1 /* Extended regular expression matching and search library, version
2 0.12. (Implements POSIX draft P10003.2/D11.2, except for
3 internationalization features.)
5 Copyright (C) 1993, 1994, 1995, 1996, 1997 Free Software Foundation, Inc.
7 This program is free software; you can redistribute it and/or modify
8 it under the terms of the GNU General Public License as published by
9 the Free Software Foundation; either version 2, or (at your option)
12 This program is distributed in the hope that it will be useful,
13 but WITHOUT ANY WARRANTY; without even the implied warranty of
14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15 GNU General Public License for more details.
17 You should have received a copy of the GNU General Public License
18 along with this program; if not, write to the Free Software
19 Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307,
22 /* AIX requires this to be the first thing in the file. */
23 #if defined (_AIX) && !defined (REGEX_MALLOC)
30 /* Converts the pointer to the char to BEG-based offset from the start. */
31 #define PTR_TO_OFFSET(d) \
32 POS_AS_IN_BUFFER (MATCHING_IN_FIRST_STRING \
33 ? (d) - string1 : (d) - (string2 - size1))
34 #define POS_AS_IN_BUFFER(p) ((p) + 1)
40 /* We need this for `regex.h', and perhaps for the Emacs include files. */
41 #include <sys/types.h>
43 /* This is for other GNU distributions with internationalized messages. */
44 #if HAVE_LIBINTL_H || defined (_LIBC)
47 # define gettext(msgid) (msgid)
51 /* This define is so xgettext can find the internationalizable
53 #define gettext_noop(String) String
56 /* The `emacs' switch turns on certain matching commands
57 that make sense only in Emacs. */
63 /* Make syntax table lookup grant data in gl_state. */
64 #define SYNTAX_ENTRY_VIA_PROPERTY
70 #define malloc xmalloc
75 /* If we are not linking with Emacs proper,
76 we can't use the relocating allocator
77 even if config.h says that we can. */
80 #if defined (STDC_HEADERS) || defined (_LIBC)
87 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
88 If nothing else has been done, use the method below. */
89 #ifdef INHIBIT_STRING_HEADER
90 #if !(defined (HAVE_BZERO) && defined (HAVE_BCOPY))
91 #if !defined (bzero) && !defined (bcopy)
92 #undef INHIBIT_STRING_HEADER
97 /* This is the normal way of making sure we have a bcopy and a bzero.
98 This is used in most programs--a few other programs avoid this
99 by defining INHIBIT_STRING_HEADER. */
100 #ifndef INHIBIT_STRING_HEADER
101 #if defined (HAVE_STRING_H) || defined (STDC_HEADERS) || defined (_LIBC)
104 #define bcmp(s1, s2, n) memcmp ((s1), (s2), (n))
107 #define bcopy(s, d, n) memcpy ((d), (s), (n))
110 #define bzero(s, n) memset ((s), 0, (n))
117 /* Define the syntax stuff for \<, \>, etc. */
119 /* This must be nonzero for the wordchar and notwordchar pattern
120 commands in re_match_2. */
125 #ifdef SWITCH_ENUM_BUG
126 #define SWITCH_ENUM_CAST(x) ((int)(x))
128 #define SWITCH_ENUM_CAST(x) (x)
133 extern char *re_syntax_table;
135 #else /* not SYNTAX_TABLE */
137 /* How many characters in the character set. */
138 #define CHAR_SET_SIZE 256
140 static char re_syntax_table[CHAR_SET_SIZE];
151 bzero (re_syntax_table, sizeof re_syntax_table);
153 for (c = 'a'; c <= 'z'; c++)
154 re_syntax_table[c] = Sword;
156 for (c = 'A'; c <= 'Z'; c++)
157 re_syntax_table[c] = Sword;
159 for (c = '0'; c <= '9'; c++)
160 re_syntax_table[c] = Sword;
162 re_syntax_table['_'] = Sword;
167 #endif /* not SYNTAX_TABLE */
169 #define SYNTAX(c) re_syntax_table[c]
171 /* Dummy macro for non emacs environments. */
172 #define BASE_LEADING_CODE_P(c) (0)
173 #define WORD_BOUNDARY_P(c1, c2) (0)
174 #define CHAR_HEAD_P(p) (1)
175 #define SINGLE_BYTE_CHAR_P(c) (1)
176 #define SAME_CHARSET_P(c1, c2) (1)
177 #define MULTIBYTE_FORM_LENGTH(p, s) (1)
178 #define STRING_CHAR(p, s) (*(p))
179 #define STRING_CHAR_AND_LENGTH(p, s, actual_len) ((actual_len) = 1, *(p))
180 #define GET_CHAR_AFTER_2(c, p, str1, end1, str2, end2) \
181 (c = ((p) == (end1) ? *(str2) : *(p)))
182 #define GET_CHAR_BEFORE_2(c, p, str1, end1, str2, end2) \
183 (c = ((p) == (str2) ? *((end1) - 1) : *((p) - 1)))
184 #endif /* not emacs */
186 /* Get the interface, including the syntax bits. */
189 /* isalpha etc. are used for the character classes. */
192 /* Jim Meyering writes:
194 "... Some ctype macros are valid only for character codes that
195 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
196 using /bin/cc or gcc but without giving an ansi option). So, all
197 ctype uses should be through macros like ISPRINT... If
198 STDC_HEADERS is defined, then autoconf has verified that the ctype
199 macros don't need to be guarded with references to isascii. ...
200 Defining isascii to 1 should let any compiler worth its salt
201 eliminate the && through constant folding." */
203 #if defined (STDC_HEADERS) || (!defined (isascii) && !defined (HAVE_ISASCII))
206 #define ISASCII(c) isascii(c)
210 #define ISBLANK(c) (ISASCII (c) && isblank (c))
212 #define ISBLANK(c) ((c) == ' ' || (c) == '\t')
215 #define ISGRAPH(c) (ISASCII (c) && isgraph (c))
217 #define ISGRAPH(c) (ISASCII (c) && isprint (c) && !isspace (c))
220 #define ISPRINT(c) (ISASCII (c) && isprint (c))
221 #define ISDIGIT(c) (ISASCII (c) && isdigit (c))
222 #define ISALNUM(c) (ISASCII (c) && isalnum (c))
223 #define ISALPHA(c) (ISASCII (c) && isalpha (c))
224 #define ISCNTRL(c) (ISASCII (c) && iscntrl (c))
225 #define ISLOWER(c) (ISASCII (c) && islower (c))
226 #define ISPUNCT(c) (ISASCII (c) && ispunct (c))
227 #define ISSPACE(c) (ISASCII (c) && isspace (c))
228 #define ISUPPER(c) (ISASCII (c) && isupper (c))
229 #define ISXDIGIT(c) (ISASCII (c) && isxdigit (c))
232 #define NULL (void *)0
235 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
236 since ours (we hope) works properly with all combinations of
237 machines, compilers, `char' and `unsigned char' argument types.
238 (Per Bothner suggested the basic approach.) */
239 #undef SIGN_EXTEND_CHAR
241 #define SIGN_EXTEND_CHAR(c) ((signed char) (c))
242 #else /* not __STDC__ */
243 /* As in Harbison and Steele. */
244 #define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
247 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
248 use `alloca' instead of `malloc'. This is because using malloc in
249 re_search* or re_match* could cause memory leaks when C-g is used in
250 Emacs; also, malloc is slower and causes storage fragmentation. On
251 the other hand, malloc is more portable, and easier to debug.
253 Because we sometimes use alloca, some routines have to be macros,
254 not functions -- `alloca'-allocated space disappears at the end of the
255 function it is called in. */
259 #define REGEX_ALLOCATE malloc
260 #define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
261 #define REGEX_FREE free
263 #else /* not REGEX_MALLOC */
265 /* Emacs already defines alloca, sometimes. */
268 /* Make alloca work the best possible way. */
270 #define alloca __builtin_alloca
271 #else /* not __GNUC__ */
274 #else /* not __GNUC__ or HAVE_ALLOCA_H */
275 #if 0 /* It is a bad idea to declare alloca. We always cast the result. */
276 #ifndef _AIX /* Already did AIX, up at the top. */
278 #endif /* not _AIX */
280 #endif /* not HAVE_ALLOCA_H */
281 #endif /* not __GNUC__ */
283 #endif /* not alloca */
285 #define REGEX_ALLOCATE alloca
287 /* Assumes a `char *destination' variable. */
288 #define REGEX_REALLOCATE(source, osize, nsize) \
289 (destination = (char *) alloca (nsize), \
290 bcopy (source, destination, osize), \
293 /* No need to do anything to free, after alloca. */
294 #define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
296 #endif /* not REGEX_MALLOC */
298 /* Define how to allocate the failure stack. */
300 #if defined (REL_ALLOC) && defined (REGEX_MALLOC)
302 #define REGEX_ALLOCATE_STACK(size) \
303 r_alloc (&failure_stack_ptr, (size))
304 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
305 r_re_alloc (&failure_stack_ptr, (nsize))
306 #define REGEX_FREE_STACK(ptr) \
307 r_alloc_free (&failure_stack_ptr)
309 #else /* not using relocating allocator */
313 #define REGEX_ALLOCATE_STACK malloc
314 #define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
315 #define REGEX_FREE_STACK free
317 #else /* not REGEX_MALLOC */
319 #define REGEX_ALLOCATE_STACK alloca
321 #define REGEX_REALLOCATE_STACK(source, osize, nsize) \
322 REGEX_REALLOCATE (source, osize, nsize)
323 /* No need to explicitly free anything. */
324 #define REGEX_FREE_STACK(arg)
326 #endif /* not REGEX_MALLOC */
327 #endif /* not using relocating allocator */
330 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
331 `string1' or just past its end. This works if PTR is NULL, which is
333 #define FIRST_STRING_P(ptr) \
334 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
336 /* (Re)Allocate N items of type T using malloc, or fail. */
337 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
338 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
339 #define RETALLOC_IF(addr, n, t) \
340 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
341 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
343 #define BYTEWIDTH 8 /* In bits. */
345 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
349 #define MAX(a, b) ((a) > (b) ? (a) : (b))
350 #define MIN(a, b) ((a) < (b) ? (a) : (b))
352 typedef char boolean;
356 static int re_match_2_internal ();
358 /* These are the command codes that appear in compiled regular
359 expressions. Some opcodes are followed by argument bytes. A
360 command code can specify any interpretation whatsoever for its
361 arguments. Zero bytes may appear in the compiled regular expression. */
367 /* Succeed right away--no more backtracking. */
370 /* Followed by one byte giving n, then by n literal bytes. */
373 /* Matches any (more or less) character. */
376 /* Matches any one char belonging to specified set. First
377 following byte is number of bitmap bytes. Then come bytes
378 for a bitmap saying which chars are in. Bits in each byte
379 are ordered low-bit-first. A character is in the set if its
380 bit is 1. A character too large to have a bit in the map is
381 automatically not in the set. */
384 /* Same parameters as charset, but match any character that is
385 not one of those specified. */
388 /* Start remembering the text that is matched, for storing in a
389 register. Followed by one byte with the register number, in
390 the range 0 to one less than the pattern buffer's re_nsub
391 field. Then followed by one byte with the number of groups
392 inner to this one. (This last has to be part of the
393 start_memory only because we need it in the on_failure_jump
397 /* Stop remembering the text that is matched and store it in a
398 memory register. Followed by one byte with the register
399 number, in the range 0 to one less than `re_nsub' in the
400 pattern buffer, and one byte with the number of inner groups,
401 just like `start_memory'. (We need the number of inner
402 groups here because we don't have any easy way of finding the
403 corresponding start_memory when we're at a stop_memory.) */
406 /* Match a duplicate of something remembered. Followed by one
407 byte containing the register number. */
410 /* Fail unless at beginning of line. */
413 /* Fail unless at end of line. */
416 /* Succeeds if at beginning of buffer (if emacs) or at beginning
417 of string to be matched (if not). */
420 /* Analogously, for end of buffer/string. */
423 /* Followed by two byte relative address to which to jump. */
426 /* Same as jump, but marks the end of an alternative. */
429 /* Followed by two-byte relative address of place to resume at
430 in case of failure. */
433 /* Like on_failure_jump, but pushes a placeholder instead of the
434 current string position when executed. */
435 on_failure_keep_string_jump,
437 /* Throw away latest failure point and then jump to following
438 two-byte relative address. */
441 /* Change to pop_failure_jump if know won't have to backtrack to
442 match; otherwise change to jump. This is used to jump
443 back to the beginning of a repeat. If what follows this jump
444 clearly won't match what the repeat does, such that we can be
445 sure that there is no use backtracking out of repetitions
446 already matched, then we change it to a pop_failure_jump.
447 Followed by two-byte address. */
450 /* Jump to following two-byte address, and push a dummy failure
451 point. This failure point will be thrown away if an attempt
452 is made to use it for a failure. A `+' construct makes this
453 before the first repeat. Also used as an intermediary kind
454 of jump when compiling an alternative. */
457 /* Push a dummy failure point and continue. Used at the end of
461 /* Followed by two-byte relative address and two-byte number n.
462 After matching N times, jump to the address upon failure. */
465 /* Followed by two-byte relative address, and two-byte number n.
466 Jump to the address N times, then fail. */
469 /* Set the following two-byte relative address to the
470 subsequent two-byte number. The address *includes* the two
474 wordchar, /* Matches any word-constituent character. */
475 notwordchar, /* Matches any char that is not a word-constituent. */
477 wordbeg, /* Succeeds if at word beginning. */
478 wordend, /* Succeeds if at word end. */
480 wordbound, /* Succeeds if at a word boundary. */
481 notwordbound /* Succeeds if not at a word boundary. */
484 ,before_dot, /* Succeeds if before point. */
485 at_dot, /* Succeeds if at point. */
486 after_dot, /* Succeeds if after point. */
488 /* Matches any character whose syntax is specified. Followed by
489 a byte which contains a syntax code, e.g., Sword. */
492 /* Matches any character whose syntax is not that specified. */
495 /* Matches any character whose category-set contains the specified
496 category. The operator is followed by a byte which contains a
497 category code (mnemonic ASCII character). */
500 /* Matches any character whose category-set does not contain the
501 specified category. The operator is followed by a byte which
502 contains the category code (mnemonic ASCII character). */
507 /* Common operations on the compiled pattern. */
509 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
511 #define STORE_NUMBER(destination, number) \
513 (destination)[0] = (number) & 0377; \
514 (destination)[1] = (number) >> 8; \
517 /* Same as STORE_NUMBER, except increment DESTINATION to
518 the byte after where the number is stored. Therefore, DESTINATION
519 must be an lvalue. */
521 #define STORE_NUMBER_AND_INCR(destination, number) \
523 STORE_NUMBER (destination, number); \
524 (destination) += 2; \
527 /* Put into DESTINATION a number stored in two contiguous bytes starting
530 #define EXTRACT_NUMBER(destination, source) \
532 (destination) = *(source) & 0377; \
533 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
538 extract_number (dest, source)
540 unsigned char *source;
542 int temp = SIGN_EXTEND_CHAR (*(source + 1));
543 *dest = *source & 0377;
547 #ifndef EXTRACT_MACROS /* To debug the macros. */
548 #undef EXTRACT_NUMBER
549 #define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
550 #endif /* not EXTRACT_MACROS */
554 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
555 SOURCE must be an lvalue. */
557 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
559 EXTRACT_NUMBER (destination, source); \
565 extract_number_and_incr (destination, source)
567 unsigned char **source;
569 extract_number (destination, *source);
573 #ifndef EXTRACT_MACROS
574 #undef EXTRACT_NUMBER_AND_INCR
575 #define EXTRACT_NUMBER_AND_INCR(dest, src) \
576 extract_number_and_incr (&dest, &src)
577 #endif /* not EXTRACT_MACROS */
581 /* Store a multibyte character in three contiguous bytes starting
582 DESTINATION, and increment DESTINATION to the byte after where the
583 character is stored. Therefore, DESTINATION must be an lvalue. */
585 #define STORE_CHARACTER_AND_INCR(destination, character) \
587 (destination)[0] = (character) & 0377; \
588 (destination)[1] = ((character) >> 8) & 0377; \
589 (destination)[2] = (character) >> 16; \
590 (destination) += 3; \
593 /* Put into DESTINATION a character stored in three contiguous bytes
594 starting at SOURCE. */
596 #define EXTRACT_CHARACTER(destination, source) \
598 (destination) = ((source)[0] \
599 | ((source)[1] << 8) \
600 | ((source)[2] << 16)); \
604 /* Macros for charset. */
606 /* Size of bitmap of charset P in bytes. P is a start of charset,
607 i.e. *P is (re_opcode_t) charset or (re_opcode_t) charset_not. */
608 #define CHARSET_BITMAP_SIZE(p) ((p)[1] & 0x7F)
610 /* Nonzero if charset P has range table. */
611 #define CHARSET_RANGE_TABLE_EXISTS_P(p) ((p)[1] & 0x80)
613 /* Return the address of range table of charset P. But not the start
614 of table itself, but the before where the number of ranges is
615 stored. `2 +' means to skip re_opcode_t and size of bitmap. */
616 #define CHARSET_RANGE_TABLE(p) (&(p)[2 + CHARSET_BITMAP_SIZE (p)])
618 /* Test if C is listed in the bitmap of charset P. */
619 #define CHARSET_LOOKUP_BITMAP(p, c) \
620 ((c) < CHARSET_BITMAP_SIZE (p) * BYTEWIDTH \
621 && (p)[2 + (c) / BYTEWIDTH] & (1 << ((c) % BYTEWIDTH)))
623 /* Return the address of end of RANGE_TABLE. COUNT is number of
624 ranges (which is a pair of (start, end)) in the RANGE_TABLE. `* 2'
625 is start of range and end of range. `* 3' is size of each start
627 #define CHARSET_RANGE_TABLE_END(range_table, count) \
628 ((range_table) + (count) * 2 * 3)
630 /* Test if C is in RANGE_TABLE. A flag NOT is negated if C is in.
631 COUNT is number of ranges in RANGE_TABLE. */
632 #define CHARSET_LOOKUP_RANGE_TABLE_RAW(not, c, range_table, count) \
635 int range_start, range_end; \
637 unsigned char *range_table_end \
638 = CHARSET_RANGE_TABLE_END ((range_table), (count)); \
640 for (p = (range_table); p < range_table_end; p += 2 * 3) \
642 EXTRACT_CHARACTER (range_start, p); \
643 EXTRACT_CHARACTER (range_end, p + 3); \
645 if (range_start <= (c) && (c) <= range_end) \
654 /* Test if C is in range table of CHARSET. The flag NOT is negated if
655 C is listed in it. */
656 #define CHARSET_LOOKUP_RANGE_TABLE(not, c, charset) \
659 /* Number of ranges in range table. */ \
661 unsigned char *range_table = CHARSET_RANGE_TABLE (charset); \
663 EXTRACT_NUMBER_AND_INCR (count, range_table); \
664 CHARSET_LOOKUP_RANGE_TABLE_RAW ((not), (c), range_table, count); \
668 /* If DEBUG is defined, Regex prints many voluminous messages about what
669 it is doing (if the variable `debug' is nonzero). If linked with the
670 main program in `iregex.c', you can enter patterns and strings
671 interactively. And if linked with the main program in `main.c' and
672 the other test files, you can run the already-written tests. */
676 /* We use standard I/O for debugging. */
679 /* It is useful to test things that ``must'' be true when debugging. */
682 static int debug = 0;
684 #define DEBUG_STATEMENT(e) e
685 #define DEBUG_PRINT1(x) if (debug) printf (x)
686 #define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
687 #define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
688 #define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
689 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
690 if (debug) print_partial_compiled_pattern (s, e)
691 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
692 if (debug) print_double_string (w, s1, sz1, s2, sz2)
695 /* Print the fastmap in human-readable form. */
698 print_fastmap (fastmap)
701 unsigned was_a_range = 0;
704 while (i < (1 << BYTEWIDTH))
710 while (i < (1 << BYTEWIDTH) && fastmap[i])
726 /* Print a compiled pattern string in human-readable form, starting at
727 the START pointer into it and ending just before the pointer END. */
730 print_partial_compiled_pattern (start, end)
731 unsigned char *start;
735 unsigned char *p = start;
736 unsigned char *pend = end;
744 /* Loop over pattern commands. */
747 printf ("%d:\t", p - start);
749 switch ((re_opcode_t) *p++)
757 printf ("/exactn/%d", mcnt);
768 printf ("/start_memory/%d/%d", mcnt, *p++);
773 printf ("/stop_memory/%d/%d", mcnt, *p++);
777 printf ("/duplicate/%d", *p++);
787 register int c, last = -100;
788 register int in_range = 0;
790 printf ("/charset [%s",
791 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
793 assert (p + *p < pend);
795 for (c = 0; c < 256; c++)
797 && (p[1 + (c/8)] & (1 << (c % 8))))
799 /* Are we starting a range? */
800 if (last + 1 == c && ! in_range)
805 /* Have we broken a range? */
806 else if (last + 1 != c && in_range)
835 case on_failure_jump:
836 extract_number_and_incr (&mcnt, &p);
837 printf ("/on_failure_jump to %d", p + mcnt - start);
840 case on_failure_keep_string_jump:
841 extract_number_and_incr (&mcnt, &p);
842 printf ("/on_failure_keep_string_jump to %d", p + mcnt - start);
845 case dummy_failure_jump:
846 extract_number_and_incr (&mcnt, &p);
847 printf ("/dummy_failure_jump to %d", p + mcnt - start);
850 case push_dummy_failure:
851 printf ("/push_dummy_failure");
855 extract_number_and_incr (&mcnt, &p);
856 printf ("/maybe_pop_jump to %d", p + mcnt - start);
859 case pop_failure_jump:
860 extract_number_and_incr (&mcnt, &p);
861 printf ("/pop_failure_jump to %d", p + mcnt - start);
865 extract_number_and_incr (&mcnt, &p);
866 printf ("/jump_past_alt to %d", p + mcnt - start);
870 extract_number_and_incr (&mcnt, &p);
871 printf ("/jump to %d", p + mcnt - start);
875 extract_number_and_incr (&mcnt, &p);
876 extract_number_and_incr (&mcnt2, &p);
877 printf ("/succeed_n to %d, %d times", p + mcnt - start, mcnt2);
881 extract_number_and_incr (&mcnt, &p);
882 extract_number_and_incr (&mcnt2, &p);
883 printf ("/jump_n to %d, %d times", p + mcnt - start, mcnt2);
887 extract_number_and_incr (&mcnt, &p);
888 extract_number_and_incr (&mcnt2, &p);
889 printf ("/set_number_at location %d to %d", p + mcnt - start, mcnt2);
893 printf ("/wordbound");
897 printf ("/notwordbound");
909 printf ("/before_dot");
917 printf ("/after_dot");
921 printf ("/syntaxspec");
923 printf ("/%d", mcnt);
927 printf ("/notsyntaxspec");
929 printf ("/%d", mcnt);
934 printf ("/wordchar");
938 printf ("/notwordchar");
950 printf ("?%d", *(p-1));
956 printf ("%d:\tend of pattern.\n", p - start);
961 print_compiled_pattern (bufp)
962 struct re_pattern_buffer *bufp;
964 unsigned char *buffer = bufp->buffer;
966 print_partial_compiled_pattern (buffer, buffer + bufp->used);
967 printf ("%d bytes used/%d bytes allocated.\n", bufp->used, bufp->allocated);
969 if (bufp->fastmap_accurate && bufp->fastmap)
971 printf ("fastmap: ");
972 print_fastmap (bufp->fastmap);
975 printf ("re_nsub: %d\t", bufp->re_nsub);
976 printf ("regs_alloc: %d\t", bufp->regs_allocated);
977 printf ("can_be_null: %d\t", bufp->can_be_null);
978 printf ("newline_anchor: %d\n", bufp->newline_anchor);
979 printf ("no_sub: %d\t", bufp->no_sub);
980 printf ("not_bol: %d\t", bufp->not_bol);
981 printf ("not_eol: %d\t", bufp->not_eol);
982 printf ("syntax: %d\n", bufp->syntax);
983 /* Perhaps we should print the translate table? */
988 print_double_string (where, string1, size1, string2, size2)
1001 if (FIRST_STRING_P (where))
1003 for (this_char = where - string1; this_char < size1; this_char++)
1004 putchar (string1[this_char]);
1009 for (this_char = where - string2; this_char < size2; this_char++)
1010 putchar (string2[this_char]);
1014 #else /* not DEBUG */
1019 #define DEBUG_STATEMENT(e)
1020 #define DEBUG_PRINT1(x)
1021 #define DEBUG_PRINT2(x1, x2)
1022 #define DEBUG_PRINT3(x1, x2, x3)
1023 #define DEBUG_PRINT4(x1, x2, x3, x4)
1024 #define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1025 #define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1027 #endif /* not DEBUG */
1029 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1030 also be assigned to arbitrarily: each pattern buffer stores its own
1031 syntax, so it can be changed between regex compilations. */
1032 /* This has no initializer because initialized variables in Emacs
1033 become read-only after dumping. */
1034 reg_syntax_t re_syntax_options;
1037 /* Specify the precise syntax of regexps for compilation. This provides
1038 for compatibility for various utilities which historically have
1039 different, incompatible syntaxes.
1041 The argument SYNTAX is a bit mask comprised of the various bits
1042 defined in regex.h. We return the old syntax. */
1045 re_set_syntax (syntax)
1046 reg_syntax_t syntax;
1048 reg_syntax_t ret = re_syntax_options;
1050 re_syntax_options = syntax;
1054 /* This table gives an error message for each of the error codes listed
1055 in regex.h. Obviously the order here has to be same as there.
1056 POSIX doesn't require that we do anything for REG_NOERROR,
1057 but why not be nice? */
1059 static const char *re_error_msgid[] =
1061 gettext_noop ("Success"), /* REG_NOERROR */
1062 gettext_noop ("No match"), /* REG_NOMATCH */
1063 gettext_noop ("Invalid regular expression"), /* REG_BADPAT */
1064 gettext_noop ("Invalid collation character"), /* REG_ECOLLATE */
1065 gettext_noop ("Invalid character class name"), /* REG_ECTYPE */
1066 gettext_noop ("Trailing backslash"), /* REG_EESCAPE */
1067 gettext_noop ("Invalid back reference"), /* REG_ESUBREG */
1068 gettext_noop ("Unmatched [ or [^"), /* REG_EBRACK */
1069 gettext_noop ("Unmatched ( or \\("), /* REG_EPAREN */
1070 gettext_noop ("Unmatched \\{"), /* REG_EBRACE */
1071 gettext_noop ("Invalid content of \\{\\}"), /* REG_BADBR */
1072 gettext_noop ("Invalid range end"), /* REG_ERANGE */
1073 gettext_noop ("Memory exhausted"), /* REG_ESPACE */
1074 gettext_noop ("Invalid preceding regular expression"), /* REG_BADRPT */
1075 gettext_noop ("Premature end of regular expression"), /* REG_EEND */
1076 gettext_noop ("Regular expression too big"), /* REG_ESIZE */
1077 gettext_noop ("Unmatched ) or \\)"), /* REG_ERPAREN */
1080 /* Avoiding alloca during matching, to placate r_alloc. */
1082 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1083 searching and matching functions should not call alloca. On some
1084 systems, alloca is implemented in terms of malloc, and if we're
1085 using the relocating allocator routines, then malloc could cause a
1086 relocation, which might (if the strings being searched are in the
1087 ralloc heap) shift the data out from underneath the regexp
1090 Here's another reason to avoid allocation: Emacs
1091 processes input from X in a signal handler; processing X input may
1092 call malloc; if input arrives while a matching routine is calling
1093 malloc, then we're scrod. But Emacs can't just block input while
1094 calling matching routines; then we don't notice interrupts when
1095 they come in. So, Emacs blocks input around all regexp calls
1096 except the matching calls, which it leaves unprotected, in the
1097 faith that they will not malloc. */
1099 /* Normally, this is fine. */
1100 #define MATCH_MAY_ALLOCATE
1102 /* When using GNU C, we are not REALLY using the C alloca, no matter
1103 what config.h may say. So don't take precautions for it. */
1108 /* The match routines may not allocate if (1) they would do it with malloc
1109 and (2) it's not safe for them to use malloc.
1110 Note that if REL_ALLOC is defined, matching would not use malloc for the
1111 failure stack, but we would still use it for the register vectors;
1112 so REL_ALLOC should not affect this. */
1113 #if (defined (C_ALLOCA) || defined (REGEX_MALLOC)) && defined (emacs)
1114 #undef MATCH_MAY_ALLOCATE
1118 /* Failure stack declarations and macros; both re_compile_fastmap and
1119 re_match_2 use a failure stack. These have to be macros because of
1120 REGEX_ALLOCATE_STACK. */
1123 /* Number of failure points for which to initially allocate space
1124 when matching. If this number is exceeded, we allocate more
1125 space, so it is not a hard limit. */
1126 #ifndef INIT_FAILURE_ALLOC
1127 #define INIT_FAILURE_ALLOC 5
1130 /* Roughly the maximum number of failure points on the stack. Would be
1131 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1132 This is a variable only so users of regex can assign to it; we never
1133 change it ourselves. */
1134 #if defined (MATCH_MAY_ALLOCATE)
1135 /* 4400 was enough to cause a crash on Alpha OSF/1,
1136 whose default stack limit is 2mb. */
1137 int re_max_failures = 20000;
1139 int re_max_failures = 2000;
1142 union fail_stack_elt
1144 unsigned char *pointer;
1148 typedef union fail_stack_elt fail_stack_elt_t;
1152 fail_stack_elt_t *stack;
1154 unsigned avail; /* Offset of next open position. */
1157 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1158 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1159 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1162 /* Define macros to initialize and free the failure stack.
1163 Do `return -2' if the alloc fails. */
1165 #ifdef MATCH_MAY_ALLOCATE
1166 #define INIT_FAIL_STACK() \
1168 fail_stack.stack = (fail_stack_elt_t *) \
1169 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1171 if (fail_stack.stack == NULL) \
1174 fail_stack.size = INIT_FAILURE_ALLOC; \
1175 fail_stack.avail = 0; \
1178 #define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1180 #define INIT_FAIL_STACK() \
1182 fail_stack.avail = 0; \
1185 #define RESET_FAIL_STACK()
1189 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1191 Return 1 if succeeds, and 0 if either ran out of memory
1192 allocating space for it or it was already too large.
1194 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1196 #define DOUBLE_FAIL_STACK(fail_stack) \
1197 ((fail_stack).size > re_max_failures * MAX_FAILURE_ITEMS \
1199 : ((fail_stack).stack = (fail_stack_elt_t *) \
1200 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1201 (fail_stack).size * sizeof (fail_stack_elt_t), \
1202 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1204 (fail_stack).stack == NULL \
1206 : ((fail_stack).size <<= 1, \
1210 /* Push pointer POINTER on FAIL_STACK.
1211 Return 1 if was able to do so and 0 if ran out of memory allocating
1213 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1214 ((FAIL_STACK_FULL () \
1215 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1217 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1220 /* Push a pointer value onto the failure stack.
1221 Assumes the variable `fail_stack'. Probably should only
1222 be called from within `PUSH_FAILURE_POINT'. */
1223 #define PUSH_FAILURE_POINTER(item) \
1224 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1226 /* This pushes an integer-valued item onto the failure stack.
1227 Assumes the variable `fail_stack'. Probably should only
1228 be called from within `PUSH_FAILURE_POINT'. */
1229 #define PUSH_FAILURE_INT(item) \
1230 fail_stack.stack[fail_stack.avail++].integer = (item)
1232 /* Push a fail_stack_elt_t value onto the failure stack.
1233 Assumes the variable `fail_stack'. Probably should only
1234 be called from within `PUSH_FAILURE_POINT'. */
1235 #define PUSH_FAILURE_ELT(item) \
1236 fail_stack.stack[fail_stack.avail++] = (item)
1238 /* These three POP... operations complement the three PUSH... operations.
1239 All assume that `fail_stack' is nonempty. */
1240 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1241 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1242 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1244 /* Used to omit pushing failure point id's when we're not debugging. */
1246 #define DEBUG_PUSH PUSH_FAILURE_INT
1247 #define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1249 #define DEBUG_PUSH(item)
1250 #define DEBUG_POP(item_addr)
1254 /* Push the information about the state we will need
1255 if we ever fail back to it.
1257 Requires variables fail_stack, regstart, regend, reg_info, and
1258 num_regs be declared. DOUBLE_FAIL_STACK requires `destination' be
1261 Does `return FAILURE_CODE' if runs out of memory. */
1263 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1265 char *destination; \
1266 /* Must be int, so when we don't save any registers, the arithmetic \
1267 of 0 + -1 isn't done as unsigned. */ \
1270 DEBUG_STATEMENT (failure_id++); \
1271 DEBUG_STATEMENT (nfailure_points_pushed++); \
1272 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1273 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1274 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1276 DEBUG_PRINT2 (" slots needed: %d\n", NUM_FAILURE_ITEMS); \
1277 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1279 /* Ensure we have enough space allocated for what we will push. */ \
1280 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1282 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1283 return failure_code; \
1285 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1286 (fail_stack).size); \
1287 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1290 /* Push the info, starting with the registers. */ \
1291 DEBUG_PRINT1 ("\n"); \
1294 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1297 DEBUG_PRINT2 (" Pushing reg: %d\n", this_reg); \
1298 DEBUG_STATEMENT (num_regs_pushed++); \
1300 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1301 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1303 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1304 PUSH_FAILURE_POINTER (regend[this_reg]); \
1306 DEBUG_PRINT2 (" info: 0x%x\n ", reg_info[this_reg]); \
1307 DEBUG_PRINT2 (" match_null=%d", \
1308 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1309 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1310 DEBUG_PRINT2 (" matched_something=%d", \
1311 MATCHED_SOMETHING (reg_info[this_reg])); \
1312 DEBUG_PRINT2 (" ever_matched=%d", \
1313 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1314 DEBUG_PRINT1 ("\n"); \
1315 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1318 DEBUG_PRINT2 (" Pushing low active reg: %d\n", lowest_active_reg);\
1319 PUSH_FAILURE_INT (lowest_active_reg); \
1321 DEBUG_PRINT2 (" Pushing high active reg: %d\n", highest_active_reg);\
1322 PUSH_FAILURE_INT (highest_active_reg); \
1324 DEBUG_PRINT2 (" Pushing pattern 0x%x: ", pattern_place); \
1325 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1326 PUSH_FAILURE_POINTER (pattern_place); \
1328 DEBUG_PRINT2 (" Pushing string 0x%x: `", string_place); \
1329 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1331 DEBUG_PRINT1 ("'\n"); \
1332 PUSH_FAILURE_POINTER (string_place); \
1334 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1335 DEBUG_PUSH (failure_id); \
1338 /* This is the number of items that are pushed and popped on the stack
1339 for each register. */
1340 #define NUM_REG_ITEMS 3
1342 /* Individual items aside from the registers. */
1344 #define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1346 #define NUM_NONREG_ITEMS 4
1349 /* We push at most this many items on the stack. */
1350 /* We used to use (num_regs - 1), which is the number of registers
1351 this regexp will save; but that was changed to 5
1352 to avoid stack overflow for a regexp with lots of parens. */
1353 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1355 /* We actually push this many items. */
1356 #define NUM_FAILURE_ITEMS \
1358 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1362 /* How many items can still be added to the stack without overflowing it. */
1363 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1366 /* Pops what PUSH_FAIL_STACK pushes.
1368 We restore into the parameters, all of which should be lvalues:
1369 STR -- the saved data position.
1370 PAT -- the saved pattern position.
1371 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1372 REGSTART, REGEND -- arrays of string positions.
1373 REG_INFO -- array of information about each subexpression.
1375 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1376 `pend', `string1', `size1', `string2', and `size2'. */
1378 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1380 DEBUG_STATEMENT (fail_stack_elt_t failure_id;) \
1382 const unsigned char *string_temp; \
1384 assert (!FAIL_STACK_EMPTY ()); \
1386 /* Remove failure points and point to how many regs pushed. */ \
1387 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1388 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1389 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1391 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1393 DEBUG_POP (&failure_id); \
1394 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1396 /* If the saved string location is NULL, it came from an \
1397 on_failure_keep_string_jump opcode, and we want to throw away the \
1398 saved NULL, thus retaining our current position in the string. */ \
1399 string_temp = POP_FAILURE_POINTER (); \
1400 if (string_temp != NULL) \
1401 str = (const char *) string_temp; \
1403 DEBUG_PRINT2 (" Popping string 0x%x: `", str); \
1404 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1405 DEBUG_PRINT1 ("'\n"); \
1407 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1408 DEBUG_PRINT2 (" Popping pattern 0x%x: ", pat); \
1409 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1411 /* Restore register info. */ \
1412 high_reg = (unsigned) POP_FAILURE_INT (); \
1413 DEBUG_PRINT2 (" Popping high active reg: %d\n", high_reg); \
1415 low_reg = (unsigned) POP_FAILURE_INT (); \
1416 DEBUG_PRINT2 (" Popping low active reg: %d\n", low_reg); \
1419 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1421 DEBUG_PRINT2 (" Popping reg: %d\n", this_reg); \
1423 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1424 DEBUG_PRINT2 (" info: 0x%x\n", reg_info[this_reg]); \
1426 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1427 DEBUG_PRINT2 (" end: 0x%x\n", regend[this_reg]); \
1429 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1430 DEBUG_PRINT2 (" start: 0x%x\n", regstart[this_reg]); \
1434 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1436 reg_info[this_reg].word.integer = 0; \
1437 regend[this_reg] = 0; \
1438 regstart[this_reg] = 0; \
1440 highest_active_reg = high_reg; \
1443 set_regs_matched_done = 0; \
1444 DEBUG_STATEMENT (nfailure_points_popped++); \
1445 } /* POP_FAILURE_POINT */
1449 /* Structure for per-register (a.k.a. per-group) information.
1450 Other register information, such as the
1451 starting and ending positions (which are addresses), and the list of
1452 inner groups (which is a bits list) are maintained in separate
1455 We are making a (strictly speaking) nonportable assumption here: that
1456 the compiler will pack our bit fields into something that fits into
1457 the type of `word', i.e., is something that fits into one item on the
1462 fail_stack_elt_t word;
1465 /* This field is one if this group can match the empty string,
1466 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1467 #define MATCH_NULL_UNSET_VALUE 3
1468 unsigned match_null_string_p : 2;
1469 unsigned is_active : 1;
1470 unsigned matched_something : 1;
1471 unsigned ever_matched_something : 1;
1473 } register_info_type;
1475 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1476 #define IS_ACTIVE(R) ((R).bits.is_active)
1477 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1478 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1481 /* Call this when have matched a real character; it sets `matched' flags
1482 for the subexpressions which we are currently inside. Also records
1483 that those subexprs have matched. */
1484 #define SET_REGS_MATCHED() \
1487 if (!set_regs_matched_done) \
1490 set_regs_matched_done = 1; \
1491 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1493 MATCHED_SOMETHING (reg_info[r]) \
1494 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1501 /* Registers are set to a sentinel when they haven't yet matched. */
1502 static char reg_unset_dummy;
1503 #define REG_UNSET_VALUE (®_unset_dummy)
1504 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1506 /* Subroutine declarations and macros for regex_compile. */
1508 static void store_op1 (), store_op2 ();
1509 static void insert_op1 (), insert_op2 ();
1510 static boolean at_begline_loc_p (), at_endline_loc_p ();
1511 static boolean group_in_compile_stack ();
1512 static reg_errcode_t compile_range ();
1514 /* Fetch the next character in the uncompiled pattern---translating it
1515 if necessary. Also cast from a signed character in the constant
1516 string passed to us by the user to an unsigned char that we can use
1517 as an array index (in, e.g., `translate'). */
1519 #define PATFETCH(c) \
1520 do {if (p == pend) return REG_EEND; \
1521 c = (unsigned char) *p++; \
1522 if (translate) c = (unsigned char) translate[c]; \
1526 /* Fetch the next character in the uncompiled pattern, with no
1528 #define PATFETCH_RAW(c) \
1529 do {if (p == pend) return REG_EEND; \
1530 c = (unsigned char) *p++; \
1533 /* Go backwards one character in the pattern. */
1534 #define PATUNFETCH p--
1537 /* If `translate' is non-null, return translate[D], else just D. We
1538 cast the subscript to translate because some data is declared as
1539 `char *', to avoid warnings when a string constant is passed. But
1540 when we use a character as a subscript we must make it unsigned. */
1542 #define TRANSLATE(d) \
1543 (translate ? (unsigned char) RE_TRANSLATE (translate, (unsigned char) (d)) : (d))
1547 /* Macros for outputting the compiled pattern into `buffer'. */
1549 /* If the buffer isn't allocated when it comes in, use this. */
1550 #define INIT_BUF_SIZE 32
1552 /* Make sure we have at least N more bytes of space in buffer. */
1553 #define GET_BUFFER_SPACE(n) \
1554 while (b - bufp->buffer + (n) > bufp->allocated) \
1557 /* Make sure we have one more byte of buffer space and then add C to it. */
1558 #define BUF_PUSH(c) \
1560 GET_BUFFER_SPACE (1); \
1561 *b++ = (unsigned char) (c); \
1565 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1566 #define BUF_PUSH_2(c1, c2) \
1568 GET_BUFFER_SPACE (2); \
1569 *b++ = (unsigned char) (c1); \
1570 *b++ = (unsigned char) (c2); \
1574 /* As with BUF_PUSH_2, except for three bytes. */
1575 #define BUF_PUSH_3(c1, c2, c3) \
1577 GET_BUFFER_SPACE (3); \
1578 *b++ = (unsigned char) (c1); \
1579 *b++ = (unsigned char) (c2); \
1580 *b++ = (unsigned char) (c3); \
1584 /* Store a jump with opcode OP at LOC to location TO. We store a
1585 relative address offset by the three bytes the jump itself occupies. */
1586 #define STORE_JUMP(op, loc, to) \
1587 store_op1 (op, loc, (to) - (loc) - 3)
1589 /* Likewise, for a two-argument jump. */
1590 #define STORE_JUMP2(op, loc, to, arg) \
1591 store_op2 (op, loc, (to) - (loc) - 3, arg)
1593 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1594 #define INSERT_JUMP(op, loc, to) \
1595 insert_op1 (op, loc, (to) - (loc) - 3, b)
1597 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1598 #define INSERT_JUMP2(op, loc, to, arg) \
1599 insert_op2 (op, loc, (to) - (loc) - 3, arg, b)
1602 /* This is not an arbitrary limit: the arguments which represent offsets
1603 into the pattern are two bytes long. So if 2^16 bytes turns out to
1604 be too small, many things would have to change. */
1605 #define MAX_BUF_SIZE (1L << 16)
1608 /* Extend the buffer by twice its current size via realloc and
1609 reset the pointers that pointed into the old block to point to the
1610 correct places in the new one. If extending the buffer results in it
1611 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1612 #define EXTEND_BUFFER() \
1614 unsigned char *old_buffer = bufp->buffer; \
1615 if (bufp->allocated == MAX_BUF_SIZE) \
1617 bufp->allocated <<= 1; \
1618 if (bufp->allocated > MAX_BUF_SIZE) \
1619 bufp->allocated = MAX_BUF_SIZE; \
1620 bufp->buffer = (unsigned char *) realloc (bufp->buffer, bufp->allocated);\
1621 if (bufp->buffer == NULL) \
1622 return REG_ESPACE; \
1623 /* If the buffer moved, move all the pointers into it. */ \
1624 if (old_buffer != bufp->buffer) \
1626 b = (b - old_buffer) + bufp->buffer; \
1627 begalt = (begalt - old_buffer) + bufp->buffer; \
1628 if (fixup_alt_jump) \
1629 fixup_alt_jump = (fixup_alt_jump - old_buffer) + bufp->buffer;\
1631 laststart = (laststart - old_buffer) + bufp->buffer; \
1632 if (pending_exact) \
1633 pending_exact = (pending_exact - old_buffer) + bufp->buffer; \
1638 /* Since we have one byte reserved for the register number argument to
1639 {start,stop}_memory, the maximum number of groups we can report
1640 things about is what fits in that byte. */
1641 #define MAX_REGNUM 255
1643 /* But patterns can have more than `MAX_REGNUM' registers. We just
1644 ignore the excess. */
1645 typedef unsigned regnum_t;
1648 /* Macros for the compile stack. */
1650 /* Since offsets can go either forwards or backwards, this type needs to
1651 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1652 typedef int pattern_offset_t;
1656 pattern_offset_t begalt_offset;
1657 pattern_offset_t fixup_alt_jump;
1658 pattern_offset_t inner_group_offset;
1659 pattern_offset_t laststart_offset;
1661 } compile_stack_elt_t;
1666 compile_stack_elt_t *stack;
1668 unsigned avail; /* Offset of next open position. */
1669 } compile_stack_type;
1672 #define INIT_COMPILE_STACK_SIZE 32
1674 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1675 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1677 /* The next available element. */
1678 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1681 /* Structure to manage work area for range table. */
1682 struct range_table_work_area
1684 int *table; /* actual work area. */
1685 int allocated; /* allocated size for work area in bytes. */
1686 int used; /* actually used size in words. */
1689 /* Make sure that WORK_AREA can hold more N multibyte characters. */
1690 #define EXTEND_RANGE_TABLE_WORK_AREA(work_area, n) \
1692 if (((work_area).used + (n)) * sizeof (int) > (work_area).allocated) \
1694 (work_area).allocated += 16 * sizeof (int); \
1695 if ((work_area).table) \
1697 = (int *) realloc ((work_area).table, (work_area).allocated); \
1700 = (int *) malloc ((work_area).allocated); \
1701 if ((work_area).table == 0) \
1702 FREE_STACK_RETURN (REG_ESPACE); \
1706 /* Set a range (RANGE_START, RANGE_END) to WORK_AREA. */
1707 #define SET_RANGE_TABLE_WORK_AREA(work_area, range_start, range_end) \
1709 EXTEND_RANGE_TABLE_WORK_AREA ((work_area), 2); \
1710 (work_area).table[(work_area).used++] = (range_start); \
1711 (work_area).table[(work_area).used++] = (range_end); \
1714 /* Free allocated memory for WORK_AREA. */
1715 #define FREE_RANGE_TABLE_WORK_AREA(work_area) \
1717 if ((work_area).table) \
1718 free ((work_area).table); \
1721 #define CLEAR_RANGE_TABLE_WORK_USED(work_area) ((work_area).used = 0)
1722 #define RANGE_TABLE_WORK_USED(work_area) ((work_area).used)
1723 #define RANGE_TABLE_WORK_ELT(work_area, i) ((work_area).table[i])
1726 /* Set the bit for character C in a list. */
1727 #define SET_LIST_BIT(c) \
1728 (b[((unsigned char) (c)) / BYTEWIDTH] \
1729 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1732 /* Get the next unsigned number in the uncompiled pattern. */
1733 #define GET_UNSIGNED_NUMBER(num) \
1737 while (ISDIGIT (c)) \
1741 num = num * 10 + c - '0'; \
1749 #define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1751 #define IS_CHAR_CLASS(string) \
1752 (STREQ (string, "alpha") || STREQ (string, "upper") \
1753 || STREQ (string, "lower") || STREQ (string, "digit") \
1754 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1755 || STREQ (string, "space") || STREQ (string, "print") \
1756 || STREQ (string, "punct") || STREQ (string, "graph") \
1757 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1759 #ifndef MATCH_MAY_ALLOCATE
1761 /* If we cannot allocate large objects within re_match_2_internal,
1762 we make the fail stack and register vectors global.
1763 The fail stack, we grow to the maximum size when a regexp
1765 The register vectors, we adjust in size each time we
1766 compile a regexp, according to the number of registers it needs. */
1768 static fail_stack_type fail_stack;
1770 /* Size with which the following vectors are currently allocated.
1771 That is so we can make them bigger as needed,
1772 but never make them smaller. */
1773 static int regs_allocated_size;
1775 static const char ** regstart, ** regend;
1776 static const char ** old_regstart, ** old_regend;
1777 static const char **best_regstart, **best_regend;
1778 static register_info_type *reg_info;
1779 static const char **reg_dummy;
1780 static register_info_type *reg_info_dummy;
1782 /* Make the register vectors big enough for NUM_REGS registers,
1783 but don't make them smaller. */
1786 regex_grow_registers (num_regs)
1789 if (num_regs > regs_allocated_size)
1791 RETALLOC_IF (regstart, num_regs, const char *);
1792 RETALLOC_IF (regend, num_regs, const char *);
1793 RETALLOC_IF (old_regstart, num_regs, const char *);
1794 RETALLOC_IF (old_regend, num_regs, const char *);
1795 RETALLOC_IF (best_regstart, num_regs, const char *);
1796 RETALLOC_IF (best_regend, num_regs, const char *);
1797 RETALLOC_IF (reg_info, num_regs, register_info_type);
1798 RETALLOC_IF (reg_dummy, num_regs, const char *);
1799 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1801 regs_allocated_size = num_regs;
1805 #endif /* not MATCH_MAY_ALLOCATE */
1807 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1808 Returns one of error codes defined in `regex.h', or zero for success.
1810 Assumes the `allocated' (and perhaps `buffer') and `translate'
1811 fields are set in BUFP on entry.
1813 If it succeeds, results are put in BUFP (if it returns an error, the
1814 contents of BUFP are undefined):
1815 `buffer' is the compiled pattern;
1816 `syntax' is set to SYNTAX;
1817 `used' is set to the length of the compiled pattern;
1818 `fastmap_accurate' is zero;
1819 `re_nsub' is the number of subexpressions in PATTERN;
1820 `not_bol' and `not_eol' are zero;
1822 The `fastmap' and `newline_anchor' fields are neither
1823 examined nor set. */
1825 /* Return, freeing storage we allocated. */
1826 #define FREE_STACK_RETURN(value) \
1828 FREE_RANGE_TABLE_WORK_AREA (range_table_work); \
1829 free (compile_stack.stack); \
1833 static reg_errcode_t
1834 regex_compile (pattern, size, syntax, bufp)
1835 const char *pattern;
1837 reg_syntax_t syntax;
1838 struct re_pattern_buffer *bufp;
1840 /* We fetch characters from PATTERN here. Even though PATTERN is
1841 `char *' (i.e., signed), we declare these variables as unsigned, so
1842 they can be reliably used as array indices. */
1843 register unsigned int c, c1;
1845 /* A random temporary spot in PATTERN. */
1848 /* Points to the end of the buffer, where we should append. */
1849 register unsigned char *b;
1851 /* Keeps track of unclosed groups. */
1852 compile_stack_type compile_stack;
1854 /* Points to the current (ending) position in the pattern. */
1855 const char *p = pattern;
1856 const char *pend = pattern + size;
1858 /* How to translate the characters in the pattern. */
1859 RE_TRANSLATE_TYPE translate = bufp->translate;
1861 /* Address of the count-byte of the most recently inserted `exactn'
1862 command. This makes it possible to tell if a new exact-match
1863 character can be added to that command or if the character requires
1864 a new `exactn' command. */
1865 unsigned char *pending_exact = 0;
1867 /* Address of start of the most recently finished expression.
1868 This tells, e.g., postfix * where to find the start of its
1869 operand. Reset at the beginning of groups and alternatives. */
1870 unsigned char *laststart = 0;
1872 /* Address of beginning of regexp, or inside of last group. */
1873 unsigned char *begalt;
1875 /* Place in the uncompiled pattern (i.e., the {) to
1876 which to go back if the interval is invalid. */
1877 const char *beg_interval;
1879 /* Address of the place where a forward jump should go to the end of
1880 the containing expression. Each alternative of an `or' -- except the
1881 last -- ends with a forward jump of this sort. */
1882 unsigned char *fixup_alt_jump = 0;
1884 /* Counts open-groups as they are encountered. Remembered for the
1885 matching close-group on the compile stack, so the same register
1886 number is put in the stop_memory as the start_memory. */
1887 regnum_t regnum = 0;
1889 /* Work area for range table of charset. */
1890 struct range_table_work_area range_table_work;
1893 DEBUG_PRINT1 ("\nCompiling pattern: ");
1896 unsigned debug_count;
1898 for (debug_count = 0; debug_count < size; debug_count++)
1899 putchar (pattern[debug_count]);
1904 /* Initialize the compile stack. */
1905 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
1906 if (compile_stack.stack == NULL)
1909 compile_stack.size = INIT_COMPILE_STACK_SIZE;
1910 compile_stack.avail = 0;
1912 range_table_work.table = 0;
1913 range_table_work.allocated = 0;
1915 /* Initialize the pattern buffer. */
1916 bufp->syntax = syntax;
1917 bufp->fastmap_accurate = 0;
1918 bufp->not_bol = bufp->not_eol = 0;
1920 /* Set `used' to zero, so that if we return an error, the pattern
1921 printer (for debugging) will think there's no pattern. We reset it
1925 /* Always count groups, whether or not bufp->no_sub is set. */
1929 /* bufp->multibyte is set before regex_compile is called, so don't alter
1931 #else /* not emacs */
1932 /* Nothing is recognized as a multibyte character. */
1933 bufp->multibyte = 0;
1936 #if !defined (emacs) && !defined (SYNTAX_TABLE)
1937 /* Initialize the syntax table. */
1938 init_syntax_once ();
1941 if (bufp->allocated == 0)
1944 { /* If zero allocated, but buffer is non-null, try to realloc
1945 enough space. This loses if buffer's address is bogus, but
1946 that is the user's responsibility. */
1947 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
1950 { /* Caller did not allocate a buffer. Do it for them. */
1951 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
1953 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
1955 bufp->allocated = INIT_BUF_SIZE;
1958 begalt = b = bufp->buffer;
1960 /* Loop through the uncompiled pattern until we're at the end. */
1969 if ( /* If at start of pattern, it's an operator. */
1971 /* If context independent, it's an operator. */
1972 || syntax & RE_CONTEXT_INDEP_ANCHORS
1973 /* Otherwise, depends on what's come before. */
1974 || at_begline_loc_p (pattern, p, syntax))
1984 if ( /* If at end of pattern, it's an operator. */
1986 /* If context independent, it's an operator. */
1987 || syntax & RE_CONTEXT_INDEP_ANCHORS
1988 /* Otherwise, depends on what's next. */
1989 || at_endline_loc_p (p, pend, syntax))
1999 if ((syntax & RE_BK_PLUS_QM)
2000 || (syntax & RE_LIMITED_OPS))
2004 /* If there is no previous pattern... */
2007 if (syntax & RE_CONTEXT_INVALID_OPS)
2008 FREE_STACK_RETURN (REG_BADRPT);
2009 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2014 /* Are we optimizing this jump? */
2015 boolean keep_string_p = false;
2017 /* 1 means zero (many) matches is allowed. */
2018 char zero_times_ok = 0, many_times_ok = 0;
2020 /* If there is a sequence of repetition chars, collapse it
2021 down to just one (the right one). We can't combine
2022 interval operators with these because of, e.g., `a{2}*',
2023 which should only match an even number of `a's. */
2027 zero_times_ok |= c != '+';
2028 many_times_ok |= c != '?';
2036 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2039 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2041 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2044 if (!(c1 == '+' || c1 == '?'))
2059 /* If we get here, we found another repeat character. */
2062 /* Star, etc. applied to an empty pattern is equivalent
2063 to an empty pattern. */
2067 /* Now we know whether or not zero matches is allowed
2068 and also whether or not two or more matches is allowed. */
2070 { /* More than one repetition is allowed, so put in at the
2071 end a backward relative jump from `b' to before the next
2072 jump we're going to put in below (which jumps from
2073 laststart to after this jump).
2075 But if we are at the `*' in the exact sequence `.*\n',
2076 insert an unconditional jump backwards to the .,
2077 instead of the beginning of the loop. This way we only
2078 push a failure point once, instead of every time
2079 through the loop. */
2080 assert (p - 1 > pattern);
2082 /* Allocate the space for the jump. */
2083 GET_BUFFER_SPACE (3);
2085 /* We know we are not at the first character of the pattern,
2086 because laststart was nonzero. And we've already
2087 incremented `p', by the way, to be the character after
2088 the `*'. Do we have to do something analogous here
2089 for null bytes, because of RE_DOT_NOT_NULL? */
2090 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2092 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2093 && !(syntax & RE_DOT_NEWLINE))
2094 { /* We have .*\n. */
2095 STORE_JUMP (jump, b, laststart);
2096 keep_string_p = true;
2099 /* Anything else. */
2100 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2102 /* We've added more stuff to the buffer. */
2106 /* On failure, jump from laststart to b + 3, which will be the
2107 end of the buffer after this jump is inserted. */
2108 GET_BUFFER_SPACE (3);
2109 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2117 /* At least one repetition is required, so insert a
2118 `dummy_failure_jump' before the initial
2119 `on_failure_jump' instruction of the loop. This
2120 effects a skip over that instruction the first time
2121 we hit that loop. */
2122 GET_BUFFER_SPACE (3);
2123 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2138 CLEAR_RANGE_TABLE_WORK_USED (range_table_work);
2140 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2142 /* Ensure that we have enough space to push a charset: the
2143 opcode, the length count, and the bitset; 34 bytes in all. */
2144 GET_BUFFER_SPACE (34);
2148 /* We test `*p == '^' twice, instead of using an if
2149 statement, so we only need one BUF_PUSH. */
2150 BUF_PUSH (*p == '^' ? charset_not : charset);
2154 /* Remember the first position in the bracket expression. */
2157 /* Push the number of bytes in the bitmap. */
2158 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2160 /* Clear the whole map. */
2161 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2163 /* charset_not matches newline according to a syntax bit. */
2164 if ((re_opcode_t) b[-2] == charset_not
2165 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2166 SET_LIST_BIT ('\n');
2168 /* Read in characters and ranges, setting map bits. */
2172 boolean escaped_char = false;
2174 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2178 /* \ might escape characters inside [...] and [^...]. */
2179 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2181 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2184 escaped_char = true;
2188 /* Could be the end of the bracket expression. If it's
2189 not (i.e., when the bracket expression is `[]' so
2190 far), the ']' character bit gets set way below. */
2191 if (c == ']' && p != p1 + 1)
2195 /* If C indicates start of multibyte char, get the
2196 actual character code in C, and set the pattern
2197 pointer P to the next character boundary. */
2198 if (bufp->multibyte && BASE_LEADING_CODE_P (c))
2201 c = STRING_CHAR_AND_LENGTH (p, pend - p, len);
2204 /* What should we do for the character which is
2205 greater than 0x7F, but not BASE_LEADING_CODE_P?
2208 /* See if we're at the beginning of a possible character
2211 else if (!escaped_char &&
2212 syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2214 /* Leave room for the null. */
2215 char str[CHAR_CLASS_MAX_LENGTH + 1];
2220 /* If pattern is `[[:'. */
2221 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2226 if (c == ':' || c == ']' || p == pend
2227 || c1 == CHAR_CLASS_MAX_LENGTH)
2233 /* If isn't a word bracketed by `[:' and `:]':
2234 undo the ending character, the letters, and
2235 leave the leading `:' and `[' (but set bits for
2237 if (c == ':' && *p == ']')
2240 boolean is_alnum = STREQ (str, "alnum");
2241 boolean is_alpha = STREQ (str, "alpha");
2242 boolean is_blank = STREQ (str, "blank");
2243 boolean is_cntrl = STREQ (str, "cntrl");
2244 boolean is_digit = STREQ (str, "digit");
2245 boolean is_graph = STREQ (str, "graph");
2246 boolean is_lower = STREQ (str, "lower");
2247 boolean is_print = STREQ (str, "print");
2248 boolean is_punct = STREQ (str, "punct");
2249 boolean is_space = STREQ (str, "space");
2250 boolean is_upper = STREQ (str, "upper");
2251 boolean is_xdigit = STREQ (str, "xdigit");
2253 if (!IS_CHAR_CLASS (str))
2254 FREE_STACK_RETURN (REG_ECTYPE);
2256 /* Throw away the ] at the end of the character
2260 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2262 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2264 int translated = TRANSLATE (ch);
2265 /* This was split into 3 if's to
2266 avoid an arbitrary limit in some compiler. */
2267 if ( (is_alnum && ISALNUM (ch))
2268 || (is_alpha && ISALPHA (ch))
2269 || (is_blank && ISBLANK (ch))
2270 || (is_cntrl && ISCNTRL (ch)))
2271 SET_LIST_BIT (translated);
2272 if ( (is_digit && ISDIGIT (ch))
2273 || (is_graph && ISGRAPH (ch))
2274 || (is_lower && ISLOWER (ch))
2275 || (is_print && ISPRINT (ch)))
2276 SET_LIST_BIT (translated);
2277 if ( (is_punct && ISPUNCT (ch))
2278 || (is_space && ISSPACE (ch))
2279 || (is_upper && ISUPPER (ch))
2280 || (is_xdigit && ISXDIGIT (ch)))
2281 SET_LIST_BIT (translated);
2284 /* Repeat the loop. */
2294 /* Because the `:' may starts the range, we
2295 can't simply set bit and repeat the loop.
2296 Instead, just set it to C and handle below. */
2301 if (p < pend && p[0] == '-' && p[1] != ']')
2304 /* Discard the `-'. */
2307 /* Fetch the character which ends the range. */
2309 if (bufp->multibyte && BASE_LEADING_CODE_P (c1))
2312 c1 = STRING_CHAR_AND_LENGTH (p, pend - p, len);
2316 if (!SAME_CHARSET_P (c, c1))
2317 FREE_STACK_RETURN (REG_ERANGE);
2320 /* Range from C to C. */
2323 /* Set the range ... */
2324 if (SINGLE_BYTE_CHAR_P (c))
2325 /* ... into bitmap. */
2328 int range_start = c, range_end = c1;
2330 /* If the start is after the end, the range is empty. */
2331 if (range_start > range_end)
2333 if (syntax & RE_NO_EMPTY_RANGES)
2334 FREE_STACK_RETURN (REG_ERANGE);
2335 /* Else, repeat the loop. */
2339 for (this_char = range_start; this_char <= range_end;
2341 SET_LIST_BIT (TRANSLATE (this_char));
2345 /* ... into range table. */
2346 SET_RANGE_TABLE_WORK_AREA (range_table_work, c, c1);
2349 /* Discard any (non)matching list bytes that are all 0 at the
2350 end of the map. Decrease the map-length byte too. */
2351 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2355 /* Build real range table from work area. */
2356 if (RANGE_TABLE_WORK_USED (range_table_work))
2359 int used = RANGE_TABLE_WORK_USED (range_table_work);
2361 /* Allocate space for COUNT + RANGE_TABLE. Needs two
2362 bytes for COUNT and three bytes for each character. */
2363 GET_BUFFER_SPACE (2 + used * 3);
2365 /* Indicate the existence of range table. */
2366 laststart[1] |= 0x80;
2368 STORE_NUMBER_AND_INCR (b, used / 2);
2369 for (i = 0; i < used; i++)
2370 STORE_CHARACTER_AND_INCR
2371 (b, RANGE_TABLE_WORK_ELT (range_table_work, i));
2378 if (syntax & RE_NO_BK_PARENS)
2385 if (syntax & RE_NO_BK_PARENS)
2392 if (syntax & RE_NEWLINE_ALT)
2399 if (syntax & RE_NO_BK_VBAR)
2406 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2407 goto handle_interval;
2413 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2415 /* Do not translate the character after the \, so that we can
2416 distinguish, e.g., \B from \b, even if we normally would
2417 translate, e.g., B to b. */
2423 if (syntax & RE_NO_BK_PARENS)
2424 goto normal_backslash;
2430 if (COMPILE_STACK_FULL)
2432 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2433 compile_stack_elt_t);
2434 if (compile_stack.stack == NULL) return REG_ESPACE;
2436 compile_stack.size <<= 1;
2439 /* These are the values to restore when we hit end of this
2440 group. They are all relative offsets, so that if the
2441 whole pattern moves because of realloc, they will still
2443 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2444 COMPILE_STACK_TOP.fixup_alt_jump
2445 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2446 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2447 COMPILE_STACK_TOP.regnum = regnum;
2449 /* We will eventually replace the 0 with the number of
2450 groups inner to this one. But do not push a
2451 start_memory for groups beyond the last one we can
2452 represent in the compiled pattern. */
2453 if (regnum <= MAX_REGNUM)
2455 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2456 BUF_PUSH_3 (start_memory, regnum, 0);
2459 compile_stack.avail++;
2464 /* If we've reached MAX_REGNUM groups, then this open
2465 won't actually generate any code, so we'll have to
2466 clear pending_exact explicitly. */
2472 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2474 if (COMPILE_STACK_EMPTY)
2475 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2476 goto normal_backslash;
2478 FREE_STACK_RETURN (REG_ERPAREN);
2482 { /* Push a dummy failure point at the end of the
2483 alternative for a possible future
2484 `pop_failure_jump' to pop. See comments at
2485 `push_dummy_failure' in `re_match_2'. */
2486 BUF_PUSH (push_dummy_failure);
2488 /* We allocated space for this jump when we assigned
2489 to `fixup_alt_jump', in the `handle_alt' case below. */
2490 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2493 /* See similar code for backslashed left paren above. */
2494 if (COMPILE_STACK_EMPTY)
2495 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2498 FREE_STACK_RETURN (REG_ERPAREN);
2500 /* Since we just checked for an empty stack above, this
2501 ``can't happen''. */
2502 assert (compile_stack.avail != 0);
2504 /* We don't just want to restore into `regnum', because
2505 later groups should continue to be numbered higher,
2506 as in `(ab)c(de)' -- the second group is #2. */
2507 regnum_t this_group_regnum;
2509 compile_stack.avail--;
2510 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2512 = COMPILE_STACK_TOP.fixup_alt_jump
2513 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2515 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2516 this_group_regnum = COMPILE_STACK_TOP.regnum;
2517 /* If we've reached MAX_REGNUM groups, then this open
2518 won't actually generate any code, so we'll have to
2519 clear pending_exact explicitly. */
2522 /* We're at the end of the group, so now we know how many
2523 groups were inside this one. */
2524 if (this_group_regnum <= MAX_REGNUM)
2526 unsigned char *inner_group_loc
2527 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2529 *inner_group_loc = regnum - this_group_regnum;
2530 BUF_PUSH_3 (stop_memory, this_group_regnum,
2531 regnum - this_group_regnum);
2537 case '|': /* `\|'. */
2538 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2539 goto normal_backslash;
2541 if (syntax & RE_LIMITED_OPS)
2544 /* Insert before the previous alternative a jump which
2545 jumps to this alternative if the former fails. */
2546 GET_BUFFER_SPACE (3);
2547 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2551 /* The alternative before this one has a jump after it
2552 which gets executed if it gets matched. Adjust that
2553 jump so it will jump to this alternative's analogous
2554 jump (put in below, which in turn will jump to the next
2555 (if any) alternative's such jump, etc.). The last such
2556 jump jumps to the correct final destination. A picture:
2562 If we are at `b', then fixup_alt_jump right now points to a
2563 three-byte space after `a'. We'll put in the jump, set
2564 fixup_alt_jump to right after `b', and leave behind three
2565 bytes which we'll fill in when we get to after `c'. */
2568 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2570 /* Mark and leave space for a jump after this alternative,
2571 to be filled in later either by next alternative or
2572 when know we're at the end of a series of alternatives. */
2574 GET_BUFFER_SPACE (3);
2583 /* If \{ is a literal. */
2584 if (!(syntax & RE_INTERVALS)
2585 /* If we're at `\{' and it's not the open-interval
2587 || ((syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2588 || (p - 2 == pattern && p == pend))
2589 goto normal_backslash;
2593 /* If got here, then the syntax allows intervals. */
2595 /* At least (most) this many matches must be made. */
2596 int lower_bound = -1, upper_bound = -1;
2598 beg_interval = p - 1;
2602 if (syntax & RE_NO_BK_BRACES)
2603 goto unfetch_interval;
2605 FREE_STACK_RETURN (REG_EBRACE);
2608 GET_UNSIGNED_NUMBER (lower_bound);
2612 GET_UNSIGNED_NUMBER (upper_bound);
2613 if (upper_bound < 0) upper_bound = RE_DUP_MAX;
2616 /* Interval such as `{1}' => match exactly once. */
2617 upper_bound = lower_bound;
2619 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
2620 || lower_bound > upper_bound)
2622 if (syntax & RE_NO_BK_BRACES)
2623 goto unfetch_interval;
2625 FREE_STACK_RETURN (REG_BADBR);
2628 if (!(syntax & RE_NO_BK_BRACES))
2630 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
2637 if (syntax & RE_NO_BK_BRACES)
2638 goto unfetch_interval;
2640 FREE_STACK_RETURN (REG_BADBR);
2643 /* We just parsed a valid interval. */
2645 /* If it's invalid to have no preceding re. */
2648 if (syntax & RE_CONTEXT_INVALID_OPS)
2649 FREE_STACK_RETURN (REG_BADRPT);
2650 else if (syntax & RE_CONTEXT_INDEP_OPS)
2653 goto unfetch_interval;
2656 /* If the upper bound is zero, don't want to succeed at
2657 all; jump from `laststart' to `b + 3', which will be
2658 the end of the buffer after we insert the jump. */
2659 if (upper_bound == 0)
2661 GET_BUFFER_SPACE (3);
2662 INSERT_JUMP (jump, laststart, b + 3);
2666 /* Otherwise, we have a nontrivial interval. When
2667 we're all done, the pattern will look like:
2668 set_number_at <jump count> <upper bound>
2669 set_number_at <succeed_n count> <lower bound>
2670 succeed_n <after jump addr> <succeed_n count>
2672 jump_n <succeed_n addr> <jump count>
2673 (The upper bound and `jump_n' are omitted if
2674 `upper_bound' is 1, though.) */
2676 { /* If the upper bound is > 1, we need to insert
2677 more at the end of the loop. */
2678 unsigned nbytes = 10 + (upper_bound > 1) * 10;
2680 GET_BUFFER_SPACE (nbytes);
2682 /* Initialize lower bound of the `succeed_n', even
2683 though it will be set during matching by its
2684 attendant `set_number_at' (inserted next),
2685 because `re_compile_fastmap' needs to know.
2686 Jump to the `jump_n' we might insert below. */
2687 INSERT_JUMP2 (succeed_n, laststart,
2688 b + 5 + (upper_bound > 1) * 5,
2692 /* Code to initialize the lower bound. Insert
2693 before the `succeed_n'. The `5' is the last two
2694 bytes of this `set_number_at', plus 3 bytes of
2695 the following `succeed_n'. */
2696 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
2699 if (upper_bound > 1)
2700 { /* More than one repetition is allowed, so
2701 append a backward jump to the `succeed_n'
2702 that starts this interval.
2704 When we've reached this during matching,
2705 we'll have matched the interval once, so
2706 jump back only `upper_bound - 1' times. */
2707 STORE_JUMP2 (jump_n, b, laststart + 5,
2711 /* The location we want to set is the second
2712 parameter of the `jump_n'; that is `b-2' as
2713 an absolute address. `laststart' will be
2714 the `set_number_at' we're about to insert;
2715 `laststart+3' the number to set, the source
2716 for the relative address. But we are
2717 inserting into the middle of the pattern --
2718 so everything is getting moved up by 5.
2719 Conclusion: (b - 2) - (laststart + 3) + 5,
2720 i.e., b - laststart.
2722 We insert this at the beginning of the loop
2723 so that if we fail during matching, we'll
2724 reinitialize the bounds. */
2725 insert_op2 (set_number_at, laststart, b - laststart,
2726 upper_bound - 1, b);
2731 beg_interval = NULL;
2736 /* If an invalid interval, match the characters as literals. */
2737 assert (beg_interval);
2739 beg_interval = NULL;
2741 /* normal_char and normal_backslash need `c'. */
2744 if (!(syntax & RE_NO_BK_BRACES))
2746 if (p > pattern && p[-1] == '\\')
2747 goto normal_backslash;
2752 /* There is no way to specify the before_dot and after_dot
2753 operators. rms says this is ok. --karl */
2761 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
2767 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
2773 BUF_PUSH_2 (categoryspec, c);
2779 BUF_PUSH_2 (notcategoryspec, c);
2786 BUF_PUSH (wordchar);
2792 BUF_PUSH (notwordchar);
2805 BUF_PUSH (wordbound);
2809 BUF_PUSH (notwordbound);
2820 case '1': case '2': case '3': case '4': case '5':
2821 case '6': case '7': case '8': case '9':
2822 if (syntax & RE_NO_BK_REFS)
2828 FREE_STACK_RETURN (REG_ESUBREG);
2830 /* Can't back reference to a subexpression if inside of it. */
2831 if (group_in_compile_stack (compile_stack, c1))
2835 BUF_PUSH_2 (duplicate, c1);
2841 if (syntax & RE_BK_PLUS_QM)
2844 goto normal_backslash;
2848 /* You might think it would be useful for \ to mean
2849 not to translate; but if we don't translate it
2850 it will never match anything. */
2858 /* Expects the character in `c'. */
2860 p1 = p - 1; /* P1 points the head of C. */
2862 if (bufp->multibyte)
2863 /* Set P to the next character boundary. */
2864 p += MULTIBYTE_FORM_LENGTH (p1, pend - p1) - 1;
2866 /* If no exactn currently being built. */
2869 /* If last exactn not at current position. */
2870 || pending_exact + *pending_exact + 1 != b
2872 /* We have only one byte following the exactn for the count. */
2873 || *pending_exact >= (1 << BYTEWIDTH) - (p - p1)
2875 /* If followed by a repetition operator. */
2876 || *p == '*' || *p == '^'
2877 || ((syntax & RE_BK_PLUS_QM)
2878 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
2879 : (*p == '+' || *p == '?'))
2880 || ((syntax & RE_INTERVALS)
2881 && ((syntax & RE_NO_BK_BRACES)
2883 : (p[0] == '\\' && p[1] == '{'))))
2885 /* Start building a new exactn. */
2889 BUF_PUSH_2 (exactn, 0);
2890 pending_exact = b - 1;
2893 /* Here, C may translated, therefore C may not equal to *P1. */
2901 /* Rest of multibyte form should be copied literally. */
2902 c = *(unsigned char *)p1;
2906 } /* while p != pend */
2909 /* Through the pattern now. */
2912 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2914 if (!COMPILE_STACK_EMPTY)
2915 FREE_STACK_RETURN (REG_EPAREN);
2917 /* If we don't want backtracking, force success
2918 the first time we reach the end of the compiled pattern. */
2919 if (syntax & RE_NO_POSIX_BACKTRACKING)
2922 free (compile_stack.stack);
2924 /* We have succeeded; set the length of the buffer. */
2925 bufp->used = b - bufp->buffer;
2930 DEBUG_PRINT1 ("\nCompiled pattern: \n");
2931 print_compiled_pattern (bufp);
2935 #ifndef MATCH_MAY_ALLOCATE
2936 /* Initialize the failure stack to the largest possible stack. This
2937 isn't necessary unless we're trying to avoid calling alloca in
2938 the search and match routines. */
2940 int num_regs = bufp->re_nsub + 1;
2942 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
2943 is strictly greater than re_max_failures, the largest possible stack
2944 is 2 * re_max_failures failure points. */
2945 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
2947 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
2950 if (! fail_stack.stack)
2952 = (fail_stack_elt_t *) xmalloc (fail_stack.size
2953 * sizeof (fail_stack_elt_t));
2956 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
2958 * sizeof (fail_stack_elt_t)));
2959 #else /* not emacs */
2960 if (! fail_stack.stack)
2962 = (fail_stack_elt_t *) malloc (fail_stack.size
2963 * sizeof (fail_stack_elt_t));
2966 = (fail_stack_elt_t *) realloc (fail_stack.stack,
2968 * sizeof (fail_stack_elt_t)));
2969 #endif /* not emacs */
2972 regex_grow_registers (num_regs);
2974 #endif /* not MATCH_MAY_ALLOCATE */
2977 } /* regex_compile */
2979 /* Subroutines for `regex_compile'. */
2981 /* Store OP at LOC followed by two-byte integer parameter ARG. */
2984 store_op1 (op, loc, arg)
2989 *loc = (unsigned char) op;
2990 STORE_NUMBER (loc + 1, arg);
2994 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
2997 store_op2 (op, loc, arg1, arg2)
3002 *loc = (unsigned char) op;
3003 STORE_NUMBER (loc + 1, arg1);
3004 STORE_NUMBER (loc + 3, arg2);
3008 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3009 for OP followed by two-byte integer parameter ARG. */
3012 insert_op1 (op, loc, arg, end)
3018 register unsigned char *pfrom = end;
3019 register unsigned char *pto = end + 3;
3021 while (pfrom != loc)
3024 store_op1 (op, loc, arg);
3028 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3031 insert_op2 (op, loc, arg1, arg2, end)
3037 register unsigned char *pfrom = end;
3038 register unsigned char *pto = end + 5;
3040 while (pfrom != loc)
3043 store_op2 (op, loc, arg1, arg2);
3047 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3048 after an alternative or a begin-subexpression. We assume there is at
3049 least one character before the ^. */
3052 at_begline_loc_p (pattern, p, syntax)
3053 const char *pattern, *p;
3054 reg_syntax_t syntax;
3056 const char *prev = p - 2;
3057 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
3060 /* After a subexpression? */
3061 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3062 /* After an alternative? */
3063 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
3067 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3068 at least one character after the $, i.e., `P < PEND'. */
3071 at_endline_loc_p (p, pend, syntax)
3072 const char *p, *pend;
3075 const char *next = p;
3076 boolean next_backslash = *next == '\\';
3077 const char *next_next = p + 1 < pend ? p + 1 : 0;
3080 /* Before a subexpression? */
3081 (syntax & RE_NO_BK_PARENS ? *next == ')'
3082 : next_backslash && next_next && *next_next == ')')
3083 /* Before an alternative? */
3084 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3085 : next_backslash && next_next && *next_next == '|');
3089 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3090 false if it's not. */
3093 group_in_compile_stack (compile_stack, regnum)
3094 compile_stack_type compile_stack;
3099 for (this_element = compile_stack.avail - 1;
3102 if (compile_stack.stack[this_element].regnum == regnum)
3109 /* Read the ending character of a range (in a bracket expression) from the
3110 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3111 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3112 Then we set the translation of all bits between the starting and
3113 ending characters (inclusive) in the compiled pattern B.
3115 Return an error code.
3117 We use these short variable names so we can use the same macros as
3118 `regex_compile' itself. */
3120 static reg_errcode_t
3121 compile_range (p_ptr, pend, translate, syntax, b)
3122 const char **p_ptr, *pend;
3123 RE_TRANSLATE_TYPE translate;
3124 reg_syntax_t syntax;
3129 const char *p = *p_ptr;
3130 int range_start, range_end;
3135 /* Even though the pattern is a signed `char *', we need to fetch
3136 with unsigned char *'s; if the high bit of the pattern character
3137 is set, the range endpoints will be negative if we fetch using a
3140 We also want to fetch the endpoints without translating them; the
3141 appropriate translation is done in the bit-setting loop below. */
3142 /* The SVR4 compiler on the 3B2 had trouble with unsigned const char *. */
3143 range_start = ((const unsigned char *) p)[-2];
3144 range_end = ((const unsigned char *) p)[0];
3146 /* Have to increment the pointer into the pattern string, so the
3147 caller isn't still at the ending character. */
3150 /* If the start is after the end, the range is empty. */
3151 if (range_start > range_end)
3152 return syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3154 /* Here we see why `this_char' has to be larger than an `unsigned
3155 char' -- the range is inclusive, so if `range_end' == 0xff
3156 (assuming 8-bit characters), we would otherwise go into an infinite
3157 loop, since all characters <= 0xff. */
3158 for (this_char = range_start; this_char <= range_end; this_char++)
3160 SET_LIST_BIT (TRANSLATE (this_char));
3166 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3167 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3168 characters can start a string that matches the pattern. This fastmap
3169 is used by re_search to skip quickly over impossible starting points.
3171 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3172 area as BUFP->fastmap.
3174 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3177 Returns 0 if we succeed, -2 if an internal error. */
3180 re_compile_fastmap (bufp)
3181 struct re_pattern_buffer *bufp;
3184 #ifdef MATCH_MAY_ALLOCATE
3185 fail_stack_type fail_stack;
3187 #ifndef REGEX_MALLOC
3190 /* We don't push any register information onto the failure stack. */
3191 unsigned num_regs = 0;
3193 register char *fastmap = bufp->fastmap;
3194 unsigned char *pattern = bufp->buffer;
3195 unsigned long size = bufp->used;
3196 unsigned char *p = pattern;
3197 register unsigned char *pend = pattern + size;
3199 /* This holds the pointer to the failure stack, when
3200 it is allocated relocatably. */
3201 fail_stack_elt_t *failure_stack_ptr;
3203 /* Assume that each path through the pattern can be null until
3204 proven otherwise. We set this false at the bottom of switch
3205 statement, to which we get only if a particular path doesn't
3206 match the empty string. */
3207 boolean path_can_be_null = true;
3209 /* We aren't doing a `succeed_n' to begin with. */
3210 boolean succeed_n_p = false;
3212 /* If all elements for base leading-codes in fastmap is set, this
3213 flag is set true. */
3214 boolean match_any_multibyte_characters = false;
3216 /* Maximum code of simple (single byte) character. */
3217 int simple_char_max;
3219 assert (fastmap != NULL && p != NULL);
3222 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3223 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3224 bufp->can_be_null = 0;
3228 if (p == pend || *p == succeed)
3230 /* We have reached the (effective) end of pattern. */
3231 if (!FAIL_STACK_EMPTY ())
3233 bufp->can_be_null |= path_can_be_null;
3235 /* Reset for next path. */
3236 path_can_be_null = true;
3238 p = fail_stack.stack[--fail_stack.avail].pointer;
3246 /* We should never be about to go beyond the end of the pattern. */
3249 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3252 /* I guess the idea here is to simply not bother with a fastmap
3253 if a backreference is used, since it's too hard to figure out
3254 the fastmap for the corresponding group. Setting
3255 `can_be_null' stops `re_search_2' from using the fastmap, so
3256 that is all we do. */
3258 bufp->can_be_null = 1;
3262 /* Following are the cases which match a character. These end
3272 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3273 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3279 /* Chars beyond end of map must be allowed. */
3280 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3283 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3284 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3290 for (j = 0; j < (1 << BYTEWIDTH); j++)
3291 if (SYNTAX (j) == Sword)
3297 for (j = 0; j < (1 << BYTEWIDTH); j++)
3298 if (SYNTAX (j) != Sword)
3303 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH - 1, p++;
3305 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3308 if (CHARSET_RANGE_TABLE_EXISTS_P (&p[-2])
3309 && match_any_multibyte_characters == false)
3311 /* Set fastmap[I] 1 where I is a base leading code of each
3312 multibyte character in the range table. */
3315 /* Make P points the range table. */
3316 p += CHARSET_BITMAP_SIZE (&p[-2]);
3318 /* Extract the number of ranges in range table into
3320 EXTRACT_NUMBER_AND_INCR (count, p);
3321 for (; count > 0; count--, p += 2 * 3) /* XXX */
3323 /* Extract the start of each range. */
3324 EXTRACT_CHARACTER (c, p);
3325 j = CHAR_CHARSET (c);
3326 fastmap[CHARSET_LEADING_CODE_BASE (j)] = 1;
3333 /* Chars beyond end of map must be allowed. End of map is
3334 `127' if bufp->multibyte is nonzero. */
3335 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3336 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH;
3337 j < simple_char_max; j++)
3340 for (j = CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH - 1, p++;
3342 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3345 if (bufp->multibyte)
3346 /* Any character set can possibly contain a character
3347 which doesn't match the specified set of characters. */
3349 set_fastmap_for_multibyte_characters:
3350 if (match_any_multibyte_characters == false)
3352 for (j = 0x80; j < 0xA0; j++) /* XXX */
3353 if (BASE_LEADING_CODE_P (j))
3355 match_any_multibyte_characters = true;
3362 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3363 for (j = 0; j < simple_char_max; j++)
3364 if (SYNTAX (j) == Sword)
3367 if (bufp->multibyte)
3368 /* Any character set can possibly contain a character
3369 whose syntax is `Sword'. */
3370 goto set_fastmap_for_multibyte_characters;
3375 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3376 for (j = 0; j < simple_char_max; j++)
3377 if (SYNTAX (j) != Sword)
3380 if (bufp->multibyte)
3381 /* Any character set can possibly contain a character
3382 whose syntax is not `Sword'. */
3383 goto set_fastmap_for_multibyte_characters;
3389 int fastmap_newline = fastmap['\n'];
3391 /* `.' matches anything (but if bufp->multibyte is
3392 nonzero, matches `\000' .. `\127' and possible multibyte
3394 if (bufp->multibyte)
3396 simple_char_max = 0x80;
3398 for (j = 0x80; j < 0xA0; j++)
3399 if (BASE_LEADING_CODE_P (j))
3401 match_any_multibyte_characters = true;
3404 simple_char_max = (1 << BYTEWIDTH);
3406 for (j = 0; j < simple_char_max; j++)
3409 /* ... except perhaps newline. */
3410 if (!(bufp->syntax & RE_DOT_NEWLINE))
3411 fastmap['\n'] = fastmap_newline;
3413 /* Return if we have already set `can_be_null'; if we have,
3414 then the fastmap is irrelevant. Something's wrong here. */
3415 else if (bufp->can_be_null)
3418 /* Otherwise, have to check alternative paths. */
3429 /* This match depends on text properties. These end with
3430 aborting optimizations. */
3431 bufp->can_be_null = 1;
3435 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3436 for (j = 0; j < simple_char_max; j++)
3437 if (SYNTAX (j) == (enum syntaxcode) k)
3440 if (bufp->multibyte)
3441 /* Any character set can possibly contain a character
3442 whose syntax is K. */
3443 goto set_fastmap_for_multibyte_characters;
3448 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3449 for (j = 0; j < simple_char_max; j++)
3450 if (SYNTAX (j) != (enum syntaxcode) k)
3453 if (bufp->multibyte)
3454 /* Any character set can possibly contain a character
3455 whose syntax is not K. */
3456 goto set_fastmap_for_multibyte_characters;
3463 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3464 for (j = 0; j < simple_char_max; j++)
3465 if (CHAR_HAS_CATEGORY (j, k))
3468 if (bufp->multibyte)
3469 /* Any character set can possibly contain a character
3470 whose category is K. */
3471 goto set_fastmap_for_multibyte_characters;
3475 case notcategoryspec:
3477 simple_char_max = bufp->multibyte ? 0x80 : (1 << BYTEWIDTH);
3478 for (j = 0; j < simple_char_max; j++)
3479 if (!CHAR_HAS_CATEGORY (j, k))
3482 if (bufp->multibyte)
3483 /* Any character set can possibly contain a character
3484 whose category is not K. */
3485 goto set_fastmap_for_multibyte_characters;
3488 /* All cases after this match the empty string. These end with
3510 case push_dummy_failure:
3515 case pop_failure_jump:
3516 case maybe_pop_jump:
3519 case dummy_failure_jump:
3520 EXTRACT_NUMBER_AND_INCR (j, p);
3525 /* Jump backward implies we just went through the body of a
3526 loop and matched nothing. Opcode jumped to should be
3527 `on_failure_jump' or `succeed_n'. Just treat it like an
3528 ordinary jump. For a * loop, it has pushed its failure
3529 point already; if so, discard that as redundant. */
3530 if ((re_opcode_t) *p != on_failure_jump
3531 && (re_opcode_t) *p != succeed_n)
3535 EXTRACT_NUMBER_AND_INCR (j, p);
3538 /* If what's on the stack is where we are now, pop it. */
3539 if (!FAIL_STACK_EMPTY ()
3540 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3546 case on_failure_jump:
3547 case on_failure_keep_string_jump:
3548 handle_on_failure_jump:
3549 EXTRACT_NUMBER_AND_INCR (j, p);
3551 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3552 end of the pattern. We don't want to push such a point,
3553 since when we restore it above, entering the switch will
3554 increment `p' past the end of the pattern. We don't need
3555 to push such a point since we obviously won't find any more
3556 fastmap entries beyond `pend'. Such a pattern can match
3557 the null string, though. */
3560 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3562 RESET_FAIL_STACK ();
3567 bufp->can_be_null = 1;
3571 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3572 succeed_n_p = false;
3579 /* Get to the number of times to succeed. */
3582 /* Increment p past the n for when k != 0. */
3583 EXTRACT_NUMBER_AND_INCR (k, p);
3587 succeed_n_p = true; /* Spaghetti code alert. */
3588 goto handle_on_failure_jump;
3605 abort (); /* We have listed all the cases. */
3608 /* Getting here means we have found the possible starting
3609 characters for one path of the pattern -- and that the empty
3610 string does not match. We need not follow this path further.
3611 Instead, look at the next alternative (remembered on the
3612 stack), or quit if no more. The test at the top of the loop
3613 does these things. */
3614 path_can_be_null = false;
3618 /* Set `can_be_null' for the last path (also the first path, if the
3619 pattern is empty). */
3620 bufp->can_be_null |= path_can_be_null;
3623 RESET_FAIL_STACK ();
3625 } /* re_compile_fastmap */
3627 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3628 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3629 this memory for recording register information. STARTS and ENDS
3630 must be allocated using the malloc library routine, and must each
3631 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3633 If NUM_REGS == 0, then subsequent matches should allocate their own
3636 Unless this function is called, the first search or match using
3637 PATTERN_BUFFER will allocate its own register data, without
3638 freeing the old data. */
3641 re_set_registers (bufp, regs, num_regs, starts, ends)
3642 struct re_pattern_buffer *bufp;
3643 struct re_registers *regs;
3645 regoff_t *starts, *ends;
3649 bufp->regs_allocated = REGS_REALLOCATE;
3650 regs->num_regs = num_regs;
3651 regs->start = starts;
3656 bufp->regs_allocated = REGS_UNALLOCATED;
3658 regs->start = regs->end = (regoff_t *) 0;
3662 /* Searching routines. */
3664 /* Like re_search_2, below, but only one string is specified, and
3665 doesn't let you say where to stop matching. */
3668 re_search (bufp, string, size, startpos, range, regs)
3669 struct re_pattern_buffer *bufp;
3671 int size, startpos, range;
3672 struct re_registers *regs;
3674 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3678 /* End address of virtual concatenation of string. */
3679 #define STOP_ADDR_VSTRING(P) \
3680 (((P) >= size1 ? string2 + size2 : string1 + size1))
3682 /* Address of POS in the concatenation of virtual string. */
3683 #define POS_ADDR_VSTRING(POS) \
3684 (((POS) >= size1 ? string2 - size1 : string1) + (POS))
3686 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3687 virtual concatenation of STRING1 and STRING2, starting first at index
3688 STARTPOS, then at STARTPOS + 1, and so on.
3690 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3692 RANGE is how far to scan while trying to match. RANGE = 0 means try
3693 only at STARTPOS; in general, the last start tried is STARTPOS +
3696 In REGS, return the indices of the virtual concatenation of STRING1
3697 and STRING2 that matched the entire BUFP->buffer and its contained
3700 Do not consider matching one past the index STOP in the virtual
3701 concatenation of STRING1 and STRING2.
3703 We return either the position in the strings at which the match was
3704 found, -1 if no match, or -2 if error (such as failure
3708 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3709 struct re_pattern_buffer *bufp;
3710 const char *string1, *string2;
3714 struct re_registers *regs;
3718 register char *fastmap = bufp->fastmap;
3719 register RE_TRANSLATE_TYPE translate = bufp->translate;
3720 int total_size = size1 + size2;
3721 int endpos = startpos + range;
3722 int anchored_start = 0;
3724 /* Nonzero if we have to concern multibyte character. */
3725 int multibyte = bufp->multibyte;
3727 /* Check for out-of-range STARTPOS. */
3728 if (startpos < 0 || startpos > total_size)
3731 /* Fix up RANGE if it might eventually take us outside
3732 the virtual concatenation of STRING1 and STRING2.
3733 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3735 range = 0 - startpos;
3736 else if (endpos > total_size)
3737 range = total_size - startpos;
3739 /* If the search isn't to be a backwards one, don't waste time in a
3740 search for a pattern that must be anchored. */
3741 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == begbuf && range > 0)
3750 /* In a forward search for something that starts with \=.
3751 don't keep searching past point. */
3752 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3754 range = PT - startpos;
3760 /* Update the fastmap now if not correct already. */
3761 if (fastmap && !bufp->fastmap_accurate)
3762 if (re_compile_fastmap (bufp) == -2)
3765 /* See whether the pattern is anchored. */
3766 if (bufp->buffer[0] == begline)
3770 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object,
3771 POS_AS_IN_BUFFER (startpos > 0
3772 ? startpos - 1 : startpos),
3776 /* Loop through the string, looking for a place to start matching. */
3779 /* If the pattern is anchored,
3780 skip quickly past places we cannot match.
3781 We don't bother to treat startpos == 0 specially
3782 because that case doesn't repeat. */
3783 if (anchored_start && startpos > 0)
3785 if (! (bufp->newline_anchor
3786 && ((startpos <= size1 ? string1[startpos - 1]
3787 : string2[startpos - size1 - 1])
3792 /* If a fastmap is supplied, skip quickly over characters that
3793 cannot be the start of a match. If the pattern can match the
3794 null string, however, we don't need to skip characters; we want
3795 the first null string. */
3796 if (fastmap && startpos < total_size && !bufp->can_be_null)
3798 if (range > 0) /* Searching forwards. */
3800 register const char *d;
3801 register int lim = 0;
3804 if (startpos < size1 && startpos + range >= size1)
3805 lim = range - (size1 - startpos);
3807 d = POS_ADDR_VSTRING (startpos);
3809 /* Written out as an if-else to avoid testing `translate'
3813 && !fastmap[(unsigned char)
3814 RE_TRANSLATE (translate, (unsigned char) *d++)])
3817 while (range > lim && !fastmap[(unsigned char) *d++])
3820 startpos += irange - range;
3822 else /* Searching backwards. */
3824 register char c = (size1 == 0 || startpos >= size1
3825 ? string2[startpos - size1]
3826 : string1[startpos]);
3828 if (!fastmap[(unsigned char) TRANSLATE (c)])
3833 /* If can't match the null string, and that's all we have left, fail. */
3834 if (range >= 0 && startpos == total_size && fastmap
3835 && !bufp->can_be_null)
3838 val = re_match_2_internal (bufp, string1, size1, string2, size2,
3839 startpos, regs, stop);
3840 #ifndef REGEX_MALLOC
3857 /* Update STARTPOS to the next character boundary. */
3860 const unsigned char *p
3861 = (const unsigned char *) POS_ADDR_VSTRING (startpos);
3862 const unsigned char *pend
3863 = (const unsigned char *) STOP_ADDR_VSTRING (startpos);
3864 int len = MULTIBYTE_FORM_LENGTH (p, pend - p);
3882 /* Update STARTPOS to the previous character boundary. */
3885 const unsigned char *p
3886 = (const unsigned char *) POS_ADDR_VSTRING (startpos);
3889 /* Find the head of multibyte form. */
3890 while (!CHAR_HEAD_P (p))
3895 if (MULTIBYTE_FORM_LENGTH (p, len + 1) != (len + 1))
3912 /* Declarations and macros for re_match_2. */
3914 static int bcmp_translate ();
3915 static boolean alt_match_null_string_p (),
3916 common_op_match_null_string_p (),
3917 group_match_null_string_p ();
3919 /* This converts PTR, a pointer into one of the search strings `string1'
3920 and `string2' into an offset from the beginning of that string. */
3921 #define POINTER_TO_OFFSET(ptr) \
3922 (FIRST_STRING_P (ptr) \
3923 ? ((regoff_t) ((ptr) - string1)) \
3924 : ((regoff_t) ((ptr) - string2 + size1)))
3926 /* Macros for dealing with the split strings in re_match_2. */
3928 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
3930 /* Call before fetching a character with *d. This switches over to
3931 string2 if necessary. */
3932 #define PREFETCH() \
3935 /* End of string2 => fail. */ \
3936 if (dend == end_match_2) \
3938 /* End of string1 => advance to string2. */ \
3940 dend = end_match_2; \
3944 /* Test if at very beginning or at very end of the virtual concatenation
3945 of `string1' and `string2'. If only one string, it's `string2'. */
3946 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
3947 #define AT_STRINGS_END(d) ((d) == end2)
3950 /* Test if D points to a character which is word-constituent. We have
3951 two special cases to check for: if past the end of string1, look at
3952 the first character in string2; and if before the beginning of
3953 string2, look at the last character in string1. */
3954 #define WORDCHAR_P(d) \
3955 (SYNTAX ((d) == end1 ? *string2 \
3956 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
3959 /* Disabled due to a compiler bug -- see comment at case wordbound */
3961 /* The comment at case wordbound is following one, but we don't use
3962 AT_WORD_BOUNDARY anymore to support multibyte form.
3964 The DEC Alpha C compiler 3.x generates incorrect code for the
3965 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
3966 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
3967 macro and introducing temporary variables works around the bug. */
3970 /* Test if the character before D and the one at D differ with respect
3971 to being word-constituent. */
3972 #define AT_WORD_BOUNDARY(d) \
3973 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
3974 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
3977 /* Free everything we malloc. */
3978 #ifdef MATCH_MAY_ALLOCATE
3979 #define FREE_VAR(var) if (var) { REGEX_FREE (var); var = NULL; } else
3980 #define FREE_VARIABLES() \
3982 REGEX_FREE_STACK (fail_stack.stack); \
3983 FREE_VAR (regstart); \
3984 FREE_VAR (regend); \
3985 FREE_VAR (old_regstart); \
3986 FREE_VAR (old_regend); \
3987 FREE_VAR (best_regstart); \
3988 FREE_VAR (best_regend); \
3989 FREE_VAR (reg_info); \
3990 FREE_VAR (reg_dummy); \
3991 FREE_VAR (reg_info_dummy); \
3994 #define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
3995 #endif /* not MATCH_MAY_ALLOCATE */
3997 /* These values must meet several constraints. They must not be valid
3998 register values; since we have a limit of 255 registers (because
3999 we use only one byte in the pattern for the register number), we can
4000 use numbers larger than 255. They must differ by 1, because of
4001 NUM_FAILURE_ITEMS above. And the value for the lowest register must
4002 be larger than the value for the highest register, so we do not try
4003 to actually save any registers when none are active. */
4004 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
4005 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
4007 /* Matching routines. */
4009 #ifndef emacs /* Emacs never uses this. */
4010 /* re_match is like re_match_2 except it takes only a single string. */
4013 re_match (bufp, string, size, pos, regs)
4014 struct re_pattern_buffer *bufp;
4017 struct re_registers *regs;
4019 int result = re_match_2_internal (bufp, NULL, 0, string, size,
4024 #endif /* not emacs */
4027 /* In Emacs, this is the string or buffer in which we
4028 are matching. It is used for looking up syntax properties. */
4029 Lisp_Object re_match_object;
4032 /* re_match_2 matches the compiled pattern in BUFP against the
4033 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
4034 and SIZE2, respectively). We start matching at POS, and stop
4037 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
4038 store offsets for the substring each group matched in REGS. See the
4039 documentation for exactly how many groups we fill.
4041 We return -1 if no match, -2 if an internal error (such as the
4042 failure stack overflowing). Otherwise, we return the length of the
4043 matched substring. */
4046 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
4047 struct re_pattern_buffer *bufp;
4048 const char *string1, *string2;
4051 struct re_registers *regs;
4057 SETUP_SYNTAX_TABLE_FOR_OBJECT (re_match_object,
4058 POS_AS_IN_BUFFER (pos > 0 ? pos - 1 : pos),
4062 result = re_match_2_internal (bufp, string1, size1, string2, size2,
4068 /* This is a separate function so that we can force an alloca cleanup
4071 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
4072 struct re_pattern_buffer *bufp;
4073 const char *string1, *string2;
4076 struct re_registers *regs;
4079 /* General temporaries. */
4083 /* Just past the end of the corresponding string. */
4084 const char *end1, *end2;
4086 /* Pointers into string1 and string2, just past the last characters in
4087 each to consider matching. */
4088 const char *end_match_1, *end_match_2;
4090 /* Where we are in the data, and the end of the current string. */
4091 const char *d, *dend;
4093 /* Where we are in the pattern, and the end of the pattern. */
4094 unsigned char *p = bufp->buffer;
4095 register unsigned char *pend = p + bufp->used;
4097 /* Mark the opcode just after a start_memory, so we can test for an
4098 empty subpattern when we get to the stop_memory. */
4099 unsigned char *just_past_start_mem = 0;
4101 /* We use this to map every character in the string. */
4102 RE_TRANSLATE_TYPE translate = bufp->translate;
4104 /* Nonzero if we have to concern multibyte character. */
4105 int multibyte = bufp->multibyte;
4107 /* Failure point stack. Each place that can handle a failure further
4108 down the line pushes a failure point on this stack. It consists of
4109 restart, regend, and reg_info for all registers corresponding to
4110 the subexpressions we're currently inside, plus the number of such
4111 registers, and, finally, two char *'s. The first char * is where
4112 to resume scanning the pattern; the second one is where to resume
4113 scanning the strings. If the latter is zero, the failure point is
4114 a ``dummy''; if a failure happens and the failure point is a dummy,
4115 it gets discarded and the next next one is tried. */
4116 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
4117 fail_stack_type fail_stack;
4120 static unsigned failure_id = 0;
4121 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
4124 /* This holds the pointer to the failure stack, when
4125 it is allocated relocatably. */
4126 fail_stack_elt_t *failure_stack_ptr;
4128 /* We fill all the registers internally, independent of what we
4129 return, for use in backreferences. The number here includes
4130 an element for register zero. */
4131 unsigned num_regs = bufp->re_nsub + 1;
4133 /* The currently active registers. */
4134 unsigned lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4135 unsigned highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4137 /* Information on the contents of registers. These are pointers into
4138 the input strings; they record just what was matched (on this
4139 attempt) by a subexpression part of the pattern, that is, the
4140 regnum-th regstart pointer points to where in the pattern we began
4141 matching and the regnum-th regend points to right after where we
4142 stopped matching the regnum-th subexpression. (The zeroth register
4143 keeps track of what the whole pattern matches.) */
4144 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4145 const char **regstart, **regend;
4148 /* If a group that's operated upon by a repetition operator fails to
4149 match anything, then the register for its start will need to be
4150 restored because it will have been set to wherever in the string we
4151 are when we last see its open-group operator. Similarly for a
4153 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4154 const char **old_regstart, **old_regend;
4157 /* The is_active field of reg_info helps us keep track of which (possibly
4158 nested) subexpressions we are currently in. The matched_something
4159 field of reg_info[reg_num] helps us tell whether or not we have
4160 matched any of the pattern so far this time through the reg_num-th
4161 subexpression. These two fields get reset each time through any
4162 loop their register is in. */
4163 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
4164 register_info_type *reg_info;
4167 /* The following record the register info as found in the above
4168 variables when we find a match better than any we've seen before.
4169 This happens as we backtrack through the failure points, which in
4170 turn happens only if we have not yet matched the entire string. */
4171 unsigned best_regs_set = false;
4172 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4173 const char **best_regstart, **best_regend;
4176 /* Logically, this is `best_regend[0]'. But we don't want to have to
4177 allocate space for that if we're not allocating space for anything
4178 else (see below). Also, we never need info about register 0 for
4179 any of the other register vectors, and it seems rather a kludge to
4180 treat `best_regend' differently than the rest. So we keep track of
4181 the end of the best match so far in a separate variable. We
4182 initialize this to NULL so that when we backtrack the first time
4183 and need to test it, it's not garbage. */
4184 const char *match_end = NULL;
4186 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
4187 int set_regs_matched_done = 0;
4189 /* Used when we pop values we don't care about. */
4190 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4191 const char **reg_dummy;
4192 register_info_type *reg_info_dummy;
4196 /* Counts the total number of registers pushed. */
4197 unsigned num_regs_pushed = 0;
4200 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
4204 #ifdef MATCH_MAY_ALLOCATE
4205 /* Do not bother to initialize all the register variables if there are
4206 no groups in the pattern, as it takes a fair amount of time. If
4207 there are groups, we include space for register 0 (the whole
4208 pattern), even though we never use it, since it simplifies the
4209 array indexing. We should fix this. */
4212 regstart = REGEX_TALLOC (num_regs, const char *);
4213 regend = REGEX_TALLOC (num_regs, const char *);
4214 old_regstart = REGEX_TALLOC (num_regs, const char *);
4215 old_regend = REGEX_TALLOC (num_regs, const char *);
4216 best_regstart = REGEX_TALLOC (num_regs, const char *);
4217 best_regend = REGEX_TALLOC (num_regs, const char *);
4218 reg_info = REGEX_TALLOC (num_regs, register_info_type);
4219 reg_dummy = REGEX_TALLOC (num_regs, const char *);
4220 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
4222 if (!(regstart && regend && old_regstart && old_regend && reg_info
4223 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
4231 /* We must initialize all our variables to NULL, so that
4232 `FREE_VARIABLES' doesn't try to free them. */
4233 regstart = regend = old_regstart = old_regend = best_regstart
4234 = best_regend = reg_dummy = NULL;
4235 reg_info = reg_info_dummy = (register_info_type *) NULL;
4237 #endif /* MATCH_MAY_ALLOCATE */
4239 /* The starting position is bogus. */
4240 if (pos < 0 || pos > size1 + size2)
4246 /* Initialize subexpression text positions to -1 to mark ones that no
4247 start_memory/stop_memory has been seen for. Also initialize the
4248 register information struct. */
4249 for (mcnt = 1; mcnt < num_regs; mcnt++)
4251 regstart[mcnt] = regend[mcnt]
4252 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
4254 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
4255 IS_ACTIVE (reg_info[mcnt]) = 0;
4256 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
4257 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
4260 /* We move `string1' into `string2' if the latter's empty -- but not if
4261 `string1' is null. */
4262 if (size2 == 0 && string1 != NULL)
4269 end1 = string1 + size1;
4270 end2 = string2 + size2;
4272 /* Compute where to stop matching, within the two strings. */
4275 end_match_1 = string1 + stop;
4276 end_match_2 = string2;
4281 end_match_2 = string2 + stop - size1;
4284 /* `p' scans through the pattern as `d' scans through the data.
4285 `dend' is the end of the input string that `d' points within. `d'
4286 is advanced into the following input string whenever necessary, but
4287 this happens before fetching; therefore, at the beginning of the
4288 loop, `d' can be pointing at the end of a string, but it cannot
4290 if (size1 > 0 && pos <= size1)
4297 d = string2 + pos - size1;
4301 DEBUG_PRINT1 ("The compiled pattern is: ");
4302 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4303 DEBUG_PRINT1 ("The string to match is: `");
4304 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4305 DEBUG_PRINT1 ("'\n");
4307 /* This loops over pattern commands. It exits by returning from the
4308 function if the match is complete, or it drops through if the match
4309 fails at this starting point in the input data. */
4312 DEBUG_PRINT2 ("\n0x%x: ", p);
4315 { /* End of pattern means we might have succeeded. */
4316 DEBUG_PRINT1 ("end of pattern ... ");
4318 /* If we haven't matched the entire string, and we want the
4319 longest match, try backtracking. */
4320 if (d != end_match_2)
4322 /* 1 if this match ends in the same string (string1 or string2)
4323 as the best previous match. */
4324 boolean same_str_p = (FIRST_STRING_P (match_end)
4325 == MATCHING_IN_FIRST_STRING);
4326 /* 1 if this match is the best seen so far. */
4327 boolean best_match_p;
4329 /* AIX compiler got confused when this was combined
4330 with the previous declaration. */
4332 best_match_p = d > match_end;
4334 best_match_p = !MATCHING_IN_FIRST_STRING;
4336 DEBUG_PRINT1 ("backtracking.\n");
4338 if (!FAIL_STACK_EMPTY ())
4339 { /* More failure points to try. */
4341 /* If exceeds best match so far, save it. */
4342 if (!best_regs_set || best_match_p)
4344 best_regs_set = true;
4347 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4349 for (mcnt = 1; mcnt < num_regs; mcnt++)
4351 best_regstart[mcnt] = regstart[mcnt];
4352 best_regend[mcnt] = regend[mcnt];
4358 /* If no failure points, don't restore garbage. And if
4359 last match is real best match, don't restore second
4361 else if (best_regs_set && !best_match_p)
4364 /* Restore best match. It may happen that `dend ==
4365 end_match_1' while the restored d is in string2.
4366 For example, the pattern `x.*y.*z' against the
4367 strings `x-' and `y-z-', if the two strings are
4368 not consecutive in memory. */
4369 DEBUG_PRINT1 ("Restoring best registers.\n");
4372 dend = ((d >= string1 && d <= end1)
4373 ? end_match_1 : end_match_2);
4375 for (mcnt = 1; mcnt < num_regs; mcnt++)
4377 regstart[mcnt] = best_regstart[mcnt];
4378 regend[mcnt] = best_regend[mcnt];
4381 } /* d != end_match_2 */
4384 DEBUG_PRINT1 ("Accepting match.\n");
4386 /* If caller wants register contents data back, do it. */
4387 if (regs && !bufp->no_sub)
4389 /* Have the register data arrays been allocated? */
4390 if (bufp->regs_allocated == REGS_UNALLOCATED)
4391 { /* No. So allocate them with malloc. We need one
4392 extra element beyond `num_regs' for the `-1' marker
4394 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4395 regs->start = TALLOC (regs->num_regs, regoff_t);
4396 regs->end = TALLOC (regs->num_regs, regoff_t);
4397 if (regs->start == NULL || regs->end == NULL)
4402 bufp->regs_allocated = REGS_REALLOCATE;
4404 else if (bufp->regs_allocated == REGS_REALLOCATE)
4405 { /* Yes. If we need more elements than were already
4406 allocated, reallocate them. If we need fewer, just
4408 if (regs->num_regs < num_regs + 1)
4410 regs->num_regs = num_regs + 1;
4411 RETALLOC (regs->start, regs->num_regs, regoff_t);
4412 RETALLOC (regs->end, regs->num_regs, regoff_t);
4413 if (regs->start == NULL || regs->end == NULL)
4422 /* These braces fend off a "empty body in an else-statement"
4423 warning under GCC when assert expands to nothing. */
4424 assert (bufp->regs_allocated == REGS_FIXED);
4427 /* Convert the pointer data in `regstart' and `regend' to
4428 indices. Register zero has to be set differently,
4429 since we haven't kept track of any info for it. */
4430 if (regs->num_regs > 0)
4432 regs->start[0] = pos;
4433 regs->end[0] = (MATCHING_IN_FIRST_STRING
4434 ? ((regoff_t) (d - string1))
4435 : ((regoff_t) (d - string2 + size1)));
4438 /* Go through the first `min (num_regs, regs->num_regs)'
4439 registers, since that is all we initialized. */
4440 for (mcnt = 1; mcnt < MIN (num_regs, regs->num_regs); mcnt++)
4442 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4443 regs->start[mcnt] = regs->end[mcnt] = -1;
4447 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4449 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4453 /* If the regs structure we return has more elements than
4454 were in the pattern, set the extra elements to -1. If
4455 we (re)allocated the registers, this is the case,
4456 because we always allocate enough to have at least one
4458 for (mcnt = num_regs; mcnt < regs->num_regs; mcnt++)
4459 regs->start[mcnt] = regs->end[mcnt] = -1;
4460 } /* regs && !bufp->no_sub */
4462 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4463 nfailure_points_pushed, nfailure_points_popped,
4464 nfailure_points_pushed - nfailure_points_popped);
4465 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4467 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4471 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4477 /* Otherwise match next pattern command. */
4478 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4480 /* Ignore these. Used to ignore the n of succeed_n's which
4481 currently have n == 0. */
4483 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4487 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4490 /* Match the next n pattern characters exactly. The following
4491 byte in the pattern defines n, and the n bytes after that
4492 are the characters to match. */
4495 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4497 /* This is written out as an if-else so we don't waste time
4498 testing `translate' inside the loop. */
4504 if ((unsigned char) RE_TRANSLATE (translate, (unsigned char) *d++)
4505 != (unsigned char) *p++)
4515 if (*d++ != (char) *p++) goto fail;
4519 SET_REGS_MATCHED ();
4523 /* Match any character except possibly a newline or a null. */
4525 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4529 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4530 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4533 SET_REGS_MATCHED ();
4534 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4535 d += multibyte ? MULTIBYTE_FORM_LENGTH (d, dend - d) : 1;
4542 register unsigned int c;
4543 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4546 /* Start of actual range_table, or end of bitmap if there is no
4548 unsigned char *range_table;
4550 /* Nonzero if there is range table. */
4551 int range_table_exists;
4553 /* Number of ranges of range table. Not in bytes. */
4556 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4559 c = (unsigned char) *d;
4561 range_table = CHARSET_RANGE_TABLE (&p[-1]); /* Past the bitmap. */
4562 range_table_exists = CHARSET_RANGE_TABLE_EXISTS_P (&p[-1]);
4563 if (range_table_exists)
4564 EXTRACT_NUMBER_AND_INCR (count, range_table);
4568 if (multibyte && BASE_LEADING_CODE_P (c))
4569 c = STRING_CHAR_AND_LENGTH (d, dend - d, len);
4571 if (SINGLE_BYTE_CHAR_P (c))
4572 { /* Lookup bitmap. */
4573 c = TRANSLATE (c); /* The character to match. */
4576 /* Cast to `unsigned' instead of `unsigned char' in
4577 case the bit list is a full 32 bytes long. */
4578 if (c < (unsigned) (CHARSET_BITMAP_SIZE (&p[-1]) * BYTEWIDTH)
4579 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4582 else if (range_table_exists)
4583 CHARSET_LOOKUP_RANGE_TABLE_RAW (not, c, range_table, count);
4585 p = CHARSET_RANGE_TABLE_END (range_table, count);
4587 if (!not) goto fail;
4589 SET_REGS_MATCHED ();
4595 /* The beginning of a group is represented by start_memory.
4596 The arguments are the register number in the next byte, and the
4597 number of groups inner to this one in the next. The text
4598 matched within the group is recorded (in the internal
4599 registers data structure) under the register number. */
4601 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4603 /* Find out if this group can match the empty string. */
4604 p1 = p; /* To send to group_match_null_string_p. */
4606 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4607 REG_MATCH_NULL_STRING_P (reg_info[*p])
4608 = group_match_null_string_p (&p1, pend, reg_info);
4610 /* Save the position in the string where we were the last time
4611 we were at this open-group operator in case the group is
4612 operated upon by a repetition operator, e.g., with `(a*)*b'
4613 against `ab'; then we want to ignore where we are now in
4614 the string in case this attempt to match fails. */
4615 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4616 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4618 DEBUG_PRINT2 (" old_regstart: %d\n",
4619 POINTER_TO_OFFSET (old_regstart[*p]));
4622 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4624 IS_ACTIVE (reg_info[*p]) = 1;
4625 MATCHED_SOMETHING (reg_info[*p]) = 0;
4627 /* Clear this whenever we change the register activity status. */
4628 set_regs_matched_done = 0;
4630 /* This is the new highest active register. */
4631 highest_active_reg = *p;
4633 /* If nothing was active before, this is the new lowest active
4635 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4636 lowest_active_reg = *p;
4638 /* Move past the register number and inner group count. */
4640 just_past_start_mem = p;
4645 /* The stop_memory opcode represents the end of a group. Its
4646 arguments are the same as start_memory's: the register
4647 number, and the number of inner groups. */
4649 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4651 /* We need to save the string position the last time we were at
4652 this close-group operator in case the group is operated
4653 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4654 against `aba'; then we want to ignore where we are now in
4655 the string in case this attempt to match fails. */
4656 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4657 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4659 DEBUG_PRINT2 (" old_regend: %d\n",
4660 POINTER_TO_OFFSET (old_regend[*p]));
4663 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4665 /* This register isn't active anymore. */
4666 IS_ACTIVE (reg_info[*p]) = 0;
4668 /* Clear this whenever we change the register activity status. */
4669 set_regs_matched_done = 0;
4671 /* If this was the only register active, nothing is active
4673 if (lowest_active_reg == highest_active_reg)
4675 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4676 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4679 { /* We must scan for the new highest active register, since
4680 it isn't necessarily one less than now: consider
4681 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4682 new highest active register is 1. */
4683 unsigned char r = *p - 1;
4684 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4687 /* If we end up at register zero, that means that we saved
4688 the registers as the result of an `on_failure_jump', not
4689 a `start_memory', and we jumped to past the innermost
4690 `stop_memory'. For example, in ((.)*) we save
4691 registers 1 and 2 as a result of the *, but when we pop
4692 back to the second ), we are at the stop_memory 1.
4693 Thus, nothing is active. */
4696 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4697 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4700 highest_active_reg = r;
4703 /* If just failed to match something this time around with a
4704 group that's operated on by a repetition operator, try to
4705 force exit from the ``loop'', and restore the register
4706 information for this group that we had before trying this
4708 if ((!MATCHED_SOMETHING (reg_info[*p])
4709 || just_past_start_mem == p - 1)
4712 boolean is_a_jump_n = false;
4716 switch ((re_opcode_t) *p1++)
4720 case pop_failure_jump:
4721 case maybe_pop_jump:
4723 case dummy_failure_jump:
4724 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4734 /* If the next operation is a jump backwards in the pattern
4735 to an on_failure_jump right before the start_memory
4736 corresponding to this stop_memory, exit from the loop
4737 by forcing a failure after pushing on the stack the
4738 on_failure_jump's jump in the pattern, and d. */
4739 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4740 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4742 /* If this group ever matched anything, then restore
4743 what its registers were before trying this last
4744 failed match, e.g., with `(a*)*b' against `ab' for
4745 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4746 against `aba' for regend[3].
4748 Also restore the registers for inner groups for,
4749 e.g., `((a*)(b*))*' against `aba' (register 3 would
4750 otherwise get trashed). */
4752 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4756 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4758 /* Restore this and inner groups' (if any) registers. */
4759 for (r = *p; r < *p + *(p + 1); r++)
4761 regstart[r] = old_regstart[r];
4763 /* xx why this test? */
4764 if (old_regend[r] >= regstart[r])
4765 regend[r] = old_regend[r];
4769 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4770 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4776 /* Move past the register number and the inner group count. */
4781 /* \<digit> has been turned into a `duplicate' command which is
4782 followed by the numeric value of <digit> as the register number. */
4785 register const char *d2, *dend2;
4786 int regno = *p++; /* Get which register to match against. */
4787 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4789 /* Can't back reference a group which we've never matched. */
4790 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4793 /* Where in input to try to start matching. */
4794 d2 = regstart[regno];
4796 /* Where to stop matching; if both the place to start and
4797 the place to stop matching are in the same string, then
4798 set to the place to stop, otherwise, for now have to use
4799 the end of the first string. */
4801 dend2 = ((FIRST_STRING_P (regstart[regno])
4802 == FIRST_STRING_P (regend[regno]))
4803 ? regend[regno] : end_match_1);
4806 /* If necessary, advance to next segment in register
4810 if (dend2 == end_match_2) break;
4811 if (dend2 == regend[regno]) break;
4813 /* End of string1 => advance to string2. */
4815 dend2 = regend[regno];
4817 /* At end of register contents => success */
4818 if (d2 == dend2) break;
4820 /* If necessary, advance to next segment in data. */
4823 /* How many characters left in this segment to match. */
4826 /* Want how many consecutive characters we can match in
4827 one shot, so, if necessary, adjust the count. */
4828 if (mcnt > dend2 - d2)
4831 /* Compare that many; failure if mismatch, else move
4834 ? bcmp_translate (d, d2, mcnt, translate)
4835 : bcmp (d, d2, mcnt))
4837 d += mcnt, d2 += mcnt;
4839 /* Do this because we've match some characters. */
4840 SET_REGS_MATCHED ();
4846 /* begline matches the empty string at the beginning of the string
4847 (unless `not_bol' is set in `bufp'), and, if
4848 `newline_anchor' is set, after newlines. */
4850 DEBUG_PRINT1 ("EXECUTING begline.\n");
4852 if (AT_STRINGS_BEG (d))
4854 if (!bufp->not_bol) break;
4856 else if (d[-1] == '\n' && bufp->newline_anchor)
4860 /* In all other cases, we fail. */
4864 /* endline is the dual of begline. */
4866 DEBUG_PRINT1 ("EXECUTING endline.\n");
4868 if (AT_STRINGS_END (d))
4870 if (!bufp->not_eol) break;
4873 /* We have to ``prefetch'' the next character. */
4874 else if ((d == end1 ? *string2 : *d) == '\n'
4875 && bufp->newline_anchor)
4882 /* Match at the very beginning of the data. */
4884 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
4885 if (AT_STRINGS_BEG (d))
4890 /* Match at the very end of the data. */
4892 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
4893 if (AT_STRINGS_END (d))
4898 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
4899 pushes NULL as the value for the string on the stack. Then
4900 `pop_failure_point' will keep the current value for the
4901 string, instead of restoring it. To see why, consider
4902 matching `foo\nbar' against `.*\n'. The .* matches the foo;
4903 then the . fails against the \n. But the next thing we want
4904 to do is match the \n against the \n; if we restored the
4905 string value, we would be back at the foo.
4907 Because this is used only in specific cases, we don't need to
4908 check all the things that `on_failure_jump' does, to make
4909 sure the right things get saved on the stack. Hence we don't
4910 share its code. The only reason to push anything on the
4911 stack at all is that otherwise we would have to change
4912 `anychar's code to do something besides goto fail in this
4913 case; that seems worse than this. */
4914 case on_failure_keep_string_jump:
4915 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
4917 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4918 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
4920 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
4924 /* Uses of on_failure_jump:
4926 Each alternative starts with an on_failure_jump that points
4927 to the beginning of the next alternative. Each alternative
4928 except the last ends with a jump that in effect jumps past
4929 the rest of the alternatives. (They really jump to the
4930 ending jump of the following alternative, because tensioning
4931 these jumps is a hassle.)
4933 Repeats start with an on_failure_jump that points past both
4934 the repetition text and either the following jump or
4935 pop_failure_jump back to this on_failure_jump. */
4936 case on_failure_jump:
4938 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
4940 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4941 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
4943 /* If this on_failure_jump comes right before a group (i.e.,
4944 the original * applied to a group), save the information
4945 for that group and all inner ones, so that if we fail back
4946 to this point, the group's information will be correct.
4947 For example, in \(a*\)*\1, we need the preceding group,
4948 and in \(zz\(a*\)b*\)\2, we need the inner group. */
4950 /* We can't use `p' to check ahead because we push
4951 a failure point to `p + mcnt' after we do this. */
4954 /* We need to skip no_op's before we look for the
4955 start_memory in case this on_failure_jump is happening as
4956 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
4958 while (p1 < pend && (re_opcode_t) *p1 == no_op)
4961 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
4963 /* We have a new highest active register now. This will
4964 get reset at the start_memory we are about to get to,
4965 but we will have saved all the registers relevant to
4966 this repetition op, as described above. */
4967 highest_active_reg = *(p1 + 1) + *(p1 + 2);
4968 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4969 lowest_active_reg = *(p1 + 1);
4972 DEBUG_PRINT1 (":\n");
4973 PUSH_FAILURE_POINT (p + mcnt, d, -2);
4977 /* A smart repeat ends with `maybe_pop_jump'.
4978 We change it to either `pop_failure_jump' or `jump'. */
4979 case maybe_pop_jump:
4980 EXTRACT_NUMBER_AND_INCR (mcnt, p);
4981 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
4983 register unsigned char *p2 = p;
4985 /* Compare the beginning of the repeat with what in the
4986 pattern follows its end. If we can establish that there
4987 is nothing that they would both match, i.e., that we
4988 would have to backtrack because of (as in, e.g., `a*a')
4989 then we can change to pop_failure_jump, because we'll
4990 never have to backtrack.
4992 This is not true in the case of alternatives: in
4993 `(a|ab)*' we do need to backtrack to the `ab' alternative
4994 (e.g., if the string was `ab'). But instead of trying to
4995 detect that here, the alternative has put on a dummy
4996 failure point which is what we will end up popping. */
4998 /* Skip over open/close-group commands.
4999 If what follows this loop is a ...+ construct,
5000 look at what begins its body, since we will have to
5001 match at least one of that. */
5005 && ((re_opcode_t) *p2 == stop_memory
5006 || (re_opcode_t) *p2 == start_memory))
5008 else if (p2 + 6 < pend
5009 && (re_opcode_t) *p2 == dummy_failure_jump)
5016 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
5017 to the `maybe_finalize_jump' of this case. Examine what
5020 /* If we're at the end of the pattern, we can change. */
5023 /* Consider what happens when matching ":\(.*\)"
5024 against ":/". I don't really understand this code
5026 p[-3] = (unsigned char) pop_failure_jump;
5028 (" End of pattern: change to `pop_failure_jump'.\n");
5031 else if ((re_opcode_t) *p2 == exactn
5032 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
5034 register unsigned int c
5035 = *p2 == (unsigned char) endline ? '\n' : p2[2];
5037 if ((re_opcode_t) p1[3] == exactn)
5039 if (!(multibyte /* && (c != '\n') */
5040 && BASE_LEADING_CODE_P (c))
5042 : (STRING_CHAR (&p2[2], pend - &p2[2])
5043 != STRING_CHAR (&p1[5], pend - &p1[5])))
5045 p[-3] = (unsigned char) pop_failure_jump;
5046 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
5051 else if ((re_opcode_t) p1[3] == charset
5052 || (re_opcode_t) p1[3] == charset_not)
5054 int not = (re_opcode_t) p1[3] == charset_not;
5056 if (multibyte /* && (c != '\n') */
5057 && BASE_LEADING_CODE_P (c))
5058 c = STRING_CHAR (&p2[2], pend - &p2[2]);
5060 /* Test if C is listed in charset (or charset_not)
5062 if (SINGLE_BYTE_CHAR_P (c))
5064 if (c < CHARSET_BITMAP_SIZE (&p1[3]) * BYTEWIDTH
5065 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
5068 else if (CHARSET_RANGE_TABLE_EXISTS_P (&p1[3]))
5069 CHARSET_LOOKUP_RANGE_TABLE (not, c, &p1[3]);
5071 /* `not' is equal to 1 if c would match, which means
5072 that we can't change to pop_failure_jump. */
5075 p[-3] = (unsigned char) pop_failure_jump;
5076 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5080 else if ((re_opcode_t) *p2 == charset)
5082 if ((re_opcode_t) p1[3] == exactn)
5084 register unsigned int c = p1[5];
5087 if (multibyte && BASE_LEADING_CODE_P (c))
5088 c = STRING_CHAR (&p1[5], pend - &p1[5]);
5090 /* Test if C is listed in charset at `p2'. */
5091 if (SINGLE_BYTE_CHAR_P (c))
5093 if (c < CHARSET_BITMAP_SIZE (p2) * BYTEWIDTH
5094 && (p2[2 + c / BYTEWIDTH]
5095 & (1 << (c % BYTEWIDTH))))
5098 else if (CHARSET_RANGE_TABLE_EXISTS_P (p2))
5099 CHARSET_LOOKUP_RANGE_TABLE (not, c, p2);
5103 p[-3] = (unsigned char) pop_failure_jump;
5104 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5108 /* It is hard to list up all the character in charset
5109 P2 if it includes multibyte character. Give up in
5111 else if (!multibyte || !CHARSET_RANGE_TABLE_EXISTS_P (p2))
5113 /* Now, we are sure that P2 has no range table.
5114 So, for the size of bitmap in P2, `p2[1]' is
5115 enough. But P1 may have range table, so the
5116 size of bitmap table of P1 is extracted by
5117 using macro `CHARSET_BITMAP_SIZE'.
5119 Since we know that all the character listed in
5120 P2 is ASCII, it is enough to test only bitmap
5123 if ((re_opcode_t) p1[3] == charset_not)
5126 /* We win if the charset_not inside the loop lists
5127 every character listed in the charset after. */
5128 for (idx = 0; idx < (int) p2[1]; idx++)
5129 if (! (p2[2 + idx] == 0
5130 || (idx < CHARSET_BITMAP_SIZE (&p1[3])
5131 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
5136 p[-3] = (unsigned char) pop_failure_jump;
5137 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5140 else if ((re_opcode_t) p1[3] == charset)
5143 /* We win if the charset inside the loop
5144 has no overlap with the one after the loop. */
5147 && idx < CHARSET_BITMAP_SIZE (&p1[3]));
5149 if ((p2[2 + idx] & p1[5 + idx]) != 0)
5153 || idx == CHARSET_BITMAP_SIZE (&p1[3]))
5155 p[-3] = (unsigned char) pop_failure_jump;
5156 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5162 p -= 2; /* Point at relative address again. */
5163 if ((re_opcode_t) p[-1] != pop_failure_jump)
5165 p[-1] = (unsigned char) jump;
5166 DEBUG_PRINT1 (" Match => jump.\n");
5167 goto unconditional_jump;
5169 /* Note fall through. */
5172 /* The end of a simple repeat has a pop_failure_jump back to
5173 its matching on_failure_jump, where the latter will push a
5174 failure point. The pop_failure_jump takes off failure
5175 points put on by this pop_failure_jump's matching
5176 on_failure_jump; we got through the pattern to here from the
5177 matching on_failure_jump, so didn't fail. */
5178 case pop_failure_jump:
5180 /* We need to pass separate storage for the lowest and
5181 highest registers, even though we don't care about the
5182 actual values. Otherwise, we will restore only one
5183 register from the stack, since lowest will == highest in
5184 `pop_failure_point'. */
5185 unsigned dummy_low_reg, dummy_high_reg;
5186 unsigned char *pdummy;
5189 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
5190 POP_FAILURE_POINT (sdummy, pdummy,
5191 dummy_low_reg, dummy_high_reg,
5192 reg_dummy, reg_dummy, reg_info_dummy);
5194 /* Note fall through. */
5197 /* Unconditionally jump (without popping any failure points). */
5200 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
5201 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
5202 p += mcnt; /* Do the jump. */
5203 DEBUG_PRINT2 ("(to 0x%x).\n", p);
5207 /* We need this opcode so we can detect where alternatives end
5208 in `group_match_null_string_p' et al. */
5210 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
5211 goto unconditional_jump;
5214 /* Normally, the on_failure_jump pushes a failure point, which
5215 then gets popped at pop_failure_jump. We will end up at
5216 pop_failure_jump, also, and with a pattern of, say, `a+', we
5217 are skipping over the on_failure_jump, so we have to push
5218 something meaningless for pop_failure_jump to pop. */
5219 case dummy_failure_jump:
5220 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
5221 /* It doesn't matter what we push for the string here. What
5222 the code at `fail' tests is the value for the pattern. */
5223 PUSH_FAILURE_POINT (0, 0, -2);
5224 goto unconditional_jump;
5227 /* At the end of an alternative, we need to push a dummy failure
5228 point in case we are followed by a `pop_failure_jump', because
5229 we don't want the failure point for the alternative to be
5230 popped. For example, matching `(a|ab)*' against `aab'
5231 requires that we match the `ab' alternative. */
5232 case push_dummy_failure:
5233 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
5234 /* See comments just above at `dummy_failure_jump' about the
5236 PUSH_FAILURE_POINT (0, 0, -2);
5239 /* Have to succeed matching what follows at least n times.
5240 After that, handle like `on_failure_jump'. */
5242 EXTRACT_NUMBER (mcnt, p + 2);
5243 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
5246 /* Originally, this is how many times we HAVE to succeed. */
5251 STORE_NUMBER_AND_INCR (p, mcnt);
5252 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p, mcnt);
5256 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
5257 p[2] = (unsigned char) no_op;
5258 p[3] = (unsigned char) no_op;
5264 EXTRACT_NUMBER (mcnt, p + 2);
5265 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
5267 /* Originally, this is how many times we CAN jump. */
5271 STORE_NUMBER (p + 2, mcnt);
5272 goto unconditional_jump;
5274 /* If don't have to jump any more, skip over the rest of command. */
5281 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
5283 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5285 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5286 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
5287 STORE_NUMBER (p1, mcnt);
5292 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
5294 /* We SUCCEED in one of the following cases: */
5296 /* Case 1: D is at the beginning or the end of string. */
5297 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5301 /* C1 is the character before D, S1 is the syntax of C1, C2
5302 is the character at D, and S2 is the syntax of C2. */
5304 int pos1 = PTR_TO_OFFSET (d - 1);
5306 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5307 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
5309 UPDATE_SYNTAX_TABLE (pos1 ? pos1 : 1);
5313 UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1);
5317 if (/* Case 2: Only one of S1 and S2 is Sword. */
5318 ((s1 == Sword) != (s2 == Sword))
5319 /* Case 3: Both of S1 and S2 are Sword, and macro
5320 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
5321 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2)))
5327 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5329 /* We FAIL in one of the following cases: */
5331 /* Case 1: D is at the beginning or the end of string. */
5332 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5336 /* C1 is the character before D, S1 is the syntax of C1, C2
5337 is the character at D, and S2 is the syntax of C2. */
5339 int pos1 = PTR_TO_OFFSET (d - 1);
5341 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5342 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
5344 UPDATE_SYNTAX_TABLE (pos1);
5348 UPDATE_SYNTAX_TABLE_FORWARD (pos1 + 1);
5352 if (/* Case 2: Only one of S1 and S2 is Sword. */
5353 ((s1 == Sword) != (s2 == Sword))
5354 /* Case 3: Both of S1 and S2 are Sword, and macro
5355 WORD_BOUNDARY_P (C1, C2) returns nonzero. */
5356 || ((s1 == Sword) && WORD_BOUNDARY_P (c1, c2)))
5362 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5364 /* We FAIL in one of the following cases: */
5366 /* Case 1: D is at the end of string. */
5367 if (AT_STRINGS_END (d))
5371 /* C1 is the character before D, S1 is the syntax of C1, C2
5372 is the character at D, and S2 is the syntax of C2. */
5374 int pos1 = PTR_TO_OFFSET (d);
5376 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
5378 UPDATE_SYNTAX_TABLE (pos1);
5382 /* Case 2: S2 is not Sword. */
5386 /* Case 3: D is not at the beginning of string ... */
5387 if (!AT_STRINGS_BEG (d))
5389 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5391 UPDATE_SYNTAX_TABLE_BACKWARD (pos1 - 1);
5395 /* ... and S1 is Sword, and WORD_BOUNDARY_P (C1, C2)
5397 if ((s1 == Sword) && !WORD_BOUNDARY_P (c1, c2))
5404 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5406 /* We FAIL in one of the following cases: */
5408 /* Case 1: D is at the beginning of string. */
5409 if (AT_STRINGS_BEG (d))
5413 /* C1 is the character before D, S1 is the syntax of C1, C2
5414 is the character at D, and S2 is the syntax of C2. */
5417 GET_CHAR_BEFORE_2 (c1, d, string1, end1, string2, end2);
5420 /* Case 2: S1 is not Sword. */
5424 /* Case 3: D is not at the end of string ... */
5425 if (!AT_STRINGS_END (d))
5427 GET_CHAR_AFTER_2 (c2, d, string1, end1, string2, end2);
5430 /* ... and S2 is Sword, and WORD_BOUNDARY_P (C1, C2)
5432 if ((s2 == Sword) && !WORD_BOUNDARY_P (c1, c2))
5440 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5441 if (PTR_CHAR_POS ((unsigned char *) d) >= PT)
5446 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5447 if (PTR_CHAR_POS ((unsigned char *) d) != PT)
5452 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5453 if (PTR_CHAR_POS ((unsigned char *) d) <= PT)
5458 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5463 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5469 int pos1 = PTR_TO_OFFSET (d);
5470 UPDATE_SYNTAX_TABLE (pos1);
5477 /* we must concern about multibyte form, ... */
5478 c = STRING_CHAR_AND_LENGTH (d, dend - d, len);
5480 /* everything should be handled as ASCII, even though it
5481 looks like multibyte form. */
5484 if (SYNTAX (c) != (enum syntaxcode) mcnt)
5488 SET_REGS_MATCHED ();
5492 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5494 goto matchnotsyntax;
5497 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5503 int pos1 = PTR_TO_OFFSET (d);
5504 UPDATE_SYNTAX_TABLE (pos1);
5511 c = STRING_CHAR_AND_LENGTH (d, dend - d, len);
5515 if (SYNTAX (c) == (enum syntaxcode) mcnt)
5519 SET_REGS_MATCHED ();
5523 DEBUG_PRINT2 ("EXECUTING categoryspec %d.\n", *p);
5530 c = STRING_CHAR_AND_LENGTH (d, dend - d, len);
5534 if (!CHAR_HAS_CATEGORY (c, mcnt))
5538 SET_REGS_MATCHED ();
5541 case notcategoryspec:
5542 DEBUG_PRINT2 ("EXECUTING notcategoryspec %d.\n", *p);
5549 c = STRING_CHAR_AND_LENGTH (d, dend - d, len);
5553 if (CHAR_HAS_CATEGORY (c, mcnt))
5557 SET_REGS_MATCHED ();
5560 #else /* not emacs */
5562 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5564 if (!WORDCHAR_P (d))
5566 SET_REGS_MATCHED ();
5571 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5575 SET_REGS_MATCHED ();
5578 #endif /* not emacs */
5583 continue; /* Successfully executed one pattern command; keep going. */
5586 /* We goto here if a matching operation fails. */
5588 if (!FAIL_STACK_EMPTY ())
5589 { /* A restart point is known. Restore to that state. */
5590 DEBUG_PRINT1 ("\nFAIL:\n");
5591 POP_FAILURE_POINT (d, p,
5592 lowest_active_reg, highest_active_reg,
5593 regstart, regend, reg_info);
5595 /* If this failure point is a dummy, try the next one. */
5599 /* If we failed to the end of the pattern, don't examine *p. */
5603 boolean is_a_jump_n = false;
5605 /* If failed to a backwards jump that's part of a repetition
5606 loop, need to pop this failure point and use the next one. */
5607 switch ((re_opcode_t) *p)
5611 case maybe_pop_jump:
5612 case pop_failure_jump:
5615 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5618 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5620 && (re_opcode_t) *p1 == on_failure_jump))
5628 if (d >= string1 && d <= end1)
5632 break; /* Matching at this starting point really fails. */
5636 goto restore_best_regs;
5640 return -1; /* Failure to match. */
5643 /* Subroutine definitions for re_match_2. */
5646 /* We are passed P pointing to a register number after a start_memory.
5648 Return true if the pattern up to the corresponding stop_memory can
5649 match the empty string, and false otherwise.
5651 If we find the matching stop_memory, sets P to point to one past its number.
5652 Otherwise, sets P to an undefined byte less than or equal to END.
5654 We don't handle duplicates properly (yet). */
5657 group_match_null_string_p (p, end, reg_info)
5658 unsigned char **p, *end;
5659 register_info_type *reg_info;
5662 /* Point to after the args to the start_memory. */
5663 unsigned char *p1 = *p + 2;
5667 /* Skip over opcodes that can match nothing, and return true or
5668 false, as appropriate, when we get to one that can't, or to the
5669 matching stop_memory. */
5671 switch ((re_opcode_t) *p1)
5673 /* Could be either a loop or a series of alternatives. */
5674 case on_failure_jump:
5676 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5678 /* If the next operation is not a jump backwards in the
5683 /* Go through the on_failure_jumps of the alternatives,
5684 seeing if any of the alternatives cannot match nothing.
5685 The last alternative starts with only a jump,
5686 whereas the rest start with on_failure_jump and end
5687 with a jump, e.g., here is the pattern for `a|b|c':
5689 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5690 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5693 So, we have to first go through the first (n-1)
5694 alternatives and then deal with the last one separately. */
5697 /* Deal with the first (n-1) alternatives, which start
5698 with an on_failure_jump (see above) that jumps to right
5699 past a jump_past_alt. */
5701 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5703 /* `mcnt' holds how many bytes long the alternative
5704 is, including the ending `jump_past_alt' and
5707 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5711 /* Move to right after this alternative, including the
5715 /* Break if it's the beginning of an n-th alternative
5716 that doesn't begin with an on_failure_jump. */
5717 if ((re_opcode_t) *p1 != on_failure_jump)
5720 /* Still have to check that it's not an n-th
5721 alternative that starts with an on_failure_jump. */
5723 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5724 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5726 /* Get to the beginning of the n-th alternative. */
5732 /* Deal with the last alternative: go back and get number
5733 of the `jump_past_alt' just before it. `mcnt' contains
5734 the length of the alternative. */
5735 EXTRACT_NUMBER (mcnt, p1 - 2);
5737 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5740 p1 += mcnt; /* Get past the n-th alternative. */
5746 assert (p1[1] == **p);
5752 if (!common_op_match_null_string_p (&p1, end, reg_info))
5755 } /* while p1 < end */
5758 } /* group_match_null_string_p */
5761 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5762 It expects P to be the first byte of a single alternative and END one
5763 byte past the last. The alternative can contain groups. */
5766 alt_match_null_string_p (p, end, reg_info)
5767 unsigned char *p, *end;
5768 register_info_type *reg_info;
5771 unsigned char *p1 = p;
5775 /* Skip over opcodes that can match nothing, and break when we get
5776 to one that can't. */
5778 switch ((re_opcode_t) *p1)
5781 case on_failure_jump:
5783 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5788 if (!common_op_match_null_string_p (&p1, end, reg_info))
5791 } /* while p1 < end */
5794 } /* alt_match_null_string_p */
5797 /* Deals with the ops common to group_match_null_string_p and
5798 alt_match_null_string_p.
5800 Sets P to one after the op and its arguments, if any. */
5803 common_op_match_null_string_p (p, end, reg_info)
5804 unsigned char **p, *end;
5805 register_info_type *reg_info;
5810 unsigned char *p1 = *p;
5812 switch ((re_opcode_t) *p1++)
5832 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5833 ret = group_match_null_string_p (&p1, end, reg_info);
5835 /* Have to set this here in case we're checking a group which
5836 contains a group and a back reference to it. */
5838 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5839 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5845 /* If this is an optimized succeed_n for zero times, make the jump. */
5847 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5855 /* Get to the number of times to succeed. */
5857 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5862 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5870 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5878 /* All other opcodes mean we cannot match the empty string. */
5884 } /* common_op_match_null_string_p */
5887 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5888 bytes; nonzero otherwise. */
5891 bcmp_translate (s1, s2, len, translate)
5892 unsigned char *s1, *s2;
5894 RE_TRANSLATE_TYPE translate;
5896 register unsigned char *p1 = s1, *p2 = s2;
5899 if (RE_TRANSLATE (translate, *p1++) != RE_TRANSLATE (translate, *p2++))
5906 /* Entry points for GNU code. */
5908 /* re_compile_pattern is the GNU regular expression compiler: it
5909 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5910 Returns 0 if the pattern was valid, otherwise an error string.
5912 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5913 are set in BUFP on entry.
5915 We call regex_compile to do the actual compilation. */
5918 re_compile_pattern (pattern, length, bufp)
5919 const char *pattern;
5921 struct re_pattern_buffer *bufp;
5925 /* GNU code is written to assume at least RE_NREGS registers will be set
5926 (and at least one extra will be -1). */
5927 bufp->regs_allocated = REGS_UNALLOCATED;
5929 /* And GNU code determines whether or not to get register information
5930 by passing null for the REGS argument to re_match, etc., not by
5934 /* Match anchors at newline. */
5935 bufp->newline_anchor = 1;
5937 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5941 return gettext (re_error_msgid[(int) ret]);
5944 /* Entry points compatible with 4.2 BSD regex library. We don't define
5945 them unless specifically requested. */
5947 #if defined (_REGEX_RE_COMP) || defined (_LIBC)
5949 /* BSD has one and only one pattern buffer. */
5950 static struct re_pattern_buffer re_comp_buf;
5954 /* Make these definitions weak in libc, so POSIX programs can redefine
5955 these names if they don't use our functions, and still use
5956 regcomp/regexec below without link errors. */
5966 if (!re_comp_buf.buffer)
5967 return gettext ("No previous regular expression");
5971 if (!re_comp_buf.buffer)
5973 re_comp_buf.buffer = (unsigned char *) malloc (200);
5974 if (re_comp_buf.buffer == NULL)
5975 return gettext (re_error_msgid[(int) REG_ESPACE]);
5976 re_comp_buf.allocated = 200;
5978 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5979 if (re_comp_buf.fastmap == NULL)
5980 return gettext (re_error_msgid[(int) REG_ESPACE]);
5983 /* Since `re_exec' always passes NULL for the `regs' argument, we
5984 don't need to initialize the pattern buffer fields which affect it. */
5986 /* Match anchors at newlines. */
5987 re_comp_buf.newline_anchor = 1;
5989 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5994 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5995 return (char *) gettext (re_error_msgid[(int) ret]);
6006 const int len = strlen (s);
6008 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
6010 #endif /* _REGEX_RE_COMP */
6012 /* POSIX.2 functions. Don't define these for Emacs. */
6016 /* regcomp takes a regular expression as a string and compiles it.
6018 PREG is a regex_t *. We do not expect any fields to be initialized,
6019 since POSIX says we shouldn't. Thus, we set
6021 `buffer' to the compiled pattern;
6022 `used' to the length of the compiled pattern;
6023 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
6024 REG_EXTENDED bit in CFLAGS is set; otherwise, to
6025 RE_SYNTAX_POSIX_BASIC;
6026 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
6027 `fastmap' and `fastmap_accurate' to zero;
6028 `re_nsub' to the number of subexpressions in PATTERN.
6030 PATTERN is the address of the pattern string.
6032 CFLAGS is a series of bits which affect compilation.
6034 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
6035 use POSIX basic syntax.
6037 If REG_NEWLINE is set, then . and [^...] don't match newline.
6038 Also, regexec will try a match beginning after every newline.
6040 If REG_ICASE is set, then we considers upper- and lowercase
6041 versions of letters to be equivalent when matching.
6043 If REG_NOSUB is set, then when PREG is passed to regexec, that
6044 routine will report only success or failure, and nothing about the
6047 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
6048 the return codes and their meanings.) */
6051 regcomp (preg, pattern, cflags)
6053 const char *pattern;
6058 = (cflags & REG_EXTENDED) ?
6059 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
6061 /* regex_compile will allocate the space for the compiled pattern. */
6063 preg->allocated = 0;
6066 /* Don't bother to use a fastmap when searching. This simplifies the
6067 REG_NEWLINE case: if we used a fastmap, we'd have to put all the
6068 characters after newlines into the fastmap. This way, we just try
6072 if (cflags & REG_ICASE)
6077 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
6078 * sizeof (*(RE_TRANSLATE_TYPE)0));
6079 if (preg->translate == NULL)
6080 return (int) REG_ESPACE;
6082 /* Map uppercase characters to corresponding lowercase ones. */
6083 for (i = 0; i < CHAR_SET_SIZE; i++)
6084 preg->translate[i] = ISUPPER (i) ? tolower (i) : i;
6087 preg->translate = NULL;
6089 /* If REG_NEWLINE is set, newlines are treated differently. */
6090 if (cflags & REG_NEWLINE)
6091 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
6092 syntax &= ~RE_DOT_NEWLINE;
6093 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
6094 /* It also changes the matching behavior. */
6095 preg->newline_anchor = 1;
6098 preg->newline_anchor = 0;
6100 preg->no_sub = !!(cflags & REG_NOSUB);
6102 /* POSIX says a null character in the pattern terminates it, so we
6103 can use strlen here in compiling the pattern. */
6104 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
6106 /* POSIX doesn't distinguish between an unmatched open-group and an
6107 unmatched close-group: both are REG_EPAREN. */
6108 if (ret == REG_ERPAREN) ret = REG_EPAREN;
6114 /* regexec searches for a given pattern, specified by PREG, in the
6117 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
6118 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
6119 least NMATCH elements, and we set them to the offsets of the
6120 corresponding matched substrings.
6122 EFLAGS specifies `execution flags' which affect matching: if
6123 REG_NOTBOL is set, then ^ does not match at the beginning of the
6124 string; if REG_NOTEOL is set, then $ does not match at the end.
6126 We return 0 if we find a match and REG_NOMATCH if not. */
6129 regexec (preg, string, nmatch, pmatch, eflags)
6130 const regex_t *preg;
6133 regmatch_t pmatch[];
6137 struct re_registers regs;
6138 regex_t private_preg;
6139 int len = strlen (string);
6140 boolean want_reg_info = !preg->no_sub && nmatch > 0;
6142 private_preg = *preg;
6144 private_preg.not_bol = !!(eflags & REG_NOTBOL);
6145 private_preg.not_eol = !!(eflags & REG_NOTEOL);
6147 /* The user has told us exactly how many registers to return
6148 information about, via `nmatch'. We have to pass that on to the
6149 matching routines. */
6150 private_preg.regs_allocated = REGS_FIXED;
6154 regs.num_regs = nmatch;
6155 regs.start = TALLOC (nmatch, regoff_t);
6156 regs.end = TALLOC (nmatch, regoff_t);
6157 if (regs.start == NULL || regs.end == NULL)
6158 return (int) REG_NOMATCH;
6161 /* Perform the searching operation. */
6162 ret = re_search (&private_preg, string, len,
6163 /* start: */ 0, /* range: */ len,
6164 want_reg_info ? ®s : (struct re_registers *) 0);
6166 /* Copy the register information to the POSIX structure. */
6173 for (r = 0; r < nmatch; r++)
6175 pmatch[r].rm_so = regs.start[r];
6176 pmatch[r].rm_eo = regs.end[r];
6180 /* If we needed the temporary register info, free the space now. */
6185 /* We want zero return to mean success, unlike `re_search'. */
6186 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
6190 /* Returns a message corresponding to an error code, ERRCODE, returned
6191 from either regcomp or regexec. We don't use PREG here. */
6194 regerror (errcode, preg, errbuf, errbuf_size)
6196 const regex_t *preg;
6204 || errcode >= (sizeof (re_error_msgid) / sizeof (re_error_msgid[0])))
6205 /* Only error codes returned by the rest of the code should be passed
6206 to this routine. If we are given anything else, or if other regex
6207 code generates an invalid error code, then the program has a bug.
6208 Dump core so we can fix it. */
6211 msg = gettext (re_error_msgid[errcode]);
6213 msg_size = strlen (msg) + 1; /* Includes the null. */
6215 if (errbuf_size != 0)
6217 if (msg_size > errbuf_size)
6219 strncpy (errbuf, msg, errbuf_size - 1);
6220 errbuf[errbuf_size - 1] = 0;
6223 strcpy (errbuf, msg);
6230 /* Free dynamically allocated space used by PREG. */
6236 if (preg->buffer != NULL)
6237 free (preg->buffer);
6238 preg->buffer = NULL;
6240 preg->allocated = 0;
6243 if (preg->fastmap != NULL)
6244 free (preg->fastmap);
6245 preg->fastmap = NULL;
6246 preg->fastmap_accurate = 0;
6248 if (preg->translate != NULL)
6249 free (preg->translate);
6250 preg->translate = NULL;
6253 #endif /* not emacs */
projects / gnulib.git / blob