mcnt] = regend[mcnt]; |
3413 |
|
|
} |
3414 |
|
|
} |
3415 |
|
|
goto fail; |
3416 |
|
|
} |
3417 |
|
|
|
3418 |
|
|
/* If no failure points, don't restore garbage. */ |
3419 |
|
|
else if (best_regs_set) |
3420 |
|
|
{ |
3421 |
|
|
restore_best_regs: |
3422 |
|
|
/* Restore best match. It may happen that `dend == |
3423 |
|
|
end_match_1' while the restored d is in string2. |
3424 |
|
|
For example, the pattern `x.*y.*z' against the |
3425 |
|
|
strings `x-' and `y-z-', if the two strings are |
3426 |
|
|
not consecutive in memory. */ |
3427 |
|
|
DEBUG_PRINT1 ("Restoring best registers.\n"); |
3428 |
|
|
|
3429 |
|
|
d = match_end; |
3430 |
|
|
dend = ((d >= string1 && d <= end1) |
3431 |
|
|
? end_match_1 : end_match_2); |
3432 |
|
|
|
3433 |
|
|
for (mcnt = 1; mcnt < num_regs; mcnt++) |
3434 |
|
|
{ |
3435 |
|
|
regstart[mcnt] = best_regstart[mcnt]; |
3436 |
|
|
regend[mcnt] = best_regend[mcnt]; |
3437 |
|
|
} |
3438 |
|
|
} |
3439 |
|
|
} /* d != end_match_2 */ |
3440 |
|
|
|
3441 |
|
|
DEBUG_PRINT1 ("Accepting match.\n"); |
3442 |
|
|
|
3443 |
|
|
/* If caller wants register contents data back, do it. */ |
3444 |
|
|
if (regs && !bufp->no_sub) |
3445 |
|
|
{ |
3446 |
|
|
/* Have the register data arrays been allocated? */ |
3447 |
|
|
if (bufp->regs_allocated == REGS_UNALLOCATED) |
3448 |
|
|
{ /* No. So allocate them with malloc. We need one |
3449 |
|
|
extra element beyond `num_regs' for the `-1' marker |
3450 |
|
|
GNU code uses. */ |
3451 |
|
|
regs->num_regs = MAX (RE_NREGS, num_regs + 1); |
3452 |
|
|
regs->start = TALLOC (regs->num_regs, regoff_t); |
3453 |
|
|
regs->end = TALLOC (regs->num_regs, regoff_t); |
3454 |
|
|
if (regs->start == NULL || regs->end == NULL) |
3455 |
|
|
return -2; |
3456 |
|
|
bufp->regs_allocated = REGS_REALLOCATE; |
3457 |
|
|
} |
3458 |
|
|
else if (bufp->regs_allocated == REGS_REALLOCATE) |
3459 |
|
|
{ /* Yes. If we need more elements than were already |
3460 |
|
|
allocated, reallocate them. If we need fewer, just |
3461 |
|
|
leave it alone. */ |
3462 |
|
|
if (regs->num_regs < num_regs + 1) |
3463 |
|
|
{ |
3464 |
|
|
regs->num_regs = num_regs + 1; |
3465 |
|
|
RETALLOC (regs->start, regs->num_regs, regoff_t); |
3466 |
|
|
RETALLOC (regs->end, regs->num_regs, regoff_t); |
3467 |
|
|
if (regs->start == NULL || regs->end == NULL) |
3468 |
|
|
return -2; |
3469 |
|
|
} |
3470 |
|
|
} |
3471 |
|
|
else |
3472 |
|
|
assert (bufp->regs_allocated == REGS_FIXED); |
3473 |
|
|
|
3474 |
|
|
/* Convert the pointer data in `regstart' and `regend' to |
3475 |
|
|
indices. Register zero has to be set differently, |
3476 |
|
|
since we haven't kept track of any info for it. */ |
3477 |
|
|
if (regs->num_regs > 0) |
3478 |
|
|
{ |
3479 |
|
|
regs->start[0] = pos; |
3480 |
|
|
regs->end[0] = (MATCHING_IN_FIRST_STRING ? d - string1 |
3481 |
|
|
: d - string2 + size1); |
3482 |
|
|
} |
3483 |
|
|
|
3484 |
|
|
/* Go through the first `min (num_regs, regs->num_regs)' |
3485 |
|
|
registers, since that is all we initialized. */ |
3486 |
|
|
for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++) |
3487 |
|
|
{ |
3488 |
|
|
if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt])) |
3489 |
|
|
regs->start[mcnt] = regs->end[mcnt] = -1; |
3490 |
|
|
else |
3491 |
|
|
{ |
3492 |
|
|
regs->start[mcnt] = POINTER_TO_OFFSET (regstart[mcnt]); |
3493 |
|
|
regs->end[mcnt] = POINTER_TO_OFFSET (regend[mcnt]); |
3494 |
|
|
} |
3495 |
|
|
} |
3496 |
|
|
|
3497 |
|
|
/* If the regs structure we return has more elements than |
3498 |
|
|
were in the pattern, set the extra elements to -1. If |
3499 |
|
|
we (re)allocated the registers, this is the case, |
3500 |
|
|
because we always allocate enough to have at least one |
3501 |
|
|
-1 at the end. */ |
3502 |
|
|
for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++) |
3503 |
|
|
regs->start[mcnt] = regs->end[mcnt] = -1; |
3504 |
|
|
} /* regs && !bufp->no_sub */ |
3505 |
|
|
|
3506 |
|
|
FREE_VARIABLES (); |
3507 |
|
|
DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n", |
3508 |
|
|
nfailure_points_pushed, nfailure_points_popped, |
3509 |
|
|
nfailure_points_pushed - nfailure_points_popped); |
3510 |
|
|
DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed); |
3511 |
|
|
|
3512 |
|
|
mcnt = d - pos - (MATCHING_IN_FIRST_STRING |
3513 |
|
|
? string1 |
3514 |
|
|
: string2 - size1); |
3515 |
|
|
|
3516 |
|
|
DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt); |
3517 |
|
|
|
3518 |
|
|
return mcnt; |
3519 |
|
|
} |
3520 |
|
|
|
3521 |
|
|
/* Otherwise match next pattern command. */ |
3522 |
|
|
#ifdef SWITCH_ENUM_BUG |
3523 |
|
|
switch ((int) ((re_opcode_t) *p++)) |
3524 |
|
|
#else |
3525 |
|
|
switch ((re_opcode_t) *p++) |
3526 |
|
|
#endif |
3527 |
|
|
{ |
3528 |
|
|
/* Ignore these. Used to ignore the n of succeed_n's which |
3529 |
|
|
currently have n == 0. */ |
3530 |
|
|
case no_op: |
3531 |
|
|
DEBUG_PRINT1 ("EXECUTING no_op.\n"); |
3532 |
|
|
break; |
3533 |
|
|
|
3534 |
|
|
|
3535 |
|
|
/* Match the next n pattern characters exactly. The following |
3536 |
|
|
byte in the pattern defines n, and the n bytes after that |
3537 |
|
|
are the characters to match. */ |
3538 |
|
|
case exactn: |
3539 |
|
|
mcnt = *p++; |
3540 |
|
|
DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt); |
3541 |
|
|
|
3542 |
|
|
/* This is written out as an if-else so we don't waste time |
3543 |
|
|
testing `translate' inside the loop. */ |
3544 |
|
|
if (translate) |
3545 |
|
|
{ |
3546 |
|
|
do |
3547 |
|
|
{ |
3548 |
|
|
PREFETCH (); |
3549 |
|
|
if (translate[(unsigned char) *d++] != (char) *p++) |
3550 |
|
|
goto fail; |
3551 |
|
|
} |
3552 |
|
|
while (--mcnt); |
3553 |
|
|
} |
3554 |
|
|
else |
3555 |
|
|
{ |
3556 |
|
|
do |
3557 |
|
|
{ |
3558 |
|
|
PREFETCH (); |
3559 |
|
|
if (*d++ != (char) *p++) goto fail; |
3560 |
|
|
} |
3561 |
|
|
while (--mcnt); |
3562 |
|
|
} |
3563 |
|
|
SET_REGS_MATCHED (); |
3564 |
|
|
break; |
3565 |
|
|
|
3566 |
|
|
|
3567 |
|
|
/* Match any character except possibly a newline or a null. */ |
3568 |
|
|
case anychar: |
3569 |
|
|
DEBUG_PRINT1 ("EXECUTING anychar.\n"); |
3570 |
|
|
|
3571 |
|
|
PREFETCH (); |
3572 |
|
|
|
3573 |
|
|
if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n') |
3574 |
|
|
|| (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000')) |
3575 |
|
|
goto fail; |
3576 |
|
|
|
3577 |
|
|
SET_REGS_MATCHED (); |
3578 |
|
|
DEBUG_PRINT2 (" Matched `%d'.\n", *d); |
3579 |
|
|
d++; |
3580 |
|
|
break; |
3581 |
|
|
|
3582 |
|
|
|
3583 |
|
|
case charset: |
3584 |
|
|
case charset_not: |
3585 |
|
|
{ |
3586 |
|
|
register unsigned char c; |
3587 |
|
|
boolean not = (re_opcode_t) *(p - 1) == charset_not; |
3588 |
|
|
|
3589 |
|
|
DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : ""); |
3590 |
|
|
|
3591 |
|
|
PREFETCH (); |
3592 |
|
|
c = TRANSLATE (*d); /* The character to match. */ |
3593 |
|
|
|
3594 |
|
|
/* Cast to `unsigned' instead of `unsigned char' in case the |
3595 |
|
|
bit list is a full 32 bytes long. */ |
3596 |
|
|
if (c < (unsigned) (*p * BYTEWIDTH) |
3597 |
|
|
&& p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
3598 |
|
|
not = !not; |
3599 |
|
|
|
3600 |
|
|
p += 1 + *p; |
3601 |
|
|
|
3602 |
|
|
if (!not) goto fail; |
3603 |
|
|
|
3604 |
|
|
SET_REGS_MATCHED (); |
3605 |
|
|
d++; |
3606 |
|
|
break; |
3607 |
|
|
} |
3608 |
|
|
|
3609 |
|
|
|
3610 |
|
|
/* The beginning of a group is represented by start_memory. |
3611 |
|
|
The arguments are the register number in the next byte, and the |
3612 |
|
|
number of groups inner to this one in the next. The text |
3613 |
|
|
matched within the group is recorded (in the internal |
3614 |
|
|
registers data structure) under the register number. */ |
3615 |
|
|
case start_memory: |
3616 |
|
|
DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]); |
3617 |
|
|
|
3618 |
|
|
/* Find out if this group can match the empty string. */ |
3619 |
|
|
p1 = p; /* To send to group_match_null_string_p. */ |
3620 |
|
|
|
3621 |
|
|
if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE) |
3622 |
|
|
REG_MATCH_NULL_STRING_P (reg_info[*p]) |
3623 |
|
|
= group_match_null_string_p (&p1, pend, reg_info); |
3624 |
|
|
|
3625 |
|
|
/* Save the position in the string where we were the last time |
3626 |
|
|
we were at this open-group operator in case the group is |
3627 |
|
|
operated upon by a repetition operator, e.g., with `(a*)*b' |
3628 |
|
|
against `ab'; then we want to ignore where we are now in |
3629 |
|
|
the string in case this attempt to match fails. */ |
3630 |
|
|
old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
3631 |
|
|
? REG_UNSET (regstart[*p]) ? d : regstart[*p] |
3632 |
|
|
: regstart[*p]; |
3633 |
|
|
DEBUG_PRINT2 (" old_regstart: %d\n", |
3634 |
|
|
POINTER_TO_OFFSET (old_regstart[*p])); |
3635 |
|
|
|
3636 |
|
|
regstart[*p] = d; |
3637 |
|
|
DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p])); |
3638 |
|
|
|
3639 |
|
|
IS_ACTIVE (reg_info[*p]) = 1; |
3640 |
|
|
MATCHED_SOMETHING (reg_info[*p]) = 0; |
3641 |
|
|
|
3642 |
|
|
/* This is the new highest active register. */ |
3643 |
|
|
highest_active_reg = *p; |
3644 |
|
|
|
3645 |
|
|
/* If nothing was active before, this is the new lowest active |
3646 |
|
|
register. */ |
3647 |
|
|
if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
3648 |
|
|
lowest_active_reg = *p; |
3649 |
|
|
|
3650 |
|
|
/* Move past the register number and inner group count. */ |
3651 |
|
|
p += 2; |
3652 |
|
|
break; |
3653 |
|
|
|
3654 |
|
|
|
3655 |
|
|
/* The stop_memory opcode represents the end of a group. Its |
3656 |
|
|
arguments are the same as start_memory's: the register |
3657 |
|
|
number, and the number of inner groups. */ |
3658 |
|
|
case stop_memory: |
3659 |
|
|
DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]); |
3660 |
|
|
|
3661 |
|
|
/* We need to save the string position the last time we were at |
3662 |
|
|
this close-group operator in case the group is operated |
3663 |
|
|
upon by a repetition operator, e.g., with `((a*)*(b*)*)*' |
3664 |
|
|
against `aba'; then we want to ignore where we are now in |
3665 |
|
|
the string in case this attempt to match fails. */ |
3666 |
|
|
old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p]) |
3667 |
|
|
? REG_UNSET (regend[*p]) ? d : regend[*p] |
3668 |
|
|
: regend[*p]; |
3669 |
|
|
DEBUG_PRINT2 (" old_regend: %d\n", |
3670 |
|
|
POINTER_TO_OFFSET (old_regend[*p])); |
3671 |
|
|
|
3672 |
|
|
regend[*p] = d; |
3673 |
|
|
DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p])); |
3674 |
|
|
|
3675 |
|
|
/* This register isn't active anymore. */ |
3676 |
|
|
IS_ACTIVE (reg_info[*p]) = 0; |
3677 |
|
|
|
3678 |
|
|
/* If this was the only register active, nothing is active |
3679 |
|
|
anymore. */ |
3680 |
|
|
if (lowest_active_reg == highest_active_reg) |
3681 |
|
|
{ |
3682 |
|
|
lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
3683 |
|
|
highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
3684 |
|
|
} |
3685 |
|
|
else |
3686 |
|
|
{ /* We must scan for the new highest active register, since |
3687 |
|
|
it isn't necessarily one less than now: consider |
3688 |
|
|
(a(b)c(d(e)f)g). When group 3 ends, after the f), the |
3689 |
|
|
new highest active register is 1. */ |
3690 |
|
|
unsigned char r = *p - 1; |
3691 |
|
|
while (r > 0 && !IS_ACTIVE (reg_info[r])) |
3692 |
|
|
r--; |
3693 |
|
|
|
3694 |
|
|
/* If we end up at register zero, that means that we saved |
3695 |
|
|
the registers as the result of an `on_failure_jump', not |
3696 |
|
|
a `start_memory', and we jumped to past the innermost |
3697 |
|
|
`stop_memory'. For example, in ((.)*) we save |
3698 |
|
|
registers 1 and 2 as a result of the *, but when we pop |
3699 |
|
|
back to the second ), we are at the stop_memory 1. |
3700 |
|
|
Thus, nothing is active. */ |
3701 |
|
|
if (r == 0) |
3702 |
|
|
{ |
3703 |
|
|
lowest_active_reg = NO_LOWEST_ACTIVE_REG; |
3704 |
|
|
highest_active_reg = NO_HIGHEST_ACTIVE_REG; |
3705 |
|
|
} |
3706 |
|
|
else |
3707 |
|
|
highest_active_reg = r; |
3708 |
|
|
} |
3709 |
|
|
|
3710 |
|
|
/* If just failed to match something this time around with a |
3711 |
|
|
group that's operated on by a repetition operator, try to |
3712 |
|
|
force exit from the ``loop'', and restore the register |
3713 |
|
|
information for this group that we had before trying this |
3714 |
|
|
last match. */ |
3715 |
|
|
if ((!MATCHED_SOMETHING (reg_info[*p]) |
3716 |
|
|
|| (re_opcode_t) p[-3] == start_memory) |
3717 |
|
|
&& (p + 2) < pend) |
3718 |
|
|
{ |
3719 |
|
|
boolean is_a_jump_n = false; |
3720 |
|
|
|
3721 |
|
|
p1 = p + 2; |
3722 |
|
|
mcnt = 0; |
3723 |
|
|
switch ((re_opcode_t) *p1++) |
3724 |
|
|
{ |
3725 |
|
|
case jump_n: |
3726 |
|
|
is_a_jump_n = true; |
3727 |
|
|
case pop_failure_jump: |
3728 |
|
|
case maybe_pop_jump: |
3729 |
|
|
case jump: |
3730 |
|
|
case dummy_failure_jump: |
3731 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
3732 |
|
|
if (is_a_jump_n) |
3733 |
|
|
p1 += 2; |
3734 |
|
|
break; |
3735 |
|
|
|
3736 |
|
|
default: |
3737 |
|
|
/* do nothing */ ; |
3738 |
|
|
} |
3739 |
|
|
p1 += mcnt; |
3740 |
|
|
|
3741 |
|
|
/* If the next operation is a jump backwards in the pattern |
3742 |
|
|
to an on_failure_jump right before the start_memory |
3743 |
|
|
corresponding to this stop_memory, exit from the loop |
3744 |
|
|
by forcing a failure after pushing on the stack the |
3745 |
|
|
on_failure_jump's jump in the pattern, and d. */ |
3746 |
|
|
if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump |
3747 |
|
|
&& (re_opcode_t) p1[3] == start_memory && p1[4] == *p) |
3748 |
|
|
{ |
3749 |
|
|
/* If this group ever matched anything, then restore |
3750 |
|
|
what its registers were before trying this last |
3751 |
|
|
failed match, e.g., with `(a*)*b' against `ab' for |
3752 |
|
|
regstart[1], and, e.g., with `((a*)*(b*)*)*' |
3753 |
|
|
against `aba' for regend[3]. |
3754 |
|
|
|
3755 |
|
|
Also restore the registers for inner groups for, |
3756 |
|
|
e.g., `((a*)(b*))*' against `aba' (register 3 would |
3757 |
|
|
otherwise get trashed). */ |
3758 |
|
|
|
3759 |
|
|
if (EVER_MATCHED_SOMETHING (reg_info[*p])) |
3760 |
|
|
{ |
3761 |
|
|
unsigned r; |
3762 |
|
|
|
3763 |
|
|
EVER_MATCHED_SOMETHING (reg_info[*p]) = 0; |
3764 |
|
|
|
3765 |
|
|
/* Restore this and inner groups' (if any) registers. */ |
3766 |
|
|
for (r = *p; r < *p + *(p + 1); r++) |
3767 |
|
|
{ |
3768 |
|
|
regstart[r] = old_regstart[r]; |
3769 |
|
|
|
3770 |
|
|
/* xx why this test? */ |
3771 |
|
|
if ((int) old_regend[r] >= (int) regstart[r]) |
3772 |
|
|
regend[r] = old_regend[r]; |
3773 |
|
|
} |
3774 |
|
|
} |
3775 |
|
|
p1++; |
3776 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
3777 |
|
|
PUSH_FAILURE_POINT (p1 + mcnt, d, -2); |
3778 |
|
|
|
3779 |
|
|
goto fail; |
3780 |
|
|
} |
3781 |
|
|
} |
3782 |
|
|
|
3783 |
|
|
/* Move past the register number and the inner group count. */ |
3784 |
|
|
p += 2; |
3785 |
|
|
break; |
3786 |
|
|
|
3787 |
|
|
|
3788 |
|
|
/* \<digit> has been turned into a `duplicate' command which is |
3789 |
|
|
followed by the numeric value of <digit> as the register number. */ |
3790 |
|
|
case duplicate: |
3791 |
|
|
{ |
3792 |
|
|
register const char *d2, *dend2; |
3793 |
|
|
int regno = *p++; /* Get which register to match against. */ |
3794 |
|
|
DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno); |
3795 |
|
|
|
3796 |
|
|
/* Can't back reference a group which we've never matched. */ |
3797 |
|
|
if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno])) |
3798 |
|
|
goto fail; |
3799 |
|
|
|
3800 |
|
|
/* Where in input to try to start matching. */ |
3801 |
|
|
d2 = regstart[regno]; |
3802 |
|
|
|
3803 |
|
|
/* Where to stop matching; if both the place to start and |
3804 |
|
|
the place to stop matching are in the same string, then |
3805 |
|
|
set to the place to stop, otherwise, for now have to use |
3806 |
|
|
the end of the first string. */ |
3807 |
|
|
|
3808 |
|
|
dend2 = ((FIRST_STRING_P (regstart[regno]) |
3809 |
|
|
== FIRST_STRING_P (regend[regno])) |
3810 |
|
|
? regend[regno] : end_match_1); |
3811 |
|
|
for (;;) |
3812 |
|
|
{ |
3813 |
|
|
/* If necessary, advance to next segment in register |
3814 |
|
|
contents. */ |
3815 |
|
|
while (d2 == dend2) |
3816 |
|
|
{ |
3817 |
|
|
if (dend2 == end_match_2) break; |
3818 |
|
|
if (dend2 == regend[regno]) break; |
3819 |
|
|
|
3820 |
|
|
/* End of string1 => advance to string2. */ |
3821 |
|
|
d2 = string2; |
3822 |
|
|
dend2 = regend[regno]; |
3823 |
|
|
} |
3824 |
|
|
/* At end of register contents => success */ |
3825 |
|
|
if (d2 == dend2) break; |
3826 |
|
|
|
3827 |
|
|
/* If necessary, advance to next segment in data. */ |
3828 |
|
|
PREFETCH (); |
3829 |
|
|
|
3830 |
|
|
/* How many characters left in this segment to match. */ |
3831 |
|
|
mcnt = dend - d; |
3832 |
|
|
|
3833 |
|
|
/* Want how many consecutive characters we can match in |
3834 |
|
|
one shot, so, if necessary, adjust the count. */ |
3835 |
|
|
if (mcnt > dend2 - d2) |
3836 |
|
|
mcnt = dend2 - d2; |
3837 |
|
|
|
3838 |
|
|
/* Compare that many; failure if mismatch, else move |
3839 |
|
|
past them. */ |
3840 |
|
|
if (translate |
3841 |
|
|
? bcmp_translate (d, d2, mcnt, translate) |
3842 |
|
|
: bcmp (d, d2, mcnt)) |
3843 |
|
|
goto fail; |
3844 |
|
|
d += mcnt, d2 += mcnt; |
3845 |
|
|
} |
3846 |
|
|
} |
3847 |
|
|
break; |
3848 |
|
|
|
3849 |
|
|
|
3850 |
|
|
/* begline matches the empty string at the beginning of the string |
3851 |
|
|
(unless `not_bol' is set in `bufp'), and, if |
3852 |
|
|
`newline_anchor' is set, after newlines. */ |
3853 |
|
|
case begline: |
3854 |
|
|
DEBUG_PRINT1 ("EXECUTING begline.\n"); |
3855 |
|
|
|
3856 |
|
|
if (AT_STRINGS_BEG (d)) |
3857 |
|
|
{ |
3858 |
|
|
if (!bufp->not_bol) break; |
3859 |
|
|
} |
3860 |
|
|
else if (d[-1] == '\n' && bufp->newline_anchor) |
3861 |
|
|
{ |
3862 |
|
|
break; |
3863 |
|
|
} |
3864 |
|
|
/* In all other cases, we fail. */ |
3865 |
|
|
goto fail; |
3866 |
|
|
|
3867 |
|
|
|
3868 |
|
|
/* endline is the dual of begline. */ |
3869 |
|
|
case endline: |
3870 |
|
|
DEBUG_PRINT1 ("EXECUTING endline.\n"); |
3871 |
|
|
|
3872 |
|
|
if (AT_STRINGS_END (d)) |
3873 |
|
|
{ |
3874 |
|
|
if (!bufp->not_eol) break; |
3875 |
|
|
} |
3876 |
|
|
|
3877 |
|
|
/* We have to ``prefetch'' the next character. */ |
3878 |
|
|
else if ((d == end1 ? *string2 : *d) == '\n' |
3879 |
|
|
&& bufp->newline_anchor) |
3880 |
|
|
{ |
3881 |
|
|
break; |
3882 |
|
|
} |
3883 |
|
|
goto fail; |
3884 |
|
|
|
3885 |
|
|
|
3886 |
|
|
/* Match at the very beginning of the data. */ |
3887 |
|
|
case begbuf: |
3888 |
|
|
DEBUG_PRINT1 ("EXECUTING begbuf.\n"); |
3889 |
|
|
if (AT_STRINGS_BEG (d)) |
3890 |
|
|
break; |
3891 |
|
|
goto fail; |
3892 |
|
|
|
3893 |
|
|
|
3894 |
|
|
/* Match at the very end of the data. */ |
3895 |
|
|
case endbuf: |
3896 |
|
|
DEBUG_PRINT1 ("EXECUTING endbuf.\n"); |
3897 |
|
|
if (AT_STRINGS_END (d)) |
3898 |
|
|
break; |
3899 |
|
|
goto fail; |
3900 |
|
|
|
3901 |
|
|
|
3902 |
|
|
/* on_failure_keep_string_jump is used to optimize `.*\n'. It |
3903 |
|
|
pushes NULL as the value for the string on the stack. Then |
3904 |
|
|
`pop_failure_point' will keep the current value for the |
3905 |
|
|
string, instead of restoring it. To see why, consider |
3906 |
|
|
matching `foo\nbar' against `.*\n'. The .* matches the foo; |
3907 |
|
|
then the . fails against the \n. But the next thing we want |
3908 |
|
|
to do is match the \n against the \n; if we restored the |
3909 |
|
|
string value, we would be back at the foo. |
3910 |
|
|
|
3911 |
|
|
Because this is used only in specific cases, we don't need to |
3912 |
|
|
check all the things that `on_failure_jump' does, to make |
3913 |
|
|
sure the right things get saved on the stack. Hence we don't |
3914 |
|
|
share its code. The only reason to push anything on the |
3915 |
|
|
stack at all is that otherwise we would have to change |
3916 |
|
|
`anychar's code to do something besides goto fail in this |
3917 |
|
|
case; that seems worse than this. */ |
3918 |
|
|
case on_failure_keep_string_jump: |
3919 |
|
|
DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump"); |
3920 |
|
|
|
3921 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
3922 |
|
|
DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt); |
3923 |
|
|
|
3924 |
|
|
PUSH_FAILURE_POINT (p + mcnt, NULL, -2); |
3925 |
|
|
break; |
3926 |
|
|
|
3927 |
|
|
|
3928 |
|
|
/* Uses of on_failure_jump: |
3929 |
|
|
|
3930 |
|
|
Each alternative starts with an on_failure_jump that points |
3931 |
|
|
to the beginning of the next alternative. Each alternative |
3932 |
|
|
except the last ends with a jump that in effect jumps past |
3933 |
|
|
the rest of the alternatives. (They really jump to the |
3934 |
|
|
ending jump of the following alternative, because tensioning |
3935 |
|
|
these jumps is a hassle.) |
3936 |
|
|
|
3937 |
|
|
Repeats start with an on_failure_jump that points past both |
3938 |
|
|
the repetition text and either the following jump or |
3939 |
|
|
pop_failure_jump back to this on_failure_jump. */ |
3940 |
|
|
case on_failure_jump: |
3941 |
|
|
on_failure: |
3942 |
|
|
DEBUG_PRINT1 ("EXECUTING on_failure_jump"); |
3943 |
|
|
|
3944 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
3945 |
|
|
DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt); |
3946 |
|
|
|
3947 |
|
|
/* If this on_failure_jump comes right before a group (i.e., |
3948 |
|
|
the original * applied to a group), save the information |
3949 |
|
|
for that group and all inner ones, so that if we fail back |
3950 |
|
|
to this point, the group's information will be correct. |
3951 |
|
|
For example, in \(a*\)*\1, we need the preceding group, |
3952 |
|
|
and in \(\(a*\)b*\)\2, we need the inner group. */ |
3953 |
|
|
|
3954 |
|
|
/* We can't use `p' to check ahead because we push |
3955 |
|
|
a failure point to `p + mcnt' after we do this. */ |
3956 |
|
|
p1 = p; |
3957 |
|
|
|
3958 |
|
|
/* We need to skip no_op's before we look for the |
3959 |
|
|
start_memory in case this on_failure_jump is happening as |
3960 |
|
|
the result of a completed succeed_n, as in \(a\)\{1,3\}b\1 |
3961 |
|
|
against aba. */ |
3962 |
|
|
while (p1 < pend && (re_opcode_t) *p1 == no_op) |
3963 |
|
|
p1++; |
3964 |
|
|
|
3965 |
|
|
if (p1 < pend && (re_opcode_t) *p1 == start_memory) |
3966 |
|
|
{ |
3967 |
|
|
/* We have a new highest active register now. This will |
3968 |
|
|
get reset at the start_memory we are about to get to, |
3969 |
|
|
but we will have saved all the registers relevant to |
3970 |
|
|
this repetition op, as described above. */ |
3971 |
|
|
highest_active_reg = *(p1 + 1) + *(p1 + 2); |
3972 |
|
|
if (lowest_active_reg == NO_LOWEST_ACTIVE_REG) |
3973 |
|
|
lowest_active_reg = *(p1 + 1); |
3974 |
|
|
} |
3975 |
|
|
|
3976 |
|
|
DEBUG_PRINT1 (":\n"); |
3977 |
|
|
PUSH_FAILURE_POINT (p + mcnt, d, -2); |
3978 |
|
|
break; |
3979 |
|
|
|
3980 |
|
|
|
3981 |
|
|
/* A smart repeat ends with `maybe_pop_jump'. |
3982 |
|
|
We change it to either `pop_failure_jump' or `jump'. */ |
3983 |
|
|
case maybe_pop_jump: |
3984 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
3985 |
|
|
DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt); |
3986 |
|
|
{ |
3987 |
|
|
register unsigned char *p2 = p; |
3988 |
|
|
|
3989 |
|
|
/* Compare the beginning of the repeat with what in the |
3990 |
|
|
pattern follows its end. If we can establish that there |
3991 |
|
|
is nothing that they would both match, i.e., that we |
3992 |
|
|
would have to backtrack because of (as in, e.g., `a*a') |
3993 |
|
|
then we can change to pop_failure_jump, because we'll |
3994 |
|
|
never have to backtrack. |
3995 |
|
|
|
3996 |
|
|
This is not true in the case of alternatives: in |
3997 |
|
|
`(a|ab)*' we do need to backtrack to the `ab' alternative |
3998 |
|
|
(e.g., if the string was `ab'). But instead of trying to |
3999 |
|
|
detect that here, the alternative has put on a dummy |
4000 |
|
|
failure point which is what we will end up popping. */ |
4001 |
|
|
|
4002 |
|
|
/* Skip over open/close-group commands. */ |
4003 |
|
|
while (p2 + 2 < pend |
4004 |
|
|
&& ((re_opcode_t) *p2 == stop_memory |
4005 |
|
|
|| (re_opcode_t) *p2 == start_memory)) |
4006 |
|
|
p2 += 3; /* Skip over args, too. */ |
4007 |
|
|
|
4008 |
|
|
/* If we're at the end of the pattern, we can change. */ |
4009 |
|
|
if (p2 == pend) |
4010 |
|
|
{ |
4011 |
|
|
/* Consider what happens when matching ":\(.*\)" |
4012 |
|
|
against ":/". I don't really understand this code |
4013 |
|
|
yet. */ |
4014 |
|
|
p[-3] = (unsigned char) pop_failure_jump; |
4015 |
|
|
DEBUG_PRINT1 |
4016 |
|
|
(" End of pattern: change to `pop_failure_jump'.\n"); |
4017 |
|
|
} |
4018 |
|
|
|
4019 |
|
|
else if ((re_opcode_t) *p2 == exactn |
4020 |
|
|
|| (bufp->newline_anchor && (re_opcode_t) *p2 == endline)) |
4021 |
|
|
{ |
4022 |
|
|
register unsigned char c |
4023 |
|
|
= *p2 == (unsigned char) endline ? '\n' : p2[2]; |
4024 |
|
|
p1 = p + mcnt; |
4025 |
|
|
|
4026 |
|
|
/* p1[0] ... p1[2] are the `on_failure_jump' corresponding |
4027 |
|
|
to the `maybe_finalize_jump' of this case. Examine what |
4028 |
|
|
follows. */ |
4029 |
|
|
if ((re_opcode_t) p1[3] == exactn && p1[5] != c) |
4030 |
|
|
{ |
4031 |
|
|
p[-3] = (unsigned char) pop_failure_jump; |
4032 |
|
|
DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n", |
4033 |
|
|
c, p1[5]); |
4034 |
|
|
} |
4035 |
|
|
|
4036 |
|
|
else if ((re_opcode_t) p1[3] == charset |
4037 |
|
|
|| (re_opcode_t) p1[3] == charset_not) |
4038 |
|
|
{ |
4039 |
|
|
int not = (re_opcode_t) p1[3] == charset_not; |
4040 |
|
|
|
4041 |
|
|
if (c < (unsigned char) (p1[4] * BYTEWIDTH) |
4042 |
|
|
&& p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH))) |
4043 |
|
|
not = !not; |
4044 |
|
|
|
4045 |
|
|
/* `not' is equal to 1 if c would match, which means |
4046 |
|
|
that we can't change to pop_failure_jump. */ |
4047 |
|
|
if (!not) |
4048 |
|
|
{ |
4049 |
|
|
p[-3] = (unsigned char) pop_failure_jump; |
4050 |
|
|
DEBUG_PRINT1 (" No match => pop_failure_jump.\n"); |
4051 |
|
|
} |
4052 |
|
|
} |
4053 |
|
|
} |
4054 |
|
|
} |
4055 |
|
|
p -= 2; /* Point at relative address again. */ |
4056 |
|
|
if ((re_opcode_t) p[-1] != pop_failure_jump) |
4057 |
|
|
{ |
4058 |
|
|
p[-1] = (unsigned char) jump; |
4059 |
|
|
DEBUG_PRINT1 (" Match => jump.\n"); |
4060 |
|
|
goto unconditional_jump; |
4061 |
|
|
} |
4062 |
|
|
/* Note fall through. */ |
4063 |
|
|
|
4064 |
|
|
|
4065 |
|
|
/* The end of a simple repeat has a pop_failure_jump back to |
4066 |
|
|
its matching on_failure_jump, where the latter will push a |
4067 |
|
|
failure point. The pop_failure_jump takes off failure |
4068 |
|
|
points put on by this pop_failure_jump's matching |
4069 |
|
|
on_failure_jump; we got through the pattern to here from the |
4070 |
|
|
matching on_failure_jump, so didn't fail. */ |
4071 |
|
|
case pop_failure_jump: |
4072 |
|
|
{ |
4073 |
|
|
/* We need to pass separate storage for the lowest and |
4074 |
|
|
highest registers, even though we don't care about the |
4075 |
|
|
actual values. Otherwise, we will restore only one |
4076 |
|
|
register from the stack, since lowest will == highest in |
4077 |
|
|
`pop_failure_point'. */ |
4078 |
|
|
unsigned dummy_low_reg, dummy_high_reg; |
4079 |
|
|
unsigned char *pdummy; |
4080 |
|
|
const char *sdummy; |
4081 |
|
|
|
4082 |
|
|
DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n"); |
4083 |
|
|
POP_FAILURE_POINT (sdummy, pdummy, |
4084 |
|
|
dummy_low_reg, dummy_high_reg, |
4085 |
|
|
reg_dummy, reg_dummy, reg_info_dummy); |
4086 |
|
|
} |
4087 |
|
|
/* Note fall through. */ |
4088 |
|
|
|
4089 |
|
|
|
4090 |
|
|
/* Unconditionally jump (without popping any failure points). */ |
4091 |
|
|
case jump: |
4092 |
|
|
unconditional_jump: |
4093 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */ |
4094 |
|
|
DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt); |
4095 |
|
|
p += mcnt; /* Do the jump. */ |
4096 |
|
|
DEBUG_PRINT2 ("(to 0x%x).\n", p); |
4097 |
|
|
break; |
4098 |
|
|
|
4099 |
|
|
|
4100 |
|
|
/* We need this opcode so we can detect where alternatives end |
4101 |
|
|
in `group_match_null_string_p' et al. */ |
4102 |
|
|
case jump_past_alt: |
4103 |
|
|
DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n"); |
4104 |
|
|
goto unconditional_jump; |
4105 |
|
|
|
4106 |
|
|
|
4107 |
|
|
/* Normally, the on_failure_jump pushes a failure point, which |
4108 |
|
|
then gets popped at pop_failure_jump. We will end up at |
4109 |
|
|
pop_failure_jump, also, and with a pattern of, say, `a+', we |
4110 |
|
|
are skipping over the on_failure_jump, so we have to push |
4111 |
|
|
something meaningless for pop_failure_jump to pop. */ |
4112 |
|
|
case dummy_failure_jump: |
4113 |
|
|
DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n"); |
4114 |
|
|
/* It doesn't matter what we push for the string here. What |
4115 |
|
|
the code at `fail' tests is the value for the pattern. */ |
4116 |
|
|
PUSH_FAILURE_POINT (0, 0, -2); |
4117 |
|
|
goto unconditional_jump; |
4118 |
|
|
|
4119 |
|
|
|
4120 |
|
|
/* At the end of an alternative, we need to push a dummy failure |
4121 |
|
|
point in case we are followed by a `pop_failure_jump', because |
4122 |
|
|
we don't want the failure point for the alternative to be |
4123 |
|
|
popped. For example, matching `(a|ab)*' against `aab' |
4124 |
|
|
requires that we match the `ab' alternative. */ |
4125 |
|
|
case push_dummy_failure: |
4126 |
|
|
DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n"); |
4127 |
|
|
/* See comments just above at `dummy_failure_jump' about the |
4128 |
|
|
two zeroes. */ |
4129 |
|
|
PUSH_FAILURE_POINT (0, 0, -2); |
4130 |
|
|
break; |
4131 |
|
|
|
4132 |
|
|
/* Have to succeed matching what follows at least n times. |
4133 |
|
|
After that, handle like `on_failure_jump'. */ |
4134 |
|
|
case succeed_n: |
4135 |
|
|
EXTRACT_NUMBER (mcnt, p + 2); |
4136 |
|
|
DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt); |
4137 |
|
|
|
4138 |
|
|
assert (mcnt >= 0); |
4139 |
|
|
/* Originally, this is how many times we HAVE to succeed. */ |
4140 |
|
|
if (mcnt > 0) |
4141 |
|
|
{ |
4142 |
|
|
mcnt--; |
4143 |
|
|
p += 2; |
4144 |
|
|
STORE_NUMBER_AND_INCR (p, mcnt); |
4145 |
|
|
DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt); |
4146 |
|
|
} |
4147 |
|
|
else if (mcnt == 0) |
4148 |
|
|
{ |
4149 |
|
|
DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2); |
4150 |
|
|
p[2] = (unsigned char) no_op; |
4151 |
|
|
p[3] = (unsigned char) no_op; |
4152 |
|
|
goto on_failure; |
4153 |
|
|
} |
4154 |
|
|
break; |
4155 |
|
|
|
4156 |
|
|
case jump_n: |
4157 |
|
|
EXTRACT_NUMBER (mcnt, p + 2); |
4158 |
|
|
DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt); |
4159 |
|
|
|
4160 |
|
|
/* Originally, this is how many times we CAN jump. */ |
4161 |
|
|
if (mcnt) |
4162 |
|
|
{ |
4163 |
|
|
mcnt--; |
4164 |
|
|
STORE_NUMBER (p + 2, mcnt); |
4165 |
|
|
goto unconditional_jump; |
4166 |
|
|
} |
4167 |
|
|
/* If don't have to jump any more, skip over the rest of command. */ |
4168 |
|
|
else |
4169 |
|
|
p += 4; |
4170 |
|
|
break; |
4171 |
|
|
|
4172 |
|
|
case set_number_at: |
4173 |
|
|
{ |
4174 |
|
|
DEBUG_PRINT1 ("EXECUTING set_number_at.\n"); |
4175 |
|
|
|
4176 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
4177 |
|
|
p1 = p + mcnt; |
4178 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p); |
4179 |
|
|
DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt); |
4180 |
|
|
STORE_NUMBER (p1, mcnt); |
4181 |
|
|
break; |
4182 |
|
|
} |
4183 |
|
|
|
4184 |
|
|
case wordbound: |
4185 |
|
|
DEBUG_PRINT1 ("EXECUTING wordbound.\n"); |
4186 |
|
|
if (AT_WORD_BOUNDARY (d)) |
4187 |
|
|
break; |
4188 |
|
|
goto fail; |
4189 |
|
|
|
4190 |
|
|
case notwordbound: |
4191 |
|
|
DEBUG_PRINT1 ("EXECUTING notwordbound.\n"); |
4192 |
|
|
if (AT_WORD_BOUNDARY (d)) |
4193 |
|
|
goto fail; |
4194 |
|
|
break; |
4195 |
|
|
|
4196 |
|
|
case wordbeg: |
4197 |
|
|
DEBUG_PRINT1 ("EXECUTING wordbeg.\n"); |
4198 |
|
|
if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1))) |
4199 |
|
|
break; |
4200 |
|
|
goto fail; |
4201 |
|
|
|
4202 |
|
|
case wordend: |
4203 |
|
|
DEBUG_PRINT1 ("EXECUTING wordend.\n"); |
4204 |
|
|
if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1) |
4205 |
|
|
&& (!WORDCHAR_P (d) || AT_STRINGS_END (d))) |
4206 |
|
|
break; |
4207 |
|
|
goto fail; |
4208 |
|
|
|
4209 |
|
|
#ifdef emacs |
4210 |
|
|
#ifdef emacs19 |
4211 |
|
|
case before_dot: |
4212 |
|
|
DEBUG_PRINT1 ("EXECUTING before_dot.\n"); |
4213 |
|
|
if (PTR_CHAR_POS ((unsigned char *) d) >= point) |
4214 |
|
|
goto fail; |
4215 |
|
|
break; |
4216 |
|
|
|
4217 |
|
|
case at_dot: |
4218 |
|
|
DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
4219 |
|
|
if (PTR_CHAR_POS ((unsigned char *) d) != point) |
4220 |
|
|
goto fail; |
4221 |
|
|
break; |
4222 |
|
|
|
4223 |
|
|
case after_dot: |
4224 |
|
|
DEBUG_PRINT1 ("EXECUTING after_dot.\n"); |
4225 |
|
|
if (PTR_CHAR_POS ((unsigned char *) d) <= point) |
4226 |
|
|
goto fail; |
4227 |
|
|
break; |
4228 |
|
|
#else /* not emacs19 */ |
4229 |
|
|
case at_dot: |
4230 |
|
|
DEBUG_PRINT1 ("EXECUTING at_dot.\n"); |
4231 |
|
|
if (PTR_CHAR_POS ((unsigned char *) d) + 1 != point) |
4232 |
|
|
goto fail; |
4233 |
|
|
break; |
4234 |
|
|
#endif /* not emacs19 */ |
4235 |
|
|
|
4236 |
|
|
case syntaxspec: |
4237 |
|
|
DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt); |
4238 |
|
|
mcnt = *p++; |
4239 |
|
|
goto matchsyntax; |
4240 |
|
|
|
4241 |
|
|
case wordchar: |
4242 |
|
|
DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n"); |
4243 |
|
|
mcnt = (int) Sword; |
4244 |
|
|
matchsyntax: |
4245 |
|
|
PREFETCH (); |
4246 |
|
|
if (SYNTAX (*d++) != (enum syntaxcode) mcnt) |
4247 |
|
|
goto fail; |
4248 |
|
|
SET_REGS_MATCHED (); |
4249 |
|
|
break; |
4250 |
|
|
|
4251 |
|
|
case notsyntaxspec: |
4252 |
|
|
DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt); |
4253 |
|
|
mcnt = *p++; |
4254 |
|
|
goto matchnotsyntax; |
4255 |
|
|
|
4256 |
|
|
case notwordchar: |
4257 |
|
|
DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n"); |
4258 |
|
|
mcnt = (int) Sword; |
4259 |
|
|
matchnotsyntax: |
4260 |
|
|
PREFETCH (); |
4261 |
|
|
if (SYNTAX (*d++) == (enum syntaxcode) mcnt) |
4262 |
|
|
goto fail; |
4263 |
|
|
SET_REGS_MATCHED (); |
4264 |
|
|
break; |
4265 |
|
|
|
4266 |
|
|
#else /* not emacs */ |
4267 |
|
|
case wordchar: |
4268 |
|
|
DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n"); |
4269 |
|
|
PREFETCH (); |
4270 |
|
|
if (!WORDCHAR_P (d)) |
4271 |
|
|
goto fail; |
4272 |
|
|
SET_REGS_MATCHED (); |
4273 |
|
|
d++; |
4274 |
|
|
break; |
4275 |
|
|
|
4276 |
|
|
case notwordchar: |
4277 |
|
|
DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n"); |
4278 |
|
|
PREFETCH (); |
4279 |
|
|
if (WORDCHAR_P (d)) |
4280 |
|
|
goto fail; |
4281 |
|
|
SET_REGS_MATCHED (); |
4282 |
|
|
d++; |
4283 |
|
|
break; |
4284 |
|
|
#endif /* not emacs */ |
4285 |
|
|
|
4286 |
|
|
default: |
4287 |
|
|
abort (); |
4288 |
|
|
} |
4289 |
|
|
continue; /* Successfully executed one pattern command; keep going. */ |
4290 |
|
|
|
4291 |
|
|
|
4292 |
|
|
/* We goto here if a matching operation fails. */ |
4293 |
|
|
fail: |
4294 |
|
|
if (!FAIL_STACK_EMPTY ()) |
4295 |
|
|
{ /* A restart point is known. Restore to that state. */ |
4296 |
|
|
DEBUG_PRINT1 ("\nFAIL:\n"); |
4297 |
|
|
POP_FAILURE_POINT (d, p, |
4298 |
|
|
lowest_active_reg, highest_active_reg, |
4299 |
|
|
regstart, regend, reg_info); |
4300 |
|
|
|
4301 |
|
|
/* If this failure point is a dummy, try the next one. */ |
4302 |
|
|
if (!p) |
4303 |
|
|
goto fail; |
4304 |
|
|
|
4305 |
|
|
/* If we failed to the end of the pattern, don't examine *p. */ |
4306 |
|
|
assert (p <= pend); |
4307 |
|
|
if (p < pend) |
4308 |
|
|
{ |
4309 |
|
|
boolean is_a_jump_n = false; |
4310 |
|
|
|
4311 |
|
|
/* If failed to a backwards jump that's part of a repetition |
4312 |
|
|
loop, need to pop this failure point and use the next one. */ |
4313 |
|
|
switch ((re_opcode_t) *p) |
4314 |
|
|
{ |
4315 |
|
|
case jump_n: |
4316 |
|
|
is_a_jump_n = true; |
4317 |
|
|
case maybe_pop_jump: |
4318 |
|
|
case pop_failure_jump: |
4319 |
|
|
case jump: |
4320 |
|
|
p1 = p + 1; |
4321 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4322 |
|
|
p1 += mcnt; |
4323 |
|
|
|
4324 |
|
|
if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n) |
4325 |
|
|
|| (!is_a_jump_n |
4326 |
|
|
&& (re_opcode_t) *p1 == on_failure_jump)) |
4327 |
|
|
goto fail; |
4328 |
|
|
break; |
4329 |
|
|
default: |
4330 |
|
|
/* do nothing */ ; |
4331 |
|
|
} |
4332 |
|
|
} |
4333 |
|
|
|
4334 |
|
|
if (d >= string1 && d <= end1) |
4335 |
|
|
dend = end_match_1; |
4336 |
|
|
} |
4337 |
|
|
else |
4338 |
|
|
break; /* Matching at this starting point really fails. */ |
4339 |
|
|
} /* for (;;) */ |
4340 |
|
|
|
4341 |
|
|
if (best_regs_set) |
4342 |
|
|
goto restore_best_regs; |
4343 |
|
|
|
4344 |
|
|
FREE_VARIABLES (); |
4345 |
|
|
|
4346 |
|
|
return -1; /* Failure to match. */ |
4347 |
|
|
} /* re_match_2 */ |
4348 |
|
|
|
4349 |
|
|
/* Subroutine definitions for re_match_2. */ |
4350 |
|
|
|
4351 |
|
|
|
4352 |
|
|
/* We are passed P pointing to a register number after a start_memory. |
4353 |
|
|
|
4354 |
|
|
Return true if the pattern up to the corresponding stop_memory can |
4355 |
|
|
match the empty string, and false otherwise. |
4356 |
|
|
|
4357 |
|
|
If we find the matching stop_memory, sets P to point to one past its number. |
4358 |
|
|
Otherwise, sets P to an undefined byte less than or equal to END. |
4359 |
|
|
|
4360 |
|
|
We don't handle duplicates properly (yet). */ |
4361 |
|
|
|
4362 |
|
|
static boolean |
4363 |
|
|
group_match_null_string_p (p, end, reg_info) |
4364 |
|
|
unsigned char **p, *end; |
4365 |
|
|
register_info_type *reg_info; |
4366 |
|
|
{ |
4367 |
|
|
int mcnt; |
4368 |
|
|
/* Point to after the args to the start_memory. */ |
4369 |
|
|
unsigned char *p1 = *p + 2; |
4370 |
|
|
|
4371 |
|
|
while (p1 < end) |
4372 |
|
|
{ |
4373 |
|
|
/* Skip over opcodes that can match nothing, and return true or |
4374 |
|
|
false, as appropriate, when we get to one that can't, or to the |
4375 |
|
|
matching stop_memory. */ |
4376 |
|
|
|
4377 |
|
|
switch ((re_opcode_t) *p1) |
4378 |
|
|
{ |
4379 |
|
|
/* Could be either a loop or a series of alternatives. */ |
4380 |
|
|
case on_failure_jump: |
4381 |
|
|
p1++; |
4382 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4383 |
|
|
|
4384 |
|
|
/* If the next operation is not a jump backwards in the |
4385 |
|
|
pattern. */ |
4386 |
|
|
|
4387 |
|
|
if (mcnt >= 0) |
4388 |
|
|
{ |
4389 |
|
|
/* Go through the on_failure_jumps of the alternatives, |
4390 |
|
|
seeing if any of the alternatives cannot match nothing. |
4391 |
|
|
The last alternative starts with only a jump, |
4392 |
|
|
whereas the rest start with on_failure_jump and end |
4393 |
|
|
with a jump, e.g., here is the pattern for `a|b|c': |
4394 |
|
|
|
4395 |
|
|
/on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6 |
4396 |
|
|
/on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3 |
4397 |
|
|
/exactn/1/c |
4398 |
|
|
|
4399 |
|
|
So, we have to first go through the first (n-1) |
4400 |
|
|
alternatives and then deal with the last one separately. */ |
4401 |
|
|
|
4402 |
|
|
|
4403 |
|
|
/* Deal with the first (n-1) alternatives, which start |
4404 |
|
|
with an on_failure_jump (see above) that jumps to right |
4405 |
|
|
past a jump_past_alt. */ |
4406 |
|
|
|
4407 |
|
|
while ((re_opcode_t) p1[mcnt-3] == jump_past_alt) |
4408 |
|
|
{ |
4409 |
|
|
/* `mcnt' holds how many bytes long the alternative |
4410 |
|
|
is, including the ending `jump_past_alt' and |
4411 |
|
|
its number. */ |
4412 |
|
|
|
4413 |
|
|
if (!alt_match_null_string_p (p1, p1 + mcnt - 3, |
4414 |
|
|
reg_info)) |
4415 |
|
|
return false; |
4416 |
|
|
|
4417 |
|
|
/* Move to right after this alternative, including the |
4418 |
|
|
jump_past_alt. */ |
4419 |
|
|
p1 += mcnt; |
4420 |
|
|
|
4421 |
|
|
/* Break if it's the beginning of an n-th alternative |
4422 |
|
|
that doesn't begin with an on_failure_jump. */ |
4423 |
|
|
if ((re_opcode_t) *p1 != on_failure_jump) |
4424 |
|
|
break; |
4425 |
|
|
|
4426 |
|
|
/* Still have to check that it's not an n-th |
4427 |
|
|
alternative that starts with an on_failure_jump. */ |
4428 |
|
|
p1++; |
4429 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4430 |
|
|
if ((re_opcode_t) p1[mcnt-3] != jump_past_alt) |
4431 |
|
|
{ |
4432 |
|
|
/* Get to the beginning of the n-th alternative. */ |
4433 |
|
|
p1 -= 3; |
4434 |
|
|
break; |
4435 |
|
|
} |
4436 |
|
|
} |
4437 |
|
|
|
4438 |
|
|
/* Deal with the last alternative: go back and get number |
4439 |
|
|
of the `jump_past_alt' just before it. `mcnt' contains |
4440 |
|
|
the length of the alternative. */ |
4441 |
|
|
EXTRACT_NUMBER (mcnt, p1 - 2); |
4442 |
|
|
|
4443 |
|
|
if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info)) |
4444 |
|
|
return false; |
4445 |
|
|
|
4446 |
|
|
p1 += mcnt; /* Get past the n-th alternative. */ |
4447 |
|
|
} /* if mcnt > 0 */ |
4448 |
|
|
break; |
4449 |
|
|
|
4450 |
|
|
|
4451 |
|
|
case stop_memory: |
4452 |
|
|
assert (p1[1] == **p); |
4453 |
|
|
*p = p1 + 2; |
4454 |
|
|
return true; |
4455 |
|
|
|
4456 |
|
|
|
4457 |
|
|
default: |
4458 |
|
|
if (!common_op_match_null_string_p (&p1, end, reg_info)) |
4459 |
|
|
return false; |
4460 |
|
|
} |
4461 |
|
|
} /* while p1 < end */ |
4462 |
|
|
|
4463 |
|
|
return false; |
4464 |
|
|
} /* group_match_null_string_p */ |
4465 |
|
|
|
4466 |
|
|
|
4467 |
|
|
/* Similar to group_match_null_string_p, but doesn't deal with alternatives: |
4468 |
|
|
It expects P to be the first byte of a single alternative and END one |
4469 |
|
|
byte past the last. The alternative can contain groups. */ |
4470 |
|
|
|
4471 |
|
|
static boolean |
4472 |
|
|
alt_match_null_string_p (p, end, reg_info) |
4473 |
|
|
unsigned char *p, *end; |
4474 |
|
|
register_info_type *reg_info; |
4475 |
|
|
{ |
4476 |
|
|
int mcnt; |
4477 |
|
|
unsigned char *p1 = p; |
4478 |
|
|
|
4479 |
|
|
while (p1 < end) |
4480 |
|
|
{ |
4481 |
|
|
/* Skip over opcodes that can match nothing, and break when we get |
4482 |
|
|
to one that can't. */ |
4483 |
|
|
|
4484 |
|
|
switch ((re_opcode_t) *p1) |
4485 |
|
|
{ |
4486 |
|
|
/* It's a loop. */ |
4487 |
|
|
case on_failure_jump: |
4488 |
|
|
p1++; |
4489 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4490 |
|
|
p1 += mcnt; |
4491 |
|
|
break; |
4492 |
|
|
|
4493 |
|
|
default: |
4494 |
|
|
if (!common_op_match_null_string_p (&p1, end, reg_info)) |
4495 |
|
|
return false; |
4496 |
|
|
} |
4497 |
|
|
} /* while p1 < end */ |
4498 |
|
|
|
4499 |
|
|
return true; |
4500 |
|
|
} /* alt_match_null_string_p */ |
4501 |
|
|
|
4502 |
|
|
|
4503 |
|
|
/* Deals with the ops common to group_match_null_string_p and |
4504 |
|
|
alt_match_null_string_p. |
4505 |
|
|
|
4506 |
|
|
Sets P to one after the op and its arguments, if any. */ |
4507 |
|
|
|
4508 |
|
|
static boolean |
4509 |
|
|
common_op_match_null_string_p (p, end, reg_info) |
4510 |
|
|
unsigned char **p, *end; |
4511 |
|
|
register_info_type *reg_info; |
4512 |
|
|
{ |
4513 |
|
|
int mcnt; |
4514 |
|
|
boolean ret; |
4515 |
|
|
int reg_no; |
4516 |
|
|
unsigned char *p1 = *p; |
4517 |
|
|
|
4518 |
|
|
switch ((re_opcode_t) *p1++) |
4519 |
|
|
{ |
4520 |
|
|
case no_op: |
4521 |
|
|
case begline: |
4522 |
|
|
case endline: |
4523 |
|
|
case begbuf: |
4524 |
|
|
case endbuf: |
4525 |
|
|
case wordbeg: |
4526 |
|
|
case wordend: |
4527 |
|
|
case wordbound: |
4528 |
|
|
case notwordbound: |
4529 |
|
|
#ifdef emacs |
4530 |
|
|
case before_dot: |
4531 |
|
|
case at_dot: |
4532 |
|
|
case after_dot: |
4533 |
|
|
#endif |
4534 |
|
|
break; |
4535 |
|
|
|
4536 |
|
|
case start_memory: |
4537 |
|
|
reg_no = *p1; |
4538 |
|
|
assert (reg_no > 0 && reg_no <= MAX_REGNUM); |
4539 |
|
|
ret = group_match_null_string_p (&p1, end, reg_info); |
4540 |
|
|
|
4541 |
|
|
/* Have to set this here in case we're checking a group which |
4542 |
|
|
contains a group and a back reference to it. */ |
4543 |
|
|
|
4544 |
|
|
if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE) |
4545 |
|
|
REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret; |
4546 |
|
|
|
4547 |
|
|
if (!ret) |
4548 |
|
|
return false; |
4549 |
|
|
break; |
4550 |
|
|
|
4551 |
|
|
/* If this is an optimized succeed_n for zero times, make the jump. */ |
4552 |
|
|
case jump: |
4553 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4554 |
|
|
if (mcnt >= 0) |
4555 |
|
|
p1 += mcnt; |
4556 |
|
|
else |
4557 |
|
|
return false; |
4558 |
|
|
break; |
4559 |
|
|
|
4560 |
|
|
case succeed_n: |
4561 |
|
|
/* Get to the number of times to succeed. */ |
4562 |
|
|
p1 += 2; |
4563 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4564 |
|
|
|
4565 |
|
|
if (mcnt == 0) |
4566 |
|
|
{ |
4567 |
|
|
p1 -= 4; |
4568 |
|
|
EXTRACT_NUMBER_AND_INCR (mcnt, p1); |
4569 |
|
|
p1 += mcnt; |
4570 |
|
|
} |
4571 |
|
|
else |
4572 |
|
|
return false; |
4573 |
|
|
break; |
4574 |
|
|
|
4575 |
|
|
case duplicate: |
4576 |
|
|
if (!REG_MATCH_NULL_STRING_P (reg_info[*p1])) |
4577 |
|
|
return false; |
4578 |
|
|
break; |
4579 |
|
|
|
4580 |
|
|
case set_number_at: |
4581 |
|
|
p1 += 4; |
4582 |
|
|
|
4583 |
|
|
default: |
4584 |
|
|
/* All other opcodes mean we cannot match the empty string. */ |
4585 |
|
|
return false; |
4586 |
|
|
} |
4587 |
|
|
|
4588 |
|
|
*p = p1; |
4589 |
|
|
return true; |
4590 |
|
|
} /* common_op_match_null_string_p */ |
4591 |
|
|
|
4592 |
|
|
|
4593 |
|
|
/* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN |
4594 |
|
|
bytes; nonzero otherwise. */ |
4595 |
|
|
|
4596 |
|
|
static int |
4597 |
|
|
bcmp_translate (s1, s2, len, translate) |
4598 |
|
|
unsigned char *s1, *s2; |
4599 |
|
|
register int len; |
4600 |
|
|
char *translate; |
4601 |
|
|
{ |
4602 |
|
|
register unsigned char *p1 = s1, *p2 = s2; |
4603 |
|
|
while (len) |
4604 |
|
|
{ |
4605 |
|
|
if (translate[*p1++] != translate[*p2++]) return 1; |
4606 |
|
|
len--; |
4607 |
|
|
} |
4608 |
|
|
return 0; |
4609 |
|
|
} |
4610 |
|
|
|
4611 |
|
|
/* Entry points for GNU code. */ |
4612 |
|
|
|
4613 |
|
|
/* re_compile_pattern is the GNU regular expression compiler: it |
4614 |
|
|
compiles PATTERN (of length SIZE) and puts the result in BUFP. |
4615 |
|
|
Returns 0 if the pattern was valid, otherwise an error string. |
4616 |
|
|
|
4617 |
|
|
Assumes the `allocated' (and perhaps `buffer') and `translate' fields |
4618 |
|
|
are set in BUFP on entry. |
4619 |
|
|
|
4620 |
|
|
We call regex_compile to do the actual compilation. */ |
4621 |
|
|
|
4622 |
|
|
const char * |
4623 |
|
|
re_compile_pattern (pattern, length, bufp) |
4624 |
|
|
const char *pattern; |
4625 |
|
|
int length; |
4626 |
|
|
struct re_pattern_buffer *bufp; |
4627 |
|
|
{ |
4628 |
|
|
reg_errcode_t ret; |
4629 |
|
|
|
4630 |
|
|
/* GNU code is written to assume at least RE_NREGS registers will be set |
4631 |
|
|
(and at least one extra will be -1). */ |
4632 |
|
|
bufp->regs_allocated = REGS_UNALLOCATED; |
4633 |
|
|
|
4634 |
|
|
/* And GNU code determines whether or not to get register information |
4635 |
|
|
by passing null for the REGS argument to re_match, etc., not by |
4636 |
|
|
setting no_sub. */ |
4637 |
|
|
bufp->no_sub = 0; |
4638 |
|
|
|
4639 |
|
|
/* Match anchors at newline. */ |
4640 |
|
|
bufp->newline_anchor = 1; |
4641 |
|
|
|
4642 |
|
|
ret = regex_compile (pattern, length, re_syntax_options, bufp); |
4643 |
|
|
|
4644 |
|
|
return re_error_msg[(int) ret]; |
4645 |
|
|
} |
4646 |
|
|
|
4647 |
|
|
/* Entry points compatible with 4.2 BSD regex library. We don't define |
4648 |
|
|
them if this is an Emacs or POSIX compilation. */ |
4649 |
|
|
|
4650 |
|
|
#if !defined (emacs) && !defined (_POSIX_SOURCE) |
4651 |
|
|
|
4652 |
|
|
/* BSD has one and only one pattern buffer. */ |
4653 |
|
|
static struct re_pattern_buffer re_comp_buf; |
4654 |
|
|
|
4655 |
|
|
char * |
4656 |
|
|
re_comp (s) |
4657 |
|
|
const char *s; |
4658 |
|
|
{ |
4659 |
|
|
reg_errcode_t ret; |
4660 |
|
|
|
4661 |
|
|
if (!s) |
4662 |
|
|
{ |
4663 |
|
|
if (!re_comp_buf.buffer) |
4664 |
|
|
return "No previous regular expression"; |
4665 |
|
|
return 0; |
4666 |
|
|
} |
4667 |
|
|
|
4668 |
|
|
if (!re_comp_buf.buffer) |
4669 |
|
|
{ |
4670 |
|
|
re_comp_buf.buffer = (unsigned char *) malloc (200); |
4671 |
|
|
if (re_comp_buf.buffer == NULL) |
4672 |
|
|
return "Memory exhausted"; |
4673 |
|
|
re_comp_buf.allocated = 200; |
4674 |
|
|
|
4675 |
|
|
re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH); |
4676 |
|
|
if (re_comp_buf.fastmap == NULL) |
4677 |
|
|
return "Memory exhausted"; |
4678 |
|
|
} |
4679 |
|
|
|
4680 |
|
|
/* Since `re_exec' always passes NULL for the `regs' argument, we |
4681 |
|
|
don't need to initialize the pattern buffer fields which affect it. */ |
4682 |
|
|
|
4683 |
|
|
/* Match anchors at newlines. */ |
4684 |
|
|
re_comp_buf.newline_anchor = 1; |
4685 |
|
|
|
4686 |
|
|
ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf); |
4687 |
|
|
|
4688 |
|
|
/* Yes, we're discarding `const' here. */ |
4689 |
|
|
return (char *) re_error_msg[(int) ret]; |
4690 |
|
|
} |
4691 |
|
|
|
4692 |
|
|
|
4693 |
|
|
int |
4694 |
|
|
re_exec (s) |
4695 |
|
|
const char *s; |
4696 |
|
|
{ |
4697 |
|
|
const int len = strlen (s); |
4698 |
|
|
return |
4699 |
|
|
0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0); |
4700 |
|
|
} |
4701 |
|
|
#endif /* not emacs and not _POSIX_SOURCE */ |
4702 |
|
|
|
4703 |
|
|
/* POSIX.2 functions. Don't define these for Emacs. */ |
4704 |
|
|
|
4705 |
|
|
#ifndef emacs |
4706 |
|
|
|
4707 |
|
|
/* regcomp takes a regular expression as a string and compiles it. |
4708 |
|
|
|
4709 |
|
|
PREG is a regex_t *. We do not expect any fields to be initialized, |
4710 |
|
|
since POSIX says we shouldn't. Thus, we set |
4711 |
|
|
|
4712 |
|
|
`buffer' to the compiled pattern; |
4713 |
|
|
`used' to the length of the compiled pattern; |
4714 |
|
|
`syntax' to RE_SYNTAX_POSIX_EXTENDED if the |
4715 |
|
|
REG_EXTENDED bit in CFLAGS is set; otherwise, to |
4716 |
|
|
RE_SYNTAX_POSIX_BASIC; |
4717 |
|
|
`newline_anchor' to REG_NEWLINE being set in CFLAGS; |
4718 |
|
|
`fastmap' and `fastmap_accurate' to zero; |
4719 |
|
|
`re_nsub' to the number of subexpressions in PATTERN. |
4720 |
|
|
|
4721 |
|
|
PATTERN is the address of the pattern string. |
4722 |
|
|
|
4723 |
|
|
CFLAGS is a series of bits which affect compilation. |
4724 |
|
|
|
4725 |
|
|
If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we |
4726 |
|
|
use POSIX basic syntax. |
4727 |
|
|
|
4728 |
|
|
If REG_NEWLINE is set, then . and [^...] don't match newline. |
4729 |
|
|
Also, regexec will try a match beginning after every newline. |
4730 |
|
|
|
4731 |
|
|
If REG_ICASE is set, then we considers upper- and lowercase |
4732 |
|
|
versions of letters to be equivalent when matching. |
4733 |
|
|
|
4734 |
|
|
If REG_NOSUB is set, then when PREG is passed to regexec, that |
4735 |
|
|
routine will report only success or failure, and nothing about the |
4736 |
|
|
registers. |
4737 |
|
|
|
4738 |
|
|
It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for |
4739 |
|
|
the return codes and their meanings.) */ |
4740 |
|
|
|
4741 |
|
|
int |
4742 |
|
|
regcomp (preg, pattern, cflags) |
4743 |
|
|
regex_t *preg; |
4744 |
|
|
const char *pattern; |
4745 |
|
|
int cflags; |
4746 |
|
|
{ |
4747 |
|
|
reg_errcode_t ret; |
4748 |
|
|
unsigned syntax |
4749 |
|
|
= (cflags & REG_EXTENDED) ? |
4750 |
|
|
RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC; |
4751 |
|
|
|
4752 |
|
|
/* regex_compile will allocate the space for the compiled pattern. */ |
4753 |
|
|
preg->buffer = 0; |
4754 |
|
|
preg->allocated = 0; |
4755 |
|
|
|
4756 |
|
|
/* Don't bother to use a fastmap when searching. This simplifies the |
4757 |
|
|
REG_NEWLINE case: if we used a fastmap, we'd have to put all the |
4758 |
|
|
characters after newlines into the fastmap. This way, we just try |
4759 |
|
|
every character. */ |
4760 |
|
|
preg->fastmap = 0; |
4761 |
|
|
|
4762 |
|
|
if (cflags & REG_ICASE) |
4763 |
|
|
{ |
4764 |
|
|
unsigned i; |
4765 |
|
|
|
4766 |
|
|
preg->translate = (char *) malloc (CHAR_SET_SIZE); |
4767 |
|
|
if (preg->translate == NULL) |
4768 |
|
|
return (int) REG_ESPACE; |
4769 |
|
|
|
4770 |
|
|
/* Map uppercase characters to corresponding lowercase ones. */ |
4771 |
|
|
for (i = 0; i < CHAR_SET_SIZE; i++) |
4772 |
|
|
preg->translate[i] = ISUPPER (i) ? tolower (i) : i; |
4773 |
|
|
} |
4774 |
|
|
else |
4775 |
|
|
preg->translate = NULL; |
4776 |
|
|
|
4777 |
|
|
/* If REG_NEWLINE is set, newlines are treated differently. */ |
4778 |
|
|
if (cflags & REG_NEWLINE) |
4779 |
|
|
{ /* REG_NEWLINE implies neither . nor [^...] match newline. */ |
4780 |
|
|
syntax &= ~RE_DOT_NEWLINE; |
4781 |
|
|
syntax |= RE_HAT_LISTS_NOT_NEWLINE; |
4782 |
|
|
/* It also changes the matching behavior. */ |
4783 |
|
|
preg->newline_anchor = 1; |
4784 |
|
|
} |
4785 |
|
|
else |
4786 |
|
|
preg->newline_anchor = 0; |
4787 |
|
|
|
4788 |
|
|
preg->no_sub = !!(cflags & REG_NOSUB); |
4789 |
|
|
|
4790 |
|
|
/* POSIX says a null character in the pattern terminates it, so we |
4791 |
|
|
can use strlen here in compiling the pattern. */ |
4792 |
|
|
ret = regex_compile (pattern, strlen (pattern), syntax, preg); |
4793 |
|
|
|
4794 |
|
|
/* POSIX doesn't distinguish between an unmatched open-group and an |
4795 |
|
|
unmatched close-group: both are REG_EPAREN. */ |
4796 |
|
|
if (ret == REG_ERPAREN) ret = REG_EPAREN; |
4797 |
|
|
|
4798 |
|
|
return (int) ret; |
4799 |
|
|
} |
4800 |
|
|
|
4801 |
|
|
|
4802 |
|
|
/* regexec searches for a given pattern, specified by PREG, in the |
4803 |
|
|
string STRING. |
4804 |
|
|
|
4805 |
|
|
If NMATCH is zero or REG_NOSUB was set in the cflags argument to |
4806 |
|
|
`regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at |
4807 |
|
|
least NMATCH elements, and we set them to the offsets of the |
4808 |
|
|
corresponding matched substrings. |
4809 |
|
|
|
4810 |
|
|
EFLAGS specifies `execution flags' which affect matching: if |
4811 |
|
|
REG_NOTBOL is set, then ^ does not match at the beginning of the |
4812 |
|
|
string; if REG_NOTEOL is set, then $ does not match at the end. |
4813 |
|
|
|
4814 |
|
|
We return 0 if we find a match and REG_NOMATCH if not. */ |
4815 |
|
|
|
4816 |
|
|
int |
4817 |
|
|
regexec (preg, string, nmatch, pmatch, eflags) |
4818 |
|
|
const regex_t *preg; |
4819 |
|
|
const char *string; |
4820 |
|
|
size_t nmatch; |
4821 |
|
|
regmatch_t pmatch[]; |
4822 |
|
|
int eflags; |
4823 |
|
|
{ |
4824 |
|
|
int ret; |
4825 |
|
|
struct re_registers regs; |
4826 |
|
|
regex_t private_preg; |
4827 |
|
|
int len = strlen (string); |
4828 |
|
|
boolean want_reg_info = !preg->no_sub && nmatch > 0; |
4829 |
|
|
|
4830 |
|
|
private_preg = *preg; |
4831 |
|
|
|
4832 |
|
|
private_preg.not_bol = !!(eflags & REG_NOTBOL); |
4833 |
|
|
private_preg.not_eol = !!(eflags & REG_NOTEOL); |
4834 |
|
|
|
4835 |
|
|
/* The user has told us exactly how many registers to return |
4836 |
|
|
information about, via `nmatch'. We have to pass that on to the |
4837 |
|
|
matching routines. */ |
4838 |
|
|
private_preg.regs_allocated = REGS_FIXED; |
4839 |
|
|
|
4840 |
|
|
if (want_reg_info) |
4841 |
|
|
{ |
4842 |
|
|
regs.num_regs = nmatch; |
4843 |
|
|
regs.start = TALLOC (nmatch, regoff_t); |
4844 |
|
|
regs.end = TALLOC (nmatch, regoff_t); |
4845 |
|
|
if (regs.start == NULL || regs.end == NULL) |
4846 |
|
|
return (int) REG_NOMATCH; |
4847 |
|
|
} |
4848 |
|
|
|
4849 |
|
|
/* Perform the searching operation. */ |
4850 |
|
|
ret = re_search (&private_preg, string, len, |
4851 |
|
|
/* start: */ 0, /* range: */ len, |
4852 |
|
|
want_reg_info ? ®s : (struct re_registers *) 0); |
4853 |
|
|
|
4854 |
|
|
/* Copy the register information to the POSIX structure. */ |
4855 |
|
|
if (want_reg_info) |
4856 |
|
|
{ |
4857 |
|
|
if (ret >= 0) |
4858 |
|
|
{ |
4859 |
|
|
unsigned r; |
4860 |
|
|
|
4861 |
|
|
for (r = 0; r < nmatch; r++) |
4862 |
|
|
{ |
4863 |
|
|
pmatch[r].rm_so = regs.start[r]; |
4864 |
|
|
pmatch[r].rm_eo = regs.end[r]; |
4865 |
|
|
} |
4866 |
|
|
} |
4867 |
|
|
|
4868 |
|
|
/* If we needed the temporary register info, free the space now. */ |
4869 |
|
|
free (regs.start); |
4870 |
|
|
free (regs.end); |
4871 |
|
|
} |
4872 |
|
|
|
4873 |
|
|
/* We want zero return to mean success, unlike `re_search'. */ |
4874 |
|
|
return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH; |
4875 |
|
|
} |
4876 |
|
|
|
4877 |
|
|
|
4878 |
|
|
/* Returns a message corresponding to an error code, ERRCODE, returned |
4879 |
|
|
from either regcomp or regexec. We don't use PREG here. */ |
4880 |
|
|
|
4881 |
|
|
size_t |
4882 |
|
|
regerror (errcode, preg, errbuf, errbuf_size) |
4883 |
|
|
int errcode; |
4884 |
|
|
const regex_t *preg; |
4885 |
|
|
char *errbuf; |
4886 |
|
|
size_t errbuf_size; |
4887 |
|
|
{ |
4888 |
|
|
const char *msg; |
4889 |
|
|
size_t msg_size; |
4890 |
|
|
|
4891 |
|
|
if (errcode < 0 |
4892 |
|
|
|| errcode >= (sizeof (re_error_msg) / sizeof (re_error_msg[0]))) |
4893 |
|
|
/* Only error codes returned by the rest of the code should be passed |
4894 |
|
|
to this routine. If we are given anything else, or if other regex |
4895 |
|
|
code generates an invalid error code, then the program has a bug. |
4896 |
|
|
Dump core so we can fix it. */ |
4897 |
|
|
abort (); |
4898 |
|
|
|
4899 |
|
|
msg = re_error_msg[errcode]; |
4900 |
|
|
|
4901 |
|
|
/* POSIX doesn't require that we do anything in this case, but why |
4902 |
|
|
not be nice. */ |
4903 |
|
|
if (! msg) |
4904 |
|
|
msg = "Success"; |
4905 |
|
|
|
4906 |
|
|
msg_size = strlen (msg) + 1; /* Includes the null. */ |
4907 |
|
|
|
4908 |
|
|
if (errbuf_size != 0) |
4909 |
|
|
{ |
4910 |
|
|
if (msg_size > errbuf_size) |
4911 |
|
|
{ |
4912 |
|
|
strncpy (errbuf, msg, errbuf_size - 1); |
4913 |
|
|
errbuf[errbuf_size - 1] = 0; |
4914 |
|
|
} |
4915 |
|
|
else |
4916 |
|
|
strcpy (errbuf, msg); |
4917 |
|
|
} |
4918 |
|
|
|
4919 |
|
|
return msg_size; |
4920 |
|
|
} |
4921 |
|
|
|
4922 |
|
|
|
4923 |
|
|
/* Free dynamically allocated space used by PREG. */ |
4924 |
|
|
|
4925 |
|
|
void |
4926 |
|
|
regfree (preg) |
4927 |
|
|
regex_t *preg; |
4928 |
|
|
{ |
4929 |
|
|
if (preg->buffer != NULL) |
4930 |
|
|
free (preg->buffer); |
4931 |
|
|
preg->buffer = NULL; |
4932 |
|
|
|
4933 |
|
|
preg->allocated = 0; |
4934 |
|
|
preg->used = 0; |
4935 |
|
|
|
4936 |
|
|
if (preg->fastmap != NULL) |
4937 |
|
|
free (preg->fastmap); |
4938 |
|
|
preg->fastmap = NULL; |
4939 |
|
|
preg->fastmap_accurate = 0; |
4940 |
|
|
|
4941 |
|
|
if (preg->translate != NULL) |
4942 |
|
|
free (preg->translate); |
4943 |
|
|
preg->translate = NULL; |
4944 |
|
|
} |
4945 |
|
|
|
4946 |
|
|
#endif /* not emacs */ |
4947 |
|
|
|
4948 |
|
|
/* |
4949 |
|
|
Local variables: |
4950 |
|
|
make-backup-files: t |
4951 |
|
|
version-control: t |
4952 |
|
|
trim-versions-without-asking: nil |
4953 |
|
|
End: |
4954 |
|
|
*/ |