1 /* Extended regular expression matching and search library,
3 (Implements POSIX draft P1003.2/D11.2, except for some of the
4 internationalization features.)
5 Copyright (C) 1993-1999, 2000 Free Software Foundation, Inc.
7 The GNU C Library is free software; you can redistribute it and/or
8 modify it under the terms of the GNU Library General Public License as
9 published by the Free Software Foundation; either version 2 of the
10 License, or (at your option) any later version.
12 The GNU C Library 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 GNU
15 Library General Public License for more details.
17 You should have received a copy of the GNU Library General Public
18 License along with the GNU C Library; see the file COPYING.LIB. If not,
19 write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
20 Boston, MA 02111-1307, USA. */
22 /* AIX requires this to be the first thing in the file. */
23 #if defined _AIX && !defined REGEX_MALLOC
35 # if defined __GNUC__ || (defined __STDC__ && __STDC__)
36 # define PARAMS(args) args
38 # define PARAMS(args) ()
40 #endif /* Not PARAMS. */
42 #if defined STDC_HEADERS && !defined emacs
45 /* We need this for `regex.h', and perhaps for the Emacs include files. */
46 # include <sys/types.h>
49 #define WIDE_CHAR_SUPPORT (HAVE_WCTYPE_H && HAVE_WCHAR_H && HAVE_BTOWC)
51 /* For platform which support the ISO C amendement 1 functionality we
52 support user defined character classes. */
53 #if defined _LIBC || WIDE_CHAR_SUPPORT
54 /* Solaris 2.5 has a bug: <wchar.h> must be included before <wctype.h>. */
60 /* We have to keep the namespace clean. */
61 # define regfree(preg) __regfree (preg)
62 # define regexec(pr, st, nm, pm, ef) __regexec (pr, st, nm, pm, ef)
63 # define regcomp(preg, pattern, cflags) __regcomp (preg, pattern, cflags)
64 # define regerror(errcode, preg, errbuf, errbuf_size) \
65 __regerror(errcode, preg, errbuf, errbuf_size)
66 # define re_set_registers(bu, re, nu, st, en) \
67 __re_set_registers (bu, re, nu, st, en)
68 # define re_match_2(bufp, string1, size1, string2, size2, pos, regs, stop) \
69 __re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
70 # define re_match(bufp, string, size, pos, regs) \
71 __re_match (bufp, string, size, pos, regs)
72 # define re_search(bufp, string, size, startpos, range, regs) \
73 __re_search (bufp, string, size, startpos, range, regs)
74 # define re_compile_pattern(pattern, length, bufp) \
75 __re_compile_pattern (pattern, length, bufp)
76 # define re_set_syntax(syntax) __re_set_syntax (syntax)
77 # define re_search_2(bufp, st1, s1, st2, s2, startpos, range, regs, stop) \
78 __re_search_2 (bufp, st1, s1, st2, s2, startpos, range, regs, stop)
79 # define re_compile_fastmap(bufp) __re_compile_fastmap (bufp)
81 # define btowc __btowc
83 /* We are also using some library internals. */
84 # include <locale/localeinfo.h>
85 # include <locale/elem-hash.h>
86 # include <langinfo.h>
89 /* This is for other GNU distributions with internationalized messages. */
90 #if HAVE_LIBINTL_H || defined _LIBC
93 # define gettext(msgid) (msgid)
97 /* This define is so xgettext can find the internationalizable
99 # define gettext_noop(String) String
102 /* The `emacs' switch turns on certain matching commands
103 that make sense only in Emacs. */
110 #else /* not emacs */
112 /* If we are not linking with Emacs proper,
113 we can't use the relocating allocator
114 even if config.h says that we can. */
117 # if defined STDC_HEADERS || defined _LIBC
124 /* When used in Emacs's lib-src, we need to get bzero and bcopy somehow.
125 If nothing else has been done, use the method below. */
126 # ifdef INHIBIT_STRING_HEADER
127 # if !(defined HAVE_BZERO && defined HAVE_BCOPY)
128 # if !defined bzero && !defined bcopy
129 # undef INHIBIT_STRING_HEADER
134 /* This is the normal way of making sure we have a bcopy and a bzero.
135 This is used in most programs--a few other programs avoid this
136 by defining INHIBIT_STRING_HEADER. */
137 # ifndef INHIBIT_STRING_HEADER
138 # if defined HAVE_STRING_H || defined STDC_HEADERS || defined _LIBC
142 # define bzero(s, n) (memset (s, '\0', n), (s))
144 # define bzero(s, n) __bzero (s, n)
148 # include <strings.h>
150 # define memcmp(s1, s2, n) bcmp (s1, s2, n)
153 # define memcpy(d, s, n) (bcopy (s, d, n), (d))
158 /* Define the syntax stuff for \<, \>, etc. */
160 /* This must be nonzero for the wordchar and notwordchar pattern
161 commands in re_match_2. */
166 # ifdef SWITCH_ENUM_BUG
167 # define SWITCH_ENUM_CAST(x) ((int)(x))
169 # define SWITCH_ENUM_CAST(x) (x)
172 #endif /* not emacs */
174 #if defined _LIBC || HAVE_LIMITS_H
179 # define MB_LEN_MAX 1
182 /* Get the interface, including the syntax bits. */
185 /* isalpha etc. are used for the character classes. */
188 /* Jim Meyering writes:
190 "... Some ctype macros are valid only for character codes that
191 isascii says are ASCII (SGI's IRIX-4.0.5 is one such system --when
192 using /bin/cc or gcc but without giving an ansi option). So, all
193 ctype uses should be through macros like ISPRINT... If
194 STDC_HEADERS is defined, then autoconf has verified that the ctype
195 macros don't need to be guarded with references to isascii. ...
196 Defining isascii to 1 should let any compiler worth its salt
197 eliminate the && through constant folding."
198 Solaris defines some of these symbols so we must undefine them first. */
200 #if defined STDC_HEADERS || (!defined isascii && !defined HAVE_ISASCII)
201 # define IN_CTYPE_DOMAIN(c) 1
203 # define IN_CTYPE_DOMAIN(c) isascii(c)
207 # define ISBLANK(c) (IN_CTYPE_DOMAIN (c) && isblank (c))
209 # define ISBLANK(c) ((c) == ' ' || (c) == '\t')
212 # define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isgraph (c))
214 # define ISGRAPH(c) (IN_CTYPE_DOMAIN (c) && isprint (c) && !isspace (c))
218 #define ISPRINT(c) (IN_CTYPE_DOMAIN (c) && isprint (c))
219 #define ISDIGIT(c) (IN_CTYPE_DOMAIN (c) && isdigit (c))
220 #define ISALNUM(c) (IN_CTYPE_DOMAIN (c) && isalnum (c))
221 #define ISALPHA(c) (IN_CTYPE_DOMAIN (c) && isalpha (c))
222 #define ISCNTRL(c) (IN_CTYPE_DOMAIN (c) && iscntrl (c))
223 #define ISLOWER(c) (IN_CTYPE_DOMAIN (c) && islower (c))
224 #define ISPUNCT(c) (IN_CTYPE_DOMAIN (c) && ispunct (c))
225 #define ISSPACE(c) (IN_CTYPE_DOMAIN (c) && isspace (c))
226 #define ISUPPER(c) (IN_CTYPE_DOMAIN (c) && isupper (c))
227 #define ISXDIGIT(c) (IN_CTYPE_DOMAIN (c) && isxdigit (c))
230 # define TOLOWER(c) _tolower(c)
232 # define TOLOWER(c) tolower(c)
236 # define NULL (void *)0
239 /* We remove any previous definition of `SIGN_EXTEND_CHAR',
240 since ours (we hope) works properly with all combinations of
241 machines, compilers, `char' and `unsigned char' argument types.
242 (Per Bothner suggested the basic approach.) */
243 #undef SIGN_EXTEND_CHAR
245 # define SIGN_EXTEND_CHAR(c) ((signed char) (c))
246 #else /* not __STDC__ */
247 /* As in Harbison and Steele. */
248 # define SIGN_EXTEND_CHAR(c) ((((unsigned char) (c)) ^ 128) - 128)
252 /* How many characters in the character set. */
253 # define CHAR_SET_SIZE 256
257 extern char *re_syntax_table;
259 # else /* not SYNTAX_TABLE */
261 static char re_syntax_table[CHAR_SET_SIZE];
271 bzero (re_syntax_table, sizeof re_syntax_table);
273 for (c = 0; c < CHAR_SET_SIZE; ++c)
275 re_syntax_table[c] = Sword;
277 re_syntax_table['_'] = Sword;
282 # endif /* not SYNTAX_TABLE */
284 # define SYNTAX(c) re_syntax_table[(unsigned char) (c)]
288 /* Should we use malloc or alloca? If REGEX_MALLOC is not defined, we
289 use `alloca' instead of `malloc'. This is because using malloc in
290 re_search* or re_match* could cause memory leaks when C-g is used in
291 Emacs; also, malloc is slower and causes storage fragmentation. On
292 the other hand, malloc is more portable, and easier to debug.
294 Because we sometimes use alloca, some routines have to be macros,
295 not functions -- `alloca'-allocated space disappears at the end of the
296 function it is called in. */
300 # define REGEX_ALLOCATE malloc
301 # define REGEX_REALLOCATE(source, osize, nsize) realloc (source, nsize)
302 # define REGEX_FREE free
304 #else /* not REGEX_MALLOC */
306 /* Emacs already defines alloca, sometimes. */
309 /* Make alloca work the best possible way. */
311 # define alloca __builtin_alloca
312 # else /* not __GNUC__ */
315 # endif /* HAVE_ALLOCA_H */
316 # endif /* not __GNUC__ */
318 # endif /* not alloca */
320 # define REGEX_ALLOCATE alloca
322 /* Assumes a `char *destination' variable. */
323 # define REGEX_REALLOCATE(source, osize, nsize) \
324 (destination = (char *) alloca (nsize), \
325 memcpy (destination, source, osize))
327 /* No need to do anything to free, after alloca. */
328 # define REGEX_FREE(arg) ((void)0) /* Do nothing! But inhibit gcc warning. */
330 #endif /* not REGEX_MALLOC */
332 /* Define how to allocate the failure stack. */
334 #if defined REL_ALLOC && defined REGEX_MALLOC
336 # define REGEX_ALLOCATE_STACK(size) \
337 r_alloc (&failure_stack_ptr, (size))
338 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
339 r_re_alloc (&failure_stack_ptr, (nsize))
340 # define REGEX_FREE_STACK(ptr) \
341 r_alloc_free (&failure_stack_ptr)
343 #else /* not using relocating allocator */
347 # define REGEX_ALLOCATE_STACK malloc
348 # define REGEX_REALLOCATE_STACK(source, osize, nsize) realloc (source, nsize)
349 # define REGEX_FREE_STACK free
351 # else /* not REGEX_MALLOC */
353 # define REGEX_ALLOCATE_STACK alloca
355 # define REGEX_REALLOCATE_STACK(source, osize, nsize) \
356 REGEX_REALLOCATE (source, osize, nsize)
357 /* No need to explicitly free anything. */
358 # define REGEX_FREE_STACK(arg)
360 # endif /* not REGEX_MALLOC */
361 #endif /* not using relocating allocator */
364 /* True if `size1' is non-NULL and PTR is pointing anywhere inside
365 `string1' or just past its end. This works if PTR is NULL, which is
367 #define FIRST_STRING_P(ptr) \
368 (size1 && string1 <= (ptr) && (ptr) <= string1 + size1)
370 /* (Re)Allocate N items of type T using malloc, or fail. */
371 #define TALLOC(n, t) ((t *) malloc ((n) * sizeof (t)))
372 #define RETALLOC(addr, n, t) ((addr) = (t *) realloc (addr, (n) * sizeof (t)))
373 #define RETALLOC_IF(addr, n, t) \
374 if (addr) RETALLOC((addr), (n), t); else (addr) = TALLOC ((n), t)
375 #define REGEX_TALLOC(n, t) ((t *) REGEX_ALLOCATE ((n) * sizeof (t)))
377 #define BYTEWIDTH 8 /* In bits. */
379 #define STREQ(s1, s2) ((strcmp (s1, s2) == 0))
383 #define MAX(a, b) ((a) > (b) ? (a) : (b))
384 #define MIN(a, b) ((a) < (b) ? (a) : (b))
386 typedef char boolean;
390 static int re_match_2_internal PARAMS ((struct re_pattern_buffer *bufp,
391 const char *string1, int size1,
392 const char *string2, int size2,
394 struct re_registers *regs,
397 /* These are the command codes that appear in compiled regular
398 expressions. Some opcodes are followed by argument bytes. A
399 command code can specify any interpretation whatsoever for its
400 arguments. Zero bytes may appear in the compiled regular expression. */
406 /* Succeed right away--no more backtracking. */
409 /* Followed by one byte giving n, then by n literal bytes. */
412 /* Matches any (more or less) character. */
415 /* Matches any one char belonging to specified set. First
416 following byte is number of bitmap bytes. Then come bytes
417 for a bitmap saying which chars are in. Bits in each byte
418 are ordered low-bit-first. A character is in the set if its
419 bit is 1. A character too large to have a bit in the map is
420 automatically not in the set. */
423 /* Same parameters as charset, but match any character that is
424 not one of those specified. */
427 /* Start remembering the text that is matched, for storing in a
428 register. Followed by one byte with the register number, in
429 the range 0 to one less than the pattern buffer's re_nsub
430 field. Then followed by one byte with the number of groups
431 inner to this one. (This last has to be part of the
432 start_memory only because we need it in the on_failure_jump
436 /* Stop remembering the text that is matched and store it in a
437 memory register. Followed by one byte with the register
438 number, in the range 0 to one less than `re_nsub' in the
439 pattern buffer, and one byte with the number of inner groups,
440 just like `start_memory'. (We need the number of inner
441 groups here because we don't have any easy way of finding the
442 corresponding start_memory when we're at a stop_memory.) */
445 /* Match a duplicate of something remembered. Followed by one
446 byte containing the register number. */
449 /* Fail unless at beginning of line. */
452 /* Fail unless at end of line. */
455 /* Succeeds if at beginning of buffer (if emacs) or at beginning
456 of string to be matched (if not). */
459 /* Analogously, for end of buffer/string. */
462 /* Followed by two byte relative address to which to jump. */
465 /* Same as jump, but marks the end of an alternative. */
468 /* Followed by two-byte relative address of place to resume at
469 in case of failure. */
472 /* Like on_failure_jump, but pushes a placeholder instead of the
473 current string position when executed. */
474 on_failure_keep_string_jump,
476 /* Throw away latest failure point and then jump to following
477 two-byte relative address. */
480 /* Change to pop_failure_jump if know won't have to backtrack to
481 match; otherwise change to jump. This is used to jump
482 back to the beginning of a repeat. If what follows this jump
483 clearly won't match what the repeat does, such that we can be
484 sure that there is no use backtracking out of repetitions
485 already matched, then we change it to a pop_failure_jump.
486 Followed by two-byte address. */
489 /* Jump to following two-byte address, and push a dummy failure
490 point. This failure point will be thrown away if an attempt
491 is made to use it for a failure. A `+' construct makes this
492 before the first repeat. Also used as an intermediary kind
493 of jump when compiling an alternative. */
496 /* Push a dummy failure point and continue. Used at the end of
500 /* Followed by two-byte relative address and two-byte number n.
501 After matching N times, jump to the address upon failure. */
504 /* Followed by two-byte relative address, and two-byte number n.
505 Jump to the address N times, then fail. */
508 /* Set the following two-byte relative address to the
509 subsequent two-byte number. The address *includes* the two
513 wordchar, /* Matches any word-constituent character. */
514 notwordchar, /* Matches any char that is not a word-constituent. */
516 wordbeg, /* Succeeds if at word beginning. */
517 wordend, /* Succeeds if at word end. */
519 wordbound, /* Succeeds if at a word boundary. */
520 notwordbound /* Succeeds if not at a word boundary. */
523 ,before_dot, /* Succeeds if before point. */
524 at_dot, /* Succeeds if at point. */
525 after_dot, /* Succeeds if after point. */
527 /* Matches any character whose syntax is specified. Followed by
528 a byte which contains a syntax code, e.g., Sword. */
531 /* Matches any character whose syntax is not that specified. */
536 /* Common operations on the compiled pattern. */
538 /* Store NUMBER in two contiguous bytes starting at DESTINATION. */
540 #define STORE_NUMBER(destination, number) \
542 (destination)[0] = (number) & 0377; \
543 (destination)[1] = (number) >> 8; \
546 /* Same as STORE_NUMBER, except increment DESTINATION to
547 the byte after where the number is stored. Therefore, DESTINATION
548 must be an lvalue. */
550 #define STORE_NUMBER_AND_INCR(destination, number) \
552 STORE_NUMBER (destination, number); \
553 (destination) += 2; \
556 /* Put into DESTINATION a number stored in two contiguous bytes starting
559 #define EXTRACT_NUMBER(destination, source) \
561 (destination) = *(source) & 0377; \
562 (destination) += SIGN_EXTEND_CHAR (*((source) + 1)) << 8; \
566 static void extract_number _RE_ARGS ((int *dest, unsigned char *source));
568 extract_number (dest, source)
570 unsigned char *source;
572 int temp = SIGN_EXTEND_CHAR (*(source + 1));
573 *dest = *source & 0377;
577 # ifndef EXTRACT_MACROS /* To debug the macros. */
578 # undef EXTRACT_NUMBER
579 # define EXTRACT_NUMBER(dest, src) extract_number (&dest, src)
580 # endif /* not EXTRACT_MACROS */
584 /* Same as EXTRACT_NUMBER, except increment SOURCE to after the number.
585 SOURCE must be an lvalue. */
587 #define EXTRACT_NUMBER_AND_INCR(destination, source) \
589 EXTRACT_NUMBER (destination, source); \
594 static void extract_number_and_incr _RE_ARGS ((int *destination,
595 unsigned char **source));
597 extract_number_and_incr (destination, source)
599 unsigned char **source;
601 extract_number (destination, *source);
605 # ifndef EXTRACT_MACROS
606 # undef EXTRACT_NUMBER_AND_INCR
607 # define EXTRACT_NUMBER_AND_INCR(dest, src) \
608 extract_number_and_incr (&dest, &src)
609 # endif /* not EXTRACT_MACROS */
613 /* If DEBUG is defined, Regex prints many voluminous messages about what
614 it is doing (if the variable `debug' is nonzero). If linked with the
615 main program in `iregex.c', you can enter patterns and strings
616 interactively. And if linked with the main program in `main.c' and
617 the other test files, you can run the already-written tests. */
621 /* We use standard I/O for debugging. */
624 /* It is useful to test things that ``must'' be true when debugging. */
629 # define DEBUG_STATEMENT(e) e
630 # define DEBUG_PRINT1(x) if (debug) printf (x)
631 # define DEBUG_PRINT2(x1, x2) if (debug) printf (x1, x2)
632 # define DEBUG_PRINT3(x1, x2, x3) if (debug) printf (x1, x2, x3)
633 # define DEBUG_PRINT4(x1, x2, x3, x4) if (debug) printf (x1, x2, x3, x4)
634 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e) \
635 if (debug) print_partial_compiled_pattern (s, e)
636 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2) \
637 if (debug) print_double_string (w, s1, sz1, s2, sz2)
640 /* Print the fastmap in human-readable form. */
643 print_fastmap (fastmap)
646 unsigned was_a_range = 0;
649 while (i < (1 << BYTEWIDTH))
655 while (i < (1 << BYTEWIDTH) && fastmap[i])
671 /* Print a compiled pattern string in human-readable form, starting at
672 the START pointer into it and ending just before the pointer END. */
675 print_partial_compiled_pattern (start, end)
676 unsigned char *start;
681 unsigned char *p = start;
682 unsigned char *pend = end;
690 /* Loop over pattern commands. */
694 printf ("%t:\t", p - start);
696 printf ("%ld:\t", (long int) (p - start));
699 switch ((re_opcode_t) *p++)
707 printf ("/exactn/%d", mcnt);
718 printf ("/start_memory/%d/%d", mcnt, *p++);
723 printf ("/stop_memory/%d/%d", mcnt, *p++);
727 printf ("/duplicate/%d", *p++);
737 register int c, last = -100;
738 register int in_range = 0;
740 printf ("/charset [%s",
741 (re_opcode_t) *(p - 1) == charset_not ? "^" : "");
743 assert (p + *p < pend);
745 for (c = 0; c < 256; c++)
747 && (p[1 + (c/8)] & (1 << (c % 8))))
749 /* Are we starting a range? */
750 if (last + 1 == c && ! in_range)
755 /* Have we broken a range? */
756 else if (last + 1 != c && in_range)
785 case on_failure_jump:
786 extract_number_and_incr (&mcnt, &p);
788 printf ("/on_failure_jump to %t", p + mcnt - start);
790 printf ("/on_failure_jump to %ld", (long int) (p + mcnt - start));
794 case on_failure_keep_string_jump:
795 extract_number_and_incr (&mcnt, &p);
797 printf ("/on_failure_keep_string_jump to %t", p + mcnt - start);
799 printf ("/on_failure_keep_string_jump to %ld",
800 (long int) (p + mcnt - start));
804 case dummy_failure_jump:
805 extract_number_and_incr (&mcnt, &p);
807 printf ("/dummy_failure_jump to %t", p + mcnt - start);
809 printf ("/dummy_failure_jump to %ld", (long int) (p + mcnt - start));
813 case push_dummy_failure:
814 printf ("/push_dummy_failure");
818 extract_number_and_incr (&mcnt, &p);
820 printf ("/maybe_pop_jump to %t", p + mcnt - start);
822 printf ("/maybe_pop_jump to %ld", (long int) (p + mcnt - start));
826 case pop_failure_jump:
827 extract_number_and_incr (&mcnt, &p);
829 printf ("/pop_failure_jump to %t", p + mcnt - start);
831 printf ("/pop_failure_jump to %ld", (long int) (p + mcnt - start));
836 extract_number_and_incr (&mcnt, &p);
838 printf ("/jump_past_alt to %t", p + mcnt - start);
840 printf ("/jump_past_alt to %ld", (long int) (p + mcnt - start));
845 extract_number_and_incr (&mcnt, &p);
847 printf ("/jump to %t", p + mcnt - start);
849 printf ("/jump to %ld", (long int) (p + mcnt - start));
854 extract_number_and_incr (&mcnt, &p);
856 extract_number_and_incr (&mcnt2, &p);
858 printf ("/succeed_n to %t, %d times", p1 - start, mcnt2);
860 printf ("/succeed_n to %ld, %d times",
861 (long int) (p1 - start), mcnt2);
866 extract_number_and_incr (&mcnt, &p);
868 extract_number_and_incr (&mcnt2, &p);
869 printf ("/jump_n to %d, %d times", p1 - start, mcnt2);
873 extract_number_and_incr (&mcnt, &p);
875 extract_number_and_incr (&mcnt2, &p);
877 printf ("/set_number_at location %t to %d", p1 - start, mcnt2);
879 printf ("/set_number_at location %ld to %d",
880 (long int) (p1 - start), mcnt2);
885 printf ("/wordbound");
889 printf ("/notwordbound");
901 printf ("/before_dot");
909 printf ("/after_dot");
913 printf ("/syntaxspec");
915 printf ("/%d", mcnt);
919 printf ("/notsyntaxspec");
921 printf ("/%d", mcnt);
926 printf ("/wordchar");
930 printf ("/notwordchar");
942 printf ("?%d", *(p-1));
949 printf ("%t:\tend of pattern.\n", p - start);
951 printf ("%ld:\tend of pattern.\n", (long int) (p - start));
957 print_compiled_pattern (bufp)
958 struct re_pattern_buffer *bufp;
960 unsigned char *buffer = bufp->buffer;
962 print_partial_compiled_pattern (buffer, buffer + bufp->used);
963 printf ("%ld bytes used/%ld bytes allocated.\n",
964 bufp->used, bufp->allocated);
966 if (bufp->fastmap_accurate && bufp->fastmap)
968 printf ("fastmap: ");
969 print_fastmap (bufp->fastmap);
973 printf ("re_nsub: %Zd\t", bufp->re_nsub);
975 printf ("re_nsub: %ld\t", (long int) bufp->re_nsub);
977 printf ("regs_alloc: %d\t", bufp->regs_allocated);
978 printf ("can_be_null: %d\t", bufp->can_be_null);
979 printf ("newline_anchor: %d\n", bufp->newline_anchor);
980 printf ("no_sub: %d\t", bufp->no_sub);
981 printf ("not_bol: %d\t", bufp->not_bol);
982 printf ("not_eol: %d\t", bufp->not_eol);
983 printf ("syntax: %lx\n", bufp->syntax);
984 /* Perhaps we should print the translate table? */
989 print_double_string (where, string1, size1, string2, size2)
1002 if (FIRST_STRING_P (where))
1004 for (this_char = where - string1; this_char < size1; this_char++)
1005 putchar (string1[this_char]);
1010 for (this_char = where - string2; this_char < size2; this_char++)
1011 putchar (string2[this_char]);
1022 #else /* not DEBUG */
1027 # define DEBUG_STATEMENT(e)
1028 # define DEBUG_PRINT1(x)
1029 # define DEBUG_PRINT2(x1, x2)
1030 # define DEBUG_PRINT3(x1, x2, x3)
1031 # define DEBUG_PRINT4(x1, x2, x3, x4)
1032 # define DEBUG_PRINT_COMPILED_PATTERN(p, s, e)
1033 # define DEBUG_PRINT_DOUBLE_STRING(w, s1, sz1, s2, sz2)
1035 #endif /* not DEBUG */
1037 /* Set by `re_set_syntax' to the current regexp syntax to recognize. Can
1038 also be assigned to arbitrarily: each pattern buffer stores its own
1039 syntax, so it can be changed between regex compilations. */
1040 /* This has no initializer because initialized variables in Emacs
1041 become read-only after dumping. */
1042 reg_syntax_t re_syntax_options;
1045 /* Specify the precise syntax of regexps for compilation. This provides
1046 for compatibility for various utilities which historically have
1047 different, incompatible syntaxes.
1049 The argument SYNTAX is a bit mask comprised of the various bits
1050 defined in regex.h. We return the old syntax. */
1053 re_set_syntax (syntax)
1054 reg_syntax_t syntax;
1056 reg_syntax_t ret = re_syntax_options;
1058 re_syntax_options = syntax;
1060 if (syntax & RE_DEBUG)
1062 else if (debug) /* was on but now is not */
1068 weak_alias (__re_set_syntax, re_set_syntax)
1071 /* This table gives an error message for each of the error codes listed
1072 in regex.h. Obviously the order here has to be same as there.
1073 POSIX doesn't require that we do anything for REG_NOERROR,
1074 but why not be nice? */
1076 static const char re_error_msgid[] =
1078 #define REG_NOERROR_IDX 0
1079 gettext_noop ("Success") /* REG_NOERROR */
1081 #define REG_NOMATCH_IDX (REG_NOERROR_IDX + sizeof "Success")
1082 gettext_noop ("No match") /* REG_NOMATCH */
1084 #define REG_BADPAT_IDX (REG_NOMATCH_IDX + sizeof "No match")
1085 gettext_noop ("Invalid regular expression") /* REG_BADPAT */
1087 #define REG_ECOLLATE_IDX (REG_BADPAT_IDX + sizeof "Invalid regular expression")
1088 gettext_noop ("Invalid collation character") /* REG_ECOLLATE */
1090 #define REG_ECTYPE_IDX (REG_ECOLLATE_IDX + sizeof "Invalid collation character")
1091 gettext_noop ("Invalid character class name") /* REG_ECTYPE */
1093 #define REG_EESCAPE_IDX (REG_ECTYPE_IDX + sizeof "Invalid character class name")
1094 gettext_noop ("Trailing backslash") /* REG_EESCAPE */
1096 #define REG_ESUBREG_IDX (REG_EESCAPE_IDX + sizeof "Trailing backslash")
1097 gettext_noop ("Invalid back reference") /* REG_ESUBREG */
1099 #define REG_EBRACK_IDX (REG_ESUBREG_IDX + sizeof "Invalid back reference")
1100 gettext_noop ("Unmatched [ or [^") /* REG_EBRACK */
1102 #define REG_EPAREN_IDX (REG_EBRACK_IDX + sizeof "Unmatched [ or [^")
1103 gettext_noop ("Unmatched ( or \\(") /* REG_EPAREN */
1105 #define REG_EBRACE_IDX (REG_EPAREN_IDX + sizeof "Unmatched ( or \\(")
1106 gettext_noop ("Unmatched \\{") /* REG_EBRACE */
1108 #define REG_BADBR_IDX (REG_EBRACE_IDX + sizeof "Unmatched \\{")
1109 gettext_noop ("Invalid content of \\{\\}") /* REG_BADBR */
1111 #define REG_ERANGE_IDX (REG_BADBR_IDX + sizeof "Invalid content of \\{\\}")
1112 gettext_noop ("Invalid range end") /* REG_ERANGE */
1114 #define REG_ESPACE_IDX (REG_ERANGE_IDX + sizeof "Invalid range end")
1115 gettext_noop ("Memory exhausted") /* REG_ESPACE */
1117 #define REG_BADRPT_IDX (REG_ESPACE_IDX + sizeof "Memory exhausted")
1118 gettext_noop ("Invalid preceding regular expression") /* REG_BADRPT */
1120 #define REG_EEND_IDX (REG_BADRPT_IDX + sizeof "Invalid preceding regular expression")
1121 gettext_noop ("Premature end of regular expression") /* REG_EEND */
1123 #define REG_ESIZE_IDX (REG_EEND_IDX + sizeof "Premature end of regular expression")
1124 gettext_noop ("Regular expression too big") /* REG_ESIZE */
1126 #define REG_ERPAREN_IDX (REG_ESIZE_IDX + sizeof "Regular expression too big")
1127 gettext_noop ("Unmatched ) or \\)") /* REG_ERPAREN */
1130 static const size_t re_error_msgid_idx[] =
1151 /* Avoiding alloca during matching, to placate r_alloc. */
1153 /* Define MATCH_MAY_ALLOCATE unless we need to make sure that the
1154 searching and matching functions should not call alloca. On some
1155 systems, alloca is implemented in terms of malloc, and if we're
1156 using the relocating allocator routines, then malloc could cause a
1157 relocation, which might (if the strings being searched are in the
1158 ralloc heap) shift the data out from underneath the regexp
1161 Here's another reason to avoid allocation: Emacs
1162 processes input from X in a signal handler; processing X input may
1163 call malloc; if input arrives while a matching routine is calling
1164 malloc, then we're scrod. But Emacs can't just block input while
1165 calling matching routines; then we don't notice interrupts when
1166 they come in. So, Emacs blocks input around all regexp calls
1167 except the matching calls, which it leaves unprotected, in the
1168 faith that they will not malloc. */
1170 /* Normally, this is fine. */
1171 #define MATCH_MAY_ALLOCATE
1173 /* When using GNU C, we are not REALLY using the C alloca, no matter
1174 what config.h may say. So don't take precautions for it. */
1179 /* The match routines may not allocate if (1) they would do it with malloc
1180 and (2) it's not safe for them to use malloc.
1181 Note that if REL_ALLOC is defined, matching would not use malloc for the
1182 failure stack, but we would still use it for the register vectors;
1183 so REL_ALLOC should not affect this. */
1184 #if (defined C_ALLOCA || defined REGEX_MALLOC) && defined emacs
1185 # undef MATCH_MAY_ALLOCATE
1189 /* Failure stack declarations and macros; both re_compile_fastmap and
1190 re_match_2 use a failure stack. These have to be macros because of
1191 REGEX_ALLOCATE_STACK. */
1194 /* Number of failure points for which to initially allocate space
1195 when matching. If this number is exceeded, we allocate more
1196 space, so it is not a hard limit. */
1197 #ifndef INIT_FAILURE_ALLOC
1198 # define INIT_FAILURE_ALLOC 5
1201 /* Roughly the maximum number of failure points on the stack. Would be
1202 exactly that if always used MAX_FAILURE_ITEMS items each time we failed.
1203 This is a variable only so users of regex can assign to it; we never
1204 change it ourselves. */
1208 # if defined MATCH_MAY_ALLOCATE
1209 /* 4400 was enough to cause a crash on Alpha OSF/1,
1210 whose default stack limit is 2mb. */
1211 long int re_max_failures = 4000;
1213 long int re_max_failures = 2000;
1216 union fail_stack_elt
1218 unsigned char *pointer;
1222 typedef union fail_stack_elt fail_stack_elt_t;
1226 fail_stack_elt_t *stack;
1227 unsigned long int size;
1228 unsigned long int avail; /* Offset of next open position. */
1231 #else /* not INT_IS_16BIT */
1233 # if defined MATCH_MAY_ALLOCATE
1234 /* 4400 was enough to cause a crash on Alpha OSF/1,
1235 whose default stack limit is 2mb. */
1236 int re_max_failures = 4000;
1238 int re_max_failures = 2000;
1241 union fail_stack_elt
1243 unsigned char *pointer;
1247 typedef union fail_stack_elt fail_stack_elt_t;
1251 fail_stack_elt_t *stack;
1253 unsigned avail; /* Offset of next open position. */
1256 #endif /* INT_IS_16BIT */
1258 #define FAIL_STACK_EMPTY() (fail_stack.avail == 0)
1259 #define FAIL_STACK_PTR_EMPTY() (fail_stack_ptr->avail == 0)
1260 #define FAIL_STACK_FULL() (fail_stack.avail == fail_stack.size)
1263 /* Define macros to initialize and free the failure stack.
1264 Do `return -2' if the alloc fails. */
1266 #ifdef MATCH_MAY_ALLOCATE
1267 # define INIT_FAIL_STACK() \
1269 fail_stack.stack = (fail_stack_elt_t *) \
1270 REGEX_ALLOCATE_STACK (INIT_FAILURE_ALLOC * sizeof (fail_stack_elt_t)); \
1272 if (fail_stack.stack == NULL) \
1275 fail_stack.size = INIT_FAILURE_ALLOC; \
1276 fail_stack.avail = 0; \
1279 # define RESET_FAIL_STACK() REGEX_FREE_STACK (fail_stack.stack)
1281 # define INIT_FAIL_STACK() \
1283 fail_stack.avail = 0; \
1286 # define RESET_FAIL_STACK()
1290 /* Double the size of FAIL_STACK, up to approximately `re_max_failures' items.
1292 Return 1 if succeeds, and 0 if either ran out of memory
1293 allocating space for it or it was already too large.
1295 REGEX_REALLOCATE_STACK requires `destination' be declared. */
1297 #define DOUBLE_FAIL_STACK(fail_stack) \
1298 ((fail_stack).size > (unsigned) (re_max_failures * MAX_FAILURE_ITEMS) \
1300 : ((fail_stack).stack = (fail_stack_elt_t *) \
1301 REGEX_REALLOCATE_STACK ((fail_stack).stack, \
1302 (fail_stack).size * sizeof (fail_stack_elt_t), \
1303 ((fail_stack).size << 1) * sizeof (fail_stack_elt_t)), \
1305 (fail_stack).stack == NULL \
1307 : ((fail_stack).size <<= 1, \
1311 /* Push pointer POINTER on FAIL_STACK.
1312 Return 1 if was able to do so and 0 if ran out of memory allocating
1314 #define PUSH_PATTERN_OP(POINTER, FAIL_STACK) \
1315 ((FAIL_STACK_FULL () \
1316 && !DOUBLE_FAIL_STACK (FAIL_STACK)) \
1318 : ((FAIL_STACK).stack[(FAIL_STACK).avail++].pointer = POINTER, \
1321 /* Push a pointer value onto the failure stack.
1322 Assumes the variable `fail_stack'. Probably should only
1323 be called from within `PUSH_FAILURE_POINT'. */
1324 #define PUSH_FAILURE_POINTER(item) \
1325 fail_stack.stack[fail_stack.avail++].pointer = (unsigned char *) (item)
1327 /* This pushes an integer-valued item onto the failure stack.
1328 Assumes the variable `fail_stack'. Probably should only
1329 be called from within `PUSH_FAILURE_POINT'. */
1330 #define PUSH_FAILURE_INT(item) \
1331 fail_stack.stack[fail_stack.avail++].integer = (item)
1333 /* Push a fail_stack_elt_t value onto the failure stack.
1334 Assumes the variable `fail_stack'. Probably should only
1335 be called from within `PUSH_FAILURE_POINT'. */
1336 #define PUSH_FAILURE_ELT(item) \
1337 fail_stack.stack[fail_stack.avail++] = (item)
1339 /* These three POP... operations complement the three PUSH... operations.
1340 All assume that `fail_stack' is nonempty. */
1341 #define POP_FAILURE_POINTER() fail_stack.stack[--fail_stack.avail].pointer
1342 #define POP_FAILURE_INT() fail_stack.stack[--fail_stack.avail].integer
1343 #define POP_FAILURE_ELT() fail_stack.stack[--fail_stack.avail]
1345 /* Used to omit pushing failure point id's when we're not debugging. */
1347 # define DEBUG_PUSH PUSH_FAILURE_INT
1348 # define DEBUG_POP(item_addr) *(item_addr) = POP_FAILURE_INT ()
1350 # define DEBUG_PUSH(item)
1351 # define DEBUG_POP(item_addr)
1355 /* Push the information about the state we will need
1356 if we ever fail back to it.
1358 Requires variables fail_stack, regstart, regend, reg_info, and
1359 num_regs_pushed be declared. DOUBLE_FAIL_STACK requires `destination'
1362 Does `return FAILURE_CODE' if runs out of memory. */
1364 #define PUSH_FAILURE_POINT(pattern_place, string_place, failure_code) \
1366 char *destination; \
1367 /* Must be int, so when we don't save any registers, the arithmetic \
1368 of 0 + -1 isn't done as unsigned. */ \
1369 /* Can't be int, since there is not a shred of a guarantee that int \
1370 is wide enough to hold a value of something to which pointer can \
1372 active_reg_t this_reg; \
1374 DEBUG_STATEMENT (failure_id++); \
1375 DEBUG_STATEMENT (nfailure_points_pushed++); \
1376 DEBUG_PRINT2 ("\nPUSH_FAILURE_POINT #%u:\n", failure_id); \
1377 DEBUG_PRINT2 (" Before push, next avail: %d\n", (fail_stack).avail);\
1378 DEBUG_PRINT2 (" size: %d\n", (fail_stack).size);\
1380 DEBUG_PRINT2 (" slots needed: %ld\n", NUM_FAILURE_ITEMS); \
1381 DEBUG_PRINT2 (" available: %d\n", REMAINING_AVAIL_SLOTS); \
1383 /* Ensure we have enough space allocated for what we will push. */ \
1384 while (REMAINING_AVAIL_SLOTS < NUM_FAILURE_ITEMS) \
1386 if (!DOUBLE_FAIL_STACK (fail_stack)) \
1387 return failure_code; \
1389 DEBUG_PRINT2 ("\n Doubled stack; size now: %d\n", \
1390 (fail_stack).size); \
1391 DEBUG_PRINT2 (" slots available: %d\n", REMAINING_AVAIL_SLOTS);\
1394 /* Push the info, starting with the registers. */ \
1395 DEBUG_PRINT1 ("\n"); \
1398 for (this_reg = lowest_active_reg; this_reg <= highest_active_reg; \
1401 DEBUG_PRINT2 (" Pushing reg: %lu\n", this_reg); \
1402 DEBUG_STATEMENT (num_regs_pushed++); \
1404 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1405 PUSH_FAILURE_POINTER (regstart[this_reg]); \
1407 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1408 PUSH_FAILURE_POINTER (regend[this_reg]); \
1410 DEBUG_PRINT2 (" info: %p\n ", \
1411 reg_info[this_reg].word.pointer); \
1412 DEBUG_PRINT2 (" match_null=%d", \
1413 REG_MATCH_NULL_STRING_P (reg_info[this_reg])); \
1414 DEBUG_PRINT2 (" active=%d", IS_ACTIVE (reg_info[this_reg])); \
1415 DEBUG_PRINT2 (" matched_something=%d", \
1416 MATCHED_SOMETHING (reg_info[this_reg])); \
1417 DEBUG_PRINT2 (" ever_matched=%d", \
1418 EVER_MATCHED_SOMETHING (reg_info[this_reg])); \
1419 DEBUG_PRINT1 ("\n"); \
1420 PUSH_FAILURE_ELT (reg_info[this_reg].word); \
1423 DEBUG_PRINT2 (" Pushing low active reg: %ld\n", lowest_active_reg);\
1424 PUSH_FAILURE_INT (lowest_active_reg); \
1426 DEBUG_PRINT2 (" Pushing high active reg: %ld\n", highest_active_reg);\
1427 PUSH_FAILURE_INT (highest_active_reg); \
1429 DEBUG_PRINT2 (" Pushing pattern %p:\n", pattern_place); \
1430 DEBUG_PRINT_COMPILED_PATTERN (bufp, pattern_place, pend); \
1431 PUSH_FAILURE_POINTER (pattern_place); \
1433 DEBUG_PRINT2 (" Pushing string %p: `", string_place); \
1434 DEBUG_PRINT_DOUBLE_STRING (string_place, string1, size1, string2, \
1436 DEBUG_PRINT1 ("'\n"); \
1437 PUSH_FAILURE_POINTER (string_place); \
1439 DEBUG_PRINT2 (" Pushing failure id: %u\n", failure_id); \
1440 DEBUG_PUSH (failure_id); \
1443 /* This is the number of items that are pushed and popped on the stack
1444 for each register. */
1445 #define NUM_REG_ITEMS 3
1447 /* Individual items aside from the registers. */
1449 # define NUM_NONREG_ITEMS 5 /* Includes failure point id. */
1451 # define NUM_NONREG_ITEMS 4
1454 /* We push at most this many items on the stack. */
1455 /* We used to use (num_regs - 1), which is the number of registers
1456 this regexp will save; but that was changed to 5
1457 to avoid stack overflow for a regexp with lots of parens. */
1458 #define MAX_FAILURE_ITEMS (5 * NUM_REG_ITEMS + NUM_NONREG_ITEMS)
1460 /* We actually push this many items. */
1461 #define NUM_FAILURE_ITEMS \
1463 ? 0 : highest_active_reg - lowest_active_reg + 1) \
1467 /* How many items can still be added to the stack without overflowing it. */
1468 #define REMAINING_AVAIL_SLOTS ((fail_stack).size - (fail_stack).avail)
1471 /* Pops what PUSH_FAIL_STACK pushes.
1473 We restore into the parameters, all of which should be lvalues:
1474 STR -- the saved data position.
1475 PAT -- the saved pattern position.
1476 LOW_REG, HIGH_REG -- the highest and lowest active registers.
1477 REGSTART, REGEND -- arrays of string positions.
1478 REG_INFO -- array of information about each subexpression.
1480 Also assumes the variables `fail_stack' and (if debugging), `bufp',
1481 `pend', `string1', `size1', `string2', and `size2'. */
1483 #define POP_FAILURE_POINT(str, pat, low_reg, high_reg, regstart, regend, reg_info)\
1485 DEBUG_STATEMENT (unsigned failure_id;) \
1486 active_reg_t this_reg; \
1487 const unsigned char *string_temp; \
1489 assert (!FAIL_STACK_EMPTY ()); \
1491 /* Remove failure points and point to how many regs pushed. */ \
1492 DEBUG_PRINT1 ("POP_FAILURE_POINT:\n"); \
1493 DEBUG_PRINT2 (" Before pop, next avail: %d\n", fail_stack.avail); \
1494 DEBUG_PRINT2 (" size: %d\n", fail_stack.size); \
1496 assert (fail_stack.avail >= NUM_NONREG_ITEMS); \
1498 DEBUG_POP (&failure_id); \
1499 DEBUG_PRINT2 (" Popping failure id: %u\n", failure_id); \
1501 /* If the saved string location is NULL, it came from an \
1502 on_failure_keep_string_jump opcode, and we want to throw away the \
1503 saved NULL, thus retaining our current position in the string. */ \
1504 string_temp = POP_FAILURE_POINTER (); \
1505 if (string_temp != NULL) \
1506 str = (const char *) string_temp; \
1508 DEBUG_PRINT2 (" Popping string %p: `", str); \
1509 DEBUG_PRINT_DOUBLE_STRING (str, string1, size1, string2, size2); \
1510 DEBUG_PRINT1 ("'\n"); \
1512 pat = (unsigned char *) POP_FAILURE_POINTER (); \
1513 DEBUG_PRINT2 (" Popping pattern %p:\n", pat); \
1514 DEBUG_PRINT_COMPILED_PATTERN (bufp, pat, pend); \
1516 /* Restore register info. */ \
1517 high_reg = (active_reg_t) POP_FAILURE_INT (); \
1518 DEBUG_PRINT2 (" Popping high active reg: %ld\n", high_reg); \
1520 low_reg = (active_reg_t) POP_FAILURE_INT (); \
1521 DEBUG_PRINT2 (" Popping low active reg: %ld\n", low_reg); \
1524 for (this_reg = high_reg; this_reg >= low_reg; this_reg--) \
1526 DEBUG_PRINT2 (" Popping reg: %ld\n", this_reg); \
1528 reg_info[this_reg].word = POP_FAILURE_ELT (); \
1529 DEBUG_PRINT2 (" info: %p\n", \
1530 reg_info[this_reg].word.pointer); \
1532 regend[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1533 DEBUG_PRINT2 (" end: %p\n", regend[this_reg]); \
1535 regstart[this_reg] = (const char *) POP_FAILURE_POINTER (); \
1536 DEBUG_PRINT2 (" start: %p\n", regstart[this_reg]); \
1540 for (this_reg = highest_active_reg; this_reg > high_reg; this_reg--) \
1542 reg_info[this_reg].word.integer = 0; \
1543 regend[this_reg] = 0; \
1544 regstart[this_reg] = 0; \
1546 highest_active_reg = high_reg; \
1549 set_regs_matched_done = 0; \
1550 DEBUG_STATEMENT (nfailure_points_popped++); \
1551 } /* POP_FAILURE_POINT */
1555 /* Structure for per-register (a.k.a. per-group) information.
1556 Other register information, such as the
1557 starting and ending positions (which are addresses), and the list of
1558 inner groups (which is a bits list) are maintained in separate
1561 We are making a (strictly speaking) nonportable assumption here: that
1562 the compiler will pack our bit fields into something that fits into
1563 the type of `word', i.e., is something that fits into one item on the
1567 /* Declarations and macros for re_match_2. */
1571 fail_stack_elt_t word;
1574 /* This field is one if this group can match the empty string,
1575 zero if not. If not yet determined, `MATCH_NULL_UNSET_VALUE'. */
1576 #define MATCH_NULL_UNSET_VALUE 3
1577 unsigned match_null_string_p : 2;
1578 unsigned is_active : 1;
1579 unsigned matched_something : 1;
1580 unsigned ever_matched_something : 1;
1582 } register_info_type;
1584 #define REG_MATCH_NULL_STRING_P(R) ((R).bits.match_null_string_p)
1585 #define IS_ACTIVE(R) ((R).bits.is_active)
1586 #define MATCHED_SOMETHING(R) ((R).bits.matched_something)
1587 #define EVER_MATCHED_SOMETHING(R) ((R).bits.ever_matched_something)
1590 /* Call this when have matched a real character; it sets `matched' flags
1591 for the subexpressions which we are currently inside. Also records
1592 that those subexprs have matched. */
1593 #define SET_REGS_MATCHED() \
1596 if (!set_regs_matched_done) \
1599 set_regs_matched_done = 1; \
1600 for (r = lowest_active_reg; r <= highest_active_reg; r++) \
1602 MATCHED_SOMETHING (reg_info[r]) \
1603 = EVER_MATCHED_SOMETHING (reg_info[r]) \
1610 /* Registers are set to a sentinel when they haven't yet matched. */
1611 static char reg_unset_dummy;
1612 #define REG_UNSET_VALUE (®_unset_dummy)
1613 #define REG_UNSET(e) ((e) == REG_UNSET_VALUE)
1615 /* Subroutine declarations and macros for regex_compile. */
1617 static reg_errcode_t regex_compile _RE_ARGS ((const char *pattern, size_t size,
1618 reg_syntax_t syntax,
1619 struct re_pattern_buffer *bufp));
1620 static void store_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc, int arg));
1621 static void store_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1622 int arg1, int arg2));
1623 static void insert_op1 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1624 int arg, unsigned char *end));
1625 static void insert_op2 _RE_ARGS ((re_opcode_t op, unsigned char *loc,
1626 int arg1, int arg2, unsigned char *end));
1627 static boolean at_begline_loc_p _RE_ARGS ((const char *pattern, const char *p,
1628 reg_syntax_t syntax));
1629 static boolean at_endline_loc_p _RE_ARGS ((const char *p, const char *pend,
1630 reg_syntax_t syntax));
1631 static reg_errcode_t compile_range _RE_ARGS ((unsigned int range_start,
1635 reg_syntax_t syntax,
1638 /* Fetch the next character in the uncompiled pattern---translating it
1639 if necessary. Also cast from a signed character in the constant
1640 string passed to us by the user to an unsigned char that we can use
1641 as an array index (in, e.g., `translate'). */
1643 # define PATFETCH(c) \
1644 do {if (p == pend) return REG_EEND; \
1645 c = (unsigned char) *p++; \
1646 if (translate) c = (unsigned char) translate[c]; \
1650 /* Fetch the next character in the uncompiled pattern, with no
1652 #define PATFETCH_RAW(c) \
1653 do {if (p == pend) return REG_EEND; \
1654 c = (unsigned char) *p++; \
1657 /* Go backwards one character in the pattern. */
1658 #define PATUNFETCH p--
1661 /* If `translate' is non-null, return translate[D], else just D. We
1662 cast the subscript to translate because some data is declared as
1663 `char *', to avoid warnings when a string constant is passed. But
1664 when we use a character as a subscript we must make it unsigned. */
1666 # define TRANSLATE(d) \
1667 (translate ? (char) translate[(unsigned char) (d)] : (d))
1671 /* Macros for outputting the compiled pattern into `buffer'. */
1673 /* If the buffer isn't allocated when it comes in, use this. */
1674 #define INIT_BUF_SIZE 32
1676 /* Make sure we have at least N more bytes of space in buffer. */
1677 #define GET_BUFFER_SPACE(n) \
1678 while ((unsigned long) (b - bufp->buffer + (n)) > bufp->allocated) \
1681 /* Make sure we have one more byte of buffer space and then add C to it. */
1682 #define BUF_PUSH(c) \
1684 GET_BUFFER_SPACE (1); \
1685 *b++ = (unsigned char) (c); \
1689 /* Ensure we have two more bytes of buffer space and then append C1 and C2. */
1690 #define BUF_PUSH_2(c1, c2) \
1692 GET_BUFFER_SPACE (2); \
1693 *b++ = (unsigned char) (c1); \
1694 *b++ = (unsigned char) (c2); \
1698 /* As with BUF_PUSH_2, except for three bytes. */
1699 #define BUF_PUSH_3(c1, c2, c3) \
1701 GET_BUFFER_SPACE (3); \
1702 *b++ = (unsigned char) (c1); \
1703 *b++ = (unsigned char) (c2); \
1704 *b++ = (unsigned char) (c3); \
1708 /* Store a jump with opcode OP at LOC to location TO. We store a
1709 relative address offset by the three bytes the jump itself occupies. */
1710 #define STORE_JUMP(op, loc, to) \
1711 store_op1 (op, loc, (int) ((to) - (loc) - 3))
1713 /* Likewise, for a two-argument jump. */
1714 #define STORE_JUMP2(op, loc, to, arg) \
1715 store_op2 (op, loc, (int) ((to) - (loc) - 3), arg)
1717 /* Like `STORE_JUMP', but for inserting. Assume `b' is the buffer end. */
1718 #define INSERT_JUMP(op, loc, to) \
1719 insert_op1 (op, loc, (int) ((to) - (loc) - 3), b)
1721 /* Like `STORE_JUMP2', but for inserting. Assume `b' is the buffer end. */
1722 #define INSERT_JUMP2(op, loc, to, arg) \
1723 insert_op2 (op, loc, (int) ((to) - (loc) - 3), arg, b)
1726 /* This is not an arbitrary limit: the arguments which represent offsets
1727 into the pattern are two bytes long. So if 2^16 bytes turns out to
1728 be too small, many things would have to change. */
1729 /* Any other compiler which, like MSC, has allocation limit below 2^16
1730 bytes will have to use approach similar to what was done below for
1731 MSC and drop MAX_BUF_SIZE a bit. Otherwise you may end up
1732 reallocating to 0 bytes. Such thing is not going to work too well.
1733 You have been warned!! */
1734 #if defined _MSC_VER && !defined WIN32
1735 /* Microsoft C 16-bit versions limit malloc to approx 65512 bytes.
1736 The REALLOC define eliminates a flurry of conversion warnings,
1737 but is not required. */
1738 # define MAX_BUF_SIZE 65500L
1739 # define REALLOC(p,s) realloc ((p), (size_t) (s))
1741 # define MAX_BUF_SIZE (1L << 16)
1742 # define REALLOC(p,s) realloc ((p), (s))
1745 /* Extend the buffer by twice its current size via realloc and
1746 reset the pointers that pointed into the old block to point to the
1747 correct places in the new one. If extending the buffer results in it
1748 being larger than MAX_BUF_SIZE, then flag memory exhausted. */
1749 #if __BOUNDED_POINTERS__
1750 # define SET_HIGH_BOUND(P) (__ptrhigh (P) = __ptrlow (P) + bufp->allocated)
1751 # define MOVE_BUFFER_POINTER(P) \
1752 (__ptrlow (P) += incr, SET_HIGH_BOUND (P), __ptrvalue (P) += incr)
1753 # define ELSE_EXTEND_BUFFER_HIGH_BOUND \
1756 SET_HIGH_BOUND (b); \
1757 SET_HIGH_BOUND (begalt); \
1758 if (fixup_alt_jump) \
1759 SET_HIGH_BOUND (fixup_alt_jump); \
1761 SET_HIGH_BOUND (laststart); \
1762 if (pending_exact) \
1763 SET_HIGH_BOUND (pending_exact); \
1766 # define MOVE_BUFFER_POINTER(P) (P) += incr
1767 # define ELSE_EXTEND_BUFFER_HIGH_BOUND
1769 #define EXTEND_BUFFER() \
1771 unsigned char *old_buffer = bufp->buffer; \
1772 if (bufp->allocated == MAX_BUF_SIZE) \
1774 bufp->allocated <<= 1; \
1775 if (bufp->allocated > MAX_BUF_SIZE) \
1776 bufp->allocated = MAX_BUF_SIZE; \
1777 bufp->buffer = (unsigned char *) REALLOC (bufp->buffer, bufp->allocated);\
1778 if (bufp->buffer == NULL) \
1779 return REG_ESPACE; \
1780 /* If the buffer moved, move all the pointers into it. */ \
1781 if (old_buffer != bufp->buffer) \
1783 int incr = bufp->buffer - old_buffer; \
1784 MOVE_BUFFER_POINTER (b); \
1785 MOVE_BUFFER_POINTER (begalt); \
1786 if (fixup_alt_jump) \
1787 MOVE_BUFFER_POINTER (fixup_alt_jump); \
1789 MOVE_BUFFER_POINTER (laststart); \
1790 if (pending_exact) \
1791 MOVE_BUFFER_POINTER (pending_exact); \
1793 ELSE_EXTEND_BUFFER_HIGH_BOUND \
1797 /* Since we have one byte reserved for the register number argument to
1798 {start,stop}_memory, the maximum number of groups we can report
1799 things about is what fits in that byte. */
1800 #define MAX_REGNUM 255
1802 /* But patterns can have more than `MAX_REGNUM' registers. We just
1803 ignore the excess. */
1804 typedef unsigned regnum_t;
1807 /* Macros for the compile stack. */
1809 /* Since offsets can go either forwards or backwards, this type needs to
1810 be able to hold values from -(MAX_BUF_SIZE - 1) to MAX_BUF_SIZE - 1. */
1811 /* int may be not enough when sizeof(int) == 2. */
1812 typedef long pattern_offset_t;
1816 pattern_offset_t begalt_offset;
1817 pattern_offset_t fixup_alt_jump;
1818 pattern_offset_t inner_group_offset;
1819 pattern_offset_t laststart_offset;
1821 } compile_stack_elt_t;
1826 compile_stack_elt_t *stack;
1828 unsigned avail; /* Offset of next open position. */
1829 } compile_stack_type;
1832 #define INIT_COMPILE_STACK_SIZE 32
1834 #define COMPILE_STACK_EMPTY (compile_stack.avail == 0)
1835 #define COMPILE_STACK_FULL (compile_stack.avail == compile_stack.size)
1837 /* The next available element. */
1838 #define COMPILE_STACK_TOP (compile_stack.stack[compile_stack.avail])
1841 /* Set the bit for character C in a list. */
1842 #define SET_LIST_BIT(c) \
1843 (b[((unsigned char) (c)) / BYTEWIDTH] \
1844 |= 1 << (((unsigned char) c) % BYTEWIDTH))
1847 /* Get the next unsigned number in the uncompiled pattern. */
1848 #define GET_UNSIGNED_NUMBER(num) \
1852 while ('0' <= c && c <= '9') \
1856 num = num * 10 + c - '0'; \
1864 #if defined _LIBC || WIDE_CHAR_SUPPORT
1865 /* The GNU C library provides support for user-defined character classes
1866 and the functions from ISO C amendement 1. */
1867 # ifdef CHARCLASS_NAME_MAX
1868 # define CHAR_CLASS_MAX_LENGTH CHARCLASS_NAME_MAX
1870 /* This shouldn't happen but some implementation might still have this
1871 problem. Use a reasonable default value. */
1872 # define CHAR_CLASS_MAX_LENGTH 256
1876 # define IS_CHAR_CLASS(string) __wctype (string)
1878 # define IS_CHAR_CLASS(string) wctype (string)
1881 # define CHAR_CLASS_MAX_LENGTH 6 /* Namely, `xdigit'. */
1883 # define IS_CHAR_CLASS(string) \
1884 (STREQ (string, "alpha") || STREQ (string, "upper") \
1885 || STREQ (string, "lower") || STREQ (string, "digit") \
1886 || STREQ (string, "alnum") || STREQ (string, "xdigit") \
1887 || STREQ (string, "space") || STREQ (string, "print") \
1888 || STREQ (string, "punct") || STREQ (string, "graph") \
1889 || STREQ (string, "cntrl") || STREQ (string, "blank"))
1892 #ifndef MATCH_MAY_ALLOCATE
1894 /* If we cannot allocate large objects within re_match_2_internal,
1895 we make the fail stack and register vectors global.
1896 The fail stack, we grow to the maximum size when a regexp
1898 The register vectors, we adjust in size each time we
1899 compile a regexp, according to the number of registers it needs. */
1901 static fail_stack_type fail_stack;
1903 /* Size with which the following vectors are currently allocated.
1904 That is so we can make them bigger as needed,
1905 but never make them smaller. */
1906 static int regs_allocated_size;
1908 static const char ** regstart, ** regend;
1909 static const char ** old_regstart, ** old_regend;
1910 static const char **best_regstart, **best_regend;
1911 static register_info_type *reg_info;
1912 static const char **reg_dummy;
1913 static register_info_type *reg_info_dummy;
1915 /* Make the register vectors big enough for NUM_REGS registers,
1916 but don't make them smaller. */
1919 regex_grow_registers (num_regs)
1922 if (num_regs > regs_allocated_size)
1924 RETALLOC_IF (regstart, num_regs, const char *);
1925 RETALLOC_IF (regend, num_regs, const char *);
1926 RETALLOC_IF (old_regstart, num_regs, const char *);
1927 RETALLOC_IF (old_regend, num_regs, const char *);
1928 RETALLOC_IF (best_regstart, num_regs, const char *);
1929 RETALLOC_IF (best_regend, num_regs, const char *);
1930 RETALLOC_IF (reg_info, num_regs, register_info_type);
1931 RETALLOC_IF (reg_dummy, num_regs, const char *);
1932 RETALLOC_IF (reg_info_dummy, num_regs, register_info_type);
1934 regs_allocated_size = num_regs;
1938 #endif /* not MATCH_MAY_ALLOCATE */
1940 static boolean group_in_compile_stack _RE_ARGS ((compile_stack_type
1944 /* `regex_compile' compiles PATTERN (of length SIZE) according to SYNTAX.
1945 Returns one of error codes defined in `regex.h', or zero for success.
1947 Assumes the `allocated' (and perhaps `buffer') and `translate'
1948 fields are set in BUFP on entry.
1950 If it succeeds, results are put in BUFP (if it returns an error, the
1951 contents of BUFP are undefined):
1952 `buffer' is the compiled pattern;
1953 `syntax' is set to SYNTAX;
1954 `used' is set to the length of the compiled pattern;
1955 `fastmap_accurate' is zero;
1956 `re_nsub' is the number of subexpressions in PATTERN;
1957 `not_bol' and `not_eol' are zero;
1959 The `fastmap' and `newline_anchor' fields are neither
1960 examined nor set. */
1962 /* Return, freeing storage we allocated. */
1963 #define FREE_STACK_RETURN(value) \
1964 return (free (compile_stack.stack), value)
1966 static reg_errcode_t
1967 regex_compile (pattern, size, syntax, bufp)
1968 const char *pattern;
1970 reg_syntax_t syntax;
1971 struct re_pattern_buffer *bufp;
1973 /* We fetch characters from PATTERN here. Even though PATTERN is
1974 `char *' (i.e., signed), we declare these variables as unsigned, so
1975 they can be reliably used as array indices. */
1976 register unsigned char c, c1;
1978 /* A random temporary spot in PATTERN. */
1981 /* Points to the end of the buffer, where we should append. */
1982 register unsigned char *b;
1984 /* Keeps track of unclosed groups. */
1985 compile_stack_type compile_stack;
1987 /* Points to the current (ending) position in the pattern. */
1988 const char *p = pattern;
1989 const char *pend = pattern + size;
1991 /* How to translate the characters in the pattern. */
1992 RE_TRANSLATE_TYPE translate = bufp->translate;
1994 /* Address of the count-byte of the most recently inserted `exactn'
1995 command. This makes it possible to tell if a new exact-match
1996 character can be added to that command or if the character requires
1997 a new `exactn' command. */
1998 unsigned char *pending_exact = 0;
2000 /* Address of start of the most recently finished expression.
2001 This tells, e.g., postfix * where to find the start of its
2002 operand. Reset at the beginning of groups and alternatives. */
2003 unsigned char *laststart = 0;
2005 /* Address of beginning of regexp, or inside of last group. */
2006 unsigned char *begalt;
2008 /* Place in the uncompiled pattern (i.e., the {) to
2009 which to go back if the interval is invalid. */
2010 const char *beg_interval;
2012 /* Address of the place where a forward jump should go to the end of
2013 the containing expression. Each alternative of an `or' -- except the
2014 last -- ends with a forward jump of this sort. */
2015 unsigned char *fixup_alt_jump = 0;
2017 /* Counts open-groups as they are encountered. Remembered for the
2018 matching close-group on the compile stack, so the same register
2019 number is put in the stop_memory as the start_memory. */
2020 regnum_t regnum = 0;
2023 DEBUG_PRINT1 ("\nCompiling pattern: ");
2026 unsigned debug_count;
2028 for (debug_count = 0; debug_count < size; debug_count++)
2029 putchar (pattern[debug_count]);
2034 /* Initialize the compile stack. */
2035 compile_stack.stack = TALLOC (INIT_COMPILE_STACK_SIZE, compile_stack_elt_t);
2036 if (compile_stack.stack == NULL)
2039 compile_stack.size = INIT_COMPILE_STACK_SIZE;
2040 compile_stack.avail = 0;
2042 /* Initialize the pattern buffer. */
2043 bufp->syntax = syntax;
2044 bufp->fastmap_accurate = 0;
2045 bufp->not_bol = bufp->not_eol = 0;
2047 /* Set `used' to zero, so that if we return an error, the pattern
2048 printer (for debugging) will think there's no pattern. We reset it
2052 /* Always count groups, whether or not bufp->no_sub is set. */
2055 #if !defined emacs && !defined SYNTAX_TABLE
2056 /* Initialize the syntax table. */
2057 init_syntax_once ();
2060 if (bufp->allocated == 0)
2063 { /* If zero allocated, but buffer is non-null, try to realloc
2064 enough space. This loses if buffer's address is bogus, but
2065 that is the user's responsibility. */
2066 RETALLOC (bufp->buffer, INIT_BUF_SIZE, unsigned char);
2069 { /* Caller did not allocate a buffer. Do it for them. */
2070 bufp->buffer = TALLOC (INIT_BUF_SIZE, unsigned char);
2072 if (!bufp->buffer) FREE_STACK_RETURN (REG_ESPACE);
2074 bufp->allocated = INIT_BUF_SIZE;
2077 begalt = b = bufp->buffer;
2079 /* Loop through the uncompiled pattern until we're at the end. */
2088 if ( /* If at start of pattern, it's an operator. */
2090 /* If context independent, it's an operator. */
2091 || syntax & RE_CONTEXT_INDEP_ANCHORS
2092 /* Otherwise, depends on what's come before. */
2093 || at_begline_loc_p (pattern, p, syntax))
2103 if ( /* If at end of pattern, it's an operator. */
2105 /* If context independent, it's an operator. */
2106 || syntax & RE_CONTEXT_INDEP_ANCHORS
2107 /* Otherwise, depends on what's next. */
2108 || at_endline_loc_p (p, pend, syntax))
2118 if ((syntax & RE_BK_PLUS_QM)
2119 || (syntax & RE_LIMITED_OPS))
2123 /* If there is no previous pattern... */
2126 if (syntax & RE_CONTEXT_INVALID_OPS)
2127 FREE_STACK_RETURN (REG_BADRPT);
2128 else if (!(syntax & RE_CONTEXT_INDEP_OPS))
2133 /* Are we optimizing this jump? */
2134 boolean keep_string_p = false;
2136 /* 1 means zero (many) matches is allowed. */
2137 char zero_times_ok = 0, many_times_ok = 0;
2139 /* If there is a sequence of repetition chars, collapse it
2140 down to just one (the right one). We can't combine
2141 interval operators with these because of, e.g., `a{2}*',
2142 which should only match an even number of `a's. */
2146 zero_times_ok |= c != '+';
2147 many_times_ok |= c != '?';
2155 || (!(syntax & RE_BK_PLUS_QM) && (c == '+' || c == '?')))
2158 else if (syntax & RE_BK_PLUS_QM && c == '\\')
2160 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2163 if (!(c1 == '+' || c1 == '?'))
2178 /* If we get here, we found another repeat character. */
2181 /* Star, etc. applied to an empty pattern is equivalent
2182 to an empty pattern. */
2186 /* Now we know whether or not zero matches is allowed
2187 and also whether or not two or more matches is allowed. */
2189 { /* More than one repetition is allowed, so put in at the
2190 end a backward relative jump from `b' to before the next
2191 jump we're going to put in below (which jumps from
2192 laststart to after this jump).
2194 But if we are at the `*' in the exact sequence `.*\n',
2195 insert an unconditional jump backwards to the .,
2196 instead of the beginning of the loop. This way we only
2197 push a failure point once, instead of every time
2198 through the loop. */
2199 assert (p - 1 > pattern);
2201 /* Allocate the space for the jump. */
2202 GET_BUFFER_SPACE (3);
2204 /* We know we are not at the first character of the pattern,
2205 because laststart was nonzero. And we've already
2206 incremented `p', by the way, to be the character after
2207 the `*'. Do we have to do something analogous here
2208 for null bytes, because of RE_DOT_NOT_NULL? */
2209 if (TRANSLATE (*(p - 2)) == TRANSLATE ('.')
2211 && p < pend && TRANSLATE (*p) == TRANSLATE ('\n')
2212 && !(syntax & RE_DOT_NEWLINE))
2213 { /* We have .*\n. */
2214 STORE_JUMP (jump, b, laststart);
2215 keep_string_p = true;
2218 /* Anything else. */
2219 STORE_JUMP (maybe_pop_jump, b, laststart - 3);
2221 /* We've added more stuff to the buffer. */
2225 /* On failure, jump from laststart to b + 3, which will be the
2226 end of the buffer after this jump is inserted. */
2227 GET_BUFFER_SPACE (3);
2228 INSERT_JUMP (keep_string_p ? on_failure_keep_string_jump
2236 /* At least one repetition is required, so insert a
2237 `dummy_failure_jump' before the initial
2238 `on_failure_jump' instruction of the loop. This
2239 effects a skip over that instruction the first time
2240 we hit that loop. */
2241 GET_BUFFER_SPACE (3);
2242 INSERT_JUMP (dummy_failure_jump, laststart, laststart + 6);
2257 boolean had_char_class = false;
2258 unsigned int range_start = 0xffffffff;
2260 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2262 /* Ensure that we have enough space to push a charset: the
2263 opcode, the length count, and the bitset; 34 bytes in all. */
2264 GET_BUFFER_SPACE (34);
2268 /* We test `*p == '^' twice, instead of using an if
2269 statement, so we only need one BUF_PUSH. */
2270 BUF_PUSH (*p == '^' ? charset_not : charset);
2274 /* Remember the first position in the bracket expression. */
2277 /* Push the number of bytes in the bitmap. */
2278 BUF_PUSH ((1 << BYTEWIDTH) / BYTEWIDTH);
2280 /* Clear the whole map. */
2281 bzero (b, (1 << BYTEWIDTH) / BYTEWIDTH);
2283 /* charset_not matches newline according to a syntax bit. */
2284 if ((re_opcode_t) b[-2] == charset_not
2285 && (syntax & RE_HAT_LISTS_NOT_NEWLINE))
2286 SET_LIST_BIT ('\n');
2288 /* Read in characters and ranges, setting map bits. */
2291 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2295 /* \ might escape characters inside [...] and [^...]. */
2296 if ((syntax & RE_BACKSLASH_ESCAPE_IN_LISTS) && c == '\\')
2298 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2306 /* Could be the end of the bracket expression. If it's
2307 not (i.e., when the bracket expression is `[]' so
2308 far), the ']' character bit gets set way below. */
2309 if (c == ']' && p != p1 + 1)
2312 /* Look ahead to see if it's a range when the last thing
2313 was a character class. */
2314 if (had_char_class && c == '-' && *p != ']')
2315 FREE_STACK_RETURN (REG_ERANGE);
2317 /* Look ahead to see if it's a range when the last thing
2318 was a character: if this is a hyphen not at the
2319 beginning or the end of a list, then it's the range
2322 && !(p - 2 >= pattern && p[-2] == '[')
2323 && !(p - 3 >= pattern && p[-3] == '[' && p[-2] == '^')
2327 = compile_range (range_start, &p, pend, translate,
2329 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2330 range_start = 0xffffffff;
2333 else if (p[0] == '-' && p[1] != ']')
2334 { /* This handles ranges made up of characters only. */
2337 /* Move past the `-'. */
2340 ret = compile_range (c, &p, pend, translate, syntax, b);
2341 if (ret != REG_NOERROR) FREE_STACK_RETURN (ret);
2342 range_start = 0xffffffff;
2345 /* See if we're at the beginning of a possible character
2348 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == ':')
2349 { /* Leave room for the null. */
2350 char str[CHAR_CLASS_MAX_LENGTH + 1];
2355 /* If pattern is `[[:'. */
2356 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2361 if ((c == ':' && *p == ']') || p == pend)
2363 if (c1 < CHAR_CLASS_MAX_LENGTH)
2366 /* This is in any case an invalid class name. */
2371 /* If isn't a word bracketed by `[:' and `:]':
2372 undo the ending character, the letters, and leave
2373 the leading `:' and `[' (but set bits for them). */
2374 if (c == ':' && *p == ']')
2376 #if defined _LIBC || WIDE_CHAR_SUPPORT
2377 boolean is_lower = STREQ (str, "lower");
2378 boolean is_upper = STREQ (str, "upper");
2382 wt = IS_CHAR_CLASS (str);
2384 FREE_STACK_RETURN (REG_ECTYPE);
2386 /* Throw away the ] at the end of the character
2390 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2392 for (ch = 0; ch < 1 << BYTEWIDTH; ++ch)
2395 if (__iswctype (__btowc (ch), wt))
2398 if (iswctype (btowc (ch), wt))
2402 if (translate && (is_upper || is_lower)
2403 && (ISUPPER (ch) || ISLOWER (ch)))
2407 had_char_class = true;
2410 boolean is_alnum = STREQ (str, "alnum");
2411 boolean is_alpha = STREQ (str, "alpha");
2412 boolean is_blank = STREQ (str, "blank");
2413 boolean is_cntrl = STREQ (str, "cntrl");
2414 boolean is_digit = STREQ (str, "digit");
2415 boolean is_graph = STREQ (str, "graph");
2416 boolean is_lower = STREQ (str, "lower");
2417 boolean is_print = STREQ (str, "print");
2418 boolean is_punct = STREQ (str, "punct");
2419 boolean is_space = STREQ (str, "space");
2420 boolean is_upper = STREQ (str, "upper");
2421 boolean is_xdigit = STREQ (str, "xdigit");
2423 if (!IS_CHAR_CLASS (str))
2424 FREE_STACK_RETURN (REG_ECTYPE);
2426 /* Throw away the ] at the end of the character
2430 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2432 for (ch = 0; ch < 1 << BYTEWIDTH; ch++)
2434 /* This was split into 3 if's to
2435 avoid an arbitrary limit in some compiler. */
2436 if ( (is_alnum && ISALNUM (ch))
2437 || (is_alpha && ISALPHA (ch))
2438 || (is_blank && ISBLANK (ch))
2439 || (is_cntrl && ISCNTRL (ch)))
2441 if ( (is_digit && ISDIGIT (ch))
2442 || (is_graph && ISGRAPH (ch))
2443 || (is_lower && ISLOWER (ch))
2444 || (is_print && ISPRINT (ch)))
2446 if ( (is_punct && ISPUNCT (ch))
2447 || (is_space && ISSPACE (ch))
2448 || (is_upper && ISUPPER (ch))
2449 || (is_xdigit && ISXDIGIT (ch)))
2451 if ( translate && (is_upper || is_lower)
2452 && (ISUPPER (ch) || ISLOWER (ch)))
2455 had_char_class = true;
2456 #endif /* libc || wctype.h */
2466 had_char_class = false;
2469 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '=')
2471 unsigned char str[MB_LEN_MAX + 1];
2474 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2480 /* If pattern is `[[='. */
2481 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2486 if ((c == '=' && *p == ']') || p == pend)
2488 if (c1 < MB_LEN_MAX)
2491 /* This is in any case an invalid class name. */
2496 if (c == '=' && *p == ']' && str[0] != '\0')
2498 /* If we have no collation data we use the default
2499 collation in which each character is in a class
2500 by itself. It also means that ASCII is the
2501 character set and therefore we cannot have character
2502 with more than one byte in the multibyte
2509 FREE_STACK_RETURN (REG_ECOLLATE);
2511 /* Throw away the ] at the end of the equivalence
2515 /* Set the bit for the character. */
2516 SET_LIST_BIT (str[0]);
2521 /* Try to match the byte sequence in `str' against
2522 those known to the collate implementation.
2523 First find out whether the bytes in `str' are
2524 actually from exactly one character. */
2525 const int32_t *table;
2526 const unsigned char *weights;
2527 const unsigned char *extra;
2528 const int32_t *indirect;
2530 const unsigned char *cp = str;
2533 /* This #include defines a local function! */
2534 # include <locale/weight.h>
2536 table = (const int32_t *)
2537 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_TABLEMB);
2538 weights = (const unsigned char *)
2539 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_WEIGHTMB);
2540 extra = (const unsigned char *)
2541 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_EXTRAMB);
2542 indirect = (const int32_t *)
2543 _NL_CURRENT (LC_COLLATE, _NL_COLLATE_INDIRECTMB);
2545 idx = findidx (&cp);
2546 if (idx == 0 || cp < str + c1)
2547 /* This is no valid character. */
2548 FREE_STACK_RETURN (REG_ECOLLATE);
2550 /* Throw away the ] at the end of the equivalence
2554 /* Now we have to go throught the whole table
2555 and find all characters which have the same
2558 XXX Note that this is not entirely correct.
2559 we would have to match multibyte sequences
2560 but this is not possible with the current
2562 for (ch = 1; ch < 256; ++ch)
2563 /* XXX This test would have to be changed if we
2564 would allow matching multibyte sequences. */
2567 int32_t idx2 = table[ch];
2568 size_t len = weights[idx2];
2570 /* Test whether the lenghts match. */
2571 if (weights[idx] == len)
2573 /* They do. New compare the bytes of
2578 && (weights[idx + 1 + cnt]
2579 == weights[idx2 + 1 + cnt]))
2583 /* They match. Mark the character as
2590 had_char_class = true;
2600 had_char_class = false;
2603 else if (syntax & RE_CHAR_CLASSES && c == '[' && *p == '.')
2605 unsigned char str[128]; /* Should be large enough. */
2608 _NL_CURRENT_WORD (LC_COLLATE, _NL_COLLATE_NRULES);
2614 /* If pattern is `[[='. */
2615 if (p == pend) FREE_STACK_RETURN (REG_EBRACK);
2620 if ((c == '.' && *p == ']') || p == pend)
2622 if (c1 < sizeof (str))
2625 /* This is in any case an invalid class name. */
2630 if (c == '.' && *p == ']' && str[0] != '\0')
2632 /* If we have no collation data we use the default
2633 collation in which each character is the name
2634 for its own class which contains only the one
2635 character. It also means that ASCII is the
2636 character set and therefore we cannot have character
2637 with more than one byte in the multibyte
2644 FREE_STACK_RETURN (REG_ECOLLATE);
2646 /* Throw away the ] at the end of the equivalence
2650 /* Set the bit for the character. */
2651 SET_LIST_BIT (str[0]);
2652 range_start = ((const unsigned char *) str)[0];
2657 /* Try to match the byte sequence in `str' against
2658 those known to the collate implementation.
2659 First find out whether the bytes in `str' are
2660 actually from exactly one character. */
2662 const int32_t *symb_table;
2663 const unsigned char *extra;
2670 _NL_CURRENT_WORD (LC_COLLATE,
2671 _NL_COLLATE_SYMB_HASH_SIZEMB);
2672 symb_table = (const int32_t *)
2673 _NL_CURRENT (LC_COLLATE,
2674 _NL_COLLATE_SYMB_TABLEMB);
2675 extra = (const unsigned char *)
2676 _NL_CURRENT (LC_COLLATE,
2677 _NL_COLLATE_SYMB_EXTRAMB);
2679 /* Locate the character in the hashing table. */
2680 hash = elem_hash (str, c1);
2683 elem = hash % table_size;
2684 second = hash % (table_size - 2);
2685 while (symb_table[2 * elem] != 0)
2687 /* First compare the hashing value. */
2688 if (symb_table[2 * elem] == hash
2689 && c1 == extra[symb_table[2 * elem + 1]]
2691 &extra[symb_table[2 * elem + 1]
2695 /* Yep, this is the entry. */
2696 idx = symb_table[2 * elem + 1];
2697 idx += 1 + extra[idx];
2705 if (symb_table[2 * elem] == 0)
2706 /* This is no valid character. */
2707 FREE_STACK_RETURN (REG_ECOLLATE);
2709 /* Throw away the ] at the end of the equivalence
2713 /* Now add the multibyte character(s) we found
2716 XXX Note that this is not entirely correct.
2717 we would have to match multibyte sequences
2718 but this is not possible with the current
2719 implementation. Also, we have to match
2720 collating symbols, which expand to more than
2721 one file, as a whole and not allow the
2722 individual bytes. */
2725 range_start = extra[idx];
2727 SET_LIST_BIT (extra[idx++]);
2730 had_char_class = false;
2740 had_char_class = false;
2745 had_char_class = false;
2751 /* Discard any (non)matching list bytes that are all 0 at the
2752 end of the map. Decrease the map-length byte too. */
2753 while ((int) b[-1] > 0 && b[b[-1] - 1] == 0)
2761 if (syntax & RE_NO_BK_PARENS)
2768 if (syntax & RE_NO_BK_PARENS)
2775 if (syntax & RE_NEWLINE_ALT)
2782 if (syntax & RE_NO_BK_VBAR)
2789 if (syntax & RE_INTERVALS && syntax & RE_NO_BK_BRACES)
2790 goto handle_interval;
2796 if (p == pend) FREE_STACK_RETURN (REG_EESCAPE);
2798 /* Do not translate the character after the \, so that we can
2799 distinguish, e.g., \B from \b, even if we normally would
2800 translate, e.g., B to b. */
2806 if (syntax & RE_NO_BK_PARENS)
2807 goto normal_backslash;
2813 if (COMPILE_STACK_FULL)
2815 RETALLOC (compile_stack.stack, compile_stack.size << 1,
2816 compile_stack_elt_t);
2817 if (compile_stack.stack == NULL) return REG_ESPACE;
2819 compile_stack.size <<= 1;
2822 /* These are the values to restore when we hit end of this
2823 group. They are all relative offsets, so that if the
2824 whole pattern moves because of realloc, they will still
2826 COMPILE_STACK_TOP.begalt_offset = begalt - bufp->buffer;
2827 COMPILE_STACK_TOP.fixup_alt_jump
2828 = fixup_alt_jump ? fixup_alt_jump - bufp->buffer + 1 : 0;
2829 COMPILE_STACK_TOP.laststart_offset = b - bufp->buffer;
2830 COMPILE_STACK_TOP.regnum = regnum;
2832 /* We will eventually replace the 0 with the number of
2833 groups inner to this one. But do not push a
2834 start_memory for groups beyond the last one we can
2835 represent in the compiled pattern. */
2836 if (regnum <= MAX_REGNUM)
2838 COMPILE_STACK_TOP.inner_group_offset = b - bufp->buffer + 2;
2839 BUF_PUSH_3 (start_memory, regnum, 0);
2842 compile_stack.avail++;
2847 /* If we've reached MAX_REGNUM groups, then this open
2848 won't actually generate any code, so we'll have to
2849 clear pending_exact explicitly. */
2855 if (syntax & RE_NO_BK_PARENS) goto normal_backslash;
2857 if (COMPILE_STACK_EMPTY)
2859 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2860 goto normal_backslash;
2862 FREE_STACK_RETURN (REG_ERPAREN);
2867 { /* Push a dummy failure point at the end of the
2868 alternative for a possible future
2869 `pop_failure_jump' to pop. See comments at
2870 `push_dummy_failure' in `re_match_2'. */
2871 BUF_PUSH (push_dummy_failure);
2873 /* We allocated space for this jump when we assigned
2874 to `fixup_alt_jump', in the `handle_alt' case below. */
2875 STORE_JUMP (jump_past_alt, fixup_alt_jump, b - 1);
2878 /* See similar code for backslashed left paren above. */
2879 if (COMPILE_STACK_EMPTY)
2881 if (syntax & RE_UNMATCHED_RIGHT_PAREN_ORD)
2884 FREE_STACK_RETURN (REG_ERPAREN);
2887 /* Since we just checked for an empty stack above, this
2888 ``can't happen''. */
2889 assert (compile_stack.avail != 0);
2891 /* We don't just want to restore into `regnum', because
2892 later groups should continue to be numbered higher,
2893 as in `(ab)c(de)' -- the second group is #2. */
2894 regnum_t this_group_regnum;
2896 compile_stack.avail--;
2897 begalt = bufp->buffer + COMPILE_STACK_TOP.begalt_offset;
2899 = COMPILE_STACK_TOP.fixup_alt_jump
2900 ? bufp->buffer + COMPILE_STACK_TOP.fixup_alt_jump - 1
2902 laststart = bufp->buffer + COMPILE_STACK_TOP.laststart_offset;
2903 this_group_regnum = COMPILE_STACK_TOP.regnum;
2904 /* If we've reached MAX_REGNUM groups, then this open
2905 won't actually generate any code, so we'll have to
2906 clear pending_exact explicitly. */
2909 /* We're at the end of the group, so now we know how many
2910 groups were inside this one. */
2911 if (this_group_regnum <= MAX_REGNUM)
2913 unsigned char *inner_group_loc
2914 = bufp->buffer + COMPILE_STACK_TOP.inner_group_offset;
2916 *inner_group_loc = regnum - this_group_regnum;
2917 BUF_PUSH_3 (stop_memory, this_group_regnum,
2918 regnum - this_group_regnum);
2924 case '|': /* `\|'. */
2925 if (syntax & RE_LIMITED_OPS || syntax & RE_NO_BK_VBAR)
2926 goto normal_backslash;
2928 if (syntax & RE_LIMITED_OPS)
2931 /* Insert before the previous alternative a jump which
2932 jumps to this alternative if the former fails. */
2933 GET_BUFFER_SPACE (3);
2934 INSERT_JUMP (on_failure_jump, begalt, b + 6);
2938 /* The alternative before this one has a jump after it
2939 which gets executed if it gets matched. Adjust that
2940 jump so it will jump to this alternative's analogous
2941 jump (put in below, which in turn will jump to the next
2942 (if any) alternative's such jump, etc.). The last such
2943 jump jumps to the correct final destination. A picture:
2949 If we are at `b', then fixup_alt_jump right now points to a
2950 three-byte space after `a'. We'll put in the jump, set
2951 fixup_alt_jump to right after `b', and leave behind three
2952 bytes which we'll fill in when we get to after `c'. */
2955 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
2957 /* Mark and leave space for a jump after this alternative,
2958 to be filled in later either by next alternative or
2959 when know we're at the end of a series of alternatives. */
2961 GET_BUFFER_SPACE (3);
2970 /* If \{ is a literal. */
2971 if (!(syntax & RE_INTERVALS)
2972 /* If we're at `\{' and it's not the open-interval
2974 || (syntax & RE_NO_BK_BRACES))
2975 goto normal_backslash;
2979 /* If got here, then the syntax allows intervals. */
2981 /* At least (most) this many matches must be made. */
2982 int lower_bound = -1, upper_bound = -1;
2984 beg_interval = p - 1;
2988 if (!(syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
2989 goto unfetch_interval;
2991 FREE_STACK_RETURN (REG_EBRACE);
2994 GET_UNSIGNED_NUMBER (lower_bound);
2998 GET_UNSIGNED_NUMBER (upper_bound);
2999 if ((!(syntax & RE_NO_BK_BRACES) && c != '\\')
3000 || ((syntax & RE_NO_BK_BRACES) && c != '}'))
3001 FREE_STACK_RETURN (REG_BADBR);
3003 if (upper_bound < 0)
3004 upper_bound = RE_DUP_MAX;
3007 /* Interval such as `{1}' => match exactly once. */
3008 upper_bound = lower_bound;
3010 if (lower_bound < 0 || upper_bound > RE_DUP_MAX
3011 || lower_bound > upper_bound)
3013 if (!(syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
3014 goto unfetch_interval;
3016 FREE_STACK_RETURN (REG_BADBR);
3019 if (!(syntax & RE_NO_BK_BRACES))
3021 if (c != '\\') FREE_STACK_RETURN (REG_EBRACE);
3028 if (!(syntax & RE_INTERVALS) && (syntax & RE_NO_BK_BRACES))
3029 goto unfetch_interval;
3031 FREE_STACK_RETURN (REG_BADBR);
3034 /* We just parsed a valid interval. */
3036 /* If it's invalid to have no preceding re. */
3039 if (syntax & RE_CONTEXT_INVALID_OPS)
3040 FREE_STACK_RETURN (REG_BADRPT);
3041 else if (syntax & RE_CONTEXT_INDEP_OPS)
3044 goto unfetch_interval;
3047 /* If the upper bound is zero, don't want to succeed at
3048 all; jump from `laststart' to `b + 3', which will be
3049 the end of the buffer after we insert the jump. */
3050 if (upper_bound == 0)
3052 GET_BUFFER_SPACE (3);
3053 INSERT_JUMP (jump, laststart, b + 3);
3057 /* Otherwise, we have a nontrivial interval. When
3058 we're all done, the pattern will look like:
3059 set_number_at <jump count> <upper bound>
3060 set_number_at <succeed_n count> <lower bound>
3061 succeed_n <after jump addr> <succeed_n count>
3063 jump_n <succeed_n addr> <jump count>
3064 (The upper bound and `jump_n' are omitted if
3065 `upper_bound' is 1, though.) */
3067 { /* If the upper bound is > 1, we need to insert
3068 more at the end of the loop. */
3069 unsigned nbytes = 10 + (upper_bound > 1) * 10;
3071 GET_BUFFER_SPACE (nbytes);
3073 /* Initialize lower bound of the `succeed_n', even
3074 though it will be set during matching by its
3075 attendant `set_number_at' (inserted next),
3076 because `re_compile_fastmap' needs to know.
3077 Jump to the `jump_n' we might insert below. */
3078 INSERT_JUMP2 (succeed_n, laststart,
3079 b + 5 + (upper_bound > 1) * 5,
3083 /* Code to initialize the lower bound. Insert
3084 before the `succeed_n'. The `5' is the last two
3085 bytes of this `set_number_at', plus 3 bytes of
3086 the following `succeed_n'. */
3087 insert_op2 (set_number_at, laststart, 5, lower_bound, b);
3090 if (upper_bound > 1)
3091 { /* More than one repetition is allowed, so
3092 append a backward jump to the `succeed_n'
3093 that starts this interval.
3095 When we've reached this during matching,
3096 we'll have matched the interval once, so
3097 jump back only `upper_bound - 1' times. */
3098 STORE_JUMP2 (jump_n, b, laststart + 5,
3102 /* The location we want to set is the second
3103 parameter of the `jump_n'; that is `b-2' as
3104 an absolute address. `laststart' will be
3105 the `set_number_at' we're about to insert;
3106 `laststart+3' the number to set, the source
3107 for the relative address. But we are
3108 inserting into the middle of the pattern --
3109 so everything is getting moved up by 5.
3110 Conclusion: (b - 2) - (laststart + 3) + 5,
3111 i.e., b - laststart.
3113 We insert this at the beginning of the loop
3114 so that if we fail during matching, we'll
3115 reinitialize the bounds. */
3116 insert_op2 (set_number_at, laststart, b - laststart,
3117 upper_bound - 1, b);
3122 beg_interval = NULL;
3127 /* If an invalid interval, match the characters as literals. */
3128 assert (beg_interval);
3130 beg_interval = NULL;
3132 /* normal_char and normal_backslash need `c'. */
3135 if (!(syntax & RE_NO_BK_BRACES))
3137 if (p > pattern && p[-1] == '\\')
3138 goto normal_backslash;
3143 /* There is no way to specify the before_dot and after_dot
3144 operators. rms says this is ok. --karl */
3152 BUF_PUSH_2 (syntaxspec, syntax_spec_code[c]);
3158 BUF_PUSH_2 (notsyntaxspec, syntax_spec_code[c]);
3164 if (syntax & RE_NO_GNU_OPS)
3167 BUF_PUSH (wordchar);
3172 if (syntax & RE_NO_GNU_OPS)
3175 BUF_PUSH (notwordchar);
3180 if (syntax & RE_NO_GNU_OPS)
3186 if (syntax & RE_NO_GNU_OPS)
3192 if (syntax & RE_NO_GNU_OPS)
3194 BUF_PUSH (wordbound);
3198 if (syntax & RE_NO_GNU_OPS)
3200 BUF_PUSH (notwordbound);
3204 if (syntax & RE_NO_GNU_OPS)
3210 if (syntax & RE_NO_GNU_OPS)
3215 case '1': case '2': case '3': case '4': case '5':
3216 case '6': case '7': case '8': case '9':
3217 if (syntax & RE_NO_BK_REFS)
3223 FREE_STACK_RETURN (REG_ESUBREG);
3225 /* Can't back reference to a subexpression if inside of it. */
3226 if (group_in_compile_stack (compile_stack, (regnum_t) c1))
3230 BUF_PUSH_2 (duplicate, c1);
3236 if (syntax & RE_BK_PLUS_QM)
3239 goto normal_backslash;
3243 /* You might think it would be useful for \ to mean
3244 not to translate; but if we don't translate it
3245 it will never match anything. */
3253 /* Expects the character in `c'. */
3255 /* If no exactn currently being built. */
3258 /* If last exactn not at current position. */
3259 || pending_exact + *pending_exact + 1 != b
3261 /* We have only one byte following the exactn for the count. */
3262 || *pending_exact == (1 << BYTEWIDTH) - 1
3264 /* If followed by a repetition operator. */
3265 || *p == '*' || *p == '^'
3266 || ((syntax & RE_BK_PLUS_QM)
3267 ? *p == '\\' && (p[1] == '+' || p[1] == '?')
3268 : (*p == '+' || *p == '?'))
3269 || ((syntax & RE_INTERVALS)
3270 && ((syntax & RE_NO_BK_BRACES)
3272 : (p[0] == '\\' && p[1] == '{'))))
3274 /* Start building a new exactn. */
3278 BUF_PUSH_2 (exactn, 0);
3279 pending_exact = b - 1;
3286 } /* while p != pend */
3289 /* Through the pattern now. */
3292 STORE_JUMP (jump_past_alt, fixup_alt_jump, b);
3294 if (!COMPILE_STACK_EMPTY)
3295 FREE_STACK_RETURN (REG_EPAREN);
3297 /* If we don't want backtracking, force success
3298 the first time we reach the end of the compiled pattern. */
3299 if (syntax & RE_NO_POSIX_BACKTRACKING)
3302 free (compile_stack.stack);
3304 /* We have succeeded; set the length of the buffer. */
3305 bufp->used = b - bufp->buffer;
3310 DEBUG_PRINT1 ("\nCompiled pattern: \n");
3311 print_compiled_pattern (bufp);
3315 #ifndef MATCH_MAY_ALLOCATE
3316 /* Initialize the failure stack to the largest possible stack. This
3317 isn't necessary unless we're trying to avoid calling alloca in
3318 the search and match routines. */
3320 int num_regs = bufp->re_nsub + 1;
3322 /* Since DOUBLE_FAIL_STACK refuses to double only if the current size
3323 is strictly greater than re_max_failures, the largest possible stack
3324 is 2 * re_max_failures failure points. */
3325 if (fail_stack.size < (2 * re_max_failures * MAX_FAILURE_ITEMS))
3327 fail_stack.size = (2 * re_max_failures * MAX_FAILURE_ITEMS);
3330 if (! fail_stack.stack)
3332 = (fail_stack_elt_t *) xmalloc (fail_stack.size
3333 * sizeof (fail_stack_elt_t));
3336 = (fail_stack_elt_t *) xrealloc (fail_stack.stack,
3338 * sizeof (fail_stack_elt_t)));
3339 # else /* not emacs */
3340 if (! fail_stack.stack)
3342 = (fail_stack_elt_t *) malloc (fail_stack.size
3343 * sizeof (fail_stack_elt_t));
3346 = (fail_stack_elt_t *) realloc (fail_stack.stack,
3348 * sizeof (fail_stack_elt_t)));
3349 # endif /* not emacs */
3352 regex_grow_registers (num_regs);
3354 #endif /* not MATCH_MAY_ALLOCATE */
3357 } /* regex_compile */
3359 /* Subroutines for `regex_compile'. */
3361 /* Store OP at LOC followed by two-byte integer parameter ARG. */
3364 store_op1 (op, loc, arg)
3369 *loc = (unsigned char) op;
3370 STORE_NUMBER (loc + 1, arg);
3374 /* Like `store_op1', but for two two-byte parameters ARG1 and ARG2. */
3377 store_op2 (op, loc, arg1, arg2)
3382 *loc = (unsigned char) op;
3383 STORE_NUMBER (loc + 1, arg1);
3384 STORE_NUMBER (loc + 3, arg2);
3388 /* Copy the bytes from LOC to END to open up three bytes of space at LOC
3389 for OP followed by two-byte integer parameter ARG. */
3392 insert_op1 (op, loc, arg, end)
3398 register unsigned char *pfrom = end;
3399 register unsigned char *pto = end + 3;
3401 while (pfrom != loc)
3404 store_op1 (op, loc, arg);
3408 /* Like `insert_op1', but for two two-byte parameters ARG1 and ARG2. */
3411 insert_op2 (op, loc, arg1, arg2, end)
3417 register unsigned char *pfrom = end;
3418 register unsigned char *pto = end + 5;
3420 while (pfrom != loc)
3423 store_op2 (op, loc, arg1, arg2);
3427 /* P points to just after a ^ in PATTERN. Return true if that ^ comes
3428 after an alternative or a begin-subexpression. We assume there is at
3429 least one character before the ^. */
3432 at_begline_loc_p (pattern, p, syntax)
3433 const char *pattern, *p;
3434 reg_syntax_t syntax;
3436 const char *prev = p - 2;
3437 boolean prev_prev_backslash = prev > pattern && prev[-1] == '\\';
3440 /* After a subexpression? */
3441 (*prev == '(' && (syntax & RE_NO_BK_PARENS || prev_prev_backslash))
3442 /* After an alternative? */
3443 || (*prev == '|' && (syntax & RE_NO_BK_VBAR || prev_prev_backslash));
3447 /* The dual of at_begline_loc_p. This one is for $. We assume there is
3448 at least one character after the $, i.e., `P < PEND'. */
3451 at_endline_loc_p (p, pend, syntax)
3452 const char *p, *pend;
3453 reg_syntax_t syntax;
3455 const char *next = p;
3456 boolean next_backslash = *next == '\\';
3457 const char *next_next = p + 1 < pend ? p + 1 : 0;
3460 /* Before a subexpression? */
3461 (syntax & RE_NO_BK_PARENS ? *next == ')'
3462 : next_backslash && next_next && *next_next == ')')
3463 /* Before an alternative? */
3464 || (syntax & RE_NO_BK_VBAR ? *next == '|'
3465 : next_backslash && next_next && *next_next == '|');
3469 /* Returns true if REGNUM is in one of COMPILE_STACK's elements and
3470 false if it's not. */
3473 group_in_compile_stack (compile_stack, regnum)
3474 compile_stack_type compile_stack;
3479 for (this_element = compile_stack.avail - 1;
3482 if (compile_stack.stack[this_element].regnum == regnum)
3489 /* Read the ending character of a range (in a bracket expression) from the
3490 uncompiled pattern *P_PTR (which ends at PEND). We assume the
3491 starting character is in `P[-2]'. (`P[-1]' is the character `-'.)
3492 Then we set the translation of all bits between the starting and
3493 ending characters (inclusive) in the compiled pattern B.
3495 Return an error code.
3497 We use these short variable names so we can use the same macros as
3498 `regex_compile' itself. */
3500 static reg_errcode_t
3501 compile_range (range_start_char, p_ptr, pend, translate, syntax, b)
3502 unsigned int range_start_char;
3503 const char **p_ptr, *pend;
3504 RE_TRANSLATE_TYPE translate;
3505 reg_syntax_t syntax;
3509 const char *p = *p_ptr;
3512 const unsigned char *collseq;
3513 unsigned int start_colseq;
3514 unsigned int end_colseq;
3522 /* Have to increment the pointer into the pattern string, so the
3523 caller isn't still at the ending character. */
3526 /* Report an error if the range is empty and the syntax prohibits this. */
3527 ret = syntax & RE_NO_EMPTY_RANGES ? REG_ERANGE : REG_NOERROR;
3530 collseq = (const unsigned char *) _NL_CURRENT (LC_COLLATE,
3531 _NL_COLLATE_COLLSEQMB);
3533 start_colseq = collseq[(unsigned char) TRANSLATE (range_start_char)];
3534 end_colseq = collseq[(unsigned char) TRANSLATE (p[0])];
3535 for (this_char = 0; this_char <= (unsigned char) -1; ++this_char)
3537 unsigned int this_colseq = collseq[(unsigned char) TRANSLATE (this_char)];
3539 if (start_colseq <= this_colseq && this_colseq <= end_colseq)
3541 SET_LIST_BIT (TRANSLATE (this_char));
3546 /* Here we see why `this_char' has to be larger than an `unsigned
3547 char' -- we would otherwise go into an infinite loop, since all
3548 characters <= 0xff. */
3549 range_start_char = TRANSLATE (range_start_char);
3550 end_char = TRANSLATE (p[0]);
3551 for (this_char = range_start_char; this_char <= end_char; ++this_char)
3553 SET_LIST_BIT (TRANSLATE (this_char));
3561 /* re_compile_fastmap computes a ``fastmap'' for the compiled pattern in
3562 BUFP. A fastmap records which of the (1 << BYTEWIDTH) possible
3563 characters can start a string that matches the pattern. This fastmap
3564 is used by re_search to skip quickly over impossible starting points.
3566 The caller must supply the address of a (1 << BYTEWIDTH)-byte data
3567 area as BUFP->fastmap.
3569 We set the `fastmap', `fastmap_accurate', and `can_be_null' fields in
3572 Returns 0 if we succeed, -2 if an internal error. */
3575 re_compile_fastmap (bufp)
3576 struct re_pattern_buffer *bufp;
3579 #ifdef MATCH_MAY_ALLOCATE
3580 fail_stack_type fail_stack;
3582 #ifndef REGEX_MALLOC
3586 register char *fastmap = bufp->fastmap;
3587 unsigned char *pattern = bufp->buffer;
3588 unsigned char *p = pattern;
3589 register unsigned char *pend = pattern + bufp->used;
3592 /* This holds the pointer to the failure stack, when
3593 it is allocated relocatably. */
3594 fail_stack_elt_t *failure_stack_ptr;
3597 /* Assume that each path through the pattern can be null until
3598 proven otherwise. We set this false at the bottom of switch
3599 statement, to which we get only if a particular path doesn't
3600 match the empty string. */
3601 boolean path_can_be_null = true;
3603 /* We aren't doing a `succeed_n' to begin with. */
3604 boolean succeed_n_p = false;
3606 assert (fastmap != NULL && p != NULL);
3609 bzero (fastmap, 1 << BYTEWIDTH); /* Assume nothing's valid. */
3610 bufp->fastmap_accurate = 1; /* It will be when we're done. */
3611 bufp->can_be_null = 0;
3615 if (p == pend || *p == succeed)
3617 /* We have reached the (effective) end of pattern. */
3618 if (!FAIL_STACK_EMPTY ())
3620 bufp->can_be_null |= path_can_be_null;
3622 /* Reset for next path. */
3623 path_can_be_null = true;
3625 p = fail_stack.stack[--fail_stack.avail].pointer;
3633 /* We should never be about to go beyond the end of the pattern. */
3636 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
3639 /* I guess the idea here is to simply not bother with a fastmap
3640 if a backreference is used, since it's too hard to figure out
3641 the fastmap for the corresponding group. Setting
3642 `can_be_null' stops `re_search_2' from using the fastmap, so
3643 that is all we do. */
3645 bufp->can_be_null = 1;
3649 /* Following are the cases which match a character. These end
3658 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3659 if (p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH)))
3665 /* Chars beyond end of map must be allowed. */
3666 for (j = *p * BYTEWIDTH; j < (1 << BYTEWIDTH); j++)
3669 for (j = *p++ * BYTEWIDTH - 1; j >= 0; j--)
3670 if (!(p[j / BYTEWIDTH] & (1 << (j % BYTEWIDTH))))
3676 for (j = 0; j < (1 << BYTEWIDTH); j++)
3677 if (SYNTAX (j) == Sword)
3683 for (j = 0; j < (1 << BYTEWIDTH); j++)
3684 if (SYNTAX (j) != Sword)
3691 int fastmap_newline = fastmap['\n'];
3693 /* `.' matches anything ... */
3694 for (j = 0; j < (1 << BYTEWIDTH); j++)
3697 /* ... except perhaps newline. */
3698 if (!(bufp->syntax & RE_DOT_NEWLINE))
3699 fastmap['\n'] = fastmap_newline;
3701 /* Return if we have already set `can_be_null'; if we have,
3702 then the fastmap is irrelevant. Something's wrong here. */
3703 else if (bufp->can_be_null)
3706 /* Otherwise, have to check alternative paths. */
3713 for (j = 0; j < (1 << BYTEWIDTH); j++)
3714 if (SYNTAX (j) == (enum syntaxcode) k)
3721 for (j = 0; j < (1 << BYTEWIDTH); j++)
3722 if (SYNTAX (j) != (enum syntaxcode) k)
3727 /* All cases after this match the empty string. These end with
3747 case push_dummy_failure:
3752 case pop_failure_jump:
3753 case maybe_pop_jump:
3756 case dummy_failure_jump:
3757 EXTRACT_NUMBER_AND_INCR (j, p);
3762 /* Jump backward implies we just went through the body of a
3763 loop and matched nothing. Opcode jumped to should be
3764 `on_failure_jump' or `succeed_n'. Just treat it like an
3765 ordinary jump. For a * loop, it has pushed its failure
3766 point already; if so, discard that as redundant. */
3767 if ((re_opcode_t) *p != on_failure_jump
3768 && (re_opcode_t) *p != succeed_n)
3772 EXTRACT_NUMBER_AND_INCR (j, p);
3775 /* If what's on the stack is where we are now, pop it. */
3776 if (!FAIL_STACK_EMPTY ()
3777 && fail_stack.stack[fail_stack.avail - 1].pointer == p)
3783 case on_failure_jump:
3784 case on_failure_keep_string_jump:
3785 handle_on_failure_jump:
3786 EXTRACT_NUMBER_AND_INCR (j, p);
3788 /* For some patterns, e.g., `(a?)?', `p+j' here points to the
3789 end of the pattern. We don't want to push such a point,
3790 since when we restore it above, entering the switch will
3791 increment `p' past the end of the pattern. We don't need
3792 to push such a point since we obviously won't find any more
3793 fastmap entries beyond `pend'. Such a pattern can match
3794 the null string, though. */
3797 if (!PUSH_PATTERN_OP (p + j, fail_stack))
3799 RESET_FAIL_STACK ();
3804 bufp->can_be_null = 1;
3808 EXTRACT_NUMBER_AND_INCR (k, p); /* Skip the n. */
3809 succeed_n_p = false;
3816 /* Get to the number of times to succeed. */
3819 /* Increment p past the n for when k != 0. */
3820 EXTRACT_NUMBER_AND_INCR (k, p);
3824 succeed_n_p = true; /* Spaghetti code alert. */
3825 goto handle_on_failure_jump;
3842 abort (); /* We have listed all the cases. */
3845 /* Getting here means we have found the possible starting
3846 characters for one path of the pattern -- and that the empty
3847 string does not match. We need not follow this path further.
3848 Instead, look at the next alternative (remembered on the
3849 stack), or quit if no more. The test at the top of the loop
3850 does these things. */
3851 path_can_be_null = false;
3855 /* Set `can_be_null' for the last path (also the first path, if the
3856 pattern is empty). */
3857 bufp->can_be_null |= path_can_be_null;
3860 RESET_FAIL_STACK ();
3862 } /* re_compile_fastmap */
3864 weak_alias (__re_compile_fastmap, re_compile_fastmap)
3867 /* Set REGS to hold NUM_REGS registers, storing them in STARTS and
3868 ENDS. Subsequent matches using PATTERN_BUFFER and REGS will use
3869 this memory for recording register information. STARTS and ENDS
3870 must be allocated using the malloc library routine, and must each
3871 be at least NUM_REGS * sizeof (regoff_t) bytes long.
3873 If NUM_REGS == 0, then subsequent matches should allocate their own
3876 Unless this function is called, the first search or match using
3877 PATTERN_BUFFER will allocate its own register data, without
3878 freeing the old data. */
3881 re_set_registers (bufp, regs, num_regs, starts, ends)
3882 struct re_pattern_buffer *bufp;
3883 struct re_registers *regs;
3885 regoff_t *starts, *ends;
3889 bufp->regs_allocated = REGS_REALLOCATE;
3890 regs->num_regs = num_regs;
3891 regs->start = starts;
3896 bufp->regs_allocated = REGS_UNALLOCATED;
3898 regs->start = regs->end = (regoff_t *) 0;
3902 weak_alias (__re_set_registers, re_set_registers)
3905 /* Searching routines. */
3907 /* Like re_search_2, below, but only one string is specified, and
3908 doesn't let you say where to stop matching. */
3911 re_search (bufp, string, size, startpos, range, regs)
3912 struct re_pattern_buffer *bufp;
3914 int size, startpos, range;
3915 struct re_registers *regs;
3917 return re_search_2 (bufp, NULL, 0, string, size, startpos, range,
3921 weak_alias (__re_search, re_search)
3925 /* Using the compiled pattern in BUFP->buffer, first tries to match the
3926 virtual concatenation of STRING1 and STRING2, starting first at index
3927 STARTPOS, then at STARTPOS + 1, and so on.
3929 STRING1 and STRING2 have length SIZE1 and SIZE2, respectively.
3931 RANGE is how far to scan while trying to match. RANGE = 0 means try
3932 only at STARTPOS; in general, the last start tried is STARTPOS +
3935 In REGS, return the indices of the virtual concatenation of STRING1
3936 and STRING2 that matched the entire BUFP->buffer and its contained
3939 Do not consider matching one past the index STOP in the virtual
3940 concatenation of STRING1 and STRING2.
3942 We return either the position in the strings at which the match was
3943 found, -1 if no match, or -2 if error (such as failure
3947 re_search_2 (bufp, string1, size1, string2, size2, startpos, range, regs, stop)
3948 struct re_pattern_buffer *bufp;
3949 const char *string1, *string2;
3953 struct re_registers *regs;
3957 register char *fastmap = bufp->fastmap;
3958 register RE_TRANSLATE_TYPE translate = bufp->translate;
3959 int total_size = size1 + size2;
3960 int endpos = startpos + range;
3962 /* Check for out-of-range STARTPOS. */
3963 if (startpos < 0 || startpos > total_size)
3966 /* Fix up RANGE if it might eventually take us outside
3967 the virtual concatenation of STRING1 and STRING2.
3968 Make sure we won't move STARTPOS below 0 or above TOTAL_SIZE. */
3970 range = 0 - startpos;
3971 else if (endpos > total_size)
3972 range = total_size - startpos;
3974 /* If the search isn't to be a backwards one, don't waste time in a
3975 search for a pattern that must be anchored. */
3976 if (bufp->used > 0 && range > 0
3977 && ((re_opcode_t) bufp->buffer[0] == begbuf
3978 /* `begline' is like `begbuf' if it cannot match at newlines. */
3979 || ((re_opcode_t) bufp->buffer[0] == begline
3980 && !bufp->newline_anchor)))
3989 /* In a forward search for something that starts with \=.
3990 don't keep searching past point. */
3991 if (bufp->used > 0 && (re_opcode_t) bufp->buffer[0] == at_dot && range > 0)
3993 range = PT - startpos;
3999 /* Update the fastmap now if not correct already. */
4000 if (fastmap && !bufp->fastmap_accurate)
4001 if (re_compile_fastmap (bufp) == -2)
4004 /* Loop through the string, looking for a place to start matching. */
4007 /* If a fastmap is supplied, skip quickly over characters that
4008 cannot be the start of a match. If the pattern can match the
4009 null string, however, we don't need to skip characters; we want
4010 the first null string. */
4011 if (fastmap && startpos < total_size && !bufp->can_be_null)
4013 if (range > 0) /* Searching forwards. */
4015 register const char *d;
4016 register int lim = 0;
4019 if (startpos < size1 && startpos + range >= size1)
4020 lim = range - (size1 - startpos);
4022 d = (startpos >= size1 ? string2 - size1 : string1) + startpos;
4024 /* Written out as an if-else to avoid testing `translate'
4028 && !fastmap[(unsigned char)
4029 translate[(unsigned char) *d++]])
4032 while (range > lim && !fastmap[(unsigned char) *d++])
4035 startpos += irange - range;
4037 else /* Searching backwards. */
4039 register char c = (size1 == 0 || startpos >= size1
4040 ? string2[startpos - size1]
4041 : string1[startpos]);
4043 if (!fastmap[(unsigned char) TRANSLATE (c)])
4048 /* If can't match the null string, and that's all we have left, fail. */
4049 if (range >= 0 && startpos == total_size && fastmap
4050 && !bufp->can_be_null)
4053 val = re_match_2_internal (bufp, string1, size1, string2, size2,
4054 startpos, regs, stop);
4055 #ifndef REGEX_MALLOC
4084 weak_alias (__re_search_2, re_search_2)
4087 /* This converts PTR, a pointer into one of the search strings `string1'
4088 and `string2' into an offset from the beginning of that string. */
4089 #define POINTER_TO_OFFSET(ptr) \
4090 (FIRST_STRING_P (ptr) \
4091 ? ((regoff_t) ((ptr) - string1)) \
4092 : ((regoff_t) ((ptr) - string2 + size1)))
4094 /* Macros for dealing with the split strings in re_match_2. */
4096 #define MATCHING_IN_FIRST_STRING (dend == end_match_1)
4098 /* Call before fetching a character with *d. This switches over to
4099 string2 if necessary. */
4100 #define PREFETCH() \
4103 /* End of string2 => fail. */ \
4104 if (dend == end_match_2) \
4106 /* End of string1 => advance to string2. */ \
4108 dend = end_match_2; \
4112 /* Test if at very beginning or at very end of the virtual concatenation
4113 of `string1' and `string2'. If only one string, it's `string2'. */
4114 #define AT_STRINGS_BEG(d) ((d) == (size1 ? string1 : string2) || !size2)
4115 #define AT_STRINGS_END(d) ((d) == end2)
4118 /* Test if D points to a character which is word-constituent. We have
4119 two special cases to check for: if past the end of string1, look at
4120 the first character in string2; and if before the beginning of
4121 string2, look at the last character in string1. */
4122 #define WORDCHAR_P(d) \
4123 (SYNTAX ((d) == end1 ? *string2 \
4124 : (d) == string2 - 1 ? *(end1 - 1) : *(d)) \
4127 /* Disabled due to a compiler bug -- see comment at case wordbound */
4129 /* Test if the character before D and the one at D differ with respect
4130 to being word-constituent. */
4131 #define AT_WORD_BOUNDARY(d) \
4132 (AT_STRINGS_BEG (d) || AT_STRINGS_END (d) \
4133 || WORDCHAR_P (d - 1) != WORDCHAR_P (d))
4136 /* Free everything we malloc. */
4137 #ifdef MATCH_MAY_ALLOCATE
4138 # define FREE_VAR(var) if (var) REGEX_FREE (var); var = NULL
4139 # define FREE_VARIABLES() \
4141 REGEX_FREE_STACK (fail_stack.stack); \
4142 FREE_VAR (regstart); \
4143 FREE_VAR (regend); \
4144 FREE_VAR (old_regstart); \
4145 FREE_VAR (old_regend); \
4146 FREE_VAR (best_regstart); \
4147 FREE_VAR (best_regend); \
4148 FREE_VAR (reg_info); \
4149 FREE_VAR (reg_dummy); \
4150 FREE_VAR (reg_info_dummy); \
4153 # define FREE_VARIABLES() ((void)0) /* Do nothing! But inhibit gcc warning. */
4154 #endif /* not MATCH_MAY_ALLOCATE */
4156 /* These values must meet several constraints. They must not be valid
4157 register values; since we have a limit of 255 registers (because
4158 we use only one byte in the pattern for the register number), we can
4159 use numbers larger than 255. They must differ by 1, because of
4160 NUM_FAILURE_ITEMS above. And the value for the lowest register must
4161 be larger than the value for the highest register, so we do not try
4162 to actually save any registers when none are active. */
4163 #define NO_HIGHEST_ACTIVE_REG (1 << BYTEWIDTH)
4164 #define NO_LOWEST_ACTIVE_REG (NO_HIGHEST_ACTIVE_REG + 1)
4166 /* Matching routines. */
4168 #ifndef emacs /* Emacs never uses this. */
4169 /* re_match is like re_match_2 except it takes only a single string. */
4172 re_match (bufp, string, size, pos, regs)
4173 struct re_pattern_buffer *bufp;
4176 struct re_registers *regs;
4178 int result = re_match_2_internal (bufp, NULL, 0, string, size,
4180 # ifndef REGEX_MALLOC
4188 weak_alias (__re_match, re_match)
4190 #endif /* not emacs */
4192 static boolean group_match_null_string_p _RE_ARGS ((unsigned char **p,
4194 register_info_type *reg_info));
4195 static boolean alt_match_null_string_p _RE_ARGS ((unsigned char *p,
4197 register_info_type *reg_info));
4198 static boolean common_op_match_null_string_p _RE_ARGS ((unsigned char **p,
4200 register_info_type *reg_info));
4201 static int bcmp_translate _RE_ARGS ((const char *s1, const char *s2,
4202 int len, char *translate));
4204 /* re_match_2 matches the compiled pattern in BUFP against the
4205 the (virtual) concatenation of STRING1 and STRING2 (of length SIZE1
4206 and SIZE2, respectively). We start matching at POS, and stop
4209 If REGS is non-null and the `no_sub' field of BUFP is nonzero, we
4210 store offsets for the substring each group matched in REGS. See the
4211 documentation for exactly how many groups we fill.
4213 We return -1 if no match, -2 if an internal error (such as the
4214 failure stack overflowing). Otherwise, we return the length of the
4215 matched substring. */
4218 re_match_2 (bufp, string1, size1, string2, size2, pos, regs, stop)
4219 struct re_pattern_buffer *bufp;
4220 const char *string1, *string2;
4223 struct re_registers *regs;
4226 int result = re_match_2_internal (bufp, string1, size1, string2, size2,
4228 #ifndef REGEX_MALLOC
4236 weak_alias (__re_match_2, re_match_2)
4239 /* This is a separate function so that we can force an alloca cleanup
4242 re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
4243 struct re_pattern_buffer *bufp;
4244 const char *string1, *string2;
4247 struct re_registers *regs;
4250 /* General temporaries. */
4254 /* Just past the end of the corresponding string. */
4255 const char *end1, *end2;
4257 /* Pointers into string1 and string2, just past the last characters in
4258 each to consider matching. */
4259 const char *end_match_1, *end_match_2;
4261 /* Where we are in the data, and the end of the current string. */
4262 const char *d, *dend;
4264 /* Where we are in the pattern, and the end of the pattern. */
4265 unsigned char *p = bufp->buffer;
4266 register unsigned char *pend = p + bufp->used;
4268 /* Mark the opcode just after a start_memory, so we can test for an
4269 empty subpattern when we get to the stop_memory. */
4270 unsigned char *just_past_start_mem = 0;
4272 /* We use this to map every character in the string. */
4273 RE_TRANSLATE_TYPE translate = bufp->translate;
4275 /* Failure point stack. Each place that can handle a failure further
4276 down the line pushes a failure point on this stack. It consists of
4277 restart, regend, and reg_info for all registers corresponding to
4278 the subexpressions we're currently inside, plus the number of such
4279 registers, and, finally, two char *'s. The first char * is where
4280 to resume scanning the pattern; the second one is where to resume
4281 scanning the strings. If the latter is zero, the failure point is
4282 a ``dummy''; if a failure happens and the failure point is a dummy,
4283 it gets discarded and the next next one is tried. */
4284 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
4285 fail_stack_type fail_stack;
4288 static unsigned failure_id;
4289 unsigned nfailure_points_pushed = 0, nfailure_points_popped = 0;
4293 /* This holds the pointer to the failure stack, when
4294 it is allocated relocatably. */
4295 fail_stack_elt_t *failure_stack_ptr;
4298 /* We fill all the registers internally, independent of what we
4299 return, for use in backreferences. The number here includes
4300 an element for register zero. */
4301 size_t num_regs = bufp->re_nsub + 1;
4303 /* The currently active registers. */
4304 active_reg_t lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4305 active_reg_t highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4307 /* Information on the contents of registers. These are pointers into
4308 the input strings; they record just what was matched (on this
4309 attempt) by a subexpression part of the pattern, that is, the
4310 regnum-th regstart pointer points to where in the pattern we began
4311 matching and the regnum-th regend points to right after where we
4312 stopped matching the regnum-th subexpression. (The zeroth register
4313 keeps track of what the whole pattern matches.) */
4314 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4315 const char **regstart, **regend;
4318 /* If a group that's operated upon by a repetition operator fails to
4319 match anything, then the register for its start will need to be
4320 restored because it will have been set to wherever in the string we
4321 are when we last see its open-group operator. Similarly for a
4323 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4324 const char **old_regstart, **old_regend;
4327 /* The is_active field of reg_info helps us keep track of which (possibly
4328 nested) subexpressions we are currently in. The matched_something
4329 field of reg_info[reg_num] helps us tell whether or not we have
4330 matched any of the pattern so far this time through the reg_num-th
4331 subexpression. These two fields get reset each time through any
4332 loop their register is in. */
4333 #ifdef MATCH_MAY_ALLOCATE /* otherwise, this is global. */
4334 register_info_type *reg_info;
4337 /* The following record the register info as found in the above
4338 variables when we find a match better than any we've seen before.
4339 This happens as we backtrack through the failure points, which in
4340 turn happens only if we have not yet matched the entire string. */
4341 unsigned best_regs_set = false;
4342 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4343 const char **best_regstart, **best_regend;
4346 /* Logically, this is `best_regend[0]'. But we don't want to have to
4347 allocate space for that if we're not allocating space for anything
4348 else (see below). Also, we never need info about register 0 for
4349 any of the other register vectors, and it seems rather a kludge to
4350 treat `best_regend' differently than the rest. So we keep track of
4351 the end of the best match so far in a separate variable. We
4352 initialize this to NULL so that when we backtrack the first time
4353 and need to test it, it's not garbage. */
4354 const char *match_end = NULL;
4356 /* This helps SET_REGS_MATCHED avoid doing redundant work. */
4357 int set_regs_matched_done = 0;
4359 /* Used when we pop values we don't care about. */
4360 #ifdef MATCH_MAY_ALLOCATE /* otherwise, these are global. */
4361 const char **reg_dummy;
4362 register_info_type *reg_info_dummy;
4366 /* Counts the total number of registers pushed. */
4367 unsigned num_regs_pushed = 0;
4370 DEBUG_PRINT1 ("\n\nEntering re_match_2.\n");
4374 #ifdef MATCH_MAY_ALLOCATE
4375 /* Do not bother to initialize all the register variables if there are
4376 no groups in the pattern, as it takes a fair amount of time. If
4377 there are groups, we include space for register 0 (the whole
4378 pattern), even though we never use it, since it simplifies the
4379 array indexing. We should fix this. */
4382 regstart = REGEX_TALLOC (num_regs, const char *);
4383 regend = REGEX_TALLOC (num_regs, const char *);
4384 old_regstart = REGEX_TALLOC (num_regs, const char *);
4385 old_regend = REGEX_TALLOC (num_regs, const char *);
4386 best_regstart = REGEX_TALLOC (num_regs, const char *);
4387 best_regend = REGEX_TALLOC (num_regs, const char *);
4388 reg_info = REGEX_TALLOC (num_regs, register_info_type);
4389 reg_dummy = REGEX_TALLOC (num_regs, const char *);
4390 reg_info_dummy = REGEX_TALLOC (num_regs, register_info_type);
4392 if (!(regstart && regend && old_regstart && old_regend && reg_info
4393 && best_regstart && best_regend && reg_dummy && reg_info_dummy))
4401 /* We must initialize all our variables to NULL, so that
4402 `FREE_VARIABLES' doesn't try to free them. */
4403 regstart = regend = old_regstart = old_regend = best_regstart
4404 = best_regend = reg_dummy = NULL;
4405 reg_info = reg_info_dummy = (register_info_type *) NULL;
4407 #endif /* MATCH_MAY_ALLOCATE */
4409 /* The starting position is bogus. */
4410 if (pos < 0 || pos > size1 + size2)
4416 /* Initialize subexpression text positions to -1 to mark ones that no
4417 start_memory/stop_memory has been seen for. Also initialize the
4418 register information struct. */
4419 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4421 regstart[mcnt] = regend[mcnt]
4422 = old_regstart[mcnt] = old_regend[mcnt] = REG_UNSET_VALUE;
4424 REG_MATCH_NULL_STRING_P (reg_info[mcnt]) = MATCH_NULL_UNSET_VALUE;
4425 IS_ACTIVE (reg_info[mcnt]) = 0;
4426 MATCHED_SOMETHING (reg_info[mcnt]) = 0;
4427 EVER_MATCHED_SOMETHING (reg_info[mcnt]) = 0;
4430 /* We move `string1' into `string2' if the latter's empty -- but not if
4431 `string1' is null. */
4432 if (size2 == 0 && string1 != NULL)
4439 end1 = string1 + size1;
4440 end2 = string2 + size2;
4442 /* Compute where to stop matching, within the two strings. */
4445 end_match_1 = string1 + stop;
4446 end_match_2 = string2;
4451 end_match_2 = string2 + stop - size1;
4454 /* `p' scans through the pattern as `d' scans through the data.
4455 `dend' is the end of the input string that `d' points within. `d'
4456 is advanced into the following input string whenever necessary, but
4457 this happens before fetching; therefore, at the beginning of the
4458 loop, `d' can be pointing at the end of a string, but it cannot
4460 if (size1 > 0 && pos <= size1)
4467 d = string2 + pos - size1;
4471 DEBUG_PRINT1 ("The compiled pattern is:\n");
4472 DEBUG_PRINT_COMPILED_PATTERN (bufp, p, pend);
4473 DEBUG_PRINT1 ("The string to match is: `");
4474 DEBUG_PRINT_DOUBLE_STRING (d, string1, size1, string2, size2);
4475 DEBUG_PRINT1 ("'\n");
4477 /* This loops over pattern commands. It exits by returning from the
4478 function if the match is complete, or it drops through if the match
4479 fails at this starting point in the input data. */
4483 DEBUG_PRINT2 ("\n%p: ", p);
4485 DEBUG_PRINT2 ("\n0x%x: ", p);
4489 { /* End of pattern means we might have succeeded. */
4490 DEBUG_PRINT1 ("end of pattern ... ");
4492 /* If we haven't matched the entire string, and we want the
4493 longest match, try backtracking. */
4494 if (d != end_match_2)
4496 /* 1 if this match ends in the same string (string1 or string2)
4497 as the best previous match. */
4498 boolean same_str_p = (FIRST_STRING_P (match_end)
4499 == MATCHING_IN_FIRST_STRING);
4500 /* 1 if this match is the best seen so far. */
4501 boolean best_match_p;
4503 /* AIX compiler got confused when this was combined
4504 with the previous declaration. */
4506 best_match_p = d > match_end;
4508 best_match_p = !MATCHING_IN_FIRST_STRING;
4510 DEBUG_PRINT1 ("backtracking.\n");
4512 if (!FAIL_STACK_EMPTY ())
4513 { /* More failure points to try. */
4515 /* If exceeds best match so far, save it. */
4516 if (!best_regs_set || best_match_p)
4518 best_regs_set = true;
4521 DEBUG_PRINT1 ("\nSAVING match as best so far.\n");
4523 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4525 best_regstart[mcnt] = regstart[mcnt];
4526 best_regend[mcnt] = regend[mcnt];
4532 /* If no failure points, don't restore garbage. And if
4533 last match is real best match, don't restore second
4535 else if (best_regs_set && !best_match_p)
4538 /* Restore best match. It may happen that `dend ==
4539 end_match_1' while the restored d is in string2.
4540 For example, the pattern `x.*y.*z' against the
4541 strings `x-' and `y-z-', if the two strings are
4542 not consecutive in memory. */
4543 DEBUG_PRINT1 ("Restoring best registers.\n");
4546 dend = ((d >= string1 && d <= end1)
4547 ? end_match_1 : end_match_2);
4549 for (mcnt = 1; (unsigned) mcnt < num_regs; mcnt++)
4551 regstart[mcnt] = best_regstart[mcnt];
4552 regend[mcnt] = best_regend[mcnt];
4555 } /* d != end_match_2 */
4558 DEBUG_PRINT1 ("Accepting match.\n");
4560 /* If caller wants register contents data back, do it. */
4561 if (regs && !bufp->no_sub)
4563 /* Have the register data arrays been allocated? */
4564 if (bufp->regs_allocated == REGS_UNALLOCATED)
4565 { /* No. So allocate them with malloc. We need one
4566 extra element beyond `num_regs' for the `-1' marker
4568 regs->num_regs = MAX (RE_NREGS, num_regs + 1);
4569 regs->start = TALLOC (regs->num_regs, regoff_t);
4570 regs->end = TALLOC (regs->num_regs, regoff_t);
4571 if (regs->start == NULL || regs->end == NULL)
4576 bufp->regs_allocated = REGS_REALLOCATE;
4578 else if (bufp->regs_allocated == REGS_REALLOCATE)
4579 { /* Yes. If we need more elements than were already
4580 allocated, reallocate them. If we need fewer, just
4582 if (regs->num_regs < num_regs + 1)
4584 regs->num_regs = num_regs + 1;
4585 RETALLOC (regs->start, regs->num_regs, regoff_t);
4586 RETALLOC (regs->end, regs->num_regs, regoff_t);
4587 if (regs->start == NULL || regs->end == NULL)
4596 /* These braces fend off a "empty body in an else-statement"
4597 warning under GCC when assert expands to nothing. */
4598 assert (bufp->regs_allocated == REGS_FIXED);
4601 /* Convert the pointer data in `regstart' and `regend' to
4602 indices. Register zero has to be set differently,
4603 since we haven't kept track of any info for it. */
4604 if (regs->num_regs > 0)
4606 regs->start[0] = pos;
4607 regs->end[0] = (MATCHING_IN_FIRST_STRING
4608 ? ((regoff_t) (d - string1))
4609 : ((regoff_t) (d - string2 + size1)));
4612 /* Go through the first `min (num_regs, regs->num_regs)'
4613 registers, since that is all we initialized. */
4614 for (mcnt = 1; (unsigned) mcnt < MIN (num_regs, regs->num_regs);
4617 if (REG_UNSET (regstart[mcnt]) || REG_UNSET (regend[mcnt]))
4618 regs->start[mcnt] = regs->end[mcnt] = -1;
4622 = (regoff_t) POINTER_TO_OFFSET (regstart[mcnt]);
4624 = (regoff_t) POINTER_TO_OFFSET (regend[mcnt]);
4628 /* If the regs structure we return has more elements than
4629 were in the pattern, set the extra elements to -1. If
4630 we (re)allocated the registers, this is the case,
4631 because we always allocate enough to have at least one
4633 for (mcnt = num_regs; (unsigned) mcnt < regs->num_regs; mcnt++)
4634 regs->start[mcnt] = regs->end[mcnt] = -1;
4635 } /* regs && !bufp->no_sub */
4637 DEBUG_PRINT4 ("%u failure points pushed, %u popped (%u remain).\n",
4638 nfailure_points_pushed, nfailure_points_popped,
4639 nfailure_points_pushed - nfailure_points_popped);
4640 DEBUG_PRINT2 ("%u registers pushed.\n", num_regs_pushed);
4642 mcnt = d - pos - (MATCHING_IN_FIRST_STRING
4646 DEBUG_PRINT2 ("Returning %d from re_match_2.\n", mcnt);
4652 /* Otherwise match next pattern command. */
4653 switch (SWITCH_ENUM_CAST ((re_opcode_t) *p++))
4655 /* Ignore these. Used to ignore the n of succeed_n's which
4656 currently have n == 0. */
4658 DEBUG_PRINT1 ("EXECUTING no_op.\n");
4662 DEBUG_PRINT1 ("EXECUTING succeed.\n");
4665 /* Match the next n pattern characters exactly. The following
4666 byte in the pattern defines n, and the n bytes after that
4667 are the characters to match. */
4670 DEBUG_PRINT2 ("EXECUTING exactn %d.\n", mcnt);
4672 /* This is written out as an if-else so we don't waste time
4673 testing `translate' inside the loop. */
4679 if ((unsigned char) translate[(unsigned char) *d++]
4680 != (unsigned char) *p++)
4690 if (*d++ != (char) *p++) goto fail;
4694 SET_REGS_MATCHED ();
4698 /* Match any character except possibly a newline or a null. */
4700 DEBUG_PRINT1 ("EXECUTING anychar.\n");
4704 if ((!(bufp->syntax & RE_DOT_NEWLINE) && TRANSLATE (*d) == '\n')
4705 || (bufp->syntax & RE_DOT_NOT_NULL && TRANSLATE (*d) == '\000'))
4708 SET_REGS_MATCHED ();
4709 DEBUG_PRINT2 (" Matched `%d'.\n", *d);
4717 register unsigned char c;
4718 boolean not = (re_opcode_t) *(p - 1) == charset_not;
4720 DEBUG_PRINT2 ("EXECUTING charset%s.\n", not ? "_not" : "");
4723 c = TRANSLATE (*d); /* The character to match. */
4725 /* Cast to `unsigned' instead of `unsigned char' in case the
4726 bit list is a full 32 bytes long. */
4727 if (c < (unsigned) (*p * BYTEWIDTH)
4728 && p[1 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
4733 if (!not) goto fail;
4735 SET_REGS_MATCHED ();
4741 /* The beginning of a group is represented by start_memory.
4742 The arguments are the register number in the next byte, and the
4743 number of groups inner to this one in the next. The text
4744 matched within the group is recorded (in the internal
4745 registers data structure) under the register number. */
4747 DEBUG_PRINT3 ("EXECUTING start_memory %d (%d):\n", *p, p[1]);
4749 /* Find out if this group can match the empty string. */
4750 p1 = p; /* To send to group_match_null_string_p. */
4752 if (REG_MATCH_NULL_STRING_P (reg_info[*p]) == MATCH_NULL_UNSET_VALUE)
4753 REG_MATCH_NULL_STRING_P (reg_info[*p])
4754 = group_match_null_string_p (&p1, pend, reg_info);
4756 /* Save the position in the string where we were the last time
4757 we were at this open-group operator in case the group is
4758 operated upon by a repetition operator, e.g., with `(a*)*b'
4759 against `ab'; then we want to ignore where we are now in
4760 the string in case this attempt to match fails. */
4761 old_regstart[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4762 ? REG_UNSET (regstart[*p]) ? d : regstart[*p]
4764 DEBUG_PRINT2 (" old_regstart: %d\n",
4765 POINTER_TO_OFFSET (old_regstart[*p]));
4768 DEBUG_PRINT2 (" regstart: %d\n", POINTER_TO_OFFSET (regstart[*p]));
4770 IS_ACTIVE (reg_info[*p]) = 1;
4771 MATCHED_SOMETHING (reg_info[*p]) = 0;
4773 /* Clear this whenever we change the register activity status. */
4774 set_regs_matched_done = 0;
4776 /* This is the new highest active register. */
4777 highest_active_reg = *p;
4779 /* If nothing was active before, this is the new lowest active
4781 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
4782 lowest_active_reg = *p;
4784 /* Move past the register number and inner group count. */
4786 just_past_start_mem = p;
4791 /* The stop_memory opcode represents the end of a group. Its
4792 arguments are the same as start_memory's: the register
4793 number, and the number of inner groups. */
4795 DEBUG_PRINT3 ("EXECUTING stop_memory %d (%d):\n", *p, p[1]);
4797 /* We need to save the string position the last time we were at
4798 this close-group operator in case the group is operated
4799 upon by a repetition operator, e.g., with `((a*)*(b*)*)*'
4800 against `aba'; then we want to ignore where we are now in
4801 the string in case this attempt to match fails. */
4802 old_regend[*p] = REG_MATCH_NULL_STRING_P (reg_info[*p])
4803 ? REG_UNSET (regend[*p]) ? d : regend[*p]
4805 DEBUG_PRINT2 (" old_regend: %d\n",
4806 POINTER_TO_OFFSET (old_regend[*p]));
4809 DEBUG_PRINT2 (" regend: %d\n", POINTER_TO_OFFSET (regend[*p]));
4811 /* This register isn't active anymore. */
4812 IS_ACTIVE (reg_info[*p]) = 0;
4814 /* Clear this whenever we change the register activity status. */
4815 set_regs_matched_done = 0;
4817 /* If this was the only register active, nothing is active
4819 if (lowest_active_reg == highest_active_reg)
4821 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4822 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4825 { /* We must scan for the new highest active register, since
4826 it isn't necessarily one less than now: consider
4827 (a(b)c(d(e)f)g). When group 3 ends, after the f), the
4828 new highest active register is 1. */
4829 unsigned char r = *p - 1;
4830 while (r > 0 && !IS_ACTIVE (reg_info[r]))
4833 /* If we end up at register zero, that means that we saved
4834 the registers as the result of an `on_failure_jump', not
4835 a `start_memory', and we jumped to past the innermost
4836 `stop_memory'. For example, in ((.)*) we save
4837 registers 1 and 2 as a result of the *, but when we pop
4838 back to the second ), we are at the stop_memory 1.
4839 Thus, nothing is active. */
4842 lowest_active_reg = NO_LOWEST_ACTIVE_REG;
4843 highest_active_reg = NO_HIGHEST_ACTIVE_REG;
4846 highest_active_reg = r;
4849 /* If just failed to match something this time around with a
4850 group that's operated on by a repetition operator, try to
4851 force exit from the ``loop'', and restore the register
4852 information for this group that we had before trying this
4854 if ((!MATCHED_SOMETHING (reg_info[*p])
4855 || just_past_start_mem == p - 1)
4858 boolean is_a_jump_n = false;
4862 switch ((re_opcode_t) *p1++)
4866 case pop_failure_jump:
4867 case maybe_pop_jump:
4869 case dummy_failure_jump:
4870 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4880 /* If the next operation is a jump backwards in the pattern
4881 to an on_failure_jump right before the start_memory
4882 corresponding to this stop_memory, exit from the loop
4883 by forcing a failure after pushing on the stack the
4884 on_failure_jump's jump in the pattern, and d. */
4885 if (mcnt < 0 && (re_opcode_t) *p1 == on_failure_jump
4886 && (re_opcode_t) p1[3] == start_memory && p1[4] == *p)
4888 /* If this group ever matched anything, then restore
4889 what its registers were before trying this last
4890 failed match, e.g., with `(a*)*b' against `ab' for
4891 regstart[1], and, e.g., with `((a*)*(b*)*)*'
4892 against `aba' for regend[3].
4894 Also restore the registers for inner groups for,
4895 e.g., `((a*)(b*))*' against `aba' (register 3 would
4896 otherwise get trashed). */
4898 if (EVER_MATCHED_SOMETHING (reg_info[*p]))
4902 EVER_MATCHED_SOMETHING (reg_info[*p]) = 0;
4904 /* Restore this and inner groups' (if any) registers. */
4905 for (r = *p; r < (unsigned) *p + (unsigned) *(p + 1);
4908 regstart[r] = old_regstart[r];
4910 /* xx why this test? */
4911 if (old_regend[r] >= regstart[r])
4912 regend[r] = old_regend[r];
4916 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
4917 PUSH_FAILURE_POINT (p1 + mcnt, d, -2);
4923 /* Move past the register number and the inner group count. */
4928 /* \<digit> has been turned into a `duplicate' command which is
4929 followed by the numeric value of <digit> as the register number. */
4932 register const char *d2, *dend2;
4933 int regno = *p++; /* Get which register to match against. */
4934 DEBUG_PRINT2 ("EXECUTING duplicate %d.\n", regno);
4936 /* Can't back reference a group which we've never matched. */
4937 if (REG_UNSET (regstart[regno]) || REG_UNSET (regend[regno]))
4940 /* Where in input to try to start matching. */
4941 d2 = regstart[regno];
4943 /* Where to stop matching; if both the place to start and
4944 the place to stop matching are in the same string, then
4945 set to the place to stop, otherwise, for now have to use
4946 the end of the first string. */
4948 dend2 = ((FIRST_STRING_P (regstart[regno])
4949 == FIRST_STRING_P (regend[regno]))
4950 ? regend[regno] : end_match_1);
4953 /* If necessary, advance to next segment in register
4957 if (dend2 == end_match_2) break;
4958 if (dend2 == regend[regno]) break;
4960 /* End of string1 => advance to string2. */
4962 dend2 = regend[regno];
4964 /* At end of register contents => success */
4965 if (d2 == dend2) break;
4967 /* If necessary, advance to next segment in data. */
4970 /* How many characters left in this segment to match. */
4973 /* Want how many consecutive characters we can match in
4974 one shot, so, if necessary, adjust the count. */
4975 if (mcnt > dend2 - d2)
4978 /* Compare that many; failure if mismatch, else move
4981 ? bcmp_translate (d, d2, mcnt, translate)
4982 : memcmp (d, d2, mcnt))
4984 d += mcnt, d2 += mcnt;
4986 /* Do this because we've match some characters. */
4987 SET_REGS_MATCHED ();
4993 /* begline matches the empty string at the beginning of the string
4994 (unless `not_bol' is set in `bufp'), and, if
4995 `newline_anchor' is set, after newlines. */
4997 DEBUG_PRINT1 ("EXECUTING begline.\n");
4999 if (AT_STRINGS_BEG (d))
5001 if (!bufp->not_bol) break;
5003 else if (d[-1] == '\n' && bufp->newline_anchor)
5007 /* In all other cases, we fail. */
5011 /* endline is the dual of begline. */
5013 DEBUG_PRINT1 ("EXECUTING endline.\n");
5015 if (AT_STRINGS_END (d))
5017 if (!bufp->not_eol) break;
5020 /* We have to ``prefetch'' the next character. */
5021 else if ((d == end1 ? *string2 : *d) == '\n'
5022 && bufp->newline_anchor)
5029 /* Match at the very beginning of the data. */
5031 DEBUG_PRINT1 ("EXECUTING begbuf.\n");
5032 if (AT_STRINGS_BEG (d))
5037 /* Match at the very end of the data. */
5039 DEBUG_PRINT1 ("EXECUTING endbuf.\n");
5040 if (AT_STRINGS_END (d))
5045 /* on_failure_keep_string_jump is used to optimize `.*\n'. It
5046 pushes NULL as the value for the string on the stack. Then
5047 `pop_failure_point' will keep the current value for the
5048 string, instead of restoring it. To see why, consider
5049 matching `foo\nbar' against `.*\n'. The .* matches the foo;
5050 then the . fails against the \n. But the next thing we want
5051 to do is match the \n against the \n; if we restored the
5052 string value, we would be back at the foo.
5054 Because this is used only in specific cases, we don't need to
5055 check all the things that `on_failure_jump' does, to make
5056 sure the right things get saved on the stack. Hence we don't
5057 share its code. The only reason to push anything on the
5058 stack at all is that otherwise we would have to change
5059 `anychar's code to do something besides goto fail in this
5060 case; that seems worse than this. */
5061 case on_failure_keep_string_jump:
5062 DEBUG_PRINT1 ("EXECUTING on_failure_keep_string_jump");
5064 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5066 DEBUG_PRINT3 (" %d (to %p):\n", mcnt, p + mcnt);
5068 DEBUG_PRINT3 (" %d (to 0x%x):\n", mcnt, p + mcnt);
5071 PUSH_FAILURE_POINT (p + mcnt, NULL, -2);
5075 /* Uses of on_failure_jump:
5077 Each alternative starts with an on_failure_jump that points
5078 to the beginning of the next alternative. Each alternative
5079 except the last ends with a jump that in effect jumps past
5080 the rest of the alternatives. (They really jump to the
5081 ending jump of the following alternative, because tensioning
5082 these jumps is a hassle.)
5084 Repeats start with an on_failure_jump that points past both
5085 the repetition text and either the following jump or
5086 pop_failure_jump back to this on_failure_jump. */
5087 case on_failure_jump:
5089 DEBUG_PRINT1 ("EXECUTING on_failure_jump");
5091 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5093 DEBUG_PRINT3 (" %d (to %p)", mcnt, p + mcnt);
5095 DEBUG_PRINT3 (" %d (to 0x%x)", mcnt, p + mcnt);
5098 /* If this on_failure_jump comes right before a group (i.e.,
5099 the original * applied to a group), save the information
5100 for that group and all inner ones, so that if we fail back
5101 to this point, the group's information will be correct.
5102 For example, in \(a*\)*\1, we need the preceding group,
5103 and in \(zz\(a*\)b*\)\2, we need the inner group. */
5105 /* We can't use `p' to check ahead because we push
5106 a failure point to `p + mcnt' after we do this. */
5109 /* We need to skip no_op's before we look for the
5110 start_memory in case this on_failure_jump is happening as
5111 the result of a completed succeed_n, as in \(a\)\{1,3\}b\1
5113 while (p1 < pend && (re_opcode_t) *p1 == no_op)
5116 if (p1 < pend && (re_opcode_t) *p1 == start_memory)
5118 /* We have a new highest active register now. This will
5119 get reset at the start_memory we are about to get to,
5120 but we will have saved all the registers relevant to
5121 this repetition op, as described above. */
5122 highest_active_reg = *(p1 + 1) + *(p1 + 2);
5123 if (lowest_active_reg == NO_LOWEST_ACTIVE_REG)
5124 lowest_active_reg = *(p1 + 1);
5127 DEBUG_PRINT1 (":\n");
5128 PUSH_FAILURE_POINT (p + mcnt, d, -2);
5132 /* A smart repeat ends with `maybe_pop_jump'.
5133 We change it to either `pop_failure_jump' or `jump'. */
5134 case maybe_pop_jump:
5135 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5136 DEBUG_PRINT2 ("EXECUTING maybe_pop_jump %d.\n", mcnt);
5138 register unsigned char *p2 = p;
5140 /* Compare the beginning of the repeat with what in the
5141 pattern follows its end. If we can establish that there
5142 is nothing that they would both match, i.e., that we
5143 would have to backtrack because of (as in, e.g., `a*a')
5144 then we can change to pop_failure_jump, because we'll
5145 never have to backtrack.
5147 This is not true in the case of alternatives: in
5148 `(a|ab)*' we do need to backtrack to the `ab' alternative
5149 (e.g., if the string was `ab'). But instead of trying to
5150 detect that here, the alternative has put on a dummy
5151 failure point which is what we will end up popping. */
5153 /* Skip over open/close-group commands.
5154 If what follows this loop is a ...+ construct,
5155 look at what begins its body, since we will have to
5156 match at least one of that. */
5160 && ((re_opcode_t) *p2 == stop_memory
5161 || (re_opcode_t) *p2 == start_memory))
5163 else if (p2 + 6 < pend
5164 && (re_opcode_t) *p2 == dummy_failure_jump)
5171 /* p1[0] ... p1[2] are the `on_failure_jump' corresponding
5172 to the `maybe_finalize_jump' of this case. Examine what
5175 /* If we're at the end of the pattern, we can change. */
5178 /* Consider what happens when matching ":\(.*\)"
5179 against ":/". I don't really understand this code
5181 p[-3] = (unsigned char) pop_failure_jump;
5183 (" End of pattern: change to `pop_failure_jump'.\n");
5186 else if ((re_opcode_t) *p2 == exactn
5187 || (bufp->newline_anchor && (re_opcode_t) *p2 == endline))
5189 register unsigned char c
5190 = *p2 == (unsigned char) endline ? '\n' : p2[2];
5192 if ((re_opcode_t) p1[3] == exactn && p1[5] != c)
5194 p[-3] = (unsigned char) pop_failure_jump;
5195 DEBUG_PRINT3 (" %c != %c => pop_failure_jump.\n",
5199 else if ((re_opcode_t) p1[3] == charset
5200 || (re_opcode_t) p1[3] == charset_not)
5202 int not = (re_opcode_t) p1[3] == charset_not;
5204 if (c < (unsigned char) (p1[4] * BYTEWIDTH)
5205 && p1[5 + c / BYTEWIDTH] & (1 << (c % BYTEWIDTH)))
5208 /* `not' is equal to 1 if c would match, which means
5209 that we can't change to pop_failure_jump. */
5212 p[-3] = (unsigned char) pop_failure_jump;
5213 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5217 else if ((re_opcode_t) *p2 == charset)
5219 /* We win if the first character of the loop is not part
5221 if ((re_opcode_t) p1[3] == exactn
5222 && ! ((int) p2[1] * BYTEWIDTH > (int) p1[5]
5223 && (p2[2 + p1[5] / BYTEWIDTH]
5224 & (1 << (p1[5] % BYTEWIDTH)))))
5226 p[-3] = (unsigned char) pop_failure_jump;
5227 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5230 else if ((re_opcode_t) p1[3] == charset_not)
5233 /* We win if the charset_not inside the loop
5234 lists every character listed in the charset after. */
5235 for (idx = 0; idx < (int) p2[1]; idx++)
5236 if (! (p2[2 + idx] == 0
5237 || (idx < (int) p1[4]
5238 && ((p2[2 + idx] & ~ p1[5 + idx]) == 0))))
5243 p[-3] = (unsigned char) pop_failure_jump;
5244 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5247 else if ((re_opcode_t) p1[3] == charset)
5250 /* We win if the charset inside the loop
5251 has no overlap with the one after the loop. */
5253 idx < (int) p2[1] && idx < (int) p1[4];
5255 if ((p2[2 + idx] & p1[5 + idx]) != 0)
5258 if (idx == p2[1] || idx == p1[4])
5260 p[-3] = (unsigned char) pop_failure_jump;
5261 DEBUG_PRINT1 (" No match => pop_failure_jump.\n");
5266 p -= 2; /* Point at relative address again. */
5267 if ((re_opcode_t) p[-1] != pop_failure_jump)
5269 p[-1] = (unsigned char) jump;
5270 DEBUG_PRINT1 (" Match => jump.\n");
5271 goto unconditional_jump;
5273 /* Note fall through. */
5276 /* The end of a simple repeat has a pop_failure_jump back to
5277 its matching on_failure_jump, where the latter will push a
5278 failure point. The pop_failure_jump takes off failure
5279 points put on by this pop_failure_jump's matching
5280 on_failure_jump; we got through the pattern to here from the
5281 matching on_failure_jump, so didn't fail. */
5282 case pop_failure_jump:
5284 /* We need to pass separate storage for the lowest and
5285 highest registers, even though we don't care about the
5286 actual values. Otherwise, we will restore only one
5287 register from the stack, since lowest will == highest in
5288 `pop_failure_point'. */
5289 active_reg_t dummy_low_reg, dummy_high_reg;
5290 unsigned char *pdummy;
5293 DEBUG_PRINT1 ("EXECUTING pop_failure_jump.\n");
5294 POP_FAILURE_POINT (sdummy, pdummy,
5295 dummy_low_reg, dummy_high_reg,
5296 reg_dummy, reg_dummy, reg_info_dummy);
5298 /* Note fall through. */
5302 DEBUG_PRINT2 ("\n%p: ", p);
5304 DEBUG_PRINT2 ("\n0x%x: ", p);
5306 /* Note fall through. */
5308 /* Unconditionally jump (without popping any failure points). */
5310 EXTRACT_NUMBER_AND_INCR (mcnt, p); /* Get the amount to jump. */
5311 DEBUG_PRINT2 ("EXECUTING jump %d ", mcnt);
5312 p += mcnt; /* Do the jump. */
5314 DEBUG_PRINT2 ("(to %p).\n", p);
5316 DEBUG_PRINT2 ("(to 0x%x).\n", p);
5321 /* We need this opcode so we can detect where alternatives end
5322 in `group_match_null_string_p' et al. */
5324 DEBUG_PRINT1 ("EXECUTING jump_past_alt.\n");
5325 goto unconditional_jump;
5328 /* Normally, the on_failure_jump pushes a failure point, which
5329 then gets popped at pop_failure_jump. We will end up at
5330 pop_failure_jump, also, and with a pattern of, say, `a+', we
5331 are skipping over the on_failure_jump, so we have to push
5332 something meaningless for pop_failure_jump to pop. */
5333 case dummy_failure_jump:
5334 DEBUG_PRINT1 ("EXECUTING dummy_failure_jump.\n");
5335 /* It doesn't matter what we push for the string here. What
5336 the code at `fail' tests is the value for the pattern. */
5337 PUSH_FAILURE_POINT (NULL, NULL, -2);
5338 goto unconditional_jump;
5341 /* At the end of an alternative, we need to push a dummy failure
5342 point in case we are followed by a `pop_failure_jump', because
5343 we don't want the failure point for the alternative to be
5344 popped. For example, matching `(a|ab)*' against `aab'
5345 requires that we match the `ab' alternative. */
5346 case push_dummy_failure:
5347 DEBUG_PRINT1 ("EXECUTING push_dummy_failure.\n");
5348 /* See comments just above at `dummy_failure_jump' about the
5350 PUSH_FAILURE_POINT (NULL, NULL, -2);
5353 /* Have to succeed matching what follows at least n times.
5354 After that, handle like `on_failure_jump'. */
5356 EXTRACT_NUMBER (mcnt, p + 2);
5357 DEBUG_PRINT2 ("EXECUTING succeed_n %d.\n", mcnt);
5360 /* Originally, this is how many times we HAVE to succeed. */
5365 STORE_NUMBER_AND_INCR (p, mcnt);
5367 DEBUG_PRINT3 (" Setting %p to %d.\n", p - 2, mcnt);
5369 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p - 2, mcnt);
5375 DEBUG_PRINT2 (" Setting two bytes from %p to no_op.\n", p+2);
5377 DEBUG_PRINT2 (" Setting two bytes from 0x%x to no_op.\n", p+2);
5379 p[2] = (unsigned char) no_op;
5380 p[3] = (unsigned char) no_op;
5386 EXTRACT_NUMBER (mcnt, p + 2);
5387 DEBUG_PRINT2 ("EXECUTING jump_n %d.\n", mcnt);
5389 /* Originally, this is how many times we CAN jump. */
5393 STORE_NUMBER (p + 2, mcnt);
5395 DEBUG_PRINT3 (" Setting %p to %d.\n", p + 2, mcnt);
5397 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p + 2, mcnt);
5399 goto unconditional_jump;
5401 /* If don't have to jump any more, skip over the rest of command. */
5408 DEBUG_PRINT1 ("EXECUTING set_number_at.\n");
5410 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5412 EXTRACT_NUMBER_AND_INCR (mcnt, p);
5414 DEBUG_PRINT3 (" Setting %p to %d.\n", p1, mcnt);
5416 DEBUG_PRINT3 (" Setting 0x%x to %d.\n", p1, mcnt);
5418 STORE_NUMBER (p1, mcnt);
5423 /* The DEC Alpha C compiler 3.x generates incorrect code for the
5424 test WORDCHAR_P (d - 1) != WORDCHAR_P (d) in the expansion of
5425 AT_WORD_BOUNDARY, so this code is disabled. Expanding the
5426 macro and introducing temporary variables works around the bug. */
5429 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
5430 if (AT_WORD_BOUNDARY (d))
5435 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5436 if (AT_WORD_BOUNDARY (d))
5442 boolean prevchar, thischar;
5444 DEBUG_PRINT1 ("EXECUTING wordbound.\n");
5445 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5448 prevchar = WORDCHAR_P (d - 1);
5449 thischar = WORDCHAR_P (d);
5450 if (prevchar != thischar)
5457 boolean prevchar, thischar;
5459 DEBUG_PRINT1 ("EXECUTING notwordbound.\n");
5460 if (AT_STRINGS_BEG (d) || AT_STRINGS_END (d))
5463 prevchar = WORDCHAR_P (d - 1);
5464 thischar = WORDCHAR_P (d);
5465 if (prevchar != thischar)
5472 DEBUG_PRINT1 ("EXECUTING wordbeg.\n");
5473 if (WORDCHAR_P (d) && (AT_STRINGS_BEG (d) || !WORDCHAR_P (d - 1)))
5478 DEBUG_PRINT1 ("EXECUTING wordend.\n");
5479 if (!AT_STRINGS_BEG (d) && WORDCHAR_P (d - 1)
5480 && (!WORDCHAR_P (d) || AT_STRINGS_END (d)))
5486 DEBUG_PRINT1 ("EXECUTING before_dot.\n");
5487 if (PTR_CHAR_POS ((unsigned char *) d) >= point)
5492 DEBUG_PRINT1 ("EXECUTING at_dot.\n");
5493 if (PTR_CHAR_POS ((unsigned char *) d) != point)
5498 DEBUG_PRINT1 ("EXECUTING after_dot.\n");
5499 if (PTR_CHAR_POS ((unsigned char *) d) <= point)
5504 DEBUG_PRINT2 ("EXECUTING syntaxspec %d.\n", mcnt);
5509 DEBUG_PRINT1 ("EXECUTING Emacs wordchar.\n");
5513 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5515 if (SYNTAX (d[-1]) != (enum syntaxcode) mcnt)
5517 SET_REGS_MATCHED ();
5521 DEBUG_PRINT2 ("EXECUTING notsyntaxspec %d.\n", mcnt);
5523 goto matchnotsyntax;
5526 DEBUG_PRINT1 ("EXECUTING Emacs notwordchar.\n");
5530 /* Can't use *d++ here; SYNTAX may be an unsafe macro. */
5532 if (SYNTAX (d[-1]) == (enum syntaxcode) mcnt)
5534 SET_REGS_MATCHED ();
5537 #else /* not emacs */
5539 DEBUG_PRINT1 ("EXECUTING non-Emacs wordchar.\n");
5541 if (!WORDCHAR_P (d))
5543 SET_REGS_MATCHED ();
5548 DEBUG_PRINT1 ("EXECUTING non-Emacs notwordchar.\n");
5552 SET_REGS_MATCHED ();
5555 #endif /* not emacs */
5560 continue; /* Successfully executed one pattern command; keep going. */
5563 /* We goto here if a matching operation fails. */
5565 if (!FAIL_STACK_EMPTY ())
5566 { /* A restart point is known. Restore to that state. */
5567 DEBUG_PRINT1 ("\nFAIL:\n");
5568 POP_FAILURE_POINT (d, p,
5569 lowest_active_reg, highest_active_reg,
5570 regstart, regend, reg_info);
5572 /* If this failure point is a dummy, try the next one. */
5576 /* If we failed to the end of the pattern, don't examine *p. */
5580 boolean is_a_jump_n = false;
5582 /* If failed to a backwards jump that's part of a repetition
5583 loop, need to pop this failure point and use the next one. */
5584 switch ((re_opcode_t) *p)
5588 case maybe_pop_jump:
5589 case pop_failure_jump:
5592 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5595 if ((is_a_jump_n && (re_opcode_t) *p1 == succeed_n)
5597 && (re_opcode_t) *p1 == on_failure_jump))
5605 if (d >= string1 && d <= end1)
5609 break; /* Matching at this starting point really fails. */
5613 goto restore_best_regs;
5617 return -1; /* Failure to match. */
5620 /* Subroutine definitions for re_match_2. */
5623 /* We are passed P pointing to a register number after a start_memory.
5625 Return true if the pattern up to the corresponding stop_memory can
5626 match the empty string, and false otherwise.
5628 If we find the matching stop_memory, sets P to point to one past its number.
5629 Otherwise, sets P to an undefined byte less than or equal to END.
5631 We don't handle duplicates properly (yet). */
5634 group_match_null_string_p (p, end, reg_info)
5635 unsigned char **p, *end;
5636 register_info_type *reg_info;
5639 /* Point to after the args to the start_memory. */
5640 unsigned char *p1 = *p + 2;
5644 /* Skip over opcodes that can match nothing, and return true or
5645 false, as appropriate, when we get to one that can't, or to the
5646 matching stop_memory. */
5648 switch ((re_opcode_t) *p1)
5650 /* Could be either a loop or a series of alternatives. */
5651 case on_failure_jump:
5653 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5655 /* If the next operation is not a jump backwards in the
5660 /* Go through the on_failure_jumps of the alternatives,
5661 seeing if any of the alternatives cannot match nothing.
5662 The last alternative starts with only a jump,
5663 whereas the rest start with on_failure_jump and end
5664 with a jump, e.g., here is the pattern for `a|b|c':
5666 /on_failure_jump/0/6/exactn/1/a/jump_past_alt/0/6
5667 /on_failure_jump/0/6/exactn/1/b/jump_past_alt/0/3
5670 So, we have to first go through the first (n-1)
5671 alternatives and then deal with the last one separately. */
5674 /* Deal with the first (n-1) alternatives, which start
5675 with an on_failure_jump (see above) that jumps to right
5676 past a jump_past_alt. */
5678 while ((re_opcode_t) p1[mcnt-3] == jump_past_alt)
5680 /* `mcnt' holds how many bytes long the alternative
5681 is, including the ending `jump_past_alt' and
5684 if (!alt_match_null_string_p (p1, p1 + mcnt - 3,
5688 /* Move to right after this alternative, including the
5692 /* Break if it's the beginning of an n-th alternative
5693 that doesn't begin with an on_failure_jump. */
5694 if ((re_opcode_t) *p1 != on_failure_jump)
5697 /* Still have to check that it's not an n-th
5698 alternative that starts with an on_failure_jump. */
5700 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5701 if ((re_opcode_t) p1[mcnt-3] != jump_past_alt)
5703 /* Get to the beginning of the n-th alternative. */
5709 /* Deal with the last alternative: go back and get number
5710 of the `jump_past_alt' just before it. `mcnt' contains
5711 the length of the alternative. */
5712 EXTRACT_NUMBER (mcnt, p1 - 2);
5714 if (!alt_match_null_string_p (p1, p1 + mcnt, reg_info))
5717 p1 += mcnt; /* Get past the n-th alternative. */
5723 assert (p1[1] == **p);
5729 if (!common_op_match_null_string_p (&p1, end, reg_info))
5732 } /* while p1 < end */
5735 } /* group_match_null_string_p */
5738 /* Similar to group_match_null_string_p, but doesn't deal with alternatives:
5739 It expects P to be the first byte of a single alternative and END one
5740 byte past the last. The alternative can contain groups. */
5743 alt_match_null_string_p (p, end, reg_info)
5744 unsigned char *p, *end;
5745 register_info_type *reg_info;
5748 unsigned char *p1 = p;
5752 /* Skip over opcodes that can match nothing, and break when we get
5753 to one that can't. */
5755 switch ((re_opcode_t) *p1)
5758 case on_failure_jump:
5760 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5765 if (!common_op_match_null_string_p (&p1, end, reg_info))
5768 } /* while p1 < end */
5771 } /* alt_match_null_string_p */
5774 /* Deals with the ops common to group_match_null_string_p and
5775 alt_match_null_string_p.
5777 Sets P to one after the op and its arguments, if any. */
5780 common_op_match_null_string_p (p, end, reg_info)
5781 unsigned char **p, *end;
5782 register_info_type *reg_info;
5787 unsigned char *p1 = *p;
5789 switch ((re_opcode_t) *p1++)
5809 assert (reg_no > 0 && reg_no <= MAX_REGNUM);
5810 ret = group_match_null_string_p (&p1, end, reg_info);
5812 /* Have to set this here in case we're checking a group which
5813 contains a group and a back reference to it. */
5815 if (REG_MATCH_NULL_STRING_P (reg_info[reg_no]) == MATCH_NULL_UNSET_VALUE)
5816 REG_MATCH_NULL_STRING_P (reg_info[reg_no]) = ret;
5822 /* If this is an optimized succeed_n for zero times, make the jump. */
5824 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5832 /* Get to the number of times to succeed. */
5834 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5839 EXTRACT_NUMBER_AND_INCR (mcnt, p1);
5847 if (!REG_MATCH_NULL_STRING_P (reg_info[*p1]))
5855 /* All other opcodes mean we cannot match the empty string. */
5861 } /* common_op_match_null_string_p */
5864 /* Return zero if TRANSLATE[S1] and TRANSLATE[S2] are identical for LEN
5865 bytes; nonzero otherwise. */
5868 bcmp_translate (s1, s2, len, translate)
5869 const char *s1, *s2;
5871 RE_TRANSLATE_TYPE translate;
5873 register const unsigned char *p1 = (const unsigned char *) s1;
5874 register const unsigned char *p2 = (const unsigned char *) s2;
5877 if (translate[*p1++] != translate[*p2++]) return 1;
5883 /* Entry points for GNU code. */
5885 /* re_compile_pattern is the GNU regular expression compiler: it
5886 compiles PATTERN (of length SIZE) and puts the result in BUFP.
5887 Returns 0 if the pattern was valid, otherwise an error string.
5889 Assumes the `allocated' (and perhaps `buffer') and `translate' fields
5890 are set in BUFP on entry.
5892 We call regex_compile to do the actual compilation. */
5895 re_compile_pattern (pattern, length, bufp)
5896 const char *pattern;
5898 struct re_pattern_buffer *bufp;
5902 /* GNU code is written to assume at least RE_NREGS registers will be set
5903 (and at least one extra will be -1). */
5904 bufp->regs_allocated = REGS_UNALLOCATED;
5906 /* And GNU code determines whether or not to get register information
5907 by passing null for the REGS argument to re_match, etc., not by
5911 /* Match anchors at newline. */
5912 bufp->newline_anchor = 1;
5914 ret = regex_compile (pattern, length, re_syntax_options, bufp);
5918 return gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
5921 weak_alias (__re_compile_pattern, re_compile_pattern)
5924 /* Entry points compatible with 4.2 BSD regex library. We don't define
5925 them unless specifically requested. */
5927 #if defined _REGEX_RE_COMP || defined _LIBC
5929 /* BSD has one and only one pattern buffer. */
5930 static struct re_pattern_buffer re_comp_buf;
5934 /* Make these definitions weak in libc, so POSIX programs can redefine
5935 these names if they don't use our functions, and still use
5936 regcomp/regexec below without link errors. */
5946 if (!re_comp_buf.buffer)
5947 return gettext ("No previous regular expression");
5951 if (!re_comp_buf.buffer)
5953 re_comp_buf.buffer = (unsigned char *) malloc (200);
5954 if (re_comp_buf.buffer == NULL)
5955 return (char *) gettext (re_error_msgid
5956 + re_error_msgid_idx[(int) REG_ESPACE]);
5957 re_comp_buf.allocated = 200;
5959 re_comp_buf.fastmap = (char *) malloc (1 << BYTEWIDTH);
5960 if (re_comp_buf.fastmap == NULL)
5961 return (char *) gettext (re_error_msgid
5962 + re_error_msgid_idx[(int) REG_ESPACE]);
5965 /* Since `re_exec' always passes NULL for the `regs' argument, we
5966 don't need to initialize the pattern buffer fields which affect it. */
5968 /* Match anchors at newlines. */
5969 re_comp_buf.newline_anchor = 1;
5971 ret = regex_compile (s, strlen (s), re_syntax_options, &re_comp_buf);
5976 /* Yes, we're discarding `const' here if !HAVE_LIBINTL. */
5977 return (char *) gettext (re_error_msgid + re_error_msgid_idx[(int) ret]);
5988 const int len = strlen (s);
5990 0 <= re_search (&re_comp_buf, s, len, 0, len, (struct re_registers *) 0);
5993 #endif /* _REGEX_RE_COMP */
5995 /* POSIX.2 functions. Don't define these for Emacs. */
5999 /* regcomp takes a regular expression as a string and compiles it.
6001 PREG is a regex_t *. We do not expect any fields to be initialized,
6002 since POSIX says we shouldn't. Thus, we set
6004 `buffer' to the compiled pattern;
6005 `used' to the length of the compiled pattern;
6006 `syntax' to RE_SYNTAX_POSIX_EXTENDED if the
6007 REG_EXTENDED bit in CFLAGS is set; otherwise, to
6008 RE_SYNTAX_POSIX_BASIC;
6009 `newline_anchor' to REG_NEWLINE being set in CFLAGS;
6010 `fastmap' to an allocated space for the fastmap;
6011 `fastmap_accurate' to zero;
6012 `re_nsub' to the number of subexpressions in PATTERN.
6014 PATTERN is the address of the pattern string.
6016 CFLAGS is a series of bits which affect compilation.
6018 If REG_EXTENDED is set, we use POSIX extended syntax; otherwise, we
6019 use POSIX basic syntax.
6021 If REG_NEWLINE is set, then . and [^...] don't match newline.
6022 Also, regexec will try a match beginning after every newline.
6024 If REG_ICASE is set, then we considers upper- and lowercase
6025 versions of letters to be equivalent when matching.
6027 If REG_NOSUB is set, then when PREG is passed to regexec, that
6028 routine will report only success or failure, and nothing about the
6031 It returns 0 if it succeeds, nonzero if it doesn't. (See regex.h for
6032 the return codes and their meanings.) */
6035 regcomp (preg, pattern, cflags)
6037 const char *pattern;
6042 = (cflags & REG_EXTENDED) ?
6043 RE_SYNTAX_POSIX_EXTENDED : RE_SYNTAX_POSIX_BASIC;
6045 /* regex_compile will allocate the space for the compiled pattern. */
6047 preg->allocated = 0;
6050 /* Try to allocate space for the fastmap. */
6051 preg->fastmap = (char *) malloc (1 << BYTEWIDTH);
6053 if (cflags & REG_ICASE)
6058 = (RE_TRANSLATE_TYPE) malloc (CHAR_SET_SIZE
6059 * sizeof (*(RE_TRANSLATE_TYPE)0));
6060 if (preg->translate == NULL)
6061 return (int) REG_ESPACE;
6063 /* Map uppercase characters to corresponding lowercase ones. */
6064 for (i = 0; i < CHAR_SET_SIZE; i++)
6065 preg->translate[i] = ISUPPER (i) ? TOLOWER (i) : i;
6068 preg->translate = NULL;
6070 /* If REG_NEWLINE is set, newlines are treated differently. */
6071 if (cflags & REG_NEWLINE)
6072 { /* REG_NEWLINE implies neither . nor [^...] match newline. */
6073 syntax &= ~RE_DOT_NEWLINE;
6074 syntax |= RE_HAT_LISTS_NOT_NEWLINE;
6075 /* It also changes the matching behavior. */
6076 preg->newline_anchor = 1;
6079 preg->newline_anchor = 0;
6081 preg->no_sub = !!(cflags & REG_NOSUB);
6083 /* POSIX says a null character in the pattern terminates it, so we
6084 can use strlen here in compiling the pattern. */
6085 ret = regex_compile (pattern, strlen (pattern), syntax, preg);
6087 /* POSIX doesn't distinguish between an unmatched open-group and an
6088 unmatched close-group: both are REG_EPAREN. */
6089 if (ret == REG_ERPAREN) ret = REG_EPAREN;
6091 if (ret == REG_NOERROR && preg->fastmap)
6093 /* Compute the fastmap now, since regexec cannot modify the pattern
6095 if (re_compile_fastmap (preg) == -2)
6097 /* Some error occurred while computing the fastmap, just forget
6099 free (preg->fastmap);
6100 preg->fastmap = NULL;
6107 weak_alias (__regcomp, regcomp)
6111 /* regexec searches for a given pattern, specified by PREG, in the
6114 If NMATCH is zero or REG_NOSUB was set in the cflags argument to
6115 `regcomp', we ignore PMATCH. Otherwise, we assume PMATCH has at
6116 least NMATCH elements, and we set them to the offsets of the
6117 corresponding matched substrings.
6119 EFLAGS specifies `execution flags' which affect matching: if
6120 REG_NOTBOL is set, then ^ does not match at the beginning of the
6121 string; if REG_NOTEOL is set, then $ does not match at the end.
6123 We return 0 if we find a match and REG_NOMATCH if not. */
6126 regexec (preg, string, nmatch, pmatch, eflags)
6127 const regex_t *preg;
6130 regmatch_t pmatch[];
6134 struct re_registers regs;
6135 regex_t private_preg;
6136 int len = strlen (string);
6137 boolean want_reg_info = !preg->no_sub && nmatch > 0;
6139 private_preg = *preg;
6141 private_preg.not_bol = !!(eflags & REG_NOTBOL);
6142 private_preg.not_eol = !!(eflags & REG_NOTEOL);
6144 /* The user has told us exactly how many registers to return
6145 information about, via `nmatch'. We have to pass that on to the
6146 matching routines. */
6147 private_preg.regs_allocated = REGS_FIXED;
6151 regs.num_regs = nmatch;
6152 regs.start = TALLOC (nmatch * 2, regoff_t);
6153 if (regs.start == NULL)
6154 return (int) REG_NOMATCH;
6155 regs.end = regs.start + nmatch;
6158 /* Perform the searching operation. */
6159 ret = re_search (&private_preg, string, len,
6160 /* start: */ 0, /* range: */ len,
6161 want_reg_info ? ®s : (struct re_registers *) 0);
6163 /* Copy the register information to the POSIX structure. */
6170 for (r = 0; r < nmatch; r++)
6172 pmatch[r].rm_so = regs.start[r];
6173 pmatch[r].rm_eo = regs.end[r];
6177 /* If we needed the temporary register info, free the space now. */
6181 /* We want zero return to mean success, unlike `re_search'. */
6182 return ret >= 0 ? (int) REG_NOERROR : (int) REG_NOMATCH;
6185 weak_alias (__regexec, regexec)
6189 /* Returns a message corresponding to an error code, ERRCODE, returned
6190 from either regcomp or regexec. We don't use PREG here. */
6193 regerror (errcode, preg, errbuf, errbuf_size)
6195 const regex_t *preg;
6203 || errcode >= (int) (sizeof (re_error_msgid_idx)
6204 / sizeof (re_error_msgid_idx[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 + re_error_msgid_idx[errcode]);
6213 msg_size = strlen (msg) + 1; /* Includes the null. */
6215 if (errbuf_size != 0)
6217 if (msg_size > errbuf_size)
6219 #if defined HAVE_MEMPCPY || defined _LIBC
6220 *((char *) __mempcpy (errbuf, msg, errbuf_size - 1)) = '\0';
6222 memcpy (errbuf, msg, errbuf_size - 1);
6223 errbuf[errbuf_size - 1] = 0;
6227 memcpy (errbuf, msg, msg_size);
6233 weak_alias (__regerror, regerror)
6237 /* Free dynamically allocated space used by PREG. */
6243 if (preg->buffer != NULL)
6244 free (preg->buffer);
6245 preg->buffer = NULL;
6247 preg->allocated = 0;
6250 if (preg->fastmap != NULL)
6251 free (preg->fastmap);
6252 preg->fastmap = NULL;
6253 preg->fastmap_accurate = 0;
6255 if (preg->translate != NULL)
6256 free (preg->translate);
6257 preg->translate = NULL;
6260 weak_alias (__regfree, regfree)
6263 #endif /* not emacs */
projects / gnulib.git / blob